DETAILED ACTION
Claims 1-10,12 and 13-20 rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Claim(s) 1,2,4,6,7,12 and 13,14,16 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) with annotated version thereof in view of Rodriguez et al. (US 2022/0309387 A1) further in view of Bera et al. (US 2025/0005925 A1) with Related U.S. Application Data: Provisional application No. 63/510,416, filed on Jun. 27, 2023:
Claim(s) 3 and 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) in view of Rodriguez et al. (US 2022/0309387 A1) further in view of Bera et al. (US 2025/0005925 A1) with Related U.S. Application Data: Provisional application No. 63/510,416, filed on Jun. 27, 2023, as applied in claims 1,2,4,6,7,12 and 13,14,16 further in view of MESUT et al. (TR 2021021147 A2) with SEARCH machine translation:
Claim(s) 5 and 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) in view of Rodriguez et al. (US 2022/0309387 A1) further in view of Bera et al. (US 2025/0005925 A1) with Related U.S. Application Data: Provisional application No. 63/510,416, filed on Jun. 27, 2023, as applied in claims 1,2,4,6,7,12 and 13,14,16, further in view of Barth et al. (US 2022/0067135 A1):
Claim(s) 8,9 and 18,19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) in view of Rodriguez et al. (US 2022/0309387 A1) further in view of Bera et al. (US 2025/0005925 A1) with Related U.S. Application Data: Provisional application No. 63/510,416, filed on Jun. 27, 2023, as applied in claims 1,2,4,6,7,12 and 13,14,16, further in view of Weisz et al. (US 10,904,488 B1):
Claim(s) 10 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) in view of Rodriguez et al. (US 2022/0309387 A1) further in view of Bera et al. (US 2025/0005925 A1) with Related U.S. Application Data: Provisional application No. 63/510,416, filed on Jun. 27, 2023, as applied in claims 1,2,4,6,7,12 and 13,14,16, further in view of Nguyen et al. (CAPSULE-FORENSICS: USING CAPSULE NETWORKS TO DETECT FORGED IMAGES AND VIDEOS) further in view of KATOH et al. (US 2022/0245405 A1):
Claim(s) 11 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) with annotated version thereof in view of Rodriguez et al. (US 2022/0309387 A1) further in view of Bera et al. (US 2025/0005925 A1) with Related U.S. Application Data: Provisional application No. 63/510,416, filed on Jun. 27, 2023, as applied in claims 1,2,4,6,7,12 and 13,14,16, further in view of Venkataraman et al. (US 2021/0322121 A1) further in view of Hohwald et al. (US 11,017,019 B1):
Response to Amendment
The amendment was received 4/3/2026. Claims 1-20 pending.
Priority
Applicant’s claim for the benefit of a prior-filed application under 35 U.S.C. 119(e) or under 35 U.S.C. 120, 121, 365(c), or 386(c) is acknowledged. Applicant has not complied with one or more conditions for receiving the benefit of an earlier filing date under 35 U.S.C. 120, 121, 365(c), or 386(c) as follows:
The later-filed application must be an application for a patent for an invention which is also disclosed in the prior application (the parent or original nonprovisional application or provisional application). The disclosure of the invention in the parent application and in the later-filed application must be sufficient to comply with the requirements of 35 U.S.C. 112(a) or the first paragraph of pre-AIA 35 U.S.C. 112, except for the best mode requirement. See Transco Products, Inc. v. Performance Contracting, Inc., 38 F.3d 551, 32 USPQ2d 1077 (Fed. Cir. 1994).
The disclosure of the prior-filed application, Application No. 63/596,326 11/06/2023, fails to provide adequate support or enablement in the manner provided by 35 U.S.C. 112(a) or pre-AIA 35 U.S.C. 112, first paragraph for one or more claims of this application. The claimed “at least one first record includes a deepfake video labelled with a ground truth indicating deepfake, and at least one second record includes an authentic video labelled with a ground truth indicating authentic” of claims 1,13 is not explicit or implicit in Application No. 63/596,326 11/06/2023.
The disclosure of the prior-filed application, Application No. 63/449,182 03/01/2023, fails to provide adequate support or enablement in the manner provided by 35 U.S.C. 112(a) or pre-AIA 35 U.S.C. 112, first paragraph for one or more claims of this application. The claimed “at least one first record includes a deepfake video labelled with a ground truth indicating deepfake, and at least one second record includes an authentic video labelled with a ground truth indicating authentic” of claims 1,13 is not explicit or implicit in Application No. 63/449,182 03/01/2023.
Accordingly, claims 1-20 are not entitled to the benefit of the prior applications.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-10,12 and 13-20 rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step zero: establish broadest reasonable interpretation, shown in the footnotes throughout this Office action;
Step 1: Claim 1 is a process; claim 13 is a machine;
Step 2A, prong 1: The claim(s) recite(s) a mental process and math, boxed-in below via representative claim 1:
PNG
media_image1.png
389
883
media_image1.png
Greyscale
Step 2A, prong 2: This mental process and math judicial exception is not integrated into a practical application because the additional elements (or non-boxed in limitations such as said “video”, “tool”123, “dataset”4, labelling-“records”5) are described at a high level and perform only generic computer functions (i.e. analyzing data, generating output, training ML models) and do not improve a technical field or technology (i.e., bringing a fake, digitally manipulated video file into a more desirable condition or more excellent condition in view of the disclosed problems of fake videos) as indicated in applicant’s disclosure, pages 1,8,9,20,25 via MPEP 216.04(d):
PNG
media_image2.png
1263
1031
media_image2.png
Greyscale
Applicant’s disclosure, pages 8,9,20,25 wherein the technology/aforementioned technical field comprises the fake, digitally manipulated video file:
PNG
media_image3.png
363
881
media_image3.png
Greyscale
PNG
media_image4.png
317
836
media_image4.png
Greyscale
PNG
media_image5.png
282
821
media_image5.png
Greyscale
In contrast, claim 11 “reflects the disclosed improvement”6 in applicant’s disclosure at page 8, ll. 8-12 and page 17,ll. 19-21:
PNG
media_image6.png
910
893
media_image6.png
Greyscale
Step 2B: The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception because each additional element, such as “video”, “tool”789, “dataset”10, labelling-“records”11, considered individually or with the mental process & math adheres to conventional practices as indicated in applicant’s specification’s background12 (page 1); thus, the claims do not recite an ‘inventive concept’ and the additional elements (such as “video”, “tool”131415, “dataset”16, labelling-“records”17 ) merely invoke a generic computing environment to perform routine, well understood and conventional operations for training machine learning models using labeled datasets:
PNG
media_image7.png
1991
883
media_image7.png
Greyscale
Response to Arguments
Applicant's arguments filed 4/3/2026 have been fully considered but they are not persuasive:
Priority
A. The Correct Legal Standard Is Possession, Not Verbatim Correspondence
Applicants state in page 8, 3rd paragraph:
Applicant notes that this principle operates symmetrically. As the Federal Circuit held in Enzo Biochem v. Gen-Probe, 323 F.3d 956 (Fed. Cir. 2002), even verbatim claim language appearing in a specification does not automatically satisfy written description if a skilled artisan cannot recognize what is claimed. The test is always possession - not the presence or absence of particular words - in either direction. Applicant raises this not to challenge the threshold sufficiency of the Examiner's rejection under Hyatt v. Dudas, 492 F.3d 1365 (Fed. Cir. 2007), but to frame the correct inquiry: the question is whether the provisional application(s) conveys possession of the claimed concept, not whether it uses the claimed words.
In response, the examiner adds further framing via MPEP 2163:
MPEP 2163 Guidelines for the Examination of Patent Applications Under the 35 U.S.C. 112(a) or Pre-AIA 35 U.S.C. 112, first paragraph, "Written Description" Requirement [R-01.2024]
II. METHODOLOGY FOR DETERMINING ADEQUACY OF WRITTEN DESCRIPTION
A. Read and Analyze the Specification for Compliance with 35 U.S.C. 112(a) or Pre-AIA 35 U.S.C. 112, first paragraph
2. Review the Entire Application to Understand How Applicant Provides Support for the Claimed Invention Including Each Element and/or Step
Prior to determining whether the disclosure provides adequate written description for the claimed subject matter, the examiner should review the claims and the entire specification, including the specific embodiments, figures, and sequence listings, to understand how applicant provides support for the various features of the claimed invention. The disclosure of an element may be critical where those of ordinary skill in the art would require it to understand that inventor was in possession of the invention. Compare Rasmussen, 650 F.2d at 1215, 211 USPQ at 327 ("one skilled in the art who read Rasmussen’s specification would understand that it is unimportant how the layers are adhered, so long as they are adhered") (emphasis in original), with Amgen, Inc. v. Chugai Pharm.Co., Ltd., 927 F.2d 1200, 1206, 18 USPQ2d 1016, 1021 (Fed. Cir. 1991) ("it is well established in our law that conception of a chemical compound requires that the inventor be able to define it so as to distinguish it from other materials, and to describe how to obtain it"). The analysis of whether the specification (Provisional application 63/449,182, filed 03/01/2023) complies with the written description requirement calls for the examiner to compare (see next pages) the scope18 of the claim (claim 1: reproduced below) with the scope of the description (63/449,182, paragraph spanning pages 18,19, reproduced below) to determine (see next pages) whether applicant has demonstrated that the inventor was in possession19 of the claimed invention. Such a review is conducted from the standpoint of one of ordinary skill in the art at the time the application was filed (see, e.g., Wang Labs., Inc. v. Toshiba Corp., 993 F.2d 858, 865, 26 USPQ2d 1767, 1774 (Fed. Cir. 1993)) and should include a determination of the field (“Artificial Intelligence (AI) technologies”, Prov. App. pg. 1, 5th para) of the invention and the level of skill and knowledge in the art. For some arts, there is an inverse correlation between the level of skill and knowledge in the art and the specificity of disclosure necessary to satisfy the written description requirement. Information which is well known in the art need not be described in detail in the specification. See, e.g., Hybritech, Inc. v. Monoclonal Antibodies, Inc., 802 F.2d 1367, 1379-80, 231 USPQ 81, 90 (Fed. Cir. 1986). However, sufficient information must be provided to show that the inventor had possession of the invention as claimed.
PNG
media_image8.png
1040
875
media_image8.png
Greyscale
The scope of claim 1 includes:
(1) “labelled”20-“video”21;
(2) “labeled”-“record”;
(3) “ground truth”22-“video”; and
(4) “ground truth”-“record”23.
The scope of Provisional Application 63/449,182, paragraph spanning pages 18,19 includes
(1) “ground truth”-“dataset”: i.e., “accurate and reliable”-“dataset”.
Thus the examiner compares/determines (in view of MPEP 2163 II.A.2.) that the inventor does not possess “labeled”-“video” and “labeled”-“record” of claim 1 in view of Provisional Application 63/449,182, paragraph spanning pages 18,19.
B. Application No. 63/449,182 Conveys Possession of the Claimed Limitation on the Merits
Applicants state on page 8, last para:
A POSITA reading page 18 of Application No. 63/449,192 would recognize that "real data records" with "ground truth" conveys the same concept as "an authentic video labelled with a ground truth indicating authentic," and that "fake data records" with "ground truth" conveys the same concept as "a deepfake video labelled with a ground truth indicating deepfake." The concepts are identical; the terminology differs. That difference is legally irrelevant under Fujikawa24 and MPEP § 2163.
In response to A POSITA reading page 18, a POSITA “would not recognize the written description (63/449,182) of the invention as providing adequate support for the claimed” “video labelled” because the disclosure’s (63/449,182) “mission is to…label…incidents25" (page 8, penult para, 4th S) or label deep fake26 incidents27 and thus is not the full scope of claim 1’s “ground-truth”-“video labelled” or “ground-truth”-“labelled”-“record” via MPEP 2163:
MPEP 2163 Guidelines for the Examination of Patent Applications Under the 35 U.S.C. 112(a) or Pre-AIA 35 U.S.C. 112, first paragraph, "Written Description" Requirement [R-01.2024]
II. METHODOLOGY FOR DETERMINING ADEQUACY OF WRITTEN DESCRIPTION
A. Read and Analyze the Specification for Compliance with 35 U.S.C. 112(a) or Pre-AIA 35 U.S.C. 112, first paragraph
Office personnel should adhere to the following procedures when reviewing patent applications for compliance with the written description requirement of 35 U.S.C. 112(a) or pre-AIA 35 U.S.C. 112, first paragraph. There is a presumption that an adequate written description of the claimed invention is present in the specification as filed, Wertheim, 541 F.2d at 262, 191 USPQ at 96, thus the examiner has the initial burden, after a thorough reading and evaluation of the content of the application, of presenting evidence or reasons why a person skilled in the art would not recognize the written description of the invention as providing adequate support for the claimed invention. To make a prima facie case, it is necessary to identify the claim limitations (claim 1 “video labelled”) that are not adequately supported, and explain why (a corresponding portion of 63/449,182, page 8, penult para, 2nd S, that discloses “label…incidents that meet the following four criteria” does not mean claim 1’s “video labelled” since the deepfake “incidents” are being labeled and thus does not result in the claimed “video labeled” since “incident” does not mean “video”) the claim is not fully supported by the disclosure. For example, in Hyatt v. Dudas, 492 F.3d 1365, 1371, 83 USPQ2d 1373, 1376-1377 (Fed. Cir. 2007), the examiner made a prima facie case by clearly and specifically explaining why applicant’s specification did not support the particular claimed combination of elements, even though applicant’s specification listed each and every element in the claimed combination. The court found the "examiner was explicit that while each element may be individually described in the specification, the deficiency was lack of adequate description of their combination" and, thus, "[t]he burden was then properly shifted to [inventor] to cite to the examiner where adequate written description could be found or to make an amendment to address the deficiency." Id.; see also Stored Value Solutions, Inc. v. Card Activation Techs., 499 Fed.App’x 5, 13-14 (Fed. Cir. 2012) (non-precedential) (Finding inadequate written support for claims drawn to a method of processing debit purchase transactions requiring three separate authorization codes because "the written description [did] not contain a method that include[d] all three codes" and "[e]ach authorization code is an important claim limitation, and the presence of multiple authorization codes in [the claim] was essential".).
In response to --The concepts28 are identical; the terminology29 differs--:
PNG
media_image9.png
306
928
media_image9.png
Greyscale
The concept (i.e., the act of expressing, as in words) of “labelled with ground truth” is in claim 1 (reproduced below).
The concept (i.e., the act of expressing, as in words) of “ground truth” is in 63/449,182, page 18 last paragraph continued to page 19 (reproduced below).
The word “label” is not in 63/449,182 at pages 18,19; thus, the concepts (i.e., the acts of expressing, as in words, of claim 1 and 63/449,182 at pages 18,19) are not identical.
Thus via MPEP 2163.02, 3rd para—terminology (claim 1’s “label”) not present in the application (63/449,182: pages 18,19) as filed (03/01/2023), involving a departure from, addition to, or deletion from the disclosure of the application (63/449,182) as filed (03/01/2023), the examiner should conclude that the claimed subject matter (claim 1’s “label”) is not described in that application (63/449,182: pages 18,19). This conclusion … result in .. denial of the benefit of the filing date of a previously filed (03/01/2023) application (63/449,182)--.
PNG
media_image10.png
642
908
media_image10.png
Greyscale
PNG
media_image11.png
1318
970
media_image11.png
Greyscale
PNG
media_image12.png
1312
943
media_image12.png
Greyscale
Applicants state on page 9, 2nd para:
This reading is further reinforced by the knowledge a POSITA brings to the disclosure. In re GPAC Inc30., 57 F.3d 1573, 1579 (Fed. Cir. 1995) (POSITA is presumed to have knowledge of the relevant technical field). Specifically, in the present case, in the context of the technical field of supervised machine learning for binary classification - i.e., authentic versus deepfake - a training dataset necessarily comprises labeled records of both classes. This is a foundational requirement of supervised learning. See, e.g., Wikipedia, "Supervised learning" as available at https://en.wikipedia.org/wiki/Supervised learning. Thus, a POSITA encountering the disclosure of "ground truth real & fake data records via 'dataset"' would understand without further elaboration that those records carry labels identifying their class.
In response, via “The inquiry is whether one of skill in the art would understand the specification itself to disclose a structure, not simply whether that person would be capable of implementing a structure”, one of ordinary skill would not understand the specification (63/449,182) itself to disclose the full scope of claim 1’s “labelled”-“record” & “video labelled” via MPEP 2163:
MPEP 2163 Guidelines for the Examination of Patent Applications Under the 35 U.S.C. 112(a) or Pre-AIA 35 U.S.C. 112, first paragraph, "Written Description" Requirement [R-01.2024]
II. METHODOLOGY FOR DETERMINING ADEQUACY OF WRITTEN DESCRIPTION
A. Read and Analyze the Specification for Compliance with 35 U.S.C. 112(a) or Pre-AIA 35 U.S.C. 112, first paragraph
3. Determine Whether There is Sufficient Written Description to Inform a Skilled Artisan That Inventor was in Possession of the Claimed Invention as a Whole at the Time the Application Was Filed
(a) Original claims, middle of 7th para:
It is not enough for the patentee simply to state or later argue that persons of ordinary skill in the art would know what structures to use to accomplish the claimed function."), quoting Atmel Corp. v. Information Storage Devices, Inc., 198 F.3d 1374, 1380, 53 USPQ2d 1225, 1229 (Fed. Cir. 1999); Biomedino, LLC v. Waters Technologies Corp., 490 F.3d 946, 953, 83 USPQ2d 1118, 1123 (Fed. Cir. 2007) ("The inquiry is whether one of skill in the art would understand the specification itself to disclose a structure, not simply whether that person would be capable of implementing a structure.").
Applicants state in page 9, 3rd para:
The present specification confirms this reading at 11[[0116]-[0117], which describes in detail the two record types - one including a deepfake video labeled with a ground truth indicating the specific deepfake tool used, and one including an authentic video labeled with a ground truth indicating absence of deepfake use. This is the same concept3132 disclosed in the provisional, expressed in the more developed language of the non-provisional claims.
The word “label”33 or the expression (as in words) thereof is not in 63/449,182 as in the present specification at [0116]-[0117]. Therefore there is no same complete label concept regarding the present specification at [0116]-[0117] and 63/449,182.
Applicants state in page 9, 4th paragraph:
Applicant additionally directs the Examiner's attention to Fig. 5 of the present application, specifically Step 512 ("Create record") and the associated training dataset 514, which illustrates the record creation step in the context of the full content- adaptive pipeline described at 1[0166]-[0190]. The provisional's disclosure of "ground truth real & fake data records via 'dataset"' corresponds to this same architectural element. A POSITA reading both documents would recognize the same concept34 in each.
I would recognize the concept of “ground truth” and “records” in the provisional (63/449,182) and in the present application (US 2024/0296698 A1). However, I don’t recognize/identify “label” or the complete concept35 or full claim scope thereof in the context with the concepts of “ground truth” and “records” in the provisional (63/449,182).
C. Application No. 63/596,326 Also Conveys Possession
Applicants state in page 9, last para:
The Examiner's characterization of Application No. 63/596,326 as disclosing only "a single name via 'file"' reads a single implementation detail in isolation from the broader disclosure. The provisional application discloses a system and method for deepfake detection that necessarily involves training data records in a manner consistent with the challenged limitation36. Read as a whole by a POSITA - as the written description standard requires, Ariad Pharms., Inc. v. Eli Lilly & Co., 598 F.3d 1336, 1351 (Fed. Cir. 2010) - the disclosure37 conveys possession of a training dataset comprising labeled records of both authentic and deepfake content. The "file" reference cited by the Examiner is one implementation detail within a broader disclosure; it is not a ceiling on what the provisional application conveys to a POSITA38.
