Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA . In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
DETAILED ACTION
This Final Office Action is in response Applicant communication filled on 01/22/2026.
Status of Claims
Claims 1,11,20 have been amended and Claims 8,18 newly canceled by Applicant.
Claims 28-29 have been newly added by Applicant.
Claims 1-3,6-7,11-13,16-17 and 20-29 are currently pending and rejected as follows.
Response to Amendments Arguments
Applicant’s 01/22/2026 amendment necessitated new grounds of rejection in this action.
Response to Applicant rebuttal arguments on 35 USC 112(a) rejection
35 USC 112(a) rejection in the previous act has been withdrawn in view of Applicant deleting the term “a feedback adjuster” as confirmed by Remarks 01/22/2026 p.13 ¶5-p.14 ¶3.
Response to Applicant rebuttal arguments on 35 USC 101 rejection
Examiner reincorporates all Examiner’s findings and rationales at Non-Final Act 10/22/2025.
Remarks 01/22/2026 ¶4 argues against using the preponderance of evidence test in analyzing whether or not a claimed invention is patent eligible. Examiner fully considered the argument but respectfully disagrees by pointing to USPTO’s Memorandum on evaluating subject matter eligibility of claims under 35 U.S.C. 101, dated August 4th, 2025, p.5 Section III, ¶1, 3rd sentence: “In order to make a rejection of a claim under any of the statutory bases (i.e., 35 U.S.C. 101, 102, 103, 112), unpatentability must be established by a preponderance of the evidence”.
Step 2A prong one: Remarks 01/22/2026 p.15 ¶4-p.16 ¶6 points to USPTO’s hypothetical Example 39, to argue that similar how the USPTO’s hypothetical Example 39 more robustly detected faces, the current claims are also directed to a more robust detecting of leads and characterize of lead data through inclusion of training and retraining of neural networks that are argued as not practically performed by humans, and distinguishing from features that involve mathematical concepts. In addition, it is argued that newly added dependent Claims 28,29 include training data set processing by, “analyzing a sample of leads data for different leads in a database and producing a buying pattern metrics for the different leads data” in a manner similar to how USPTO’s hypothetical Example 39 included processing or training set of for facial image analysis.
Examiner fully considered the Step 2A prong 1 argument but respectfully disagrees finding it unpersuasive because all USPTO’s 101 examples are hypothetical and non-precedential. See USPTO “2019 PEG, 101 Examples 37-42 document entitled “Subject Matter Eligibility Examples: Abstract Ideas” p.1 ¶1 2nd sentence. “The examples below are hypothetical and only intended to be illustrative of the claim analysis under the 2019 PEG” corroborating “May 2016 Update: Memorandum- Formulating a Subject Matter Eligibility Rejection and Evaluating the Applicant’s Response to a Subject Matter Eligibility Rejection”, p.5 ¶2 Section C: “USPTO issued examples in conjunction with the Interim Eligibility Guidance, including […] July 2015 Update Appendix I: Examples […]; These examples, many of which are hypothetical, were drafted to show exemplary analyses under the Interim Eligibility Guidance and are intended to be illustrative of the analysis only. While some of the fact patterns draw from U.S. Supreme Court and U.S. Court of Appeals for the Federal Circuit decisions, the examples do not carry the weight of court decisions. Therefore, the examples should not be used as a basis for a subject matter eligibility rejection.
Also, aside from the fact that the USPTO’s Example 39 is hypothetical and non-precedential, the Examiner also submits that the training and retraining of neural networks to allegedly detect leads and characterize business or lead data more robustly, as argued by Remarks 01/22/2026 p.16 ¶6, with respect to the currently amended independent Claims 1,11,20, and the “analyzing a sample of leads data for different leads in a database and producing a buying pattern metrics for the different leads data”, as newly argued by Remarks 01/22/2026 ¶8, with respect to newly dependent Claims 28,29, do not preclude said claims from reciting, describing or setting forth the fundamental sale economic practices or principles [MPEP 2106.04(a)(2) IIA] of the abstract grouping of Certain Methods of Organizing Human Activities. Clearly here, the detection and characterization of sales lead data, as alleged by Applicant above bears no resemblance and is irreconcilably different than the technological details of machine learning on facial and non-facial images provided by the hypothetical and nonprecedential USPTO’s Example 39 in “creating a first training set comprising the collected set of digital facial images, the modified set of digital facial images, and a set of digital non-facial images; training the neural network in a first stage using the first training set; creating a second training set for a second stage of training comprising the first training set and digital non-facial images that are incorrectly detected as facial images after the first stage of training; and training the neural network in a second stage using the second training set. At most here, the asserted robustness in detecting and characterizes the sales lead data (independent Claims 1,11,20) and the produc[ed] buying patterns (dependent Claims 28,29) as argued by Remarks 01/22/2026 p.16 ¶6, ¶8, are entrepreneurial and abstract rather than technological and eligible. While the Applicant’s solution for achieving such sales leads data does propose use of trained and retrained of machine learning or neural networks, the Examiner finds that such solution still solves an abstract problem in sales lead data, not a technological problem and certainly not a technological solution to a technological problem. That is; no matter in which manner the claimed data is being trained, retrained or manipulated, it still undisputedly achieves an economic, entrepreneurial or abstract goal or objective in determining and characterizing the business or sales lead data. This finding is especially important because MPEP 2106.04(a)(2) III C #2 cited, among others, FairWarning IP, LLC v. Iatric Sys., Inc., 839 F.3d 1089, 120 USPQ2d 1293 (Fed. Cir. 2016), to stress that performing an abstract process in a computer environment, does not precluded its underlining claims from reciting the abstract idea. Here, such abstract process is set forth by determining and characterizing business or sales lead data, and associated scor[ing], while the computer environment being represented by the trained and retrained mahine learning or neural networks, as raised by Remarks 01/22/2026 p.16 ¶6, ¶8. Equally important, in FairWarning IP, LLC v. Iatric Sys., Inc., 839 F.3d 1089, 120 USPQ2d 1293 (Fed. Cir. 2016) as cited by MPEP 2106.04(a)(2) III C #2 above, the Federal Circuit ruled that even the inability for the human mind to perform each claim step does not alone confer patentability. Specifically in FairWarning supra, the Court found that accessing, compiling and combining data from disparate information sources that made it possible to generate a full picture of a user's activity, identity, frequency of activity, and the like in a computer environment, represented a concept of merely selecting information, by content or source, for collection, analysis, and announcement which did not differentiate from mental processes, whose implicit exclusion from 101 undergirds the information-based category of abstract ideas. While the Examiner does not necessarily concede that a human would not be able to train a neural network by hand1, the Examiner nevertheless asserts that use of such trained and retrained neural networks as aids such as computer tools [MPEP 2106.04(a)(2) III C #3] and/or computer environment [MPEP 2106.04(a)(2) III C #2] to achieve the argued sales lead data, does not preclude the claims from reciting, describing or setting forth the abstract exception. Examiner further justifies such rationale by pointing to SAP Am, Inc v InvestPic as cited by MPEP 2106.04(a)(2) I. C (i). Specifically, similar to how Applicant argues in favor of a more robust detection and characterization of business or sales lead data (Remarks 01/22/2026 p.16 ¶6), the ‘291 patent of SAP supra, described an analogous need for improving upon existing practices that performed rudimentary statistical functions not useful to investors in forecasting the behavior of financial markets because they relied upon assumptions that the probability distribution function (‘PDF’) for the financial data followed a normal or Gaussian distribution.” (’291 patent, col. 1, lines 24–36). Yet, it was found that “the PDF for financial market data is heavy tailed (i.e., the histograms of financial market data typically involve many outliers containing important information),” rather than symmetric like a normal distribution. Id., col. 1, lines 36– 37, 41–44. To remedy those deficiencies, the patent in “SAP” proposed utilization of resampled statistical methods for analysis of financial data, which did not assume a normal probability distribution. One such method is a bootstrap method, which estimates distribution of data in a pool (a sample space) by repeated sampling of the data in the pool. A sample space in a boot-strap method can be defined by selecting a specific investment or a particular period of time. Data samples are drawn from the sample space with replacement: samples are drawn from the sample space and then returned to the pool before next sample is drawn. Yet, the Federal Circuit noted: “Dependent method claims 2-7 and 10 add limitations… [that] require the resampling method to be a bootstrap method." SAP, 260 F. Supp. 3d at 715 . Likewise, "[c]laims 8 and 9 add limitations that the statistical method is a jackknife method and a cross validation method." Id. at 716. Because bootstrap, jack-knife, and cross-validation methods are all "particular methods of resampling," those features simply provide further narrowing of what are still mathematical operations. They add nothing outside the abstract realm. See Mayo, 566 U.S. at 88-89 (stating that narrow embodiments of ineligible matter, citing mathematical ideas as an example, are still ineligible); buySAFE, 765 F.3d at 1353 (same). Dependent method claims 12-21 are no different”. “Here, the focus of the claims is not any improved computer or network, but the improved mathematical analysis”. Similarly, the Supreme Court also found that an iterative formula for computing an alarm limit, by reputedly substituting the model with a most recent model, remained ineligible. see Parker v. Flook, 437 U.S. 584, 585, 198 USPQ 193, 195 (1978), as cited by MPEP 2106.04(a)(2) I. Specifically, in Flook, the process was repeated at the selected time intervals, and in each updating computation, the most recently calculated alarm base and the current measurement of the process variable was substituted for the corresponding numbers in the original calculation.
