DETAILED ACTION
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 .
Double Patenting
The nonstatutory double patenting rejection is based on a judicially created doctrine grounded in public policy (a policy reflected in the statute) so as to prevent the unjustified or improper timewise extension of the “right to exclude” granted by a patent and to prevent possible harassment by multiple assignees. A nonstatutory double patenting rejection is appropriate where the conflicting claims are not identical, but at least one examined application claim is not patentably distinct from the reference claim(s) because the examined application claim is either anticipated by, or would have been obvious over, the reference claim(s). See, e.g., In re Berg, 140 F.3d 1428, 46 USPQ2d 1226 (Fed. Cir. 1998); In re Goodman, 11 F.3d 1046, 29 USPQ2d 2010 (Fed. Cir. 1993); In re Longi, 759 F.2d 887, 225 USPQ 645 (Fed. Cir. 1985); In re Van Ornum, 686 F.2d 937, 214 USPQ 761 (CCPA 1982); In re Vogel, 422 F.2d 438, 164 USPQ 619 (CCPA 1970); In re Thorington, 418 F.2d 528, 163 USPQ 644 (CCPA 1969).
A timely filed terminal disclaimer in compliance with 37 CFR 1.321(c) or 1.321(d) may be used to overcome an actual or provisional rejection based on nonstatutory double patenting provided the reference application or patent either is shown to be commonly owned with the examined application, or claims an invention made as a result of activities undertaken within the scope of a joint research agreement. See MPEP § 717.02 for applications subject to examination under the first inventor to file provisions of the AIA as explained in MPEP § 2159. See MPEP § 2146 et seq. for applications not subject to examination under the first inventor to file provisions of the AIA . A terminal disclaimer must be signed in compliance with 37 CFR 1.321(b).
The filing of a terminal disclaimer by itself is not a complete reply to a nonstatutory double patenting (NSDP) rejection. A complete reply requires that the terminal disclaimer be accompanied by a reply requesting reconsideration of the prior Office action. Even where the NSDP rejection is provisional the reply must be complete. See MPEP § 804, subsection I.B.1. For a reply to a non-final Office action, see 37 CFR 1.111(a). For a reply to final Office action, see 37 CFR 1.113(c). A request for reconsideration while not provided for in 37 CFR 1.113(c) may be filed after final for consideration. See MPEP §§ 706.07(e) and 714.13.
The USPTO Internet website contains terminal disclaimer forms which may be used. Please visit www.uspto.gov/patent/patents-forms. The actual filing date of the application in which the form is filed determines what form (e.g., PTO/SB/25, PTO/SB/26, PTO/AIA /25, or PTO/AIA /26) should be used. A web-based eTerminal Disclaimer may be filled out completely online using web-screens. An eTerminal Disclaimer that meets all requirements is auto-processed and approved immediately upon submission. For more information about eTerminal Disclaimers, refer to www.uspto.gov/patents/apply/applying-online/eterminal-disclaimer.
Claims 1-2, 4-5, 7-9, 11-12, 14-16 and 18-19 are rejected on the ground of nonstatutory double patenting as being unpatentable over claims 1, 5-7 and 9-10 of co-pending application no. 18/449,426. Although the claims at issue are not identical, they are not patentably distinct from each other because claims 1-2, 4-5, 7-9, 11-12, 14-16 and 18-19 under examination are anticipated, respectively, by claims 1, 5-7 and 9-10 of the co-pending application no. 18/449,426.
In regards to claim 1 and analogous claims 8 and 15,
Examiner notes the differences between the claims are highlighted below by bolding all limitations that differ, italicizing additional limitations, and underlining limitations that will be addressed below.
Instant Application 18/506,017
Co-pending Application 18/449,426
1. A method comprising: executing a machine learning model on at least one computer comprising a processor and a memory;
providing a data transformation module of the machine learning model, wherein the data transformation module accepts an original dataset as input to a first neural network and a second neural network and outputs a transformed dataset;
providing a sensitive attribute suppression module of the machine learning model, wherein the sensitive attribute suppression module accepts the transformed dataset as input to a third neural network and calculates, for each attribute of a plurality of annotated sensitive attributes, a sensitive attribute suppression loss;
providing an annotated useful attribute preservation module of the machine learning model, wherein the annotated useful attribute preservation module accepts the transformed dataset as input to a fourth neural network, and calculates, for each attribute of a plurality of annotated useful attributes, a useful attribute preservation loss;
providing a generic feature suppression module of the machine learning model that accepts parameters of a distribution of a latent variable from the first neural network and calculates, for an unannotated generic attribute, a generic feature suppression loss;
combining the sensitive attribute suppression loss, the useful attribute preservation loss, and the generic feature suppression loss into a total loss;
and training the first neural network and the second neural network with the total loss.
1. A method comprising: executing a machine learning model on at least one computer comprising a processor and a memory;
providing a data transformation module of the machine learning model, wherein the data transformation module accepts a raw dataset as input to a neural network 0, and wherein the neural network 0 outputs a transformed dataset;
providing a sensitive attribute suppression module of the machine learning model, wherein the sensitive attribute suppression module accepts the raw dataset as input to a neural network #, accepts the transformed dataset as input to a neural network #', and calculates, for each attribute of a plurality of annotated sensitive attributes S, a sensitive attribute suppression loss;
providing an annotated useful attribute preservation module of the machine learning model, wherein the annotated useful attribute preservation module accepts the raw dataset as input to a neural network l/J, accepts the transformed dataset as input to a neural network l/J', and calculates, for each attribute of a plurality of annotated useful attributes U, an annotated useful attribute preservation loss;
providing an unannotated useful attribute preservation module of the machine learning model, wherein the unannotated useful attribute preservation module accepts the transformed dataset and the raw dataset as input to a neural network n, and calculates, for an unannotated useful attribute F, an unannotated useful attribute preservation loss;
combining the sensitive attribute suppression loss, the annotated useful attribute preservation loss, and the unannotated useful attribute preservation loss into a total loss; and
training the neural network 0 and the neural network n using the total loss.