D. If the Examiner Maintains This Rejection, Applicant Requests Substantive Engagement on the Record
Applicants state on page 10, 2nd para:
Applicant has identified the specific provisional disclosure - page 18 of Application No. 63/449,182 - and mapped each element of the challenged limitation to that disclosure. Pursuant to MPEP @ 707.07(f) and 37 C.F.R. § 1.104, if the Examiner maintains this rejection, Applicant respectfully requests that the next Office Action identify on the record:
1. What concept3940 - not what words - is absent from the disclosure at page 18 of
Application No. 63/449,182; and
2. Why the cited passages fail to convey possession of the claimed limitation to a
POSITA, given that those passages explicitly disclose "ground truth real &
fake data records via 'dataset."'
Regarding 1, “label”41 or expressed as a word (“deepfake”) or phrase (“video not deepfake”) indicating that what follows belongs in a particular category (or deepfake group) or classification (or authentic class) is missing from the disclosure at page 18 of Application No. 63/449,182. Regarding 2, the inventor has possession of “ground truth real & fake data records via 'dataset."'.
E. The Priority Date Is Dispositive as to Bera
Applicants state in page 10,11:
Applicant notes that Bera et al. (US 2025/0005925 A1) claims priority to Provisional Application No. 63/510,416, filed June 27, 2023 - which postdates Applicant's provisional Application No. 63/449,182, filed March 1, 2023. If Applicant's priority claim to the March 1, 2023 provisional is upheld, Bera's effective filing date is later than Applicant's, and Bera does not qualify as prior art under 35 U.S.C. § 102(a)(2). This would require withdrawal of every § 103 rejection that relies on Bera, which encompasses all pending claims 1-20. Applicant respectfully requests that the Examiner address this issue directly in any subsequent Office Action and, if the Examiner intends to maintain Bera as prior art, articulate on the record the specific basis for that determination.
Applicant respectfully requests withdrawal of the priority rejection and confirmation of Applicant's entitlement to the benefit of at least Application No. 63/449,182, filed March 1, 2023.
Based on the above remarks A-D above, the examiner maintain Bera as prior art as articulated in the above remarks A-D regarding the lack of the complete concept42 of a “label” at said page 18 of 63/449,182.
Claim Objections
Claims 1 and 13 are objected to for informalities as using the abbreviation "ML" whereas the corresponding full term does not appear in those claims, as described in Applicant's disclosure at page 6.
In response, Applicant amended the claims in accordance with the Examiner's remarks. To vit, Applicant has amended each of Claims 1, 11, 12, and 13 to recite "machine learning (ML)" upon first use, consistent with the disclosure at page 6 of the specification. The abbreviation "ML" as used in dependent Claims 2-10 and 14-20 is now properly supported by the defined term introduced in the amended independent claims. Applicant respectfully submits that the objection has been fully addressed as to all affected claims and requests withdrawal thereof.
The objection is withdrawn.
35 USC 101 Rejections
NO MENTAL PROCESS OR MATHEMATICAL CONCEPTS RECITED
Applicants state in page 12, last para:
The Examiner characterizes the bulk of the claim recitations as "mental
process and math." Applicant respectfully disagrees. The claimed features cannot be practically performed mentally or with pen and paper. Specifically:
The examiner respectfully disagrees since the boxed portion (Office action 12/03/2025, page 10), representing the abstract idea, is less than half of claim 1. Thus the non-boxed portions are additional elements (such as “deepfake tool”):
PNG
media_image13.png
1004
960
media_image13.png
Greyscale
Applicant's arguments filed 4/3/2026, pages 12-15 have been fully considered but they are not persuasive:
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 12, 2nd bullet: “the output…is a computationally synthesized video file”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 12, 3rd bullet: “iterative algorithmic optimization (e.g., backpropagation, gradient decent) over large datasets”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
INTEGRATION INTO PRACTICAL APPLICATION PROVIDED
Applicant's arguments filed 4/3/2026, pages 13,14 have been fully considered but they are not persuasive:
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 13, 2nd para: “content-specific, dynamically generated training data”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., pages 13,14 last para: “Retrieves authentic videos matched to those specific features…Dynamically generates deepfake videos…Creates a balanced, labeled training dataset…calibrated”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 14 2nd para: “closed-loop, content-adaptive technical pipeline”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Applicants state in page 14, 2nd & 3rd paragraphs:
This is a closed-loop, content-adaptive technical pipeline that produces a
measurably improved deepfake detector for a specific input video. The specification at <]{[0165] explicitly states:
"The customized detector ML model may more accurately determine whether the input video is a deepfake video in comparison, for example, to using a general deepfake detector ML model designed to analyze multiple different videos."
This is a concrete, quantifiable technical improvement - increased detection accuracy for content-specific deepfake detection. This improvement is not merely an improvement to a business practice or an abstract concept; it is an improvement to the technical performance of a machine learning system for video authenticity detection.
Regarding [0165], there is a “bare assertion of an improvement without the detail necessary”; thus, “the examiner should not determine the claim improves technology” via MPEP 2106.04(d)(1) Evaluating Improvements in the Functioning of a Computer, or an Improvement to Any Other Technology or Technical Field in Step 2A Prong Two [R-10.2019], 2nd para:
The courts have not provided an explicit test for this consideration, but have instead illustrated how it is evaluated in numerous decisions. These decisions, and a detailed explanation of how examiners should evaluate this consideration are provided in MPEP § 2106.05(a). In short, first the specification should be evaluated to determine if the disclosure provides sufficient details such that one of ordinary skill in the art would recognize the claimed invention as providing an improvement. The specification need not explicitly set forth the improvement, but it must describe the invention such that the improvement would be apparent to one of ordinary skill in the art. Conversely, if the specification (US 2024/0296698 A1 [0165]) explicitly sets forth an improvement ([0165] 2nd S: “The customized detector ML model may more accurately determine whether the input video is a deepfake video in comparison, for example, to using a general deepfake detector ML model designed to analyze multiple different videos.”) but in a conclusory manner (i.e., a bare assertion of an improvement without the detail necessary to be apparent to a person of ordinary skill in the art), the examiner should not determine the claim improves technology. Second, if the specification sets forth an improvement in technology, the claim must be evaluated to ensure that the claim itself reflects the disclosed improvement. That is, the claim includes the components or steps of the invention that provide the improvement described in the specification. The claim itself does not need to explicitly recite the improvement described in the specification (e.g., "thereby increasing the bandwidth of the channel").
Applicants state in pages 14,15:
The Examiner's characterization of the claimed elements as "generic" is
incorrect. The Examiner identifies "video," "tool," "dataset," and "labelling-records"
as generic elements. However, the claims do not merely recite the use of generic
videos, tools, or datasets. They recite a specific functional relationship among these
elements:
• The authentic videos are not generic - they are specifically retrieved based on
features extracted from the particular input video under examination;
• The deepfake tool is not used generically - it is applied to those specifically
retrieved authentic videos to generate training data that mirrors the input
video's content domain;
• The training dataset is not generic - it is dynamically constructed for the
specific input video, with both labeled deepf ake and authentic records
balanced for training purposes.
This specific, ordered combination43 of steps - where the input video's features
drive the entire training data generation and model customization process - is not a
generic use of computers. It is a specific technical architecture that produces a
technically superior result.
In response to “ordered combination” via MPEP 2106.05(a), 4th para:
“The full scope4445 of the claim under the BRI should be considered to determine if the claim reflects an improvement in technology (e.g., the improvement described in the specification). In making this determination, it is critical that examiners look at the claim "as a whole," in other words, the claim should be evaluated "as an ordered combination, without ignoring the requirements of the individual steps.":
1. (Currently Amended) A computer implemented method of training a customized detector machine learning (ML) model for use on an input video, comprising:
analyzing the input video to identify a plurality of features associated with the input video;
obtaining a plurality of authentic videos associated with the plurality of features;
feeding the plurality of authentic videos into at least one deepfake tool;
obtaining a plurality of deepfake videos as an outcome46 of the at least one deepfake tool (this “obtaining” limitation does not reflect the improvement in [0052] under BRI);
creating a training dataset comprising a plurality of records, wherein at least one first record includes a deepfake video labelled with a ground truth indicating deepfake, and at least one second record includes an authentic video labelled with a ground truth indicating authentic; and
training the customized detector ML model on the training dataset.
Thus does claim 1 reflect the improvement of the disclosure (US 2024/0296698 A1: paragraph [0052]) under the broadest reasonable interpretation (BRI)? No, the disclosed “generated by a deepfake tool fed authentic videos” is not reflected in claim 1 under BRI:
PNG
media_image14.png
629
922
media_image14.png
Greyscale
Applicants state in page 15, 3rd para:
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 15, 3rd para, last S: “dynamically generating content-specific training data from the input video's own features and using it to train a customized detector” ) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 15, last bullets:
The specific combination of:
• Feature extraction from an unknown input video;
• Feature-matched authentic video retrieval;
• On-demand deepfake generation using those specific authentic videos;
• Balanced, labeled training dataset construction; and
• Customized detector training via fine-tuning/domain shift (see, e.g., Claim 6)
are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Applicants state in page 15:
Even under Step 2B of the Alice/Mayo framework (MPEP § 2106.05), the
claims at hand recite significantly more than any alleged abstract idea. The specific
combination of:
• Feature extraction from an unknown input video;
• Feature-matched authentic video retrieval;
• On-demand deepfake generation using those specific authentic videos;
• Balanced, labeled training dataset construction; and
• Customized detector training via fine-tuning/domain shift (see, e.g., Claim 6)
represents an inventive concept -a specific technical solution that was not
routine, well-understood, or conventional in the field of deepfake detection. The prior
art cited by the Examiner in the § 103 rejections itself confirms this: the Examiner
requires five to seven separate references to piece together the claimed combination,
demonstrating that no single reference or routine practice teaches or suggests this
specific pipeline.
IN response, “the search for an inventive concept is different from an obviousness analysis under 35 U.S.C. 103” and “patentability of the claimed invention under 35 U.S.C. 102 and 103 with respect to the prior art is neither required for, nor a guarantee of, patent eligibility under 35 U.S.C. 101” via:
MPEP 2106.05 Eligibility Step 2B: Whether a Claim Amounts to Significantly More [R-07.2022]
I. THE SEARCH FOR AN INVENTIVE CONCEPT, 4th para:
Although the courts often evaluate considerations such as the conventionality of an additional element in the eligibility analysis, the search for an inventive concept should not be confused with a novelty or non-obviousness determination. See Mayo, 566 U.S. at 91, 101 USPQ2d at 1973 (rejecting "the Government’s invitation to substitute §§ 102, 103, and 112 inquiries for the better established inquiry under § 101 "). As made clear by the courts, the "‘novelty’ of any element or steps in a process, or even of the process itself, is of no relevance in determining whether the subject matter of a claim falls within the § 101 categories of possibly patentable subject matter." Intellectual Ventures I v. Symantec Corp., 838 F.3d 1307, 1315, 120 USPQ2d 1353, 1358 (Fed. Cir. 2016) (quoting Diamond v. Diehr, 450 U.S. at 188–89, 209 USPQ at 9). See also Synopsys, Inc. v. Mentor Graphics Corp., 839 F.3d 1138, 1151, 120 USPQ2d 1473, 1483 (Fed. Cir. 2016) ("a claim for a new abstract idea is still an abstract idea. The search for a § 101 inventive concept is thus distinct from demonstrating § 102 novelty."). In addition, the search for an inventive concept is different from an obviousness analysis under 35 U.S.C. 103. See, e.g., BASCOM Global Internet v. AT&T Mobility LLC, 827 F.3d 1341, 1350, 119 USPQ2d 1236, 1242 (Fed. Cir. 2016) ("The inventive concept inquiry requires more than recognizing that each claim element, by itself, was known in the art. . . . [A]n inventive concept can be found in the non-conventional and non-generic arrangement of known, conventional pieces."). Specifically, lack of novelty under 35 U.S.C. 102 or obviousness under 35 U.S.C. 103 of a claimed invention does not necessarily indicate that additional elements are well-understood, routine, conventional elements. Because they are separate and distinct requirements from eligibility, patentability of the claimed invention under 35 U.S.C. 102 and 103 with respect to the prior art is neither required for, nor a guarantee of, patent eligibility under 35 U.S.C. 101. The distinction between eligibility (under 35 U.S.C. 101 ) and patentability over the art (under 35 U.S.C. 102 and/or 103 ) is further discussed in MPEP § 2106.05(d).
35 USC 103 Rejections
II. RANA DOES NOT TEACH THE CLAIMED CONCEPT AND CANNOT SERVE AS A PROPER PRIMARY REFERENCE
B. Rana Fails to Teach the Claimed Feature of "Analyzing the Input
Video to Identify a Plurality of Features Associated With the Input Video"
Applicant's arguments filed 4/3/2026, pages 19 have been fully considered but they are not persuasive.
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 19, section B, 1st para, last S: “a content-domain characterization step”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 20, 1st para, last S: “does not serve this function, does not drive any retrieval process, and does not characterize the content domain of an input video for the purpose of training data generation.”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 20, 3rd para, 3rd S: “ It is not a step of identifying which specific people appear in the video as content-domain identifiers for driving authentic video retrieval”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
C. Rana Fails to Teach the Claimed Feature of "Obtaining a Plurality of
Authentic Videos Associated With the Plurality of Features"
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 21, 2nd para, 3rd S:
The claimed feature requires a dynamic, input-video-driven retrieval process: the specific features identified from the specific input video under examination in preceding action(s) as recited in the Claims determine which authentic videos are obtained. The authentic videos must be associated with those specific features - they must, for example, depict the same people, use the same camera type,
contain the same language, show the same background as the input video, and/or the like.
) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 21, 3rd para, 2nd to 4th Ss:
They are assembled in advance, without reference to any specific input video. There is no feature-driven retrieval process in Rana. There is no step or procedure in which the features of a specific input video under examination are extracted and used to select which authentic videos to obtain.
) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 21, 4th para:
This distinction is architecturally fundamental and not merely terminological. Some of technical contributions and effects of the claimed subject matter at the least depend inter alia on this dynamic, input-video-feature-driven retrieval: it is what makes the resulting training dataset content-specific to the input video, and it is what makes the trained detector a "customized" detector rather than a general-purpose one. Rana's static corpus approach does not teach, suggest, or render obvious this dynamic
retrieval architecture.
) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
D. Rana Fails to Teach the Claimed "Domain Shift" of Claim 6
Applicants state in page 22:
These are two different technical disciplines using the word "domain" in entirely different senses. Frequency-domain signal processing and ML domain adaptation share no technical relationship. The Examiner's mapping is a terminological coincidence, not a technical correspondence.
IN response, “domain Shift” “does not limit” and “is taught” given that “one of the alternatives” (“fine-tuning”) “is taught by the prior art” (Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE): Office action 12/03/2025, page 26, reproduced below) via MPEP 2143.03 All Claim Limitations Must Be Considered [R-01.2024], 3rd para, 2nd & 3rd Ss:
Language (“fine-tuning and/or domain shift”) that suggests or makes a feature or step optional (“or” is alternative language) but does not require that feature or step does not limit the scope of a claim under the broadest reasonable claim interpretation. In addition, when a claim requires selection of an element from a list of alternatives, the prior art teaches the element if one of the alternatives is taught by the prior art. See, e.g., Fresenius USA, Inc. v. Baxter Int’l, Inc., 582 F.3d 1288, 1298, 92 USPQ2d 1163, 1171 (Fed. Cir. 2009).
PNG
media_image15.png
1185
839
media_image15.png
Greyscale
E. Rana Fails to Teach the Claimed "Iterating" of Claim 10
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 22, 4th para: “the entire pipeline” & “pipeline-level iteration…within a neural network”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
III. RODRIGUEZ IS NOT ANALOGOUS ART, DOES NOT TEACH THE CLAIMED "CUSTOMIZED" CONCEPT, AND ANY MODIFICATION OF RANA BY IT IS BOTH INSUFFICENT AS FAILING TO TEACH THE CLAIM LIMITATION AND IMPROPER FOR LACKING MOTIVATION TO COMBINE
A. Rodriguez Is Not Analogous Art
Applicants state in page 23, 3rd para:
Under the two-part test for analogous art, In re Bigio, 381 F.3d 1320, 1325
(Fed. Cir. 2004), a reference must either be in the same field of endeavor as the
claimed invention or be reasonably pertinent to the particular problem addressed by
the claimed invention. Rodriguez satisfies neither prong:
Regarding “relevant field of endeavor”, the examiner respectfully disagrees since the claimed invention and Rodriguez (US 2022/0309387 A1) endeavor (i.e. make a strenuous effort) in the same field of endeavor47 of classification via MPEP 2141.01(a):
Applicant’s disclosure US 2024/0296698 A1: [0107]- [0114]:
[0107] At 302, a class of the specific deepfake tool is identified. The class may be defined according to one or more, and/or combinations of adaptations performed by the specific deepfake tool on an input video. Examples of classes include: adapting a face (from a certain specific person to another specific person), adapting what a person says such as to anything possible by lip synchronization, adapting a body of a person, replacing an object, adapting a background, and removing an object.
[0108] There may be multiple different deepfake tools, in which case, the class is determined for each deepfake tool.
[0109] The class of the specific deepfake tool may be identified, for example, based on metadata and/or a tag associated with the deepfake tool indicating what the deepfake tool is trained to do. In another example, the class of the deepfake tool may be identified by feeding an authentic sample video into the deepfake tool to obtain a sample deepfake video, and comparing the deepfake video to the authentic sample video to detect the change, and determining the class according to the change. The comparison may be done, for example, by subtracting frames of the sample deepfake video from the authentic sample video to identify non-zero values as indicating the change, and/or using other processes that analyze two videos to detect changes between them.
[0110] At 304, one or more sample authentic videos corresponding to the class are obtained. In the case of multiple classes, different sample authentic videos are obtained for each class.
[0111] A variety of different authentic videos, of the same class, may be obtained, for example, a variety of different scenes, variety of different cameras used to create the authentic videos, and the like. The variety may help avoid bias during training of the detection ML model.
[0112] The sample authentic videos corresponding to the class may be automatically identified by a class detector ML model. The class detector model may be trained on a training dataset of training videos labelled with ground truth labels indicating presence of a feature associated with the class.
[0113] At 306, the sample authentic video(s) corresponding to the class are fed into the specific deepfake tool that modifies the sample authentic videos based on the class.
[0114] At 308, a deepfake video(s) modified based on the class is obtained as an outcome of the specific deepfake tool.