Since the computerized resampled statistical model in SAP supra, and the iterative or repetitive process of model substitution in Flook supra, did not save their underlining claims from patent ineligibility, the Examiner analogously reasons that here, a similar iterative or repetitive process of training and retraining of neural networks to robustly detect leads and characterize lead data, as argued by Applicant at Remarks 01/22/2026 p.16 ¶6, would also not preclude the current claims from reciting, describing or setting forth the abstract fundamental practices and its associated mathematical manipulations. Also, MPEP 2106.04(a)(2) II A ¶2 is clear that the term "fundamental" is not used in the sense of necessarily being old or well-known2, but rather as a building block of modern economy. Here, the purported robustness in detecting leads and characterizing lead data, as alleged by Applicant at Remarks 01/22/2026 p.16 ¶6 represents such fundamental, building block of modern economy regardless of whether or not such robustness in the sales lead data is old or well-known. Yet, MPEP 2106.04(d)(1) is clear that “improvement in the judicial exception itself is not an improvement in technology” and MPEP 2106.04 I ¶3 is also clear that claims directed to narrow laws that have limited applications, remain patent ineligible.
Also, MPEP 2106.04 I, cites Myriad, 569 U.S at 591, 106 USPQ2d at 1979 to stress that even a “groundbreaking, innovative, or even brilliant discovery does not by itself satisfy the 101 inquiry”. The “Myriad” rationale was corroborated by SAP Am Inc v InvestPic cited by MPEP 2106.04(a)(2) I.C(i). Digging deeper into the Court’s rationale in SAP supra, Examiner finds the Court ruled that, “even if one assumes that the techniques claimed are groundbreaking, innovative, or even brilliant those features are not enough for eligibility because their innovation is innovation in ineligible subject matter. An advance of that nature is ineligible for patenting”. That is, “no matter how much of an advance in the field the claims” [would] “recite the advance” [would still] “lie entirely in the realm of abstract ideas” with no plausibly alleged innovation in non-abstract application realm.
Here, as in SAP Am Inc v. InvestPic, LLC, 890 F.3d 1016, 126 USPQ.2d 1638 (Fed. Cir. 2018), no matter how much of an advance or roboustness in the sales lead data the claims would recite, said advance would still lie entirely within the realm of Certain Methods of Organizing Human Activities with no plausibly of the alleged innovation entering the non-abstract realm. The “SAP” findings were corroborated by Versata Dev Grp Inc v SAP Am Inc 115 USPQ2d 1681 Fed Cir 2015 again undelaying the difference between improvement to entrepreneurial goal objective versus improvement to actual technology. see MPEP 2106.04.
Here, the robustness in the sales lead data as argued by Remarks 01/22/2026 p.16 ¶6, and the “analyzing a sample of leads data for different leads in a database and producing a buying pattern metrics for the different leads data” as argued by Remarks 01/22/2026 p.16 ¶8, vis-à-vis newly added dependent Claims 28,29, would correspond to such abstract, mathematical analysis to achieve an equally abstract fundamental or economic concept of “business” or “sales lead data”, as a fundamental building block, which, no matter of its alleged robustness, would be ineligible following the legal tests of SAP, Flook, Myriad, and Versata as cited by MPEP supra.
Accordingly, there is a preponderance of legal evidence showing that the claims’ character as a whole remains undeniably abstract. Thus, the Step 2A prong one argument is unpersuasive.
Also, as it will be further revealed in the analysis, at the subsequent steps below, the level of machine learning, automation or computerization used for performing such abstract business concepts of sales lead data and associated mathematical manipulations of characteriz[ing] “the current business lead based on the individual different data sets”; “determining a probability of occurrence”, “determining a probability of conversion”, scor[ing], “producing” “pattern metrics” etc. would at best represent a mere application of the abstract exception [MPEP 2106.05(f)(2)(i)] and/or a narrowing of said abstract exception to a field of use or technological environment [MPEP 2106.05(h)], none of which integrate the abstract exception into a practical application, and for the same reasons none of which provide significantly more than what already found abstract.
Thus, it will be shown that the subject matter eligibility arguments below are unpersuasive and the claims remain patent ineligible.
Step 2A prong two: Remarks 01/22/2026 p.16 last ¶-p.17 ¶3 argues that the limitation
“responsive to the error signal, selectively adjusting weights of particular neurons of the at least one machine learning model to address the divergence; and retraining with the adjusted weights to reduce the error signal” as amended at each of independent Claims 1,11,20 is similar to training machine learning models that improve computer functionality in Ex Parte Desjardins, Appeal No. 2024-000567 (PTAB September 26,2025, Appeals Review Panel Decision). Remarks 01/22/2026 p.17 ¶4-p.18 ¶3 argues that when considering the claim as a whole, the “trend analyzer”, “machine learning models”, math model, and the steps for training and retraining improve accuracy and reliability, to produce accurate and reliable information for previously inaccurate lead data, which Remarks 01/22/2026 p.18 ¶ 2 asserts to be a particular transformation.
Examiner fully considered the Step 2A prong 2 argument but respectfully disagrees finding it unpersuasive, stressing that in Ex Parte Desjardins, Appeal No. 2024-000567 (PTAB September 26, 2025, Appeals Review Panel Decision), the specification identified the improvement to machine learning technology by explaining how the machine learning model is trained to learn new tasks while protecting knowledge about previous tasks to overcome the problem of “catastrophic forgetting,” and that the claims reflected the improvement identified in the specification. Indeed, the enumerated improvements identified in the Desjardins specification included disclosures of the effective learning of new tasks in succession in connection with specifically protecting knowledge concerning previously accomplished tasks; allowing the system to reduce use of storage capacity; and the enablement of reduced complexity in the system. Such improvements were tantamount to how the machine learning model itself would function in operation and therefore not subsumed in the identified mathematical calculation.
Here however, as demonstrated by the Examiner above, the Applicant has not been the first to invent machine learning nor is the Applicant alleging as much. At most, the Applicant uses machine learning to aid or perform an entrepreneurial and abstract concept of determining “sales lead data” and characteriz[ing] “the current business lead based on the individual different data sets”; “determining a probability of occurrence”, “determining a probability of conversion”, scor[ing], “producing” “pattern metrics” etc. Thus here, any benefit, in the application or use of retrain[ed] machine learning, as raised by Applicant at Remarks 01/22/2026 p16 last ¶-p17 ¶3, would at most be a benefit ensuing from improving upon the abstract determination in “sales lead data”. This is clearly distinct from Ex Parte Desjardins supra which improved how the machine learning model itself would function, and clearly a case of improvement of the abstract exception itself, which, according to MPEP 2106.05(a) II ¶2, is not improvement in actual technology. Such finding is also consistent with MPEP 2106.04(a)(2) II A paragraph 2 explaining that the term fundamental is not used in the sense of being old or well-known, but rather as a building block of modern economy. Here, the determination and scoring associated with the sales lead data represents such building block of modern economy ineligible for patent protection, no matter how novel, old or well-known its claimed process is, and no matter of whether or not its implementation or application in using retrain[ed] machine learning produces accurate or inaccurate lead data. Such use or application of the abstract exception, even when more granularly tested at the level of additional computer-based elements, would correspond to a mere invocation of machines or computer functionality to apply an abstract or business process and its underlining mathematical algorithms, which according to MPEP 2106.05(f)(2) (i) does not integrate said abstract process into a practical application. Such use or application of the abstract exception by retrained machine learning can also perhaps be argued as a technological environment narrowing of the combination of collection (here “retrieving”), analysis (here “determining”, “estimating” etc.) and displaying certain results of the collection of analysis, when tested per MPEP 2106.05(h) vi.