As shown in the mapping above, claim 1 of the reference application includes most the limitations of claim 1 of the instant application, while also reciting further limitations.
Claim 1 of the instant application differs from claim 1 of the reference patent in that it recites:
“providing a generic feature suppression module of the machine learning model that accepts parameters of a distribution of a latent variable from the first neural network and calculates, for an unannotated generic attribute, a generic feature suppression loss.”
Wherein claim 1 of the reference application recites:
“providing an unannotated useful attribute preservation module of the machine learning model, wherein the unannotated useful attribute preservation module accepts the transformed dataset and the raw dataset as input to a neural network n, and calculates, for an unannotated useful attribute F, an unannotated useful attribute preservation loss.”
The BRI of the respective modules is merely software. Further, the instant disclosure at paragraph [0008] shows that a “generic feature suppression loss” falls within the broadest reasonable interpretation of the category encompassed by “loss;” as “the generic feature suppression loss may be a estimation of an upper bound of mutual information between the generic feature and the transformed dataset.” Examiner determines the recitation of “accepts parameters of a distribution of a latent variable from the first neural network” in claim 1 on the instant application to be analogous to the transformed dataset wherein the transformed dataset was output by the neural network o of the copending application “wherein the neural network 0 outputs a transformed dataset” and claim 1 of the instant application previously recited “original dataset as input to a first neural network and a second neural network and outputs a transformed dataset.” Lastly, the additional neural networks accepting raw datasets in the copending application reads on “accepts an original dataset as input to… a second neural network” in claim 1 of the instant application.
Claims 8 (machine) and 15 (manufacture) are rejected on the same grounds under nonstatutory double patenting as claim 1 as they are substantially similar, respectively, Mutatis mutandis.
Regarding claim 2 and analogous claims 9 and 16, which depends upon claim 1:
Claim 2 of the present application corresponds to claim 7 of the copending application, as both recite the same invention. As such this claim is rejected as nonstatutory double patenting.
Examiner notes the differences between the claims are highlighted below by bolding all limitations that differ, italicizing additional limitations, and underlining limitations that will be addressed below.
The recitation of “trained… at a same time as the trainings of the neural network 0 and the neural network n” in claim 7 of the co-pending application reads on “trained jointly with the training of the first neural network and the second neural network” of claim 2 of the instant application.
The recitation of “trained… using the total loss” of claim 7 of the co-pending application reads on “trained… using the sensitive attribute suppression loss” of claim 2 of the instant application as the total loss comprises of the “sensitive attribute suppression loss;” thus, using it.
Instant Application 18/506,017
Co-pending Application 18/449,426
2. The method of claim 1,
wherein the third neural network is trained jointly with the training of the first neural network and the second neural network using the sensitive attribute suppression loss, and wherein the third neural network is trained using supervised learning.
7. The method of claim 1,
wherein the neural network l/J' is trained using a traditional supervised learning method at a same time as the training of the neural network 0 and the neural network n, using the total loss.
Claims 9 (machine) and 16 (manufacture) are rejected on the same grounds under nonstatutory double patenting as claim 2 as they are substantially similar, respectively, Mutatis mutandis.
Regarding claim 4 and analogous claims 11 and 18, which depends upon claim 1:
Claim 4 of the present application corresponds to claim 9 of the copending application, as both recite the same invention. As such this claim is rejected as nonstatutory double patenting.
Examiner notes the differences between the claims are highlighted below by bolding all limitations that differ, italicizing additional limitations, and underlining limitations that will be addressed below.
Instant Application 18/506,017
Co-pending Application 18/449,426
4. The method of claim 1,
wherein the sensitive attribute suppression loss is a constraint to an estimation of mutual information between each attribute of the plurality of annotated sensitive attributes and the transformed dataset.
9. The method of claim 1,
wherein the sensitive attribute suppression loss is a constraint to an estimation of mutual information between each attribute of the plurality of annotated sensitive attributes S and the transformed dataset.
Claims 11 (machine) and 18 (manufacture) are rejected on the same grounds under nonstatutory double patenting as claim 4 as they are substantially similar, respectively, Mutatis mutandis.
Regarding claim 5 and analogous claims 12 and 19, which depends upon claim 1:
Claim 5 of the present application corresponds to claim 10 of the copending application, as both recite the same invention. As such this claim is rejected as nonstatutory double patenting.
Examiner notes the differences between the claims are highlighted below by bolding all limitations that differ, italicizing additional limitations, and underlining limitations that will be addressed below.
Instant Application 18/506,017
Co-pending Application 18/449,426
5. The method of claim 1,
wherein the useful attribute preservation loss is a constraint to an estimation of mutual information between each attribute of a plurality of annotated useful attributes and the transformed dataset.
10. The method of claim 1, wherein the annotated useful attribute preservation loss is a constraint to an estimation of mutual information between each attribute of a plurality of annotated useful attributes U and the transformed dataset.
Claims 12 (machine) and 19 (manufacture) are rejected on the same grounds under nonstatutory double patenting as claim 5 as they are substantially similar, respectively, Mutatis mutandis.
Regarding claim 7 and analogous claim 14, which depends upon claim 1:
Claim 