Rodriguez’s disclosure [0047]:
[0047] In some embodiments, the anomaly model 282 may be a classification model. Here, for example, such classification model may be utilized to predict a discrete label for each input scan of a user's metadata. Various classification models, such as models characterized as, without limitation, discrete tree classifiers, random tree classifiers, neural networks, support vector machine, naive Bayes classifiers, and the like, may be generated as an anomaly model. In some embodiments, a gradient boost classification model is generated. Gradient boost classification is able to predict a probability with each label which enables the risk levels to be ranked. In some embodiments, the anomaly model 282 may comprise one or more cascade-based models for detecting anomalies via multiple stages. Each stage may be associated with a stage specific model and a stage specific detection threshold such as risk levels. More details of the cascade-based models are described below. In some embodiments, a detection threshold may be set as at least one statistical standard deviation from a mean value. In some embodiments, a detection threshold may be set as at least two statistical standard deviations from a mean value. In some embodiments, a detection threshold may be set as at least three statistical standard deviations from a mean value. In some embodiments, a detection threshold may be an arbitrary value set by the user.
Thus based on the above “relevant field of endeavor” analysis Rodriguez is analogous art.
MPEP 2141.01(a) Analogous and Nonanalogous Art [R-01.2024]
I. TO RELY ON A REFERENCE UNDER 35 U.S.C. 103, IT MUST BE ANALOGOUS ART TO THE CLAIMED INVENTION, 2nd & 3rd paragraphs:
When determining whether the "relevant field of endeavor" test is met, the examiner should consider "explanations of the invention’s subject matter in the patent application, including the embodiments, function, and structure of the claimed invention." Airbus S.A.S. v. Firepass Corp., 941 F.3d 1374, 1380, 2019 USPQ2d 430083 (Fed. Cir. 2019) (quoting Bigio, 381 F.3d at 1325, 72 USPQ2d at 1212). When determining whether a prior art reference meets the "same field of endeavor" test for the analogous art, the primary focus is on what the reference discloses. Airbus, 41 F.3d at 1380. The examiner must consider the disclosure of each reference "in view of the ‘the reality of the circumstances.’" Airbus, 41 F.3d at 1380 (quoting Bigio, 381 F.3d at 1326, 72 USPQ2d at 1212). These circumstances are to be weighed "from the vantage point of the common sense likely to be exerted by one of ordinary skill in the art in assessing the scope of the endeavor." Airbus, 41 F.3d at 1380. See also Donner Technology, LLC v. Pro Stage Gear, LLC, 979 F.3d 1353, 2020 USPQ2d 11335 (Fed. Cir. 2020); Sanofi-Aventis, 66 F.4th at 1378; and Netflix, Inc. v. DivX, LLC, 80 F.4th 1352, 1358-59, 2023 USPQ2d 1057 (Fed. Cir. 2023) ("The field of endeavor is ‘not limited to the specific point of novelty, the narrowest possible conception of the field, or the particular focus within a given field.’") (quoting Unwired Planet, LLC v. Google Inc., 841 F.3d 995, 1001, 120 USPQ2d 1593, 1597 (Fed. Cir. 2016)).
As for the "reasonably pertinent" test, the examiner should consider the problem faced by the inventor, as reflected - either explicitly or implicitly - in the specification. In order for a reference to be "reasonably pertinent" to the problem, it must "logically [] have commended itself to an inventor's attention in considering his problem." In re ICON Health and Fitness, Inc., 496 F.3d 1374, 1379-80 (Fed. Cir. 2007) (quoting In re Clay, 966 F.2d 656,658, 23 USPQ2d 1058, 1061 (Fed. Cir. 1992)). See also In re Klein, 647 F.3d 1343, 1348, 98 USPQ2d 1991, 1993 (Fed. Cir. 2011) An inventor is not expected to have been aware of all prior art outside of the field of endeavor. Airbus, 41 F.3d at 1380-82. A reference outside of the field of endeavor is reasonably pertinent if a person of ordinary skill would have consulted it and applied its teachings when faced with the problem that the inventor was trying to solve. Airbus, 41 F.3d at 1380-82. In order to support a determination that a reference is reasonably pertinent, it may be appropriate to include a statement of the examiner's understanding of the problem. The question of whether a reference is reasonably pertinent often turns on how the problem to be solved is perceived. If the problem to be solved is viewed in a narrow or constrained way, and such a view is not consistent with the specification, the scope of available prior art may be inappropriately limited. It may be necessary for the examiner to explain why an inventor seeking to solve the identified problem would have looked to the reference in an attempt to find a solution to the problem, i.e., factual reasons why the prior art is pertinent to the identified problem. See Donner Tech., LLC v. Pro Stage Gear, LLC, 979 F.3d 1353, 1359, 2020 USPQ2d 11335 (Fed. Cir. 2020) ("Thus, when addressing whether a reference is analogous art with respect to the claimed invention under a reasonable-pertinence theory, the problems to which both relate must be identified and compared.").
B. Rodriguez's "Customized" Concept Is Technically Different From the Claimed "Customized"
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 24, 4th para: “dynamically generated, input-video-feature-matched training dataset”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Applicant state in page 24, last para:
As already noted herein, MPEP § 2111 requires that claim limitations be given meaning consistent with their use in the specification and drawings. The specification at <]{[0165] defines "customized" in the context of a detector trained specifically for a particular input video's content domain. Rodriguez's user-preference recognition customization does not correspond to this concept.
In response the specification [0165] does not clearly set forth a special definition of “customized” via MPEP 2111.01 Plain Meaning [R-01.2024] IV. APPLICANT MAY BE OWN LEXICOGRAPHER AND/OR MAY DISAVOW CLAIM SCOPE
The only exceptions to giving the words in a claim their ordinary and customary meaning in the art are (1) when the applicant acts as their own lexicographer; and (2) when the applicant disavows or disclaims the full scope of a claim term in the specification. To act as their own lexicographer, the applicant must clearly set forth a special definition of a claim term in the specification that differs from the plain and ordinary meaning it would otherwise possess. CCS Fitness, Inc. v. Brunswick Corp., 288 F.3d 1359, 1366, 62 USPQ2d 1658, 1662 (Fed. Cir. 2002). The specification may also include an intentional disclaimer, or disavowal, of claim scope. In both of these cases, "the inventor’s intention, as expressed in the specification, is regarded as dispositive." Phillips v. AWH Corp., 415 F.3d 1303, 1316 (Fed. Cir. 2005) (en banc). See also Starhome GmbH v. AT&T Mobility LLC, 743 F.3d 849, 857, 109 USPQ2d 1885, 1890-91 (Fed. Cir. 2014) (holding that the term "gateway" should be given its ordinary and customary meaning of "a connection between different networks" because nothing in the specification indicated a clear intent to depart from that ordinary meaning); Thorner v. Sony Computer Entm’t Am. LLC, 669 F.3d 1362, 1367-68, 101 USPQ2d 1457, 1460 (Fed. Cir. 2012) (The asserted claims of the patent were directed to a tactile feedback system for video game controllers comprising a flexible pad with a plurality of actuators "attached to said pad." The court held that the claims were not limited to actuators attached to the external surface of the pad, even though the specification used the word "attached" when describing embodiments affixed to the external surface of the pad but the word "embedded" when describing embodiments affixed to the internal surface of the pad. The court explained that the plain and ordinary meaning of "attached" includes both external and internal attachments. Further, there is no clear and explicit statement in the specification to redefine "attached" or disavow the full scope of the term.).
C. The Reasoning Stated For Combining Rana and Rodriguez Fails to Provide Proper and Sufficient Rationale
The Examiner's stated motivation - that "one of skill in the art of models can
make Rana's be as Rodriguez's predictably recognizing the change accurately resulting
in 'updatable and customizable categories for the purposes of recognizing' deepfakes
according to personal preference" - does not satisfy the requirement for a specific,
articulated reason with rational underpinning. KSR International Co. v. Teleflex Inc.,
550 U.S. 398,418 (2007); In re Kahn, 441 F.3d 977, 988 (Fed. Cir. 2006).
Specifically, the stated motivation:
1. Does not identify a specific technical problem that combining Rana and
Rodriguez would solve;
2. Does not explain how48 Rodriguez's user-preference-driven recognition category
customization would be implemented in Rana's deepfake detection pipeline;
3. Does not explain how the combination would produce the claimed content-adaptive,
input-video-feature-driven training data generation pipeline;
4. Rests on the generic observation that both references involve ML models -
which is precisely the type of conclusory reasoning KSR and In re
Kahn require to be supported with specific findings, yet in the present case the
Examiner has provided none.
Regarding bullets “1.” & “2.” & “4.”, the examiner notes that applicants are referencing “‘identify[ing] a reason that would have prompted a person of ordinary skill in the relevant field to combine the elements in the way the claimed new invention does’"” via MPEP 2141 III., 5th para:
MPEP 2141 Examination Guidelines for Determining Obviousness Under 35 U.S.C. 103 [R-01.2024]
EXAMINATION GUIDELINES FOR DETERMINING OBVIOUSNESS UNDER 35 U.S.C. 103
III. RATIONALES TO SUPPORT REJECTIONS UNDER 35 U.S.C. 103, 5th para:
The key to supporting any rejection under 35 U.S.C. 103 is the clear articulation of the reason(s) why the claimed invention would have been obvious. The Supreme Court in KSR noted that the analysis supporting a rejection under 35 U.S.C. 103 should be made explicit. The Court quoting In re Kahn, 441 F.3d 977, 988, 78 USPQ2d 1329, 1336 (Fed. Cir. 2006), stated that "‘[R]ejections on obviousness cannot be sustained by mere conclusory statements; instead, there must be some articulated reasoning with some rational underpinning to support the legal conclusion of obviousness.’" KSR, 550 U.S. at 418, 82 USPQ2d at 1396. See also Adapt Pharma Operations Ltd. v. Teva Pharms. USA, Inc., 25 F.4th 1354, 1365, 2022 USPQ2d 144 (Fed. Cir. 2022) (stating that a determination of obviousness "requires ‘identify[ing] a reason that would have prompted a person of ordinary skill in the relevant field to combine the elements in the way the claimed new invention does’" (quoting KSR, 550 U.S. at 418, 82 USPQ2d at 1395). Examples of rationales that may support a conclusion of obviousness include: [AltContent: rect]
(A) Combining prior art elements according to known methods to yield predictable results;
(B) Simple substitution of one known element for another to obtain predictable results;
(C) Use of known technique to improve similar devices (methods, or products) in the same way;
(D) Applying a known technique to a known device (method, or product) ready for improvement to yield predictable results;
(E) "Obvious to try" – choosing from a finite number of identified, predictable solutions, with a reasonable expectation of success;
(F) Known work in one field of endeavor may prompt variations of it for use in either the same field or a different one based on design incentives or other market forces if the variations are predictable to one of ordinary skill in the art;
(G) Some teaching, suggestion, or motivation in the prior art that would have led one of ordinary skill to modify the prior art reference or to combine prior art reference teachings to arrive at the claimed invention.
Thus there should be a cause for one of skill in the art to combine Rana with Rodriguez in view of MPEP 2141 III. 5th para. Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE), page 106 provides a cause/reason to combine by teaching problems (such as “many computing resources”) with machine learning/deep learning:
PNG
media_image16.png
1359
970
media_image16.png
Greyscale
Rodriguez (US 20220309387 A1) teaches the same problem “more computing resources” via [0073]:
[0073] In some embodiments, step 412 may further comprise scanning the particular set of metadata items of the at least one particular electronic document associated with the account of the particular user of the plurality of users to detect one or more changes in the particular set of metadata items. In various embodiments, such scanning may be performed in any suitable manner or order. In some embodiments, the scanning may be performed in completion such that all of the resultant changes in the metadata items are provided to the data anomaly detection model for processing. In some embodiments, the scanning may be performed in phased operations such that categories of metadata of priority may be first checked upon, after which the resultant data is immediately provided to the data anomaly detection model without waiting for the rest of the metadata being scanned. This way, the scanning may be configured to unearth metadata changes in a prioritized manner, especially given that, one or more types of changes in metadata of specific categories may be dispositive for the data anomaly detection model to conclude there is an occurrence of an anomaly. Accordingly, in some embodiments, when the data anomaly detection model determines that there are one or more anomalies in the account of the user based on the partial scan results, the scanning for the set of documents associated with the user may conclude without incurring more computing resources, thereby achieving the monitoring of a larger population of users' activities in a real time, or near real time fashion.
Thus said one of skill would be “prompted” (MPEP 2141 III. 5th para) to combine to solve the resource problem as shown in the 35 USC 103 rejection of claim 1 in the Office action of 12/03/2025, page 20.
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 25, bullet “3.” : “content-content-adaptive, input-video-feature-driven training data generation pipeline”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
D. The Combination of Rana and Rodriguez Does Not Produce the Claimed Feature(s)
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 25:
• Analyzes a specific input video to extract its content-domain features;
• Dynamically retrieves authentic videos matched to those specific features;
• Feeds those specific authentic videos into deepfake tools to generate content-matched
training data; and
• Trains a detector specifically on that dynamically generated, content-matched
dataset.
) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
IV. BERA DOES NOT QUALIFY AS PRIOR ART AND, EVEN IF IT DID, DOES NOT TEACH THE CLAIMED FEATURE(S)
A. Bera's Effective Filing Date Is Later Than Applicant's Priority Date
Applicant respectfully requests that the Examiner address this dispositive issue directly. If the Examiner intends to maintain Bera as prior art, the Examiner must first resolve the priority dispute in Applicant's disfavor and articulate on the record the specific basis for that determination.
The priority dispute has been resolved in the Priority section, sub-sections A,B,C,D regarding the absence of the full scope of “labelled” as applied to claim 1’s:
“at least one first record” (“labelled”)
“deepfake video” (“labelled”);
“second record” (“labelled”); and
“authentic video” (“labelled”).
B. Even If Bera Qualified as Prior Art, It Does Not Teach the Claimed Pipeline
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 26, last para: “the claimed content-adaptive pipeline in which the features of a specific input video drive the retrieval of feature matched authentic videos and the on-demand generation of content-specific deepfake training data”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
V. TRAVERSAL OF SPECIFIC REJECTIONS
A. Claims 1, 2, 4, 6, 7, 12, 13, 14, and 16 - Rana+ Rodriguez+ Bera
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 27, 4th para: “identify content-domain features for the purpose of driving feature-matched authentic video retrieval… input-video-feature-driven training data customization… content-adaptive pipeline… closed-loop, input-video-driven training data generation and model customization architecture and/or approach ”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Claim 2 and Claim 14:
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 28, 2nd para: “trains a customized detector specifically for an input video” ) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993). In contrast, claim 1, last line says “training the customized detector ML model on the training dataset”.
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 28, 3rd para: “a closed loop for a single input video of unknown authenticity” ) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993). In contrast, claim 1, last line says “training the customized detector ML model on the training dataset”.
Claim 4 and Claim 16:
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 28, 4th para: “used as content-domain identifiers for driving feature-matched authentic video retrieval” ) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993). In contrast, claim 1, last line says “training the customized detector ML model on the training dataset”.
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 28, 5th para: “semantic content-domain descriptors” ) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993). In contrast, claim 1, last line says “training the customized detector ML model on the training dataset”.
Applicants state in page 28, 6th para:
Moreover, as further noted in Section 11.B, the Examiner provides no mapping
for camera used and language spoken - the two features for which the gap between
Rana's disclosure and the claim is most apparent. These features are entirely absent
from Rana's disclosure in the context of content-domain characterization for training
data retrieval.
In response the claimed “camera used and language spoken” “does not limit the scope”49 of claims 4,16 and thus “is taught” given that “one of the alternatives is taught by the prior art” (“people depicted in the input video” is one of the alternative taught by Rana as shown in the Office action of 12/03/2025, page 25, reproduced below) via MPEP 2143.03, 3rd para.
PNG
media_image17.png
1175
831
media_image17.png
Greyscale
Claim 6:
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 28, last para: “ML domain adaptation… the machine learning technique of adapting an existing model to a new data distribution using the dynamically generated content-specific training dataset”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Claim 7:
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 29, 1st para: “ a dynamically generated, input-video-specific dataset … The claimed balanced dataset is specifically balanced between deepfake videos generated from feature-matched authentic videos and the authentic videos themselves - a specific structural relationship that is a direct consequence of the content-adaptive pipeline”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Claim 12:
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 29, 2nd para: “specific content-adaptive training pipeline”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
B. Claims 3 and 15 - Rana + Rodriguez + Bera + Mesut
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 29, 4th para: “specific content-adaptive pipeline… a detector specifically trained on dynamically generated, input-video-feature-matched training data”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Applicants state in page 29, last para:
The Examiner's stated motivation to combine Mesut - based on "financial services" market forces - does not identify a specific technical reason why a POSITA in deepfake detection would consult Mesut's financial fraud detection system to arrive at the claimed use of a content-specifically trained customized detector for detecting deepfake tool usage on a specific input video. Market forces in financial services do
not provide a technical motivation to implement the specific architecture and/or approach of Claims 3 and 15.
In response, the reason or cause why a POSITA in deepfake detection would consult Mesut's financial fraud detection system is at Rana’s page 70, section 7.3 (reproduced below) that teaches the problem of detecting deepfakes is “challenging” and the motivation (as shown in the section title “7.3 Motivation to Analyze and Detect Deepfake”) to detect deepfakes; thus, one of skill in the art would refer to Mesut’s (TR 20211021147 A2) teachings of detecting deepfakes and combine as shown in the Office action of 12/03/2025, page 35:
PNG
media_image18.png
1367
951
media_image18.png
Greyscale
C. Claims 5 and 17 - Rana + Rodriguez + Bera + Barth
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 30, 5th para: “a retrieval search… match features identified from the input video, for the purpose of retrieving training data”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993). In contrast claim 5 says:
PNG
media_image19.png
160
759
media_image19.png
Greyscale
Applicants state in page 30, last para:
The Examiner's stated motivation to combine - that the combination "may improve the performance of the method and reduce the cost for its implementation" (Barth [0028]) - is a general performance improvement rationale that does not specifically address why a POSITA would apply Barth's authentication search methodology to the claimed training data retrieval problem. This is insufficient
under KSR and In re Kahn.
In response, the reason to combine is as discussed above regarding said Rana’s, pg. 70, section: 7.3 Motivation to Analyze and Detect Deepfake. Thus a POSITA would apply Barth's authentication search methodology to Rana’s deepfake detection and combine as shown in the Office action of 12/03/2025, page 38.
D. Claims 8, 9, 18, and 19- Rana+ Rodriguez+ Bera+ Weisz
Applicants state in page 31, last para:
In sharp contrast, Weisz's system:
• Replaces a live video feed with a synthetic video for privacy/appearance
purposes in video conferencing;
• Does not operate "in response to absence of a feature from an authentic video";
• Does not feed authentic videos into a deep fake tool for the purpose of
generating labeled training data;
• Does not address the problem of training data construction for deepfake
detection.
In response to applicant's arguments against the references individually, one cannot show nonobviousness by attacking references individually where the rejections are based on combinations (illustrated below) of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
PNG
media_image20.png
1277
884
media_image20.png
Greyscale
PNG
media_image21.png
2809
1263
media_image21.png
Greyscale
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 32, 2nd para: “the claimed technical context - training data augmentation for deepfake detection” ) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Applicants state in page 32, 2nd para:
Weisz's technical context - real-time video conferencing synthetic replacement- is categorically different from the claimed technical context - training data augmentation for deepfake detection. The Examiner's stated motivation to combine, based on "personal preferences" and "situational context" in video conferencing (Weisz, c.11, 11. 27-27), does not identify a technical reason why a POSITA working on deepf ake detection training data generation would consult a video conferencing privacy tool. These references address different technical problems in different technical fields.