Further, MPEP 2106.04 I paragraph 3 is also clear: claims directed to narrow laws that have limited applications, remain ineligible. Here, the limited applications of machine learning to determine “sales lead data” remains ineligible under auspices of MPEP 2106.04 I paragraph 3.
For additional details, the Examiner also reincorporates all Examiner’s findings and rationales above emphasizing the inability of the analogous reinforcement or learning algorithm of SAP to render eligible its underlining abstract idea even when attempting to similarly improve accuracy and reliability of the financial information, and the inability of the iterative algorithm in Flook to render eligible the underlining abstract idea. Indeed, as corroborated by MPEP 2106.05 (c) ¶5, the mere manipulation of mathematical constricts has not been deemed as a patent eligible transformation3. Also, MPEP 2106.05 (c) states Examiners may find it helpful to evaluate other considerations such as mere instructions to apply exception consideration (MPEP 2106.05(f)), insignificant extra-solution activity consideration (MPEP 2106.05(g)), and the field of use and technological environment consideration (MPEP 2106.05(h)), when making a determination of whether a claim satisfies the particular transformation consideration. Examiner abides by such guidance and reincorporates all the Examiner’s findings and rationales above, as tested per MPEP 2106.05(f) and (h), showing the level of automation or computerization characterized by the retrained machine learning does not render the determination of the sales lead data less abstract and eligible. Based on the preponderance of legal evidence the Examiner, submits that the level of automation or computerization, as argued by Applicant above at Remarks 01/22/2026 p16 last ¶-p17 ¶3, does not integrate the abstract exception into a practical application.
Step 2B: Remarks 01/22/2026 p.18 ¶5 argues that “training the three or more machine learning models further includes: generating the training data sets by analyzing a sample of leads data for different leads in a database and producing a buying pattern metrics for the different leads data” as recited at the newly added dependent Claims 28, 29 add further to the improvement of computer functionality by generating specific training data sets
Examiner fully considered the Step 2B argument but respectfully disagrees finding it unpersuasive by reincorporating herein all the Examiner’s findings and rationales above.
For example, “generating the training data sets by analyzing a sample of leads data for different leads in a database and producing a buying pattern metrics for the different leads data” of dependent Claims 28,29 can be argued as not meaningfully different than the “SAP” proposed utilization of resampled statistical methods for analysis of financial data, such as bootstrap method, which estimates distribution of data in a pool (a sample space) by repeated sampling of the data in the pool. A sample space in a boot-strap method can be defined by selecting a specific investment or a particular period of time. Data samples are drawn from the sample space with replacement: samples are drawn from the sample space and then returned to the pool before next sample is drawn. Yet, the Federal Circuit noted: “Dependent method claims 2-7 and 10 add limitations… [that] require the resampling method to be a bootstrap method." SAP, 260 F. Supp. 3d at 715. Likewise, "[c]laims 8 and 9 add limitations that the statistical method is a jackknife method and a cross validation method." Id. at 716. Because bootstrap, jack-knife, and cross-validation methods are all "particular methods of resampling," those features simply provide further narrowing of what are still mathematical operations. They add nothing outside the abstract realm. See Mayo, 566 U.S. at 88-89 (stating that narrow embodiments of ineligible matter, citing mathematical ideas as an example, are still ineligible); buySAFE, 765 F.3d at 1353 (same). Dependent method claims 12-21 are no different.
Thus, a case can be made that “generating the training data sets by analyzing a sample of leads data for different leads in a database and producing a buying pattern metrics for the different leads data” of dependent Claims 28,29, is abstract right from the onset with the level of computerization or automation representative of a computer environment [MPEP 2106.04(A)(2) III C #2] or tool [MPEP 2106.04(a)(2) III C #3] incapable to render the claim patent eligible. Even when more granularly tested, at Step 2B, as part of an additional computer-based elements, such “generating the training data sets by analyzing a sample of leads data for different leads in a database and producing a buying pattern metrics for the different leads data” corresponds to a mathematical algorithm being applied on the computer, which according to MPEP 2106.5(f)(2)(i) does not provide significantly more. Thus, the Step 2B argument is unpersuasive.
Based on the preponderance of evidence above, the Examiner submits that the claims still recite, describe or set forth the abstract exception (Step 2A prong one), with no additional elements capable to either alone, or together, integrate the abstract exception not a practical application (Step 2A prong two) or provide significantly more (Step 2B). The claims are ineligible.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), first paragraph:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1-3,6-7,11-13,16-17 and 20-29 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for pre-AIA the inventor(s), at the time the application was filed, had possession of the claimed invention.
Claims 1,11,20 are independent and have been amended to each recite
- “retraining with the adjusted weights to reduce the error signal” [bolded emphasis added]
Remarks 01/22/2026 p.13 ¶2 points as support to Original Specification ¶ [0041]-¶ [0042]. Yet, none of Original Specification ¶ [0041]-¶ [0042], or any section of the Original Specification provides clear, deliberate and sufficient support to show that Applicant had possession for the newly added matter of retraining with the adjusted weights to reduce the error signal. In fact, the Original Specification ¶ [0042] 2nd sentence appears to be teaching away the retraining, by language such as “without over training” [interpreted as without re-training] “the neural nets 24”.
Claims 2,3,6,7,21,23,26,28 are dependent and rejected based on rejected parent Claim 1.
Claims 12,13,16,17,22,24,27,29 are dependent and rejected based on rejected parent Claim 11
Claim 25 is dependent and rejected based on rejected parent Claim 20.
Clarification and/or correction is/are required.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(B) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-3,6-7,11-13,16-17 and 20-29 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), ¶2, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor, or for pre-AIA the applicant regards as the invention.
Claims 1,11,20 as amended each preponderantly, consistently and repeatedly recite “three or more machine learning models”, except for once at the “generating” limitation, where each of claims 1,11,20 recite “the machine learning models”. Thus, independent Claims 1,11,20 are rendered vague and indefinite because it is unclear if “the machine learning models” as subsequently recited at “generating” limitation, relate back, in the antecedent, to “three or more machine learning models” as newly amended at the “training” limitation.
Claims 1,11,20 are recommended to be amended to each recite, among others, and as an example only: generating an error signal for at least one of the three or more machine learning models in response divergence of output from a preferred output. Furthermore,
Claims 11,20 have been amended to each recite, among others:
- “training three or more machine learning models including:” …
- inputting the individual data sets and respective buying patten metrics into three or more machine learning models to provide the-respective output scores,
Claims 11, 20 are rendered vague and indefinite because it is unclear if “three or more machine learning models” as subsequently recited at “inputting” limitation relate back to “three or more machine learning models” as antecedently amended at “training limitation”
Claims 11,20 are recommended to be amended to each recite, among others, and as an example only:
- inputting the individual data sets and respective buying patten metrics into the three or more machine learning models to provide the-respective output scores,
Claims 2-3,6,7,21,23,26,28 are dependent and rejected based on rejected parent Claim 1.
Claims 12,13,16,17,22,24,27,29 are dependent and rejected based on rejected parent Claim 11.
Claim 25 is dependent and rejected based on rejected parent Claim 20.
Clarification and/or correction is/are required.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
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 1-3,6-7,11-13,16-17 and 20-29 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea, here abstract idea) without significantly more. The claim(s) recite(s) describe or set forth the abstract grouping(s)4 of Certain Method of Organizing Human Activities implemented through equally abstract Mental Processes and Mathematical Relationships expressed in words. Here, independent Claims 1,11,20 still recite fundamental economic or commercial practices, principles and commercial or behavioral concepts set forth by recitations of: “retrieving” “individual data sets associated with different behaviors or preferences of a current business lead”; “analyzing” “the individual data sets to determine respective buying pattern metrics for each individual data set”; “determining a probability of occurrence of each of the respective output scores” that “characterizes the current business lead” and “determining a probability of conversion for each of the respective output scores” that “characterizes the current business lead” “by dividing a number of leads” “in predicting that a given lead results in a purchase” as recited at each of independent Claims 1,11,20, narrowed at dependent Claims 6,16 to “inputs describing or characterizing the current business lead” and narrowed at dependent Claims 7,17 to “inputs” [which] “may include one or more of: website views associated with the current business lead, measurements of website click-through behaviors, data from email responses pertaining to the product or service, data with chats from sales representatives, and measurements of engagement at one or more events”, narrowed to “lead data” “selected from the group” “comprising: digital body language of a lead, footprint of a lead at a given enterprise, engagement at events and/or with representatives, downloads” at dependent Claims 26,27. All these set forth to the abstract “Certain Method of Organizing Human Activities” grouping including fundamental economic practices or principles. Also, the fact that the composite score is displayed “on a graphical user interface (GUI)” (independent Claims 1,11,20), does not necessarily preclude the claims from reciting, describing or setting forth Certain Method of Organizing Human Activities, because per MPEP 2106.04(a)(2) II ¶6, 4th sentence certain activity between a person and a computer may still fall within certain methods of organizing human activity.