In response, the reason to combine is as discussed above regarding said Rana’s, pg. 70, section: 7.3 Motivation to Analyze and Detect Deepfake. Thus a POSITA would apply Weisz's deepfake detection to Rana’s deepfake detection and combine as shown in the Office action of 12/03/2025, page 41.
E. Claims 10 and 20 - Rana+ Rodriguez+ Bera+ Nguyen + Katoh
Regarding Nguyen:
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 32, last para: “pipeline level…pipeline-level process”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Regarding Katoh:
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 33, 2nd para: “applying multiple ML domain adaptation operations… each calibrated to a different content domain”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Applicants state in page 33, 3rd para:
The Examiner's stated motivation to combine Katoh - that the combination
would "suppress the deterioration in accuracy of the entire models" (Katoh [0031]) - is
directed to Katoh's specific problem of hardware sensor degradation. It does not
provide a technical reason why a POSITA in deepfake detection would apply Katoh's
sensor-condition robustness training methodology to the problem of creating content-domain-
specific customized deepfake detectors through pipeline-level iteration with ML domain shifts.
In response, technical reason why a POSITA in deepfake detection would apply Katoh's sensor-condition robustness training methodology is at Rana’s page 108 teaching a motivation for using machine learning (ML):
PNG
media_image22.png
1386
1081
media_image22.png
Greyscale
Thus said POSITA would of reasonably referred to the machine learning of Katoh and combine as shown in the 35 USC 103 rejection of claims 10,20 in the Office action of 12/03/2025, page 53.
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 33, last para: “the claimed pipeline-level iteration architecture”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
F. Claim 11 - Rana + Rodriguez + Bera + Venkataraman + Hohwald
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 34, 1st para: “Claim l's specific content-adaptive training pipeline”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Regarding Venkataraman:
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 34, 3rd para: “characterizes a video's content features - e.g., people, language, camera, background etc. - for the purpose of selecting appropriate content-specific deepfake detectors…annotations of…content types”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 34, penult para: “selecting content-domain-specific deepfake detectors based on content annotations of an input video of unknown authenticity”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Regarding Hohwald:
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 35, 2nd para: “trained on content-specific training datasets generated by the pipeline of Claim 1”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 35, 2nd para: “selecting content-domain-specific deepfake detectors based on content annotations of an input video of unknown authenticity”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 35, 3rd para: “using content specifically trained customized detectors”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Applicants state in page 25, 3rd para:
The Examiner's stated motivation to combine Hohwald - that it would "continually improve accuracy" and "match user preferences" (Hohwald, c.5, 11. 25-30) - is directed to Hohwald's online learning and user preference refinement system. It does not identify a technical reason why a POSIT A would integrate Hohwald's image search engine style class aggregation into a deepfake detection system using content-specifically trained customized detectors.
The technical reason why a POSITA would integrate Hohwald's image search engine style class aggregation into a deepfake detection system at Rana’s, pg. 92, section 11.2 Challenges in Deepfake Detection (reproduced below), regarding classifying images and citing to another reference [24]:
24. W. Fan, L. Zhao, J. Wang, Y. Chen, F. Wu and Y. Liu, “FamDroid: Learning-Based
Android Malware Family Classification Using Static Analysis,” arXiv:2101.03965v2,
2021.
Thus, a POSITA would have reasonably referred to the reference [24] or other references (Hohwald US 11,017,019) regarding classifying images as genuine (authentic) or computer-generated (non-authentic) and combine as shown in the Office action of 12/03/2025, page 65:
PNG
media_image23.png
1365
973
media_image23.png
Greyscale
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 35, 4th para: “the claimed method of annotation-driven selection of content- specifically trained customized detectors”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
VI. THE EXAMINER'S COMBINATIONS COLLECTIVELY DEMONSTRATE NON-OBVIOUSNESS
Applicant’s state in page 36, 2nd para:
The breadth and diversity of the technical fields from which the Examiner has drawn references to reconstruct the claimed subject matter is objective evidence that the claimed combination was not obvious to a POSITA. Such a POSITA working on deepfake detection would not have had reason to consult references on surgical robotics, video conferencing privacy, sensor condition robustness, and financial fraud detection to arrive at the claimed content-adaptive deepfake detection pipeline. The
Examiner's combination is a product of hindsight reconstruction using Applicant's disclosure as a guide - precisely the methodology condemned in W.L. Gore, 721 F.2d at 1553.
In response to applicant's argument that the examiner's conclusion of obviousness is based upon improper hindsight reasoning, it must be recognized that any judgment on obviousness is in a sense necessarily a reconstruction based upon hindsight reasoning. But so long as it takes into account only knowledge which was within the level of ordinary skill at the time the claimed invention was made, and does not include knowledge gleaned only from the applicant's disclosure, such a reconstruction is proper. See In re McLaughlin, 443 F.2d 1392, 170 USPQ 209 (CCPA 1971).
In response to applicant's argument that the examiner has combined an excessive number of references, reliance on a large number of references in a rejection does not, without more, weigh against the obviousness of the claimed invention. See In re Gorman, 933 F.2d 982, 18 USPQ2d 1885 (Fed. Cir. 1991).
VII. DEMAND FOR RESPONSE TO THE SUBSTANCE OF APPLICANT'S ARGUMENT
In response to applicant's argument that the references fail to show certain features of the invention, it is noted that the features upon which applicant relies (i.e., page 36, bullet 1.: “the claimed content-adaptive, input-video-feature-driven training data generation pipeline… of using the features of a specific input video of unknown authenticity to dynamically retrieve feature-matched authentic videos for on-demand deepfake generation and content-specific detector training”) are not recited in the rejected claim(s). Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
Applicants state in page 36,37:
Applicant has now identified, with specificity, the technical deficiencies in each of the Examiner's reference mappings and combination rationales. Pursuant to MPEP §§ 707.07(f) and 706.03, if the Examiner maintains any of these rejections, Applicant respectfully requests that the Examiner shall identify on the record, for each maintained rejection:
1. Which specific passage in each cited reference teaches the claimed content-adaptive,
input-video-feature-driven training data generation pipeline -not a superficially similar phrase, but the specific technical concept of using the features of a specific input video of unknown authenticity to dynamically retrieve feature-matched authentic videos for on-demand deepfake generation and content-specific detector training;
2. The specific technical reason50 - beyond the generic observation that references involve machine learning models - why51 a POSITA in deepfake detection would have considered the cited references pertinent52 to the problem sought to be solved, as well as53 would have been motivated5455 to combine56 the cited references in the specific manner proposed; and
3. How the proposed combination, if implemented as described, would actually produce the claimed subject matter at hand, rather than those architectures and/or approaches as described in the cited references themselves.
A rejection maintained without addressing these specific points would not meet the examiner's obligation to respond to the substance of Applicant's traversal under 37 C.F.R. § 1.104 and MPEP § 707.07(f).
Regarding bullet “2.” ’s “technical reason” , Official Notice is not taken in view of MPEP 2144.03 B. If Official Notice Is Taken of a Fact, Unsupported by Documentary Evidence, the Technical Line of Reasoning Underlying a Decision To Take Such Notice Must Be Clear and Unmistakable.
Bullet “2.” ’s “technical reason” is interpreted as “
Bullet “2.” is ambiguous and does not reflect the MPEP’s discussion of finding/identifying motivation. Bullet “2.” is interpreted as (wherein underline is addition and
2. The specific technical reason as well as would have been motivated to combine the cited references in the specific manner proposed - beyond the generic observation that references involve machine learning models – why a POSITA in deepfake detection would have considered the cited references pertinent to the problem sought to be solved
Further regarding bullet “2.” with reference to claims 1,3,5,8,10,11 the specific Rodriguez (US 2022/0309387 A1) & Bera (US 2025/0005925 A1) & MESUT (TR 2021021147 A2) & Barth (US 2022/0067125 A1) & Weisz (US 10,904,488 B1) & Nguyen (CAPSULE-FORENSICS: USING CAPSULE NETWORKS TO DETECT FORGED IMAGES AND VIDEOS) & KATOH (US 2022/02445405 A1) & Venkataraman (US 2021/0322121 A1) & Hohwald (US 11,017,019 B1)] pertinent to the problem sought to be solved is at Rana’s (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) pages 67 and 108:
page 67, section 7.1 Introduction (reproduced below), last S, that discusses “Various approaches have since been described in the literature to deal with the problems raised by Deepfake”. Thus, one of skill in the art would have a reason or cause or motive or a prompt to look at other deepfake references [such as Bera (US 2025/0005925 A1) & MESUT (TR 2021021147 A2) & Barth (US 2022/0067125 A1) & Weisz (US 10,904,488 B1) & Nguyen (CAPSULE-FORENSICS: USING CAPSULE NETWORKS TO DETECT FORGED IMAGES AND VIDEOS) & Hohwald (US 11,017,019 B1)] for a solution regarding the problem of deepfakes and respectively combine as shown in the Office action of 12/03/2025, pages 22,35,41,47,65.
7.1 Introduction
Rapid progress in AI, ML, and DL has resulted in various technologies and tools for manipulating multimedia. Though most applications developed are for legitimate purposes such as entertainment, education, etc., malicious users can also exploit them for unlawful or nefarious purposes. For example, high-quality and realistic fake videos, images, or audios have been created to spread misinformation and propaganda, foment political discord and hate, or even harass and blackmail people; these manipulated, high quality and realistic videos became known recently as Deepfake. Various approaches have since been described in the literature to deal with the problems raised by Deepfake.
; and
Rana’s page 108 (reproduced below), 1st paragraph that discusses experimenting with classical machine learning method in detecting Deepfakes and advantages of the machine learning method prompting/causing one of skill in the art to refer to other machine learning teachings [such as Rodriguez (US 2022/0309387 A1) & KATOH (US 2022/02445405 A1) & Venkataraman (US 2021/0322121 A1)] and combine with Rana of the combination of Rana,Rodriguez for detecting/dealing with the problem raised by deepfakes as shown in the rejection of claim 1 in the Office action of 12/03/2025, pages 20 (showing the combination of Rodriguez),53 (showing the combination of Katoh), and page 62 (showing the combination of Venkataraman).
PNG
media_image24.png
1318
915
media_image24.png
Greyscale
Further regarding claims 1,3,5,8,10,11 and bullet “2.”’s The specific - beyond the generic observation that references involve machine learning models - why5758 a POSITA in deepfake detection would have considered the cited references pertinent to the problem sought to be solved, as well as59 would have been motivated606162 to63 combine64 the cited references [Rodriguez (US 2022/0309387 A1) & Bera (US 2025/0005925 A1) & MESUT (TR 2021021147 A2) & Barth (US 2022/0067125 A1) & Weisz (US 10,904,488 B1) & Nguyen (CAPSULE-FORENSICS: USING CAPSULE NETWORKS TO DETECT FORGED IMAGES AND VIDEOS) & KATOH (US 2022/02445405 A1) & Venkataraman (US 2021/0322121 A1) & Hohwald (US 11,017,019 B1)] in the specific manner proposed is discussed above regarding reasons a POSITA would consider the above pertinent references to a problem to be solved as well as (i.e., in addition to) the respective motivational statements (i.e., provided goals) for each combination of the references to achieve in the respective 35 USC 103 rejections of:
Rodriguez (US 2022/0309387 A1) with identified (Rodriguez: [0015] 2nd S: accurate classification) motivation in pages 20 & 49 & 59 in the Office action of 12/03/2025 &
Bera (US 2025/0005925 A1) with identified (Bera [0052): better than other deepfake detection] motivation in page 22 &
MESUT (TR 2021021147 A2) with identified [MESUT: pg. 3, 4th txt blk: a factually accurate “threshold”] motivation in page 35 &
Barth (US 2022/0067125 A1) with identified (Barth [0058]: improved performance and reduced cost) motivation in page 38 &
Weisz (US 10,904,488 B1) with identified (Weisz, c.11,ll. 27: teaches a synthetic video due to personal preference during a video conference) motivation in page 41 &
Nguyen (CAPSULE-FORENSICS: USING CAPSULE NETWORKS TO DETECT FORGED IMAGES AND VIDEOS) with identified (Nguyen: pg. 4, left col, last S: being effective for multiple forged image attacks) motivation in page 47 &
KATOH (US 2022/02445405 A1) with identified (Katoh [0031] last S: maintains accuracy of models) motivation in page 53 &
Venkataraman (US 2021/0322121 A1) with identified (Venkataraman [0070] 3rd S: teaching more accurate prediction for video) motivation in page 62 &
Hohwald (US 11,017,019 B1) with identified (Hohwald: c.5,ll. 25-30: improving accuracy of authentication of a user and hence helping Rana prove video authenticity) motivation in page 65.
3. How65 the proposed combination, if implemented as described, would actually66 produce67 the claimed subject matter at hand, rather than those architectures and/or approaches as described in the cited references themselves.
A rejection maintained without addressing these specific points would not meet the examiner's obligation to respond to the substance of Applicant's traversal under 37 C.F.R. § 1.104 and MPEP § 707.07(f).
Regarding bullet “3.”, the proposed combination, if implemented as described, would actually “produce” (MPEP 2143.01: 2nd para, 2nd S) the claimed subject matter at hand by mapping to the references by column/line number (e.g., page 18 of the Office action of 12/03/2025 mapping the facts of Rana to claim 1) as a finding of actual fact or “Factual findings”68 “wherein there is some teaching, suggestion, or motivation” (MPEP 2143.01: 2nd para, 2nd S), as mapped for bullet “2.” above, for each combination.
PNG
media_image25.png
1273
852
media_image25.png
Greyscale
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 1,2,4,6,7,12 and 13,14,16 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) with annotated version thereof in view of Rodriguez et al. (US 2022/0309387 A1) further in view of Bera et al. (US 2025/0005925 A1) with Related U.S. Application Data: Provisional application No. 63/510,416, filed on Jun. 27, 2023:
PNG
media_image26.png
720
536
media_image26.png
Greyscale
Re 1. (Currently Amended), Rana teaches A computer implemented method of training a customized 69 machine learning70 (ML) model (or a larger “effective Deepfake detector”-“stacking ensemble neural network (called DFC) model”, pg.132: fig. 12.3) for use on an input video, comprising:
analyzing the input video to identify a plurality of (face-“edge”, pg. 110) features associated with the input video (fig. 12.3: “Input Video”);
obtaining a plurality of authentic videos (that “defines the authenticity of the newly generated images”, pg. 68, 1st S) associated with the plurality of (face-“edge”, pg. 110) features;
feeding the plurality of authentic videos into at least one deepfake (“encoder”, pg. 74, 3rd bullet) tool;
obtaining a plurality of deepfake videos (via section 8.2 Deepfake Generation Pipeline, pg. 74) as an outcome of the at least one deepfake tool;
creating (via section 8.2 Deepfake Generation Pipeline, pg. 74) a training dataset (via “Algorithm DeepfakeStack Classifier (DFSC)”) below: “Input: Training data”, pg. 101) comprising a plurality of (data-“folder…file”71, pg. 100) records, wherein at least one first record includes a deepfake video labelled (“Deepfake”-“category”, pg. 100) with a ground truth indicating deepfake, and at least one second record includes an authentic video labelled (“Original”72-“category”, pg. 100) with a ground truth indicating authentic; and
training the customized st bullet) on the training dataset (via:
PNG
media_image27.png
481
1092
media_image27.png
Greyscale
PNG
media_image28.png
680
914
media_image28.png
Greyscale
Rana does not teach the difference of claim 1 of:
A) “customized (detector ML model)…
B) a ground truth…
C) a ground truth”.
Rodriguez teaches the difference of claim 1:
A) customized (“detection”-“machine learned model implementations herein” [0025]) (detector ML model)…
B) a ground truth…
C) a ground truth.
Since73 Rana teaches motivation to “experiment” on a machine learning (ML) model, pg. 108, 1st para (i.e., the reason to combine references):
PNG
media_image29.png
1333
920
media_image29.png
Greyscale
, one of skill in the art of machine learning models would have been motivated to experiment on and to reasonably refer to other teachings of machine learning models in order to detect deepfakes and can make Rana’s be as Rodriguez’s predictably recognizing the change accurately resulting in “updatable74 and customizable75 categories for the purposes of recognizing”, Rodriguez [0015] 2nd S, deepfakes according to personal preference by:
a) creating deepfake images/videos;
b) inputting as training data the deepfake images/videos along with the corresponding originals into Rodriguez’s fig. 2B:286: “Anomaly Model Generation Engine”;
c) verify that Rodriguez’s fig. 2B:286: “Anomaly Model Generation Engine” correctly predicted a fake image verses the original;
d) make feedback adjustments to Rodriguez’s fig. 2B:286: “Anomaly Model Generation Engine” if necessary to correctly detect a deepfake;
e) deploy/stack Rodriguez’s trained machine learning model as Ranna’s teachings (such as said Rana’s page 108 of a machine learning model) of machine learning models such as fig. 1 that stacks deepfake detectors.
PNG
media_image30.png
1175
1091
media_image30.png
Greyscale
The combination (illustrated above) of Rana,Rodriguez does not teach the remaining difference of claim 1:
B) a ground truth (indicating deepfake)76…77
C) a ground truth (indicating authentic).
Bera teaches, referring to 63/510,416, the remaining difference of claim 1:B) a ground truth (“indicating whether a respective training video is or isn't a deepfake video” [0024]) (indicating deepfake)…
C) a ground truth (“indicating whether a respective training video is or isn't a deepfake video” [0024]) (indicating authentic).
Since78 Rana of The combination (illustrated above) of Rana,Rodriguez teaches deepfakes and “problems raised by Deepfake”: Rana page 67, 1st para, last S:
PNG
media_image31.png
1340
966
media_image31.png
Greyscale
, one of skill in the art would have a reason to look at the other teachings successfully dealing79 with the problem raised by deepfakes to find a solution to the deepfake problem and thus can make Rana’s of The combination (illustrated above) of Rana,Rodriguez be as Bera’s predictably recognizing the change “showcasing its superior performance”, Bera [0052], relative to others of deepfake detection by stacking Bera’s vision and audio and lip encoders of Bera’s fig. 1 with Rana’s deepfake detector stack of Rana’s fig. 1:
PNG
media_image32.png
2037
1300
media_image32.png
Greyscale
Re 2. (Original), Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches The computer implemented method of claim 1, wherein80
A) it is81 (via the above illustrated-combination figure) undetermined (until reaching, via the illustrated combination figure, “Real” or “Deepfaked” or “Fake”) whether the input video is (via the above illustrated-combination figure) authentic or deepfake (as shown in the above illustrated figure), and82
it is83 undetermined (until reaching, via the illustrated combination figure, “Real” or “Deepfaked” or “Fake”)
B) whether the plurality of features are (via the above illustrated-combination figure) authentic of fake (as shown in the above illustrated figure),
C) whether the customized st bullet as further modified via the combination) is84 (via the above illustrated-combination figure) for determining whether the input video is (via the above illustrated-combination figure) authentic or deepfake (as shown in the above illustrated figure).
Re 4. (Original), Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches The computer implemented method of claim 1, wherein the plurality of (face-“edge”, pg. 110) features include85:
A) (“eliminating image frames containing multiple”, Rana, pg. 75) people depicted in the input video,
B) camera used to capture the input video,
C) language spoken in the input video, and86
D) background depicted in the input video87.
Re 6. (Original), Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches The computer implemented method of claim 1, wherein training (via “Algorithm DeepfakeStack Classifier (DFSC)”, below: “Input: Training data”, pg. 101) the customized ML model (fig. 12.3: “DL-based models” “for detecting such manipulated videos”, pg. 71, 1st bullet) comprises applying88
A) fine-tuning89 (expressed as actions: “adjust” and “For example…adapt”, pg. 102) and/or
B) (“frequency”, pg. 79) domain shift90
on91 an (“already been”) existing (“DeepfakeStackClassifier (DFC)”, pg. 103) deepfake detection model (i.e., already existing video-detecting DFC “architecture as base-learners, and…models”, pg. 102, “for detecting such manipulated videos”, pg. 71, 1st bullet) using the training dataset (via “Algorithm DeepfakeStack Classifier (DFSC)”, below: “Input: Training data”, pg. 101) for creating the customized ML model (fig. 12.3: “DL-based models” “for detecting such manipulated videos”, pg. 71, 1st bullet).
Re 7. (Original), Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches The computer implemented method of claim 1, wherein a plurality of second (“Original”92-“category”, pg. 100) records are for inclusion in the training dataset for balancing the training dataset (“to make the balanced dataset”, pg. 100, 1st bullet), the plurality of second (“Original”93-“category”, pg. 100) records including the plurality of authentic videos (that “defines the authenticity of the newly generated images”, pg. 68, 1st S) labelled (in an “Original”94-“category”, pg. 100) with a ground truth (via The combination (illustrated above) of Rana,Rodriguez,Bera) indicating absence of deepfake and/or authentic.
Re 12. (Currently Amended), Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches A computer implemented method of using a customized 95 machine learning (ML) model (fig. 12.3: “DL-based models” “for detecting such manipulated videos”, pg. 71, 1st bullet as further modified via the combination) on an input (face-“edge”, pg. 110) video, comprising:
feeding an input video into the customized st bullet as further modified via the combination) trained according to claim 1; and
obtaining an indication (via Rodriguez of The combination (illustrated above) of Rana,Rodriguez,Bera) of whether the input (face-“edge”, pg. 110) video is a deepfake video as an outcome of the customized st bullet as further modified via the combination).
Claim 13 is rejected like claim 1:
13. (Currently Amended) A system for training a customized machine learning (ML) model for use on an input video, comprising:
at least one processor executing a code for:
analyzing the input video to identify a plurality of features associated with the input video;
obtaining a plurality of authentic videos associated with the plurality of features; feeding the plurality of authentic videos into at least one deepfake tool;
obtaining a plurality of deepfake videos as an outcome of the at least one deepfake tool;
creating a training dataset comprising a plurality of records, wherein at least one first record includes a deepfake video labelled with a ground truth indicating deepfake, and at least one second record includes an authentic video labelled with a ground truth indicating authentic; and
training the customized
Claim 14 is rejected like claim 2:
14. (Original) The system of claim 13, wherein
A) it is undetermined whether the input video is authentic or deepfake, and
it is undetermined
B) whether the plurality of features are authentic of fake,
C) whether the customized
Re 2., Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches The computer implemented method of claim 1, wherein96
A) it is97 (via the above illustrated-combination figure) undetermined (until reaching, via the illustrated combination figure, “Real” or “Deepfaked” or “Fake”) whether the input video is (via the above illustrated-combination figure) authentic or deepfake (as shown in the above illustrated figure), and98
it is99 undetermined (until reaching, via the illustrated combination figure, “Real” or “Deepfaked” or “Fake”)
B) whether the plurality of features are (via the above illustrated-combination figure) authentic of fake (as shown in the above illustrated figure),
C) whether the customized detector ML model (fig. 12.3: “DL-based models” “for detecting such manipulated videos”, pg. 71, 1st bullet as further modified via the combination) is100 (via the above illustrated-combination figure) for determining whether the input video is (via the above illustrated-combination figure) authentic or deepfake (as shown in the above illustrated figure).
Claim 16 is rejected like claim 4:
16. (Original) The system of claim 13, wherein the plurality of features include:
A) people depicted in the input video,
B) camera used to capture the input video,
C) language spoken in the input video, and
D) background depicted in the input video (via the rejection of claim 4:
Re 4., Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches The computer implemented method of claim 1, wherein the plurality of (face-“edge”, pg. 110) features include101:
A) (“eliminating image frames containing multiple”, Rana, pg. 75) people depicted in the input video,
B) camera used to capture the input video,
C) language spoken in the input video, and102
D) background depicted in the input video103.
Claim(s) 3 and 15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) in view of Rodriguez et al. (US 2022/0309387 A1) further in view of Bera et al. (US 2025/0005925 A1) with Related U.S. Application Data: Provisional application No. 63/510,416, filed on Jun. 27, 2023, as applied in claims 1,2,4,6,7,12 and 13,14,16 further in view of MESUT et al. (TR 2021021147 A2) with SEARCH machine translation:
PNG
media_image33.png
720
563
media_image33.png
Greyscale
Re 3. (Original), Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches The computer implemented method of claim 1, further comprising feeding the (face-“edge”, pg. 110) input video into the customized st bullet as further modified via the combination) for detecting use of a deepfake (“encoder”, pg. 74, 3rd bullet) tool on the (face-“edge”, pg. 110) input video.
Rana of The combination (illustrated above) of Rana,Rodriguez,Bera does not teach the difference of claim 3: “detecting use”.
Mesut teaches the difference of claim 3 of “detecting use” via “detecting the use”, Mesut, Abstract.
Since104 Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches deepfake as discussed above in claim 1, one of skill in the art of deepfakes can make Rana’s of The combination (illustrated above) of Rana,Rodriguez,Bera be as Mesut’s predictably recognizing the change Prompting Variations of It for Use in Either the Same Field or a Different One Based on Design Incentives or Other Market Forces, MPEP 2143 I.F.:
F. Known Work in One Field of Endeavor May Prompt Variations of It for Use in Either the Same Field or a Different One Based on Design Incentives or Other Market Forces if the Variations Are Predictable to One of Ordinary Skill in the Art:
F. Known Work in One Field of Endeavor May Prompt Variations of It for Use in Either the Same Field or a Different One Based on Design Incentives or Other Market Forces if the Variations Are Predictable to One of Ordinary Skill in the Art
To reject a claim (claim 3) based on this rationale, Office personnel must resolve the Graham factual inquiries (shown above). Then, Office personnel must articulate the following:
(1) a finding [Mesut, abstract: “deepfake”] that the scope and content of the prior art (MESUT et al. (TR 2021021147 A2)), whether in the same field of endeavor as that of the applicant’s invention or a different field of endeavor, included a similar or analogous device (method, or product);
(2) a finding (“financial services”, MESUT: pg. 1, 9th txt blk) that there were design incentives or market forces which would have prompted adaptation of the known device (method, or product) [Mesut, abstract: “deepfake”];
(3) a finding (shown above in the rejection of claim 3) that the differences (claim 3’s “detecting use”) between the claimed invention and the prior art (MESUT et al. (TR 2021021147 A2)) were encompassed in known variations or in a principle known in the prior art (MESUT et al. (TR 2021021147 A2));
(4) a finding that one of ordinary skill in the art, in view of the identified design incentives or other market forces (“financial services”, MESUT: pg. 1, 9th txt blk), could have implemented the claimed variation of the prior art (MESUT et al. (TR 2021021147 A2)), and the claimed variation would have been predictable (by being factually accurate via a “threshold”105, MESUT pg. 3, 4th txt blk) to one of ordinary skill in the art; and
(5) whatever additional findings based on the Graham factual inquiries may be necessary, in view of the facts of the case under consideration, to explain a conclusion of obviousness.
The rationale to support a conclusion that the claimed invention would have been obvious is that design incentives or other market forces could have prompted one of ordinary skill in the art to vary the prior art in a predictable manner to result in the claimed invention. If any of these findings cannot be made, then this rationale cannot be used to support a conclusion that the claim would have been obvious to one of ordinary skill in the art.
by:
a) obtaining Rana’s “probability of correctly detecting the Deepfake”, Rana: pg. 96. last para, 2nd S; and
b) creating an accuracy truth threshold program with the accuracy truth threshold set according to MESUT’s “threshold”, NESUT, pg. 3, 4th txt blk;
c) inputting Rana’s probability of correctly detecting the Deepfake into the accuracy truth threshold program
d) using the true and accurate probability as Rana’s teachings of probability:
PNG
media_image34.png
2967
1300
media_image34.png
Greyscale
Claim 15 is rejected like claim 3:
15. (Original) The system of claim 13, further comprising code for feeding the input video into the customized
Re 3., Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches The computer implemented method of claim 1, further comprising feeding the (face-“edge”, pg. 110) input video into the customized detector ML model (fig. 12.3: “DL-based models” “for detecting such manipulated videos”, pg. 71, 1st bullet as further modified via the combination) for detecting use of a deepfake (“encoder”, pg. 74, 3rd bullet) tool on the (face-“edge”, pg. 110) input video.
.
Claim(s) 5 and 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) in view of Rodriguez et al. (US 2022/0309387 A1) further in view of Bera et al. (US 2025/0005925 A1) with Related U.S. Application Data: Provisional application No. 63/510,416, filed on Jun. 27, 2023, as applied in claims 1,2,4,6,7,12 and 13,14,16, further in view of Barth et al. (US 2022/0067135 A1):
PNG
media_image35.png
720
563
media_image35.png
Greyscale
Re 5. (Original), Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches The computer implemented method of claim 1, wherein the obtaining the plurality of authentic videos (that “defines the authenticity of the newly generated images”, pg. 68, 1st S) comprises searching
A) the internet (“with internet connection”, pg. 67) and/or
B) at least one database (with “Database Name”, pg. 87)
for the plurality of authentic videos (that “defines the authenticity of the newly generated images”, pg. 68, 1st S) according to at least one (face-“edge”, pg. 110) feature of the plurality of (face-“edge”, pg. 110) features.
Rana of The combination (illustrated above) of Rana,Rodriguez,Bera does not teach the difference of claim 5:
“searching” (the Markush element).
Barth teaches the difference of claim 5:
searching (“for matches of the characteristic features 21 as predefined in the feature database 53 within the authenticating image or video data 45” [0058], last S) (the Markush element).
Since106 Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches/suggests selecting a database, page 97: Table 10.1: “The List of Deepfake datasets”:
PNG
media_image36.png
1545
1055
media_image36.png
Greyscale
, one of skill in the art of databases would have a reason to look at various database for deepfakes and pick a dataset for deepfakes and thus can make Rana’s database of The combination (illustrated above) of Rana,Rodriguez,Bera be as Barth’s predictably recognizing the change “may improve the performance of the method and reduce the cost for its implementation”, Barth [0028] last S by:
a) installing Barth’s search function (fig. 3:140: “step 140 searches for matches of the characteristic features 21 as predefined in the feature database 53 within the authenticating image or video data 45”, Barth [0058]) into an internet connected computer;
a) using Barth’s installed search function to search the deepfake databases;
b) selecting a deepfake database after performing the search;
c) downloading the selected dataset to the internet connected computer;
c) using the selected, downloaded deepfake database as Rana’s teaching of a database.
Claim 17 is rejected like claim 5:
17. (Original) The system of claim 13, wherein the obtaining the plurality of authentic videos comprises searching the
A) internet and/or
B) at least one database
for the plurality of authentic videos according to at least one feature of the plurality of features (via the rejection of claim 5:
Re 5., Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches The computer implemented method of claim 1, wherein the obtaining the plurality of authentic videos (that “defines the authenticity of the newly generated images”, pg. 68, 1st S) comprises searching
A) the internet (“with internet connection”, pg. 67) and/or
B) at least one database (with “Database Name”, pg. 87)
for the plurality of authentic videos (that “defines the authenticity of the newly generated images”, pg. 68, 1st S) according to at least one (face-“edge”, pg. 110) feature of the plurality of (face-“edge”, pg. 110) features.
.
Claim(s) 8,9 and 18,19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) in view of Rodriguez et al. (US 2022/0309387 A1) further in view of Bera et al. (US 2025/0005925 A1) with Related U.S. Application Data: Provisional application No. 63/510,416, filed on Jun. 27, 2023, as applied in claims 1,2,4,6,7,12 and 13,14,16, further in view of Weisz et al. (US 10,904,488 B1):
PNG
media_image37.png
720
563
media_image37.png
Greyscale
Re 8. (Original), Rana of The combination (illustrated above) of Rana, Rodriguez, Bera teaches The computer implemented method of claim 1, further comprising:
in response to absence of a (face-“edge”, pg. 110) feature from an authentic video (that “defines the authenticity of the newly generated images”, pg. 68, 1st S),
feeding the authentic video (that “defines the authenticity of the newly generated images”, pg. 68, 1st S) into the at least one deepfake (“encoder”, pg. 74, 3rd bullet) tool for including the (face-“edge”, pg. 110) feature in association with a deepfake video generated by at least one deepfake (“encoder”, pg. 74, 3rd bullet) tool.
Rana of The combination (illustrated above) of Rana,Rodriguez,Bera does not teach the difference of claim 8 :
“in response to absence”.
Weisz teaches the difference of claim 8:
in response to absence (“the deep net 308 may generate video output 310 to replace live video transmission with a synthetic video transmission based at least in part on the participant model 304.”, c.7,ll. 35-45).
Since107 Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches/suggests generating deepfakes and challenges thereof, pg. 90, section 11.1 Challenges in Deepfake Generation, 1st para:
PNG
media_image38.png
1535
1056
media_image38.png
Greyscale
, one of skill in the art of deepfakes would have a reason to look to other teachings of generating deepfakes to address such challenges and thus can make Rana’s deepfake generation of The combination (illustrated above) of Rana,Rodriguez,Bera be as Weisz’s predictably recognizing the change “to transmit the synthetic video transmission due to personal preferences, based on situational context (e.g., whether the participant believes a video conference requires more formal dress than what the participant is currently wearing), etc.”, Weisz, c. 11,ll. 27-27 by training Rana’s deepfake generation according to Weisz’s training: “The trained participant model is usable to generate a synthetic video representation of the participant.”, Weisz, c. 3,ll. 15-20.
Claim 9 is rejected like claim 8:
9. (Original) The computer implemented method of claim 1, further comprising:
in response to a (face-“edge”, pg. 110) feature being associated with the authentic video (that “defines the authenticity of the newly generated images”, pg. 68, 1st S), the (face-“edge”, pg. 110) feature representing an authentic (face-“edge”, pg. 110) feature, feeding the authentic video (that “defines the authenticity of the newly generated images”, pg. 68, 1st S) into the at least one deepfake (“encoder”, pg. 74, 3rd bullet) tool for replacing the authentic (face-“edge”, pg. 110) feature with a deepfake (face-“edge”, pg. 110) feature associated with the deepfake video generated by at least one deepfake (“encoder”, pg. 74, 3rd bullet) tool.
Like claim 8:
Rana of The combination (illustrated above) of Rana, Rodriguez, Bera does not teach the difference of claim 8 (9) :
“in response to absence”
“replacing”.
Weisz teaches the difference of claim 8(9):
in response to absence (“the deep net 308 may generate video output 310 to replace live video transmission with a synthetic video transmission based at least in part on the participant model 304.”, c.7,ll. 35-45).
Since Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches deepfakes, one of skill in the art of deepfakes can make Rana’s of The combination (illustrated above) of Rana,Rodriguez,Bera be as Weisz’s predictably recognizing the change “to transmit the synthetic video transmission due to personal preferences, based on situational context (e.g., whether the participant believes a video conference requires more formal dress than what the participant is currently wearing), etc.”, Weisz, c. 11,ll. 27-27.
Claim 18 is rejected like claim 8:
18. (Original) The system of claim 13, further comprising code for:
in response to absence of a feature from an authentic video,
feeding the authentic video into the at least one deepfake tool for including the feature in association with a deepfake video generated by at least one deepfake tool (via the rejection of claim 8:
Re 8., Rana of The combination (illustrated above) of Rana,Rodriguez,Bera teaches The computer implemented method of claim 1, further comprising:
in response to absence of a (face-“edge”, pg. 110) feature from an authentic video (that “defines the authenticity of the newly generated images”, pg. 68, 1st S),
feeding the authentic video (that “defines the authenticity of the newly generated images”, pg. 68, 1st S) into the at least one deepfake (“encoder”, pg. 74, 3rd bullet) tool for including the (face-“edge”, pg. 110) feature in association with a deepfake video generated by at least one deepfake (“encoder”, pg. 74, 3rd bullet) tool.).
Claim 19 is rejected like claim 9:
19. The system of claim 13, further comprising code for:
in response to a feature being associated with the authentic video, the feature representing an authentic feature, feeding the authentic video into the at least one deepfake tool for replacing the authentic feature with a deepfake feature associated with the deepfake video generated by at least one deepfake tool (via the rejection of claim 9:
Claim 9 is rejected like claim 8:
9. The computer implemented method of claim 1, further comprising:
in response to a (face-“edge”, pg. 110) feature being associated with the authentic video (that “defines the authenticity of the newly generated images”, pg. 68, 1st S), the (face-“edge”, pg. 110) feature representing an authentic (face-“edge”, pg. 110) feature, feeding the authentic video (that “defines the authenticity of the newly generated images”, pg. 68, 1st S) into the at least one deepfake (“encoder”, pg. 74, 3rd bullet) tool for replacing the authentic (face-“edge”, pg. 110) feature with a deepfake (face-“edge”, pg. 110) feature associated with the deepfake video generated by at least one deepfake (“encoder”, pg. 74, 3rd bullet) tool.).
Claim(s) 10 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) in view of Rodriguez et al. (US 2022/0309387 A1) further in view of Bera et al. (US 2025/0005925 A1) with Related U.S. Application Data: Provisional application No. 63/510,416, filed on Jun. 27, 2023, as applied in claims 1,2,4,6,7,12 and 13,14,16, further in view of Nguyen et al. (CAPSULE-FORENSICS: USING CAPSULE NETWORKS TO DETECT FORGED IMAGES AND VIDEOS) further in view of KATOH et al. (US 2022/0245405 A1):
PNG
media_image39.png
724
674
media_image39.png
Greyscale
Re 10. (Original), Rana of The combination (illustrated above) of Rana, Rodriguez, Bera The computer implemented method of claim 1, further comprising:
iterating (via “a few iterations”, pg. 86: Methods: “Classifiers” “CapsuleNetwork [19]”: “CNN”)
the (face-“edge”, pg. 110) analyzing,
the obtaining the plurality of authentic videos (that “defines the authenticity of the newly generated images”, pg. 68, 1st S),
the (“encoder”, pg. 74, 3rd bullet) feeding,
the obtaining (via section 8.2 Deepfake Generation Pipeline, pg. 74) the plurality of deepfake videos,
the creating the training dataset (via “Algorithm DeepfakeStack Classifier (DFSC)”: algorithm line 5: “Crate new dataset from D”, pg. 101), and
the (fig. 12.3: “DL-based models” “for detecting such manipulated videos”, pg. 71, 1st bullet) training,
by applying (via “Apply a Gaussian filter to diffuse the mask boundary area further”, pg. 76: description of fig. 8.3) a plurality of domain shifts to create a plurality of customized ML models (via “Algorithm DeepfakeStack Classifier (DFSC)”: algorithm line 1: “Generate base learners/classifiers, c1,c2…,cT”, pg. 101).