Further, MPEP 2106.04 (a)(2) II C cites the same BSG Tech. LLC v. Buyseasons, Inc., 899 F.3d 1281, 1286, 127 USPQ2d 1688, 1691 (Fed. Cir. 2018) and Interval Licensing LLC, v. AOL, Inc., 896 F.3d 1335, 127 USPQ2d 1553, 1555 (Fed. Cir. 2018) to state that considering historical usage information while inputting data , acquiring content from an information source, controlling the timing of the display of acquired content, displaying the content, and acquiring an updated version of the previously-acquired content when the information source updates its content to allow providing information to a person without interfering with the person’s primary activity still falls within the abstract confines of the certain methods of organizing human activities It then follows that here, functions such as “analyzing, by a trend analyzer, the individual data sets to determine respective buying pattern metrics for each individual data set, as real time data for the current business lead is populated into the respective databases”, “monitoring the composite score in real time as the composite score is generated to determine that the composite score is less accurate based, at least in part, on historical data including previous composite score performance in predicting that a given lead results in a purchase; and in response to determining that the composite score is less accurate, selectively adjusting the combining method in real time based on the real time data populated into the respective databases, so as to improve the accuracy of the composite score”, would similarly not preclude the claims from reciting, describing or setting forth the abstract idea and certainly would not provide an actual improvement in actual technology to integrate the abstract exception into a practical application when tested per MPEP 2106.05(f)(2) and MPEP 2106.05(f)(1).
Further, said fundamental economic or commercial practices or principles that correspond to the “Certain Method of Organizing Human Activities” can be argued by Examiner as practically implementable through equally abstract computer-aided evaluation (throughout Claims 1-3, 6-8, 11-13,16-18, 20-27) and judgement (here “based on the respective measurements of impact” “providing feedback loop to selectively scale the probability of conversions when performing the combining method” and “in response to determining that the composite score is less accurate, selectively adjusting the combining method… so as to improve the accuracy of the composite score” at independent Claims 1,11,20), of the abstract “Mental Processes” grouping. Such “Mental Processes” grouping further include an example of wide-area real-time performance monitoring system for monitoring and assessing dynamic stability of [technological environment] listed by MPEP 2106.04(a)(2) III D citation of Electric Power Group, 830 F.3d at 1351 and n.1, 119 USPQ2d at 1740 and n.1. It then follows that here, the amended feature of “analyzing, by a trend analyzer, the individual data sets to determine respective buying pattern metrics for each individual data set, as real time data for the current business lead is populated into the respective databases”, and “real time” “monitoring the composite score” and “estimating” “accura[cy]”, as well as “selectively adjusting the combining method in real time based on the real time data populated into the respective databases” would not necessarily preclude the argued claims from reciting, describing or setting forth the abstract exception in a manner analogous to Electric Power Group. Also here, as in Electric Power Group v. Alstom, S.A., 830 F.3d 1350, 1353-54, 119 USPQ2d 1739,1741-42 (Fed. Cir. 2016), cited by MPEP 2106.04(a)(2) IIIA, 5th bullet point, it is the combination of collecting information, analyzing it, and displaying certain results of collection and analysis5 that corresponds to the abstract “Mental Processes” grouping.
The collection is set forth as “retrieving” “individual data sets associated with different behaviors or preferences of a current business lead”, “inputting the individual data sets and respective buying patten metrics” “to provide the-respective output scores”, “selecting a plurality of output scores” "based, at least in part, on the each estimated accuracy of the output scores” at Claims 1,11,20 and “inputs” “describ[e] or characterize[e] a business lead” at Claims 6,7,16,17.
The analysis is set forth by repeatedly recited analyzing, determining, estimating, combining, weighting, adding estimates at Claims 1,2,8,11,12,18,20. For example “determining a probability of occurrence of each of the respective output scores”, “by dividing a number of leads in a sample associated with the respective output score, by a first total number of leads in the sample”; “determining a probability of conversion for each of the respective output scores by dividing a number of leads associated with a given one of the respective output scores”, “which converts into an opportunity within a predefined time period, by a second total number of leads associated with the given one of the respective output scores” and “estimating an accuracy of each of the output scores based, at least in part on historical performance of past corresponding lead scores”, “determining respective measurements of impact of each of the plurality of select output scores on accuracy of the composite score by analyzing combinations of the respective output score”; “selectively adjusting the combining method in real time based on the real time data populated into the respective databases, so as to improve the accuracy of the composite score” at independent Claims 1,11,20 fall well within arithmetic cognitive capabilities of one of ordinary skills. Same rationale applies to “weighting and adding estimates using one or more weighted probability distributions” at dependent Claims 2,12; “linear combination of a normalized probability of conversion with the probability occurrence of each of the respective output scores”, “wherein the normalized probability of conversion represents the probability of conversion for a given one of the respective output scores divided by a sum of probabilities of conversion of a model” at dependent Claims 21,22, “analyzing a sample of leads data for different leads in a database and producing a buying pattern metrics for the different leads data” at dependent Claims 28,29.
The certain results of collection and analysis6 are set forth by “displaying” “improved composite score as a single source of truth of the lead data rather than displaying the disparate lead data” at independent Claims 1,11,20; “displaying analytics of historical composite scores versus historical probability of conversions” at dependent Claims 23,24,25.
Equally important, MPEP 2106.04(a)(2) III C states that: #1. Performing mental process on generic computer, #2. Performing mental process in computer environment, #3. Using computer as tool to perform a mental process does not preclude a claim from reciting the abstract exception.
It then follows that here, as in MPEP 2106.04(a)(2) III C, the nominal recitation of: “via a web server or application server from multiple disparate databases” capable of “retrieving individual data sets”, and nominal recitation of “by a trend analyzer” “analyzing” “the individual data sets to determine respective buying pattern metrics for each individual data set, as real time data for the current business lead is populated into the respective databases”; and nominal recitation of “performing” “monitoring the composite score in real time” and nominal recitation and use of “three or more machine learning models” would not necessarily preclude the claims, from reciting, describing or setting froth the abstract exception under examples #1,2,3 above. Also, such computer aided evaluation and judgment uses equally abstract Mathematical Relationships expressed in words, such as “math model” for “determining a probability” “of” “occurrence” and “conversion” respectively by “dividing a number of leads” to achieve the aforementioned “business lead” analytics. The fact that the independent Claims 1,11,20 call for “three or more” “models” and “different data sets” does not preclude the claims from describing or setting forth the abstract idea because MPEP 2106.04 II A1 found that a mathematical equation in repetitively calculating step7 [akin here to “selectively combining the plurality of select output scores to generate a composite score via a combining method using a sum of the probability of occurrences and the probability of conversions, including : determining respective measurements of impact of each of the plurality of select output scores on accuracy of the composite score by analyzing combinations of the respective output scores”; and “providing a feedback loop to selectively scale the probability of conversions when performing the combining methods” -independent Claims 1,11,20] still set forth the abstract exception. MPEP 2106.04(a)(2) IA iv is also clear that generating first and second data by taking existing information, manipulating the data using mathematical functions such as correlations, and organizing this information into a new form set forth the abstract exception. Here, the current independent Claims 1,11,120 are similarly “generating or estimating an accuracy of each of the output scores based, at least in part on”, [existing] “historical performance of past corresponding lead scores”, for subsequent mathematical manipulation to organize the information into a “composite score via a combining method using a sum of the probability of occurrences and the probability of conversions”, as a “single source of truth of the lead data”.