Rana of The combination (illustrated above) of Rana, Rodriguez, Bera does not teach the difference of claim 10 of:
A) “iterating…
B) a plurality of domain shifts to…
C) customized
Nguyen teaches A) of the difference of claim 10:
A) (“We slightly improved the algorithm of Sabour et al. [15] by adding Gaussian random noise to the 3-D weight tensor W and applying one additional squash (equation 1) before routing by”--3rd page, 3rd sentence below: “Algorithm 1 Dynamic routing between capsules”: 6th line: “for r iterations do”--) iterating…
B) a plurality of domain shifts to…
C) customized
Since108 Rana of The combination (illustrated above) of Rana, Rodriguez, Bera cites/”specifically refers to the other”109 (via “a few iterations”, pg. 86: Methods: “Classifiers” “CapsuleNetwork [19]”: “CNN”) to Nguyen, one of skill in the art can make Rana’s of The combination (illustrated above) of Rana, Rodriguez, Bera be as Nguyen’s predictably recognizing the change as “effective for a wide range of forged image and video attacks”, Nguyen, pg. 4 left col, last S.
The combination (illustrated above) of Rana, Rodriguez, Bera,Nguyen does not
teach the remaining difference of claim 10:
B) a plurality of domain shifts to…
C) customized
Rodriguez teaches C) of the difference of claim 10 as discussed in the rejection of claim 1:
B) a plurality of domain shifts to…
C) customized
Similar to the rejection of claim 1, Since110 Rodriguez of The combination (illustrated above) of Rana, Rodriguez, Bera teaches a model, one of skill in the art of models can make the combination’s of The combination (illustrated above) of Rana, Rodriguez, Bera be as Rodriguez’s predictably recognizing the change accurately resulting in “updatable111 and customizable112 categories for the purposes of recognizing”, Rodriguez [0015] 2nd S, deepfakes according to personal preference:
PNG
media_image40.png
2967
1300
media_image40.png
Greyscale
PNG
media_image41.png
2301
1300
media_image41.png
Greyscale
The combination (illustrated above) of Rana, Rodriguez, Bera,Nguyen does not
teach the remaining difference of claim 10:
B) a plurality of domain shifts to (create a plurality of customized ML models)113 …114
Katoh teaches the remaining difference of claim 10:
B) a plurality of (“expected”) domain shifts to (“each model” or “for each model. For example, according to the first learning method, deterioration of a sensor of a camera for capturing data to be estimated, positional deviation of the camera, and an increase in noise at a time of imaging are assumed, and model learning115 is executed with those statuses assumed in advance.” [0058]) (create a plurality of customized detector ML models)116 …117 (as indicated in fig. 1: MODEL 1-3:
PNG
media_image42.png
690
1069
media_image42.png
Greyscale
Since118 Rana of The combination (illustrated above) of Rana, Rodriguez, Bera, Nguyen teaches motivation for machine learning as discussed in the rejection of claim 1, one of skill in the art of models would of have a reason to look to other machine learning models for deepfake detection and thus can make Rana’s of The combination (illustrated above) of Rana, Rodriguez, Bera, Nguyen be as Katoh’s predictably recognizing the change “to suppress the deterioration in accuracy of the entire models”, Katoh [0031] last S by:
a) obtaining Katoh’s machine learning models;
b) inputting deepfakes and corresponding originals into Katoh’s machine learning models;
b) training Katoh’s machine learning models to recognize deepfakes from the originals;
c) inputting the trained models in the deepfake detector stack of Rana’s fig. 1:
PNG
media_image43.png
2037
1300
media_image43.png
Greyscale
Claim 20 is rejected like claim 10:
20. (Original) The system of claim 13, further comprising code for:
iterating
the analyzing,
the obtaining the plurality of authentic videos,
the feeding,
the obtaining the plurality of deepfake videos,
the creating the training dataset, and the training,
by applying a plurality of domain shifts to create a plurality of customized
Re 10., Rana of The combination (illustrated above) of Rana, Rodriguez, Bera The computer implemented method of claim 1, further comprising:
iterating (via “a few iterations”, pg. 86: Methods: “Classifiers” “CapsuleNetwork [19]”: “CNN”)
the (face-“edge”, pg. 110) analyzing,
the obtaining the plurality of authentic videos (that “defines the authenticity of the newly generated images”, pg. 68, 1st S),
the (“encoder”, pg. 74, 3rd bullet) feeding,
the obtaining (via section 8.2 Deepfake Generation Pipeline, pg. 74) the plurality of deepfake videos,
the creating the training dataset (via “Algorithm DeepfakeStack Classifier (DFSC)”: algorithm line 5: “Crate new dataset from D”, pg. 101), and
the (fig. 12.3: “DL-based models” “for detecting such manipulated videos”, pg. 71, 1st bullet) training,
by applying (via “Apply a Gaussian filter to diffuse the mask boundary area further”, pg. 76: description of fig. 8.3) a plurality of domain shifts to create a plurality of customized detector ML models (via “Algorithm DeepfakeStack Classifier (DFSC)”: algorithm line 1: “Generate base learners/classifiers, c1,c2…,cT”, pg. 101).
Claim(s) 11 is/are rejected under 35 U.S.C. 103 as being unpatentable over Rana (ANALYZING AND DETECTING ANDROID MALWARE AND DEEPFAKE) with annotated version thereof in view of Rodriguez et al. (US 2022/0309387 A1) further in view of Bera et al. (US 2025/0005925 A1) with Related U.S. Application Data: Provisional application No. 63/510,416, filed on Jun. 27, 2023, as applied in claims 1,2,4,6,7,12 and 13,14,16, further in view of Venkataraman et al. (US 2021/0322121 A1) further in view of Hohwald et al. (US 11,017,019 B1):
PNG
media_image44.png
724
783
media_image44.png
Greyscale
Re 11. (Currently Amended), Rana of the combination (illustrated above) of Rana, Rodriguez, Bera teaches A computer implemented method of detecting a deepfake video, comprising:
analyzing a video (“of one’s face”, pg. 73);
annotating the video according to the analysis (“of one’s face”, pg. 73);
selecting a plurality of customized machine learning (ML) models (i.e., “Model A (DL-based models)”119), pg. 83) according120 to the annotations,
wherein the plurality of customized “Algorithm DeepfakeStack Classifier (DFSC)”: algorithm line 1: “Generate base learners/classifiers, c1,c2…,cT”, pg. 101) according to claim 1 (via the combination (illus9trated above) of Rana, Rodriguez, Bera);
aggregating (via “aggregate the temporal and spatial consistency”, pg. 86: “Table 9.2 (continued)”: row: “FakeCatcher [76]” and “a combined121 prediction”, pg. 99, description of fig. 12.2) a plurality of outcomes (that “suggest that deep learning-based methods are effective in Deepfake detection”, pg. 83) of the plurality of customized
determining a probability that the video was created (via a “Deepfake”122123 “probability”, pg. 96) by a deepfake (“encoder”, pg. 74, 3rd bullet) tool according to the aggregation.
Rana of the combination (illustrated above) of Rana, Rodriguez, Bera does not teach the difference of claim 11 of:
A) “annotating…according…
B) selecting…
C) customized
D) according to the annotations…
E) customized
F) aggregating…
G) customized
H) according to the aggregation”.
Similar to the rejection of claim 10, Rodriguez teaches C) & E) & G) of the difference of claim 11 as discussed in the rejection of claim 1:
C) customized
E) customized
G) customized
Similar to the rejection of claim 1, Since124 Rodriguez of The combination (illustrated above) of Rana, Rodriguez, Bera teaches a model, one of skill in the art of models can make the combination’s of The combination (illustrated above) of Rana, Rodriguez, Bera be as Rodriguez’s predictably recognizing the change accurately resulting in “updatable125 and customizable126 categories for the purposes of recognizing”, Rodriguez [0015] 2nd S, deepfakes according to personal preference:
Rana of the combination (illustrated above) of Rana, Rodriguez, Bera does not teach the difference of claim 11 of:
A) annotating (the video)127…128according (to the analysis)…
B) selecting (a plurality129 of customized detector ML models)…
D) according to the annotations…
F) aggregating (a plurality of outcomes )…
H) according to the aggregation.
Venkataranan teaches A) and B) and D) of the remaining difference of claim 11:
A) annotating (via fig. 2:206: “FOR EACH OF THE EXTRACTED VIDEO SEGMENTS, ANNOTATE VIDEO FRAMES CONTAINING THE IMAGES OF THE TARGET TYPE OF TOOL-TISSUE INTERACTION ACCORDING TO A SET OF PREDETERMINED STRENGTH LEVELS”) (the video)130…131according (“a set of predetermined strength levels” [0039] at listed item “3)” and “by132…analytic process” [0017]) (to the analysis)…
B) selecting133134 (expressing:
(1) the action thereof via “selects a machine learning model from a set of machine learning models” [0024] 4th S: fig. 3: 306: “SELECT A VISUAL HAPTIC MODEL FROM A SET OF MODELS OF THE DISCLOSED VISUAL-HAPTIC FEEDBACK SYSTEM BASED ON THE DETECTED SURGICAL TASK” &
(2) the result thereof in fig. 3:308: “APPLY THE SELECTED VISUAL HAPTIC MODEL--”) (a plurality of customized detector ML models)…
D) (“training the selected machine learning model”) according135136137 to (or “by:”138) the annotations (i.e., “4) using the annotated video images” [0039])139…
F) aggregating (a plurality of outcomes )…
H) according to the aggregation.
Since140 Rana of the combination (illustrated above) of Rana, Rodriguez, Bera teaches a model as discussed in the rejection of claim 1, one of skill in the art of models would of have a reason to refer to other teachings of machine learning for detecting deepfakes and thus can make Rana’s of the combination (illustrated above) of Rana, Rodriguez, Bera be as Venkataraman’s predictably recognizing the change “collectively allows for identifying correlations among the sequence of video images/frames to facilitate generating a more accurate prediction for the sequence of video images/frames.”, Venkataraman [0070] 3rd S, by:
a) obtaining Venkataraman’s machine learning model;
b) inputting deepfakes and corresponding originals into Venkataraman’s machine learning model;
c) training/experimenting on Venkataraman’s machine learning models to detect deepfakes from the originals;
d) inputting Venkataraman’s trained, experimentation model into Rana stack of deepfake detectors.
The combination (illustrated above) of Rana, Rodriguez, Bera, Venkataraman does not teach the remaining limitation of claim 11:
F) aggregating (a plurality of outcomes )141…142
H) (determining a probability143…144) according145 to the aggregation.
Hohwald teaches the remaining difference of claim 11:
F) aggregating (resulting in an aggregate via “gather”146, c.15,ll. 14-20: fig. 2:242: “IMAGE SEARCH ENGINE”: fig. 5:506: “Image Authenticity Probability”147) (a plurality of outcomes )…
H) (determining a probability…) (“determine an image authenticity probability 506”, c.13,ll. 45-50) according148 to (i.e., “based on149”, ibid) the aggregation (or “the combined style class probabilities” 504, ibid).
PNG
media_image45.png
803
1113
media_image45.png
Greyscale
Since150 Rana of the combination (illustrated above) of Rana, Rodriguez, Bera, Venkataraman teaches experimenting on a machine learning model for detecting deepfakes as discussed in the rejection of claim 1, one of skill in the art of models would have a reason to refer to other teachings of machine learning for experimenting thereon for detecting deepfakes and thus can make Rana’s of the combination (illustrated above) of Rana, Rodriguez, Bera, Venkataraman be as Hohwald’s predictably recognizing the change to “continually improve its accuracy (and keep up with trending differences of what authentic looks like) by incorporating online learning into its authenticity model. By presenting results to the user, and identifying the media that a user interacts with (indicating positive results), and examining the results that a user ignores (indicating negative results), the system can continually learn and refine itself to match user preferences.”, Hohwald, c. 5,ll. 25-30, contributing to Rana’s teaching of providing “some critical features helpful in proving its [video contents’] authenticity”, Rana, pg. 82 by:
a) obtaining Venkataraman’s experimented on machine learning predictions (i.e., probabilities) for detecting deepfakes; and
b) inputting the deepfake probabilities into Hohwald’s fig. 5:505: “Logistic Regression Model”.
Thus the combination (illustrated above) of Rana, Rodriguez, Bera, Venkataraman, Holwald teaches claim 11.
Conclusion
The prior art “nearest to the subject matter defined in the claims” (MPEP 707.05) made of record and not relied upon is considered pertinent to applicant's disclosure.
The following table lists several references that are relevant to the subject matter claimed and disclosed in this Application. The references are not relied on by the Examiner, but are provided to assist the Applicant in responding to this Office action:
Citation
Relevance
Malik et al. (US 2025/0182510 A1)
Malik teaches a custom machine learning model discriminating between genuine and forged data via [0118][0119]:
“for effective discrimination between genuine and forged data…extracted sequences may then be passed through identical DNN models 503 and 504, which could be any deep architecture such as ResNet-18 or a custom-built model”:
PNG
media_image46.png
579
1094
media_image46.png
Greyscale
as the closest to the claimed “customized detector machine learning (ML) model” of claim 1.
Dar et al. (Efficient-SwishNet Based System for Facial Emotion Recognition)
Dar teaches using Facial Recognition (i.e., “our customized Efficient SwishNet model” for “FER”) to detect deepfakes:
“FER has many applications in various domains i.e., face authentication systems, e-learning, detecting emotions of drivers while driving, systems helping the disabled person, healthcare, entertainment, deepfakes detection, etc., [3]... In the pre-processing step, all the images of each dataset are resized to 224 × 224 with three channels. This resolution for the input image is the requirement of our model. After preprocessing, images are fed to our customized Efficient SwishNet model to extract the reliable features and later classify the emotions of seven different categories.”
as the closest to the claimed “customized detector machine learning (ML) model” of claim 1.
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DENNIS ROSARIO whose telephone number is (571)272-7397. The examiner can normally be reached Monday-Friday, 9AM-5PM EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Henok Shiferaw can be reached at 571-272-4637. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/DENNIS ROSARIO/ Examiner, Art Unit 2676
/MATTHEW C BELLA/ Supervisory Patent Examiner, Art Unit 2667
1 tool: any instrument of manual operation. (Dictionary.com)
2 tool: anything used as a means of performing an operation or achieving an end (Dictionary.com)
3 The claimed “tool” (“means”) doesnot invoke 35 USC 112(f).
4 dataset: Computers. a collection of data records for computer processing. (Dictionary.com)
5 record: Computers., a group of related fields, or a single field, treated as a unit and comprising part of a file or data set, for purposes of input, processing, output, or storage by a computer. (Dictionary.com)
6 MPEP 2106.04(d)(1) Evaluating Improvements in the Functioning of a Computer, or an Improvement to Any Other Technology or Technical Field in Step 2A Prong Two [R-10.2019], 2nd para, last Ss:
--Second, if the specification sets forth an improvement in technology, the claim must be evaluated to ensure that the claim itself reflects the disclosed improvement. That is, the claim includes the components or steps of the invention that provide the improvement described in the specification. The claim itself does not need to explicitly recite the improvement described in the specification (e.g., "thereby increasing the bandwidth of the channel").--
7 tool: any instrument of manual operation. (Dictionary.com)
8 tool: anything used as a means of performing an operation or achieving an end (Dictionary.com)
9 The claimed “tool” (“means”) does not invoke 35 USC 112(f).
10 dataset: Computers. a collection of data records for computer processing. (Dictionary.com)
11 record: Computers., a group of related fields, or a single field, treated as a unit and comprising part of a file or data set, for purposes of input, processing, output, or storage by a computer. (Dictionary.com)
12 background: one's origin, education, experience, etc., in relation to one's present character, status, etc., wherein experience is defined: knowledge or practical wisdom gained from what one has observed, encountered, or undergone, wherein practical is defined: of or relating to practice or action, wherein practice is defined: custom, wherein custom is defined: convention, wherein convention is defined: conventionalism, wherein conventionalism is defined: adherence to or advocacy of conventional attitudes or practices (Dictionary.com)
13 tool: any instrument of manual operation. (Dictionary.com)
14 tool: anything used as a means of performing an operation or achieving an end (Dictionary.com)
15 The claimed “tool” (“means”) does not invoke 35 USC 112(f).
16 dataset: Computers. a collection of data records for computer processing. (Dictionary.com)
17 record: Computers., a group of related fields, or a single field, treated as a unit and comprising part of a file or data set, for purposes of input, processing, output, or storage by a computer. (Dictionary.com)
18 scope: Linguistics, Logic. the range of words or elements of an expression (For example: claim 1: “at least one first record includes a deepfake video labelled with a ground truth indicating deepfake”) over which a modifier (patent examiner) or operator (or me) has control. (Dictionary.com)
19 possession: the act or fact of possessing, wherein possess is defined: to have knowledge of. . (Dictionary.com)
20 label: a word or phrase indicating that what follows belongs in a particular category or classification. (Dictionary.com)
21 CLAIM SCOPE: I modify “video” with “labelled”
22 ground truth: semantic sense 2b: 1977- Information obtained by direct observation of a real system, as opposed to a model or simulation; a set of data that is considered to be accurate and reliable, and is used to calibrate a model, algorithm, procedure, etc. Also: (spec. in image recognition technologies) information obtained by direct visual examination, esp. as used to check or calibrate an automated recognition system. (OED.com): Oxford English Dictionary
23 CLAIM SCOPE: I modify “record” with “ground truth”
24 MPEP 2163 Guidelines for the Examination of Patent Applications Under the 35 U.S.C. 112(a) or Pre-AIA 35 U.S.C. 112, first paragraph, "Written Description" Requirement [R-01.2024]
I. GENERAL PRINCIPLES GOVERNING COMPLIANCE WITH THE "WRITTEN DESCRIPTION" REQUIREMENT FOR APPLICATIONS
A. Original Claims, 3rd para:
Written description issues may also arise if the knowledge and level of skill in the art would not have permitted the ordinary artisan to immediately envisage the claimed product arising from the disclosed process. See, e.g., Fujikawa v. Wattanasin, 93 F.3d 1559, 1571, 39 USPQ2d 1895, 1905 (Fed. Cir. 1996) (a "laundry list" disclosure of every possible moiety does not necessarily constitute a written description of every species in a genus because it would not "reasonably lead" those skilled in the art to any particular species); In re Ruschig, 379 F.2d 990, 995, 154 USPQ 118, 123 (CCPA 1967) ("If n-propylamine had been used in making the compound instead of n-butylamine, the compound of claim 13 would have resulted. Appellants submit to us, as they did to the board, an imaginary specific example patterned on specific example 6 by which the above butyl compound is made so that we can see what a simple change would have resulted in a specific supporting disclosure being present in the present specification. The trouble is that there is no such disclosure, easy though it is to imagine it." (emphasis in original)); Purdue Pharma L.P. v. Faulding Inc., 230 F.3d 1320, 1328, 56 USPQ2d 1481, 1487 (Fed. Cir. 2000) ("[T]he specification does not clearly disclose to the skilled artisan that the inventors ... considered the ratio... to be part of their invention .... There is therefore no force to Purdue’s argument that the written description requirement was satisfied because the disclosure revealed a broad invention from which the [later-filed] claims carved out a patentable portion").
25 “incidents” is the grammatical object of the verb “label”
26 deepfake: a fake, digitally manipulated video or audio file produced by using deep learning, an advanced type of machine learning, and typically featuring a person’s likeness and voice in a situation that did not actually occur. (Dictionary.com)
27 “incidents” is the grammatical object of the verb “label”
28 concept: a general notion or idea; conception, wherein conception is defined: the act of conceiving; the state of being conceived, wherein conceive is defined: to express, as in words. (Dictionary.com)
29 MPEP 2163.02 Standard for Determining Compliance With the Written Description Requirement [R-07.2022], 3rd para:
The subject matter of the claim need not be described literally (i.e., using the same terms or in haec verba) in order for the disclosure to satisfy the description requirement. If a claim (claim 1) is amended to include subject matter, limitations, or terminology (“label”) not present in the application (63/449,182: pages 18,19) as filed (03/01/2023) , involving a departure from, addition to, or deletion from the disclosure of the application (63/449,182: pages 18,19) as filed (03/01/2023), the examiner should conclude that the claimed subject matter is not described in that application. This conclusion will result in the rejection of the claims affected under 35 U.S.C. 112(a) or pre-AIA 35 U.S.C.112, first paragraph - description requirement, or denial of the benefit of the filing date of a previously filed application, as appropriate.
30 In re GPAC Inc., 57 F.3d 1573, 1579 (Fed. Cir. 1995):
--Standard of Review
Whether a reference or a combination of references renders a claimed invention obvious under 35 U.S.C. § 103 is a question of law subject to full and independent review in this court. Gardner v. TEC Sys., Inc., 725 F.2d 1338, 1344, 220 USPQ 777, 782 (Fed. Cir.), cert. denied, 469 U.S. 830, 105 S. Ct. 116, 83 L. Ed. 2d 60 (1984). We review for clear error the underlying factual findings leading to an obviousness conclusion. In re Woodruff, 919 F.2d 1575, 1577, 16 USPQ2d 1934, 1935 (Fed. Cir. 1990). Furthermore, under this standard of review we disturb a factual finding of the Board only if definitely and firmly convinced that the Board has erred. In re Caveney, 761 F.2d 671, 674, 226 USPQ 1, 3 (Fed. Cir. 1985); see United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S. Ct. 525, 541-42, 76 USPQ 430, 443 (1948). These factual findings include: (1) the scope and content of the prior art; (2) the level of ordinary skill in the art at the time of the invention; (3) objective evidence of nonobviousness; and (4) the differences between the prior art and the claimed subject matter. Specialty Composites v. Cabot Corp., 845 F.2d 981, 989, 6 USPQ2d 1601, 1607 (Fed. Cir. 1988) (citing Graham v. John Deere Co., 383 U.S. 1, 17-18, 86 S. Ct. 684, 693-94, 148 USPQ 459, 467 (1966)). In determining the scope and content of the prior art, " [w]hether a reference ... is 'analogous' is a fact question" that we review for clear error. In re Clay, 966 F.2d 656, 658, 23 USPQ2d 1058, 1060 (Fed. Cir. 1992)….
Level of Ordinary Skill in the Art
The person of ordinary skill in the art is a hypothetical person who is presumed to know the relevant prior art. Custom Accessories, Inc. v. Jeffrey-Allan Indus., Inc., 807 F.2d 955, 962, 1 USPQ2d 1196, 1201 (Fed. Cir. 1986). In determining this skill level, the court may consider various factors including "type of problems encountered in the art; prior art solutions to those problems; rapidity with which innovations are made; sophistication of the technology; and educational level of active workers in the field." Id. In a given case, every factor may not be present, and one or more factors may predominate. Id. at 962-63, 1 USPQ2d at 1201.--
31 concept: a general notion or idea; conception, wherein conception is defined: the act of conceiving; the state of being conceived, wherein conceive is defined: to express, as in words. (Dictionary.com)
32 MPEP 2163 Guidelines for the Examination of Patent Applications Under the 35 U.S.C. 112(a) or Pre-AIA 35 U.S.C. 112, first paragraph, "Written Description" Requirement [R-01.2024]
II. METHODOLOGY FOR DETERMINING ADEQUACY OF WRITTEN DESCRIPTION
A. Read and Analyze the Specification for Compliance with 35 U.S.C. 112(a) or Pre-AIA 35
U.S.C. 112, first paragraph
3. Determine Whether There is Sufficient Written Description to Inform a Skilled Artisan That Inventor was in Possession of the Claimed Invention as a Whole at the Time the Application Was Filed
(a) Original claims, 1st para:
Possession may be shown in many ways. For example, possession may be shown by describing an actual reduction to practice of the claimed invention. Possession may also be shown by a clear depiction of the invention in detailed drawings or in structural chemical formulas which permit a person skilled in the art to clearly recognize that inventor had possession of the claimed invention. An adequate written description of the invention may be shown by any description of sufficient, relevant, identifying characteristics so long as a person skilled in the art would recognize that the inventor had possession of the claimed invention. See, e.g., Purdue Pharma L.P. v. Faulding Inc., 230 F.3d 1320, 1323, 56 USPQ2d 1481, 1483 (Fed. Cir. 2000) (the written description "inquiry is a factual one and must be assessed on a case-by-case basis"); see also Pfaff v. Wells Elec., Inc., 55 U.S. at 66, 119 S.Ct. at 311, 48 USPQ2d at 1646 ("The word ‘invention’ must refer to a concept that is complete, rather than merely one that is ‘substantially complete.’ It is true that reduction to practice ordinarily provides the best evidence that an invention is complete. But just because reduction to practice is sufficient evidence of completion, it does not follow that proof of reduction to practice is necessary in every case. Indeed, both the facts of the Telephone Cases and the facts of this case demonstrate that one can prove that an invention is complete and ready for patenting before it has actually been reduced to practice.").
33 label: a word or phrase indicating that what follows belongs in a particular category or classification. (Dictionary.com)
34 concept: a general notion or idea; conception, wherein conception is defined: the act of conceiving; the state of being conceived, wherein conceive is defined: to express, as in words. (Dictionary.com)
label: a word or phrase indicating that what follows belongs in a particular category or classification. (Dictionary.com)
35 MPEP 2163 Guidelines for the Examination of Patent Applications Under the 35 U.S.C. 112(a) or Pre-AIA 35 U.S.C. 112, first paragraph, "Written Description" Requirement [R-01.2024]
II. METHODOLOGY FOR DETERMINING ADEQUACY OF WRITTEN DESCRIPTION
A. Read and Analyze the Specification for Compliance with 35 U.S.C. 112(a) or Pre-AIA 35
U.S.C. 112, first paragraph
3. Determine Whether There is Sufficient Written Description to Inform a Skilled Artisan That Inventor was in Possession of the Claimed Invention as a Whole at the Time the Application Was Filed
(a) Original claims, 1st para:
Possession may be shown in many ways. For example, possession may be shown by describing an actual reduction to practice of the claimed invention. Possession may also be shown by a clear depiction of the invention in detailed drawings or in structural chemical formulas which permit a person skilled in the art to clearly recognize that inventor had possession of the claimed invention. An adequate written description of the invention may be shown by any description of sufficient, relevant, identifying characteristics so long as a person skilled in the art would recognize that the inventor had possession of the claimed invention. See, e.g., Purdue Pharma L.P. v. Faulding Inc., 230 F.3d 1320, 1323, 56 USPQ2d 1481, 1483 (Fed. Cir. 2000) (the written description "inquiry is a factual one and must be assessed on a case-by-case basis"); see also Pfaff v. Wells Elec., Inc., 55 U.S. at 66, 119 S.Ct. at 311, 48 USPQ2d at 1646 ("The word ‘invention’ must refer to a concept that is complete, rather than merely one that is ‘substantially complete.’ It is true that reduction to practice ordinarily provides the best evidence that an invention is complete. But just because reduction to practice is sufficient evidence of completion, it does not follow that proof of reduction to practice is necessary in every case. Indeed, both the facts of the Telephone Cases and the facts of this case demonstrate that one can prove that an invention is complete and ready for patenting before it has actually been reduced to practice.").
36 I agree
37 Which disclosure conveys possession of a training dataset comprising labeled records of both authentic and deepfake content? 63/596,326? (63/596,326 does not have the word “label” or the expression (in words) thereof) 63/449,182? US 2024/0296698 A1?
38 I agree
39 concept: a general notion or idea; conception, wherein conception is defined: the act of conceiving; the state of being conceived, wherein conceive is defined: to express, as in words. (Dictionary.com)
40 MPEP 2163 II. A. 3. (a) Original claims, 1st para:
Possession may be shown in many ways. For example, possession may be shown by describing an actual reduction to practice of the claimed invention. Possession may also be shown by a clear depiction of the invention in detailed drawings or in structural chemical formulas which permit a person skilled in the art to clearly recognize that inventor had possession of the claimed invention. An adequate written description of the invention may be shown by any description of sufficient, relevant, identifying characteristics so long as a person skilled in the art would recognize that the inventor had possession of the claimed invention. See, e.g., Purdue Pharma L.P. v. Faulding Inc., 230 F.3d 1320, 1323, 56 USPQ2d 1481, 1483 (Fed. Cir. 2000) (the written description "inquiry is a factual one and must be assessed on a case-by-case basis"); see also Pfaff v. Wells Elec., Inc., 55 U.S. at 66, 119 S.Ct. at 311, 48 USPQ2d at 1646 ("The word ‘invention’ must refer to a concept that is complete, rather than merely one that is ‘substantially complete.’ It is true that reduction to practice ordinarily provides the best evidence that an invention is complete. But just because reduction to practice is sufficient evidence of completion, it does not follow that proof of reduction to practice is necessary in every case. Indeed, both the facts of the Telephone Cases and the facts of this case demonstrate that one can prove that an invention is complete and ready for patenting before it has actually been reduced to practice.").
41 label: a word or phrase indicating that what follows belongs in a particular category or classification. (Dictionary.com)
42 MPEP 2163 II. A. 3. (a) Original claims, 1st para:
Possession may be shown in many ways. For example, possession may be shown by describing an actual reduction to practice of the claimed invention. Possession may also be shown by a clear depiction of the invention in detailed drawings or in structural chemical formulas which permit a person skilled in the art to clearly recognize that inventor had possession of the claimed invention. An adequate written description of the invention may be shown by any description of sufficient, relevant, identifying characteristics so long as a person skilled in the art would recognize that the inventor had possession of the claimed invention. See, e.g., Purdue Pharma L.P. v. Faulding Inc., 230 F.3d 1320, 1323, 56 USPQ2d 1481, 1483 (Fed. Cir. 2000) (the written description "inquiry is a factual one and must be assessed on a case-by-case basis"); see also Pfaff v. Wells Elec., Inc., 55 U.S. at 66, 119 S.Ct. at 311, 48 USPQ2d at 1646 ("The word ‘invention’ must refer to a concept that is complete, rather than merely one that is ‘substantially complete.’ It is true that reduction to practice ordinarily provides the best evidence that an invention is complete. But just because reduction to practice is sufficient evidence of completion, it does not follow that proof of reduction to practice is necessary in every case. Indeed, both the facts of the Telephone Cases and the facts of this case demonstrate that one can prove that an invention is complete and ready for patenting before it has actually been reduced to practice.").
43 MPEP 2106.05(a) Improvements to the Functioning of a Computer or To Any Other Technology or Technical Field [R-07.2022], 4th para:
After the examiner has consulted the specification and determined that the disclosed invention improves technology, the claim must be evaluated to ensure the claim itself reflects the disclosed improvement in technology. Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1316, 120 USPQ2d 1353, 1359 (Fed. Cir. 2016) (patent owner argued that the claimed email filtering system improved technology by shrinking the protection gap and mooting the volume problem, but the court disagreed because the claims themselves did not have any limitations that addressed these issues). That is, the claim must include the components or steps of the invention that provide the improvement described in the specification. However, the claim itself does not need to explicitly recite the improvement described in the specification (e.g., "thereby increasing the bandwidth of the channel"). The full scope of the claim under the BRI should be considered to determine if the claim reflects an improvement in technology (e.g., the improvement described in the specification). In making this determination, it is critical that examiners look at the claim "as a whole," in other words, the claim should be evaluated "as an ordered combination, without ignoring the requirements of the individual steps." When performing this evaluation, examiners should be "careful to avoid oversimplifying the claims" by looking at them generally and failing to account for the specific requirements of the claims. McRO, 837 F.3d at 1313, 120 USPQ2d at 1100.
44 Applicant’s disclosure, US 20240296698 A1: [0226] Although the invention has been described in conjunction with specific embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art. Accordingly, it is intended to embrace all such alternatives, modifications and variations that fall within the spirit and broad scope of the appended claims.
45 scope: Linguistics, Logic. the range of words or elements of an expression (claim 1) over which a modifier (a patent examiner) or operator (or me) has control. (Dictionary.com)
46 outcome: a final product or end result; consequence; issue, wherein product is defined: a person or thing produced by or resulting from a process, as a natural, social, or historical one; result. The alternative “result” is “taken” (MPEP 2111.01 III, 4th para, 1st S) as the meaning of “outcome” under BRI. Thus the alternatives “product…produced by…a process” is not “taken” under BRI and in view said applicant’s disclosure US 20240296698 A1 [0226] regarding “all such alternatives…fall within the spirit and broad scope of the appended claims” (Dictionary.com)
47 endeavor: a strenuous effort; attempt. (Dictionary.com)
48 how: about the manner, condition, or way in which (Dictionary.com)
49 MPEP 2143.03 All Claim Limitations Must Be Considered [R-01.2024], 3rd para:
As a general matter, the grammar and ordinary meaning of terms as understood by one having ordinary skill in the art used in a claim will dictate whether, and to what extent, the language limits the claim scope. Language that suggests or makes a feature or step optional but does not require that feature or step does not limit the scope of a claim under the broadest reasonable claim interpretation. In addition, when a claim requires selection of an element from a list of alternatives, the prior art teaches the element if one of the alternatives is taught by the prior art. See, e.g., Fresenius USA, Inc. v. Baxter Int’l, Inc., 582 F.3d 1288, 1298, 92 USPQ2d 1163, 1171 (Fed. Cir. 2009).
50 MPEP
2144 Supporting a Rejection Under 35 U.S.C. 103 [R-01.2024]
2144.01 Implicit Disclosure [R-10.2019]
2144.02 Reliance on Scientific Theory [R-08.2012]
2144.03 Reliance on Common Knowledge in the Art or "Well Known" Prior Art [R-01.2024]
PROCEDURE FOR RELYING ON COMMON KNOWLEDGE OR TAKING OFFICIAL NOTICE
B. If Official Notice Is Taken of a Fact, Unsupported by Documentary Evidence, the Technical Line of Reasoning Underlying a Decision To Take Such Notice Must Be Clear and Unmistakable.
In certain older cases, official notice has been taken of a fact that is asserted to be "common knowledge" without specific reliance on documentary evidence where the fact noticed was readily verifiable, such as when other references of record supported the noticed fact, or where there was nothing of record to contradict it. See In re Soli, 317 F.2d 941, 945-46, 137 USPQ 797, 800 (CCPA 1963) (accepting the examiner’s assertion that the use of "a control is standard procedure throughout the entire field of bacteriology" because it was readily verifiable and disclosed in references of record not cited by the Office); In re Chevenard, 139 F.2d 711, 713, 60 USPQ 239, 241 (CCPA 1943) (accepting the examiner’s finding that a brief heating at a higher temperature was the equivalent of a longer heating at a lower temperature where there was nothing in the record to indicate the contrary and where the applicant never demanded that the examiner produce evidence to support his statement). If such notice is taken, the basis for such reasoning must be set forth explicitly. The examiner must provide specific factual findings predicated on sound technical and scientific reasoning to support the conclusion of common knowledge. See Soli, 317 F.2d at 946, 37 USPQ at 801; Chevenard, 139 F.2d at 713, 60 USPQ at 241. The applicant should be presented with the explicit basis on which the examiner regards the matter as subject to official notice so as to allow the applicant an opportunity to adequately traverse the rejection in the next reply after the Office action in which the "common knowledge" statement was made.
51 why: for what cause or reason, wherein what is defined: (used interrogatively to inquire the reason or purpose of something, usually used in elliptical constructions), wherein elliptical is defined: (of a style of speaking or writing) tending to be ambiguous (I interpreted bullet “2.” as a request for a reason to be motivated; however, the MPEP requires “the PTO to identify record evidence of a teaching, suggestion, or motivation to combine references” [MPEP 2143 I. E.: Example 9, 5th para, 5th S], which I have done in the respective rejections), cryptic, or obscure. (Dictionary.com)
52 MPEP 2141.01(a) Analogous and Nonanalogous Art [R-01.2024]
53 as well as IDIOMS: In addition to, as in The editors as well as the proofreaders are working overtime . [c. 1700]” Dictionary.com
54 MPEP 2144 Supporting a Rejection Under 35 U.S.C. 103 [R-01.2024]
IV. RATIONALE DIFFERENT FROM APPLICANT’S IS PERMISSIBLE
The reason or motivation to modify the reference may often suggest what the inventor has done, but for a different purpose or to solve a different problem. It is not necessary that the prior art suggest the combination to achieve the same advantage or result discovered by applicant. See, e.g., In re Kahn, 441 F.3d 977, 987, 78 USPQ2d 1329, 1336 (Fed. Cir. 2006) (motivation question arises in the context of the general problem confronting the inventor rather than the specific problem solved by the invention); Cross Med. Prods., Inc. v. Medtronic Sofamor Danek, Inc., 424 F.3d 1293, 1323, 76 USPQ2d 1662, 1685 (Fed. Cir. 2005) ("One of ordinary skill in the art need not see the identical problem addressed in a prior art reference to be motivated to apply its teachings."); In re Lintner, 458 F.2d 1013, 173 USPQ 560 (CCPA 1972) (discussed below); In re Dillon, 919 F.2d 688, 16 USPQ2d 1897 (Fed. Cir. 1990), cert. denied, 500 U.S. 904 (1991) (discussed below).