Also, MPEP 2106.04 I. C. i states that performing resampled statistical analysis to generate resampled distribution8 sets forth the abstract idea. It follows that here the analogous “inputting the individual data sets and respective buying metrics” “to provide the-respective output scores”, “determining a probability of occurrence of each of the respective output scores”, “by dividing a number of leads in a sample associated with the respective output score, by a first total number of leads in the sample”; “determining a probability of conversion for each of the respective output scores by dividing a number of leads associated with a given one of the respective output scores from a particular one of the two” “learning models, which converts into an opportunity within a predefined time period, by a second total number of leads associated with the given one of the respective output scores”; “estimating an accuracy of each of the output scores based, at least in part, on historical performance of past corresponding lead scores”; “selecting a plurality of the output scores” “based, at least in part, on the each estimated accuracy of the output scores” and “selectively combining the plurality of select output scores to generate a composite score via combining method using a sum of the probability of occurrences and the probability of conversions, including determining respective measurements of impact of each of the plurality of select output scores on accuracy of the composite score, by analyzing combinations of the plurality of output scores”; and “based on the respective measurements of impact, providing a feedback loop to selectively scale the probability of conversions when performing the combining method”, “in response to determining that the composite score is less accurate” “selectively adjusting the combining method so as to maintain or improve the accuracy of the composite score” at independent Claims 1,11,20, and “selectively weighting and adding estimates using one or more weighted probability distributions” at Claims 2,12, “a linear combination of a normalized probability of conversion with the probability occurrence of each of the respective output scores”, “wherein the normalized probability of conversion represents the probability of conversion for a given one of the respective output scores divided by a sum of probabilities of conversion of a model” at Claims 21,22 are not be meaningfully different than said resampled statistical analysis to generate resampled distribution9, and thus similarly set forth the abstract exception.
Another example, cited by MPEP 2106.04(a)(2) I ¶2 is “Flook” where the mathematical formula for computing alarm limits was characterized by the Supreme Court as part of the abstract mathematical relationships and formulas. Here too, following “Flook” example as cited by MPEP 2106.04(a)(2) I ¶2, the Examiner reasons that “use” “data sets to provide” “the respective output scores”, “estimating an accuracy of each of the output scores based, at least in part, on historical performance of past corresponding lead scores”; “selecting a plurality of the output scores” “based, at least in part, on the each estimated first level of accuracy of the output scores” and “selectively combining the plurality of select output scores to generate a composite score via a combining methods using a sum of the probability of occurrences and the probability of conversions” “including” “determining respective measurements of impact of each of the plurality of select output scores on accuracy of the composite score, by analyzing combinations of the respective output scores”; “based on the respective measurements of impact, providing a feedback loop to selectively scale the probability of conversions when performing the combining methods”; and “in response to determining that the composite score is less accurate” “selectively adjusting the combining method in real time based on the real time data populated into the respective databases, so as to maintain or improve the accuracy of the composite score” at independent Claims 1,11,20, and “selectively weighting and adding estimates using one or more weighted probability distributions” at dependent Claims 2,12 are not meaningfully different than “Flook’s” “method for updating the value of at least one alarm limit on at least one process variable involved in a process comprising the catalytic chemical conversion of hydrocarbons wherein said alarm limit has a current value off Bo+K wherein Bo is the current alarm base and K is a predetermined alarm offset which comprises: (1) Determining the present value of said process variable, said present value being defined as PVL; (2) Determining a new alarm base B1, using the following equation: B1=Bo(1.0-F)+PVL(F) where F is a predetermined number greater than zero and less than 1.0; (3) Determining an updated alarm limit which is defined as B1+K; and thereafter "(4) Adjusting said alarm limit to said updated alarm limit value”… “The process is repeated at the selected time intervals. In each updating computation, the most recently calculated alarm base and the current measurement of the process variable will be substituted for the corresponding numbers in the original calculation, but the alarm offset and the weighting factor will remain constant”. [bolded emphasis added]. Since the calculations in “Flook” were deemed by the Supreme Court as abstract, the Examiner analogously reasons that here, the similar claimed models’ calculations, including their repetitive calculations, combinations, updates of adjustment as introduced by the “feedback loop” of independent Claims 1,11,20 should also be treated as abstract. Also looking closer at “Flook” supra, the Examiner finds that the Court found that use of computers for automatic monitoring-alarming including monitoring of chemical process variables, along with the recomputing and readjusting of alarm limits at selected time intervals, where in each updating computation, the most recently calculated alarm base and the current measurement of the process variable will be substituted for the corresponding numbers in the original calculation, did not save the claims from patent ineligibility. It would then follow that here, the analogous “monitoring the composite score in real time as the composite score is generated, to determine that the composite score is less accurate based, at least in part, on historical data including previous composite score performance in predicting that a given lead results in a purchase” “and” subsequent “selectively adjusting” “in response to determining that the composite score is less accurate” would at most represent an improvement in the abstract solution and thus not render the claims eligible. Also, since MPEP 2106.04(a)(2) I A, B found that the Arrhenius equation10, falls within the abstract exception, Examiner reasons that here too, the analogous use of Monte Carlo method at dependent Claims 3,13 in the combining method would similarly describe or set forth the abstract idea.
Therefore, Examiner submits that there is a preponderance of evidence for the claims’ character as a whole reciting, describing or setting forth the abstract idea. Step 2A prong one.
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
This judicial exception is not integrated into a practical application because, per Step 2A prong two, the individual or combination of the additional, computer-based elements are/is found, to merely apply the above abstract idea [MPEP 2106.05(f)] and/or narrow the abstract character of the claims to a field of use or technological environment [MPEP 2106.05(h)].
Here, some of the computerized elements in the “instruct[ed]” / “encoded logic” “execut[able]” “processor(s)” of Claims 1,20, “trend analyzer”, “machine learning models” of Claims 1,8,11, 18 , 20, and “databases”, “web server, or application server” and “graphical user interface” of independent Claims 1,11,20 / “graphical user interface (GUI) having UI display screen in communication with a composite score generator” at independent Claim 20, were tested above as computer aids at the prior step. Examiner now submits that even if now more granularly tested as additional computer-based elements, at Step 2A prong two of the analysis, they would still represent under MPEP 2106.05(f)(2) mere tools to apply the abstract idea and its underlining algorithm11 and to perform economic tasks [as recited and mapped above] or other tasks to receive and transmit data12 [here “database inputs” at dependent Claims 6,7,16,17; “output(s)” at Claims 1,8,11,18,20, and “individual data sets” retriev[ed] via “a web server or application server” “from disparate databases”] and to require use of software other computer components to tailor information13 [“displaying on a graphical user interface (GUI), the improved composite score as a single source of truth of the lead data, rather than displaying the disparate lead data” at independent Claims 1,11,20, and possibly; “displaying analytics of historical composite scores versus historical probability of conversions” at dependent Claims 23,24,25], as well as to monitor audit log data executed on a general-purpose computer14 [here “monitoring the composite score in real time as the composite score is generated, to determine that the composite score is less accurate based, at least in part, on historical data including previous composite score performance in predicting that a given lead results in a purchase” at independent Claims 1,11,20.
Also, per narrowing of “machine learning models” of Claims 1,8,11,16,18,28,29 and “math model” of Claims 1,11,20, Examiner points to MPEP 2106.05(f)(2) citing Alice Corp Pty Ltd V CLS Bank Int’l, 573 US 208,223, 110 USPQ2d 1976, 1983 (2014), Gottschalk v. Benson, 409 U.S. 63, 64, 175 USPQ 673, 674 (1972); and Versata Dev. Group Inc. v SAP Am, Inc, 793 F.3d 1306, 1334,115 USPQ2d 1681,1701 (Fed Cir 2015); to state that a mathematical algorithm applied on computer, is example of additional elements instructed to apply the exception. This corresponds to computerization in “training three or more machine learning models including: inputting different collections of training data sets into each of the three or more machine learning models; generating an error signal for at least one of the machine learning models in response divergence of output from a preferred output; responsive to the error signal, selectively adjusting weights of particular neurons of the at least one machine learning model to address the divergence; and retraining with the adjusted weights to reduce the error signal” at independent Claims 1,11,20 and “wherein training the three or more machine learning models further includes: generating the training data sets by analyzing a sample of leads data for different leads in a database and producing a buying pattern metrics for the different leads data” at dependent Claims 28,29.