55 motivate: to provide with a motive, or a cause or reason to act; incite; impel, where motive is defined: the goal or object of a person's actions.(Dictionary.com)
56 MPEP 2141 Examination Guidelines for Determining Obviousness Under 35 U.S.C. 103 [R-01.2024]
EXAMINATION GUIDELINES FOR DETERMINING OBVIOUSNESS UNDER 35 U.S.C. 103
III. RATIONALES TO SUPPORT REJECTIONS UNDER 35 U.S.C. 103, 5th para:
The key to supporting any rejection under 35 U.S.C. 103 is the clear articulation of the reason(s) why the claimed invention would have been obvious. The Supreme Court in KSR noted that the analysis supporting a rejection under 35 U.S.C. 103 should be made explicit. The Court quoting In re Kahn, 441 F.3d 977, 988, 78 USPQ2d 1329, 1336 (Fed. Cir. 2006), stated that "‘[R]ejections on obviousness cannot be sustained by mere conclusory statements; instead, there must be some articulated reasoning with some rational underpinning to support the legal conclusion of obviousness.’" KSR, 550 U.S. at 418, 82 USPQ2d at 1396. See also Adapt Pharma Operations Ltd. v. Teva Pharms. USA, Inc., 25 F.4th 1354, 1365, 2022 USPQ2d 144 (Fed. Cir. 2022) (stating that a determination of obviousness "requires ‘identify[ing] a reason that would have prompted a person of ordinary skill in the relevant field to combine the elements in the way the claimed new invention does’" (quoting KSR, 550 U.S. at 418, 82 USPQ2d at 1395). Examples of rationales that may support a conclusion of obviousness include:…
57 why: for what cause or reason, wherein what is defined: (used interrogatively to inquire the reason or purpose of something, usually used in elliptical constructions), wherein elliptical is defined: (of a style of speaking or writing) tending to be ambiguous (I interpreted bullet “2.” as a request for a reason to be motivated; however, the MPEP requires “the PTO to identify record evidence of a teaching, suggestion, or motivation to combine references” [MPEP 2143 I. E.: Example 9, 5th para, 5th S], which I have done in the respective rejections), cryptic, or obscure. (Dictionary.com)
58 why: for what cause or reason, wherein what is defined: (used interrogatively as a request for specific information) (Dictionary.com)
59 as well as IDIOMS In addition to, as in The editors as well as the proofreaders are working overtime . [c. 1700] (Dictionary.com)
60 MPEP 2144 Supporting a Rejection Under 35 U.S.C. 103 [R-01.2024]
IV. RATIONALE DIFFERENT FROM APPLICANT’S IS PERMISSIBLE
The reason or motivation to modify the reference may often suggest what the inventor has done, but for a different purpose or to solve a different problem. It is not necessary that the prior art suggest the combination to achieve the same advantage or result discovered by applicant. See, e.g., In re Kahn, 441 F.3d 977, 987, 78 USPQ2d 1329, 1336 (Fed. Cir. 2006) (motivation question arises in the context of the general problem confronting the inventor rather than the specific problem solved by the invention); Cross Med. Prods., Inc. v. Medtronic Sofamor Danek, Inc., 424 F.3d 1293, 1323, 76 USPQ2d 1662, 1685 (Fed. Cir. 2005) ("One of ordinary skill in the art need not see the identical problem addressed in a prior art reference to be motivated to apply its teachings."); In re Lintner, 458 F.2d 1013, 173 USPQ 560 (CCPA 1972) (discussed below); In re Dillon, 919 F.2d 688, 16 USPQ2d 1897 (Fed. Cir. 1990), cert. denied, 500 U.S. 904 (1991) (discussed below).
61 MPEP 2143 Examples of Basic Requirements of a Prima Facie Case of Obviousness [R-01.2024]
I. EXAMPLES OF RATIONALES
G. Some Teaching, Suggestion, or Motivation in the Prior Art That Would Have Led One of Ordinary Skill To Modify the Prior Art Reference or To Combine Prior Art Reference Teachings To Arrive at the Claimed Invention
The rationale to support a conclusion that the claim would have been obvious is that "a person of ordinary skill in the art would have been motivated to combine the prior art to achieve the claimed invention and whether there would have been a reasonable expectation of success in doing so." DyStar Textilfarben GmbH & Co. Deutschland KG v. C.H. Patrick Co., 464 F.3d 1356, 1360, 80 USPQ2d 1641, 1645 (Fed. Cir. 2006). If any of these findings cannot be made, then this rationale cannot be used to support a conclusion that the claim would have been obvious to one of ordinary skill in the art.
62 motivate: to provide with a motive, or a cause or reason to act; incite; impel, where motive is defined: the goal or object of a person's actions.(Dictionary.com)
63 to: (used for expressing aim, purpose, or intention), wherein purpose is defined: an intended or desired result; end; aim; goal. (Dictionary.com) .
64 MPEP 2141 Examination Guidelines for Determining Obviousness Under 35 U.S.C. 103 [R-01.2024]
EXAMINATION GUIDELINES FOR DETERMINING OBVIOUSNESS UNDER 35 U.S.C. 103
III. RATIONALES TO SUPPORT REJECTIONS UNDER 35 U.S.C. 103, 5th para:
The key to supporting any rejection under 35 U.S.C. 103 is the clear articulation of the reason(s) why the claimed invention would have been obvious. The Supreme Court in KSR noted that the analysis supporting a rejection under 35 U.S.C. 103 should be made explicit. The Court quoting In re Kahn, 441 F.3d 977, 988, 78 USPQ2d 1329, 1336 (Fed. Cir. 2006), stated that "‘[R]ejections on obviousness cannot be sustained by mere conclusory statements; instead, there must be some articulated reasoning with some rational underpinning to support the legal conclusion of obviousness.’" KSR, 550 U.S. at 418, 82 USPQ2d at 1396. See also Adapt Pharma Operations Ltd. v. Teva Pharms. USA, Inc., 25 F.4th 1354, 1365, 2022 USPQ2d 144 (Fed. Cir. 2022) (stating that a determination of obviousness "requires ‘identify[ing] a reason that would have prompted a person of ordinary skill in the relevant field to combine the elements in the way the claimed new invention does’" (quoting KSR, 550 U.S. at 418, 82 USPQ2d at 1395). Examples of rationales that may support a conclusion of obviousness include:…
65 how: in what way or manner; by what means?. (Dictionary.com)
66 actually: as an actual or existing fact; really. (Dictionary.com)
67 MPEP 2143 Examples of Basic Requirements of a Prima Facie Case of Obviousness [R-01.2024]
MPEP 2143.01 Suggestion or Motivation To Modify the References [R-01.2024], 2nd para:
Obviousness can be established by combining or modifying the teachings of the prior art to produce the claimed invention where there is some teaching, suggestion, or motivation to do so. In re Kahn, 441 F.3d 977, 986, 78 USPQ2d 1329, 1335 (Fed. Cir. 2006) (discussing rationale underlying the motivation-suggestion-teaching test as a guard against using hindsight in an obviousness analysis). Axonics, Inc. v. Medtronic, Inc., 73 F.4th 950, 957-58, 2023 USPQ2d 795 (Fed. Cir. 2023) (the court found an erroneous framing of the motivation inquiry led to an incorrect conclusion of nonobviousness). A "motivation to combine may be found explicitly or implicitly in market forces; design incentives; the ‘interrelated teachings of multiple patents’; ‘any need or problem known in the field of endeavor at the time of invention and addressed by the patent’; and the background knowledge, creativity, and common sense of the person of ordinary skill." Zup v. Nash Mfg., 896 F.3d 1365, 1371, 127 USPQ2d 1423, 1427 (Fed. Cir. 2018) (quoting Plantronics, Inc. v. Aliph, Inc., 724 F.3d 1343, 1354 [107 USPQ2d 1706] (Fed. Cir. 2013) (citing Perfect Web Techs., Inc. v. InfoUSA, Inc., 587 F.3d 1324, 1328 [92 USPQ2d 1849] (Fed. Cir. 2009) (quoting KSR, 550 U.S. at 418-21)). See MPEP § 2143 regarding the need to provide a reasoned explanation even in situations involving common sense or ordinary ingenuity. See also MPEP § 2144.05, subsection II, B.
68 MPEP Examination Guidelines for Determining Obviousness Under 35 U.S.C. 103 [R-01.2024]
EXAMINATION GUIDELINES FOR DETERMINING OBVIOUSNESS UNDER 35 U.S.C. 103
II. THE BASIC FACTUAL INQUIRIES OF GRAHAM v. JOHN DEERE CO.
Office Personnel As Factfinders, 1st para:
Office personnel fulfill the critical role of factfinder when resolving the Graham inquiries. It must be remembered that while the ultimate determination of obviousness is a legal conclusion, the underlying Graham inquiries are factual. When making an obviousness rejection, Office personnel must therefore ensure that the written record includes findings of fact concerning the state of the art and the teachings of the references applied. In certain circumstances, it may also be important to include explicit findings as to how a person of ordinary skill would have understood prior art teachings, or what a person of ordinary skill would have known or could have done. Factual findings made by Office personnel are the necessary underpinnings to establish obviousness.
69 The crossed-out text “is taught” and “does not limit the scope of” claim 1 “under the broadest reasonable claim interpretation” via MPEP 2143.03, 3rd para (reproduced below).
70 MPEP 2143.03 All Claim Limitations Must Be Considered [R-01.2024], 3rd para:
As a general matter, the grammar (such as coordinate-adjectives: see Nordquist: Coordinate Adjectives: Definition and Examples, wherein coordinate is defined: Grammar. of the same rank in grammatical construction, as Jack and Jill in the phrase Jack and Jill, or got up and shook hands in the sentence He got up and shook hands. (Dictionary.com)) and ordinary meaning of terms as understood by one having ordinary skill in the art used in a claim will dictate whether, and to what extent, the language limits the claim scope. Language (“a customized detector machine learning (ML) model”) that suggests or makes a feature or step optional (the “detector machine learning” are coordinate-adjectives meaning that each adjective “detector” & “machine learning” coordinates with the other equally and individually to modify “model”: --detector model—or “machine learning model” or –machine learning and detector model (went up the hill, To fetch a pail of water)—or detector and machine learning model (went up the hill, To fetch a pail of water), wherein “and” is defined: (used to connect alternatives). (Dictionary.com)) but does not require that feature or step does not limit the scope of a claim under the broadest reasonable claim interpretation. In addition, when a claim requires selection of an element from a list of alternatives (--detector model—or “machine learning model”), the prior art teaches the element if one of the alternatives is taught by the prior art. See, e.g., Fresenius USA, Inc. v. Baxter Int’l, Inc., 582 F.3d 1288, 1298, 92 USPQ2d 1163, 1171 (Fed. Cir. 2009).
71file: Computers., a collection of related data or program records stored on some input/output or auxiliary storage medium. (Dictionary.com)
72 original: the first and genuine form of something, from which others are derived, wherein genuine is defined: not fake or counterfeit; original; real; authentic (Dictionary.com)
73 since: because; inasmuch as, wherein because is defined: for the reason that; due to the fact that. (Dictionary.com)
74 update: Computers., to incorporate new or more accurate information in (a database, program, procedure, etc.). (Dictionary.com)
75 customize: to modify or build according to individual or personal specifications or preference. (Dictionary.com)
76 (Italics) represent other claim limitations already mapped above
77 Ellipsis (…) represent claim other limitations already mapped above
78 since: because; inasmuch as, wherein because is defined: for the reason that; due to the fact that. (Dictionary.com)
79 dealing: to be able to handle competently or successfully; cope (followed by with ). (Dictionary.com)
80 Markush element of Markush alternatives follows: [A and (B or C)]
81 (“is” (are, be) essentially means look at a figure: Dictionary.com)
82 and: (used to connect [Markush] alternatives). (Dictionary.com)
83 (“is” (are, be) essentially means look at a figure: Dictionary.com)
84 (“is” (are, be) essentially means look at a figure: Dictionary.com)
85 Markush element follows: [A,B,C and D]
86 and: (used to connect [Markush] alternatives). (Dictionary.com)
87 Since Markush alternative A is taught, the Markush element [A,B,C and D] is taught; hence, Markush alternatives B,C and D are taught under the broadest reasonable interpretation of claim 4.
88 Markush elements (I & II) follow: I:[A and/or B] & II:[(1) or (2)]
89 -ing of (“tuning”)” a suffix (“ing”) of nouns (“tuning”) formed from verbs (“tune”), expressing
(1) the action of the verb (“tune”) or
(2) its (“tuning”) result, product, material, etc. (the art of building; a new building; cotton wadding ), wherein expressing is defined: to put (thought) into words (i.e., claim 6); utter or state. (Dictionary.com)
90 Since Markush element A is taught, the Markush element [A and/or B] is taught
91 on: in connection, association, or cooperation with; as a part or element of. (Dictionary.com)
92 original: the first and genuine form of something, from which others are derived, wherein genuine is defined: not fake or counterfeit; original; real; authentic (Dictionary.com)
93 original: the first and genuine form of something, from which others are derived, wherein genuine is defined: not fake or counterfeit; original; real; authentic (Dictionary.com)
94 original: the first and genuine form of something, from which others are derived, wherein genuine is defined: not fake or counterfeit; original; real; authentic (Dictionary.com)
95 The crossed-out text “is taught” and “does not limit the scope of” claim 1 “under the broadest reasonable claim interpretation” via MPEP 2143.03, 3rd para (reproduced in the rejection of claim 1).
96 Markush element of Markush alternatives follows: [A and (B or C)]
97 (“is” (are, be) essentially means look at a figure: Dictionary.com)
98 and: (used to connect [Markush] alternatives). (Dictionary.com)
99 (“is” (are, be) essentially means look at a figure: Dictionary.com)
100 (“is” (are, be) essentially means look at a figure: Dictionary.com)
101 Markush element follows: [A,B,C and D]
102 and: (used to connect [Markush] alternatives). (Dictionary.com)
103 Since Markush alternative A is taught, the Markush element [A,B,C and D] is taught; hence, Markush alternatives B,C and D are taught under the broadest reasonable interpretation of claim 4.
104 since: because; inasmuch as, wherein because is defined: for the reason that; due to the fact that. (Dictionary.com)
105 threshold: a level or point at which something would happen, would cease to happen, or would take effect, become true, etc, wherein true is defined: not false, fictional, or illusory; factual or factually accurate; conforming with reality (Dictionary.com)
106 since: because; inasmuch as, wherein because is defined: for the reason that; due to the fact that. (Dictionary.com)
107 since: because; inasmuch as, wherein because is defined: for the reason that; due to the fact that. (Dictionary.com)
108 since: because; inasmuch as, wherein because is defined: for the reason that; due to the fact that. (Dictionary.com)
109 MPEP 2144.08 Obviousness of Species When Prior Art Teaches Genus [R-01.2024]
II. DETERMINE WHETHER THE CLAIMED SPECIES OR SUBGENUS WOULD HAVE BEEN OBVIOUS TO ONE OF ORDINARY SKILL IN THE PERTINENT ART AT THE RELEVANT TIME
4. Determine Whether One of Ordinary Skill in the Art Would Have Had a Reason To Select the Claimed Species or Subgenus
(b) Consider the Express Teachings
If the prior art reference expressly teaches a particular reason to select the claimed species or subgenus, Office personnel should point out the express disclosure and explain why it would have been obvious to one of ordinary skill in the art to select the claimed invention. An express teaching may be based on a statement in the prior art reference such as an art recognized equivalence. For example, see Merck & Co. v. Biocraft Labs., 874 F.2d 804, 807, 10 USPQ2d 1843, 1846 (Fed. Cir. 1989) (holding claims directed to diuretic compositions comprising a specific mixture of amiloride and hydrochlorothiazide were obvious over a prior art reference expressly teaching that amiloride was a pyrazinoylguanidine which could be coadministered with potassium excreting diuretic agents, including hydrochlorothiazide which was a named example, to produce a diuretic with desirable sodium and potassium eliminating properties). See also, In re Kemps, 97 F.3d 1427, 1430, 40 USPQ2d 1309, 1312 (Fed. Cir. 1996) (holding it would have been obvious to combine teachings of prior art to achieve claimed invention where one reference specifically refers to the other).
110 since: because; inasmuch as, wherein because is defined: for the reason that; due to the fact that. (Dictionary.com)
111 update: Computers., to incorporate new or more accurate information in (a database, program, procedure, etc.). (Dictionary.com)
112 customize: to modify or build according to individual or personal specifications or preference. (Dictionary.com)
113 (italics) represent claim limitations already taught by the prior art
114 ellipsis (…) represents claim limitations already taught by the prior art
115 learn:(intr; often foll by of or about) to become informed; know, wherein become is defined: (copula) to come to be; develop or grow into, wherein develop is defined: to come or bring into existence; generate or be generated (Dictionary.com)
116 (italics) represent claim limitations already taught by the prior art
117 ellipsis (…) represents claim limitations already taught by the prior art
118 since: because; inasmuch as, wherein because is defined: for the reason that; due to the fact that. (Dictionary.com)
119 deep learning: Computers. an advanced type of machine learning that uses multilayered neural networks to establish nested hierarchical models for data processing and analysis, as in image recognition or natural language processing, with the goal of self-directed information processing. (Dictionary.com)
120 participle
121 combined: taken as a whole or considered together; in the aggregate. (Dictionary.com)
122 deepfake: “8.1 Definition A combination of "Deep Learning" and "Fake" can be called Deepfake that refers to any photo-realistic audiovisual content formed with the help of DL.”, Rana, pg. 73, wherein formed is defined: to be formed or produced, wherein produced is defined: to be made, as specified, wherein made is defined: artificially produced, wherein produced is defined: to create, bring forth, or yield offspring, products, etc.. (Dictionary.com)
123 deepfake: a fake, digitally manipulated video or audio file produced by using deep learning, an advanced type of machine learning, and typically featuring a person’s likeness and voice in a situation that did not actually occur, wherein produced is defined: to be made, as specified, wherein made is defined: artificially produced, wherein produced is defined: to create, bring forth, or yield offspring, products, etc.. (Dictionary.com)
124 since: because; inasmuch as, wherein because is defined: for the reason that; due to the fact that. (Dictionary.com)
125 update: Computers., to incorporate new or more accurate information in (a database, program, procedure, etc.). (Dictionary.com)
126 customize: to modify or build according to individual or personal specifications or preference. (Dictionary.com)
127 (Italics) represent claim limitations already taught above
128 Ellipses (…) represent claim limitations already taught above
129 plurality: state or fact of being plural wherein plural is defined: being one of such a plurality. (Dictionary.com)
130 (Italics) represent claim limitations already taught above
131 Ellipses (…) represent claim limitations already taught above
132 by: according to; in conformity with. (Dictionary.com)
133 “selecting” comprises an implicit Markush element—[(1) or (2)]—as detailed in the next “-ing” footnote
134 -ing (of “selecting”): a suffix (-ing) of nouns (“selecting”) formed from verbs (“select”), expressing (1) the action of the verb or (2) its result, product, material, etc. (the art of building; a new building; cotton wadding ), wherein express is defined: to put (thought) into words (i.e., claim 11); utter or state. (Dictionary.com)
135 “according” comprises an implicit Markush element—[(A) or (B)]—as detailed in the next “-ing” footnote
136 -ing (of “according”): a suffix of nouns formed from verbs (accord), expressing (A) the action of the verb (“accord”) or (B) its result (“training the selected machine learning model”: trained model), product, material, etc. (the art of building; a new building; cotton wadding ), wherein express is defined: to put (thought) into words; utter or state. (Dictionary.com)
137 “according” is a participle participating in the action of the (gerund) noun “selecting”
138 by: according to; in conformity with. (Dictionary.com)
139 annotation(s): the act(s) of annotating (Dictionary.com)
140 since: because; inasmuch as, wherein because is defined: for the reason that; due to the fact that. (Dictionary.com)
141 (italics) represent claim limitations already taught above
142 Ellipses (…) represent claim limitations already mapped above
143 (italics) represent claim limitations already taught above
144 Ellipses (…) represent claim limitations already mapped above
145 “according” is a participle contributing to the action of “(determining a probability…)”.
146 gather: to bring together into one group, collection, or place, wherein group is defined: any collection or assemblage of persons or things; cluster; aggregation. (Dictionary.com)
147 probability: Statistics. the relative possibility that an event will occur, as expressed by the ratio of the number of actual occurrences to the total number of possible occurrences, wherein total is defined: constituting or comprising the whole; entire; whole, wherein whole is defined: comprising the full quantity, amount, extent, number, etc., without diminution or exception; entire, full, or total, wherein total is defined: the total amount; sum; aggregate (Dictionary.com)
148 “according” is a participle contributing to the action of “(determining a probability…)”.
149 on: by the agency or means of, wherein by is defined: according to; in conformity with. (Dictionary.com)
150 since: because; inasmuch as, wherein because is defined: for the reason that; due to the fact that. (Dictionary.com)
Prosecution Insights