Similarly, per MPEP 2106.05(h), the same rationales apply to specifying that the abstract idea of monitoring audit log data relates to transactions or activities that are executed in a computer environment15, and limiting the combination of collecting information, analyzing it, and displaying certain results of the collection and analysis to data related to a field of use or technological environment16. In a similar vein, MPEP 2106.05(h) cites Parker v. Flook, to state that calculating an updated value for an alarm limit (a numerical limit on a process variable such as temperature, pressure or flow rate) according to a mathematical formula in a process comprising the catalytic chemical conversion of hydrocarbons, represented mere examples of limiting the abstract idea to a field of use limitation which does not integrate the abstract idea into a practical application. Looking closer at Parker v. Flook, 437 U.S. 584, 98 S. Ct. 2522,57 L. Ed. 2d 451,198 USPQ 193 (1978), Examiners find that the claims were directed to an analogous computational process repeated at selected time intervals, where in each updating computation, the most recently calculated alarm base and the current measurement of the process variable will be substituted for the corresponding numbers in the original calculation. Here too, the Claims 1,11,20 similarly call for “three or more machine learning models”, “retraining with the adjusted weights to reduce the error signal”, use “individual data sets” to “provide respective output scores, wherein the output includes plural scores, wherein each of the output scores characterizes the current business lead based on the individual different data set” and “selectively combining the plurality of select output scores to generate a composite score via a combining method using a sum of the probability of occurrences and the probability of conversions”. Since, MPEP 2106.05(h) found such calculating of an updated value for a limit, does not integrate the abstract idea into a practical application, Examiner similarly reasons that current “the machine learning algorithms” as generally recited would similarly not integrate the abstract idea into a practical application, not even when performing repeated computations as reflected by language such as “retraining with the adjusted weights to reduce the error signal” and “based on the respective measurements of impact providing a feedback loop to selectively scale the probability of conversions when performing the combined method” at independent Claims 1,11,20.
As per the recitation of “displaying on a graphical user interface (GUI), the improved composite score as a single source of truth of the lead data, rather than displaying the disparate lead data”, at independent Claims 1,11,20 the Examiner points to MPEP 2106.05(h) again citing Electric Power Group, LLC v Alstom S.A 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed Cir. 2016), where Federal Circuit ruled that “even as to the claim requirement of "displaying concurrent visualization" of two or more types of information, '710 patent, col. 31, line 37, is understood to require time-synchronized display would offer anything but readily available computer components. As stressed by the Federal Circuit “We have repeatedly held that such invocations of computers and networks that are not even arguably inventive are "insufficient to pass the test of an inventive concept in the application" of an abstract idea”. buySAFE, 765 F.3d at 1353,1355; Mortg. Grader, Inc. v. First Choice Loan Servs. Inc., 811 F.3d 1314, 1324-25 (Fed. Cir. 2016); Intellectual Ventures I LLC v. Capital One Bank (USA), 792 F.3d 1363, 1370 (Fed. Cir. 2015); Internet Patents, 790 F.3d at 1348-49; Content Extraction, 776 F.3d at 1347-48”. Examiner applies a similar rationale as in “Electric Power Group” to reason that here, “displaying on a graphical user interface (GUI), the improved composite score as a single source of truth of the lead data, rather than displaying the disparate lead data”, at independent Claims 1,11,20, would similarly not integrate the abstract idea. In fact, the Federal Circuit in FairWarning IP, LLC v. Iatric Sys., 839 F.3d 1089, 1095, 120 USPQ2d 1293, 1296 (Fed. Cir. 2016), as cited by each of MPEP 2106.05(a),(f),(h), found that despite FairWarning contention that its system allowed for the compilation and combination of [*1097] these disparate information sources and that the patented method "made it possible to generate a full picture of a user's activity, identity, frequency of activity, and the like in a computer environment." Id. at 10, the Federal Circuit responded that the mere combination of data sources, does not make the claims patent eligible. As we have explained, merely selecting information, by content or source, for collection, analysis, and [announcement] does nothing significant to differentiate a process from ordinary mental processes, whose implicit exclusion from § 101 undergirds the information-based category of abstract ideas." Elec. Power, 830 F.3d 1350, [2016 BL 247416], 2016 WL 4073318, at *4. Thus, Examiner submits that here there is still a preponderance of legal evidence showing that that the additional computer-based elements identified above, do not integrate the abstract idea into a practical application. Step 2A prong two.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
The claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception because as shown above, the additional computer-based elements merely apply the already recited abstract idea [MPEP 2106.05(f)] and/or provide a narrowing of the abstract idea to a field of user or technological environment [MPEP 2106.05(h)]. Examiner follows MPEP 2106.05 (d) II and carries over the findings at MPEP 2106.05 (f) and (h) as a sufficient option for evidence that the additional computer-based elements also do not provide significantly more, without relying on conventionality test of MPEP 2106.05(d).
For these reasons, said computer-based additional elements similarly do not provide significantly more than the abstract idea itself in light of MPEP 2106.05(f) and/or (h) as sufficient option(s) for evidence. For example, Examiner investigated FairWarning IP, LLC v. Iatric Sys., 839 F.3d 1089, 1095, 120 USPQ2d 1293, 1296 (Fed. Cir. 2016), as cited by MPEP 2106.05(f)(2), and found that capabilities of the additional, computer-based elements to monitor audit log data executed on a general-purpose compute, are examples of applying the abstract idea, which does not integrate it into a practical application. Step 2A prong two. Further, upon closer investigation of “FairWarning” supra, the Examiner finds that Federal Circuit found unpersuasive an argument of compilation, combination and accessing of disparate information sources to make it possible to generate a full picture of frequency of activity and the like in a computer environment. Thus, the Examiner reasons that accessing disparate learning models and their compilation and combination as a “composite score” would similarly be unpersuasive in rendering the claims eligible. As the Federal Circuit explained in “FairWarning”, selecting information, by content or source, for collection, analysis and announcement, does nothing significant to differentiate a process from ordinary mental processes, whose implicit exclusion from 101 undergirds the information-based category of abstract ideas. Also here, similar to how “Fairwarning” claimed rules as reflected in audit log data, during a pre-determined time interval and in excess of a specific volume, the independent Claims 1,11,20 are directed to an analogous “generating an error signal for at least one of the machine learning models in response divergence of output from a preferred output”; “estimating an accuracy of each of the output scores based at least in part of historical performance of past corresponding lead scores”, “selectively combining the plurality of select output scores to generate a composite score” “monitor[ed]” “in real time” “in response to determining that the composite score is less accurate” [than the threshold or volume] for “selectively adjusting the combining method so as to improve the accuracy of the composite score”. Such computerized “selectively adjusting the combining method so as to improve the accuracy of the composite score in real time based on the real time data populated into the respective databases” is analogous to the generated “full picture of user activity, identity, frequency of activity and the like”, in FairWarning supra and consistent with the current Original Specification ¶ [83] last sentence: “This can lead to enhanced situational awareness, which can be particularly important for enterprise sales representatives, account managers, and so on”.
Alternatively, when tested per MPEP 2106.05(h), the same rationales apply to specifying that the abstract idea of monitoring audit log data relates to transactions or activities that are executed in a computer environment17, and limiting the combination of collecting information, analyzing it, and displaying certain results of the collection and analysis to data related to a field of use or technological environment18 .
Yet, even assuming arguendo that further evidence would be require to demonstrate conventionality of the additional, computer-based elements, the Examiner would point to MPEP 2106.05(d) to demonstrate that said additional elements remain well-understood, routine, conventional as evidenced by at least the Applicant’s own Specification and/or by case law.
Per MPEP 2106.05(d)(I)(2) Examiner points to Applicant’s own Specification as follows:
* Original Specification ¶ [34] “For clarity, certain well-known components, such as hard drives, processors, operating systems, power supplies, Internet Service Providers (ISPs), Application Programming Interfaces (APIs), web services, websites, and so on, are not necessarily explicitly called out in the figures. However, those skilled in the art with access to the present teachings will know which components to implement and how to implement them to meet the needs of a given implementation”.
* Original Specification ¶ [41] 2nd sentence, reciting at high level of generality: “For the purposes of the present discussion, a machine learning model may be any computer software program or algorithm that can be trained (e.g., via supervised learning) or self-trained (e.g., via unsupervised learning), such that it can make improvements responsive to mistakes, e.g., as indicated by error signals, cost functions, or other means of gauging divergence of outputs from preferred outputs”
* Original Specification [163] reciting at high level of generality: A "processor" includes any suitable hardware and/or software system, mechanism or component that processes data, signals or other information. A processor can include a system with a general-purpose central processing unit, multiple processing units, dedicated circuitry for achieving functionality, or other systems. Processing need not be limited to a geographic location, or have temporal limitations. For example, a processor can perform its functions in real time, offline, in a batch mode etc Portions of processing can be performed at different times and at different locations, by different (or the same) processing systems. Examples of processing systems can include servers, clients, end user devices, routers, switches, networked storage, etc. A computer may be any processor in communication with a memory. The memory may be any suitable processor-readable storage medium, such as random-access memory (RAM), read-only memory (ROM), magnetic or optical disk, or other non-transitory media suitable for storing instructions for execution by the processor.
Per MPEP 2106.05(d)(II), the additional computer-based elements, can also be viewed as performing the well-understood, routine or conventional functions of:
* gather statistics19 [here “wherein the database inputs may include one or more of the following: website views associated with the lead, measurements of website click-through behaviors, data from email responses pertaining to the product or service, data with chats from sales representatives, and measurements of engagement at one or more events” at dependent Claims 7, 17] as well as electronically extract data20 / electronic recordkeeping21 / store and retrieve information in memory22 , receiving or transmitting data over a network, including utilizing an intermediary computer to forward information23 [here “retrieving via a web server or applications server from disparate databases, individual data sets” at independent Claims 1, 11, 20] / performing repetitive calculations24 and arrange a hierarchy of groups and sort information25 [here “selectively adjusting the combining method” at independent Claims 1, 11, 20]
With respect to the use of “trained neural networks” at dependent Claims 8,18 and with respect to “the three or more machine learning model” at independent Claims 1,11,20, the Examiner points to the conventionality of ensemble machine learning demonstrated by at least:
* US 20200134716 A1 ¶ [0088] “Generally, the MLAs that are trained and/or deployed as described herein may comprise any one, or some combination (e.g., in an ensemble), of a number of known machine learning techniques, which may include, without limitation: linear/logistic regression models, classification models, time-series models, clustering algorithms, nearest neighbor methods, decision trees, support vector machines, graphical models, neural networks, boosting, bagging, random forests, other ensemble methods, and/or any other type of function, algorithm, and/or model. In certain implementations, some of the noted algorithms might not use specifically engineered features”.
* US 20210192388 A1 ¶ [0070] last sentence: “There are several known techniques, including (but not limited to): convolutional neural networks (CNNs), recurrent neural networks (RNNs), ensemble learning methods such as adaptive boosting (e.g., Adaboost) learning, decision trees, support vector machines (SVMs), and other supervised learning techniques”.
* US 20200210894 A1 ¶ [0089] “The creation method for the prediction model by formula (3) is one example, and the prediction model may be derived by a generally known method such as regularization, decision tree, ensemble learning, neural networks, and Bayesian networks”.
* US 20200160460 A1 ¶ [0070] 4th sentence: “the machine learning classifier may include any machine learning classifier known in the art including, but not limited to, a conditional generative adversarial network (CGAN), a convolutional neural network (CNN) (e.g., GoogleNet, AlexNet, and the like), an ensemble learning classifier”
* US 20200134716 A1 ¶ [0088] “Generally, the MLAs that are trained and/or deployed as described herein may comprise any one, or some combination (e.g., in an ensemble), of a number of known machine learning techniques, which may include, without limitation: linear/logistic regression models, classification models, time-series models, clustering algorithms, nearest neighbor methods, decision trees, support vector machines, graphical models, neural networks, boosting, bagging, random forests, other ensemble methods, and/or any other type of function, algorithm, and/or model. In certain implementations, some of the noted algorithms might not use specifically engineered features”.
* US 20190289826 A1 ¶ [0115] system 100 may utilize one or more machine learning techniques to carry out one or more functions of the present disclosure. It is contemplated herein that system 100 may be configured to carry out any type of deep learning technique and/or machine learning algorithm/classifier known in the art including, but not limited to, a convolutional neural network, an ensemble learning classifier, a random forest classifier, an artificial neural network, and the like. In this regard, the one or more processors 12, 138, 146 may be configured to train one or more machine learning classifiers configured to carry out the one or more functions of the present disclosure.
* US 20210192388 A1 ¶ [0070] last sentence: “There are several known techniques, including (but not limited to): convolutional neural networks (CNNs), recurrent neural networks (RNNs), ensemble learning methods such as adaptive boosting (e.g., Adaboost) learning, decision trees, support vector machines (SVMs), and other supervised learning techniques”.
* US 20200210894 A1 ¶ [0089] “The creation method for the prediction model by formula (3) is one example, and the prediction model may be derived by a generally known method such as regularization, decision tree, ensemble learning, neural networks, and Bayesian networks”.
* US 20200143528 A1 ¶ [0051] 1st sentence: “It is further noted herein that the machine learning classifier generated in step 202 may include any type of machine learning algorithm/classifier and/or deep learning technique or classifier known in the art including, but not limited to, a random forest classifier, a support vector machine (SVM) classifier, an ensemble learning classifier, an artificial neural network (ANN), and the like”
With respect to the use of “Monte-Carlo” at dependent Claims 3,13, Examiner points to
* US 20090210246 A1 ¶ [0052] last two sentences: The means and the variances of the distribution are estimated using Markov-Chain Monte-Carlo methods common to empirical Bayes estimation. The inverse of this estimated variance matrix is used as the initial precision matrix Pn.
* US 20050222939 A1 ¶ [0017] second to last sentence: The method (an analytical approximative approach, i.e., a "closed form" nearby solution) described in embodiments herein may be able to quickly calculate NPV and related key figures in comparison to the time required to calculate similar values using common Monte Carlo based methods, while still providing comparable accuracy of the results.
With respect to the “linear combination of normalized probabilities with probabilities of occurrences of scores in the three or more machine learning models” at dependent Claims 21, 22, the Examiner points, in addition to the above findings to the preponderance of evidence demonstrated by at least the following:
* US 5479576 A column 8 lines 44-52: The form of the probability density function P(w; x,y) is in accordance with formula (3), and probability density function of formula (13) indicates the linear combination of probabilities of the normal distributions with respect to the input vector x and output vector y, in Fig.5.
* US 20200125820 A1 ¶ [0116] In Equation 4, only a linear combination of a normalized first component vector v1 and a normalized second component vector v2 will be described for convenience of description. ¶ [0117] In Equation 4, v is the embedding vector 550 in which the weighted component vectors 521 are linearly combined.
* US 20180172879 A1 ¶ [0208] In a particular implementation of the invention, the probability relationship is a normal relationship or a linear combination of normal relationships.
When tested per MPEP 2106.05(d), all of these fail to provide anything significantly more than what was already identified as the abstract exception.
In conclusion, Claims 1-3,6-7,11-13,16-17 and 20-29 although directed to statutory categories (here “non-transitory processor-readable medium” at Claims 1-3,6-7,21,23,26,28; “method” or process at Claims 11-13,16-17,22,24,27,29 “apparatus” or machine at Claims 20,25) they still recite, or at least set forth the abstract idea (Step 2A prong one), with their additional, computer-based elements not integrating the abstract idea into a practical application (Step 2A prong two) or providing significantly more than abstract idea (Step 2B).
Claims 1-3,6-7,11-13,16-17 and 20-29 are thus patent ineligible.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Allowability notice
- reasons for overcoming the prior art -
Examiner applies similar findings and rationales as those at Non-Final Act 10/22/2025 p.25-p.26 ¶1, Final Act 11/05/2024 p.8 ¶3-p.9 ¶6 and Non-Final Act 09/06/2024 p.21- p.34 ¶1 to submit that the closest prior art on record remains:
Xu et al, US 20200027157 A1 hereinafter Xu, in view of
Gottin et al, US 20190303799 A1 hereinafter Gottin.
Melzer; Glenn Ward US 20180247325 A1 hereinafter Melzer.
Godfrey; Luke US 11334790 B1 hereinafter Godfrey.
Watanabe et al US 5479576 A, hereinafter Watanabe
Moreover, with respect to the division features of independent Claims 1,11,20, the Examiner clarifies while Xu ¶ [0073] goes so far to recite: “Additionally, the acceptance rate is defined as rate acc= nA/N which is the proportion of accepted leads over all leads in the dataset. In one or more embodiments, the acceptance rate is limited by the available resources for engagement. Furthermore, rA represents marketing qualified leads (“MQL”) and rR represents inquiries rate (“INQ”); neither Xu nor any of the prior art on record teaches either alone or together, with adequate rationales, the combination of:
- I. “determining a probability of occurrence of each of the respective output scores for each of the three or more machine learning models, by dividing a number of leads in a sample associated with the respective output score, by a first total number of leads in the sample”;
- II. “determining a probability of conversion for each of the respective output scores by dividing a number of leads associated with a given one of the respective output scores from a particular one of the three or more machine learning models, which converts into an opportunity within a predefined time period, by a second total number of leads associated with the given one of the respective output scores”;
- III. “estimating an accuracy of each of the output scores based, at least in part, on historical performance of past corresponding lead scores”;
- IV. “selecting a plurality of the output scores from select two or more of the three or more machine learning models based, at least in part, on the each estimated accuracy of the output scores”;
- V. “selectively combining the plurality of select output scores to generate a composite score via a combining method using a sum of the probability of occurrences and the probability of conversions including”
- VI. “monitoring the composite score in real time as the composite score is generated, to determine that the composite score is less accurate based, at least in part, on historical data including previous composite score performance in predicting that a given lead results in a purchase” as recited at each of independent Claims 1,11,20.
Claims 2,3,6-7,21,23,26,28 are dependent and overcome the prior art based on parent claim 1.
Claims 12,13,16-17,22,24,27,29 are dependent and overcome prior art based on parent claim 11
Claim 25 is dependent and overcomes the prior art by dependency to parent Claim 20.
To be clear, novelty (35 USC 102) and non-obviousness (35 USC 103) still pertain to features that are mostly abstract that do not render the claims patent eligible (35 USC 101). Simply said the novel and non-obviousness rationale above do not necessarily render the claims patent eligible. See for example MPEP 2106.04 I ¶5, 3rd sentence citing Mayo, 566 U.S. 71, 101 USPQ2d at 1965); Flook, 437 U.S. at 591-92, 198 USPQ2d at 198 "the novelty of the mathematical algorithm is not a determining factor at all”.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Conclusion
The following art is made of record and considered pertinent to Applicant's disclosure:
Narayanan et al, Nearest Centroid and Normalized Neural Network based Lead Generation in Social Media Marketing, Revista geintec-gestao inovacao e tecnologias,11 n4, 2583-2602, Jul 22, 2021
EP 2991009 A1 entitled Inventory optimization tool
US 20030004777 A1 ¶ [0228] 4. After a longer period of time the low level of exploratory activity will compromise the ability of the regression model of the Targeted Presentation Sub-system to maintain accuracy (assuming that there are changes in the true response behaviour of visitors over time).
US 20190339950 A1 ¶ [0051] the neural network 260 is able to learn from existing task UI layouts created in the past (“training layouts” 308) by developers 250, for example, or from newly created task UI layout implemented by a Workflow developer (online learning). In one or more embodiments, after the neural network 260 has learned from training layouts 308, the neural network 260 may be considered “trained”. Over time, the neural network 260 learns continuously which properties are semantically similar due to an increasing amount of training data (e.g., every time a new task UI layout is input to the neural network 260, the neural network 260 learns new semantical groups and updates its weights, which leads then to updated property vectors accordingly). In one or more embodiments, the neural network 260 may calculate semantical similarities between properties which it may not have been explicitly trained with. For example, if property X is often placed nearby property Y in a task UI layout, and property Z is often placed nearby property Y, the neural network 260 may detect a semantical similarity between property X and Z.
US 20230123300 A1 mid-¶[0103] Each of the plurality of neural network models is trained using k-fold cross validation, resulting in a score that predicts the skill of each model in extracting the set of concepts that capture the associations and relations present in the set of words and phrases in unseen (future) data.
US 20220197306 A1 ¶ [1539] For every input in a training dataset, the output of the artificial neural network may be observed and compared with the expected output, and the error between the expected output and the observed output may be propagated back to the previous layer.
US 20180247325 A1 Determining valuation information for a package of multiple components
US 20160063419 A1 Inventory optimization tool reciting at ¶ [0007]: determining, based on the inventory data, probability distributions for each of lead time and demand for the one or more inventories to be simulated; determining a lead time demand probability distribution of the one or more inventories to be simulated based on the determined probability distributions for the lead time and the demand; determining a predictive state of the one or more inventories to be simulated based on the lead time demand probability distribution; and outputting one or more evaluation parameters associated with the predictive state of the one or more inventories to be simulated.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). 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 OCTAVIAN ROTARU whose telephone number is (571)270-7950. The examiner can normally be reached on 571.270.7950 from 9AM to 6PM. If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, PATRICIA H MUNSON, can be reached at telephone number (571)270-5396. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300. Information regarding the status of an application may be obtained from Patent Center. Status information for published applications may be obtained from Patent Center. Status information for unpublished applications is available through Patent Center for authorized users only. Should you have questions about access to Patent Center, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). 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) Form at https://www.uspto.gov/patents/uspto-automated- interview-request-air-form.
/Octavian Rotaru/
Primary Examiner, Art Unit 3624 A
February 21st, 2026
1 Brandan Artley, Training a Neural Network by Hand, towardsdatascience webpages, Jun 23, 2022
2 OIP Techs., Inc. v. Amazon.com, Inc., 788 F.3d 1359, 1364, 115 U.S.P.Q.2d 1090, 1092 (Fed Cir. 2015) (a new method of price optimization was found to be a fundamental economic concept);
3 CyberSource v. Retail Decisions, 654 F.3d 1366, 1372 n.2, 99 USPQ2d 1690, 1695 n.2 (Fed. Cir. 2011) (quoting In re Warmerdam, 33 F.3d 1354, 1355, 1360, 31 USPQ2d 1754, 1755, 1759 (Fed. Cir. 1994)).
4 MPEP 2106.04(a): “examiners should identify at least one abstract idea grouping, but preferably identify all groupings to the extent possible”.
5 Electric Power Group v. Alstom, S.A., 830 F.3d 1350, 1353-54, 119 USPQ2d 1739, 1741-42 (Fed. Cir. 2016)
6 Electric Power Group v. Alstom, S.A., 830 F.3d 1350, 1353-54, 119 USPQ2d 1739, 1741-42 (Fed. Cir. 2016)
7 Diehr, 450 U.S. at 178 n. 2, 179 n.5, 191-92, 209 USPQ at 4-5 (1981)
8 SAP America, Inc. v. Investpic, LLC, 898 F.3d 1161, 1163-65, 127 USPQ2d 1597, 1598-1600 (Fed. Cir. 2018), modifying SAP America, Inc. v. Investpic, LLC, 890 F.3d 1016, 126 USPQ2d 1638 (Fed. Cir. 2018);
9 SAP America, Inc. v. Investpic, LLC, 898 F.3d 1161, 1163-65, 127 USPQ2d 1597, 1598-1600 (Fed. Cir. 2018), modifying SAP America, Inc. v. Investpic, LLC, 890 F.3d 1016, 126 USPQ2d 1638 (Fed. Cir. 2018);
10 Diamond v. Diehr; 450 U.S. at 178 n. 2, 179 n.5, 191-92, 209 USPQ at 4-5 (1981);
11 Alice Corp. Pty. Ltd. V. CLS Bank Int’l, 573 U.S. 208, 223, 110 USPQ2d 1976, 1983 (2014); Gottschalk v. Benson, 409 U.S. 63, 64, 175 USPQ 673, 674 (1972); Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015);
12 Affinity Labs v. DirecTV, 838 F.3d 1253, 1262, 120 USPQ2d 1201, 1207 (Fed. Cir. 2016);
TLI Communications LLC v. AV Auto, LLC, 823 F.3d 607, 613, 118 USPQ2d 1744, 1748 (Fed. Cir. 2016),
Intellectual Ventures I LLC v. Capital One Bank (USA), 792 F.3d 1363, 1367, 115 USPQ2d 1636, 1639 (Fed. Cir. 2015)
13 Intellectual Ventures I LLC v. Capital One Bank (USA), 792 F.3d 1363, 1370-71, 115 USPQ2d 1636, 1642 (Fed. Cir. 2015);
14 FairWarning IP, LLC v. Iatric Sys., 839 F.3d 1089, 1095, 120 USPQ2d 1293, 1296 (Fed. Cir. 2016)
15 FairWarning v. Iatric Sys., 839 F.3d 1089, 1094-95, 120 USPQ2d 1293, 1295 (Fed. Cir. 2016);
16 Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016);
17 FairWarning v. Iatric Sys., 839 F.3d 1089, 1094-95, 120 USPQ2d 1293, 1295 (Fed. Cir. 2016);
18 Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016);
19 OIP Techs., 788 F.3d at 1362-63, 115 USPQ2d at 1092-93
20 Content Extraction and Transmission, LLC v. Wells Fargo Bank, 776 F.3d 1343, 1348, 113 USPQ2d 1354, 1358 (Fed. Cir. 2014)
21 Alice Corp. Pty. Ltd. v. CLS Bank Int'l, 573 U.S. 208, 225, 110 USPQ2d 1984 (2014)
Ultramercial, 772 F.3d at 716, 112 USPQ2d at 1755
22 Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015); OIP Techs., 788 F.3d at 1363, 115 USPQ2d at 1092-93
23 Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362
24 Flook, 437 U.S. at 594, 198 USPQ2d at 199 (recomputing or readjusting alarm limit values);
Bancorp Services v. Sun Life, 687 F.3d 1266, 1278, 103 USPQ2d 1425, 1433 (Fed. Cir. 2012)
25 Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1331, 115 USPQ2d 1681, 1699 (Fed. Cir. 2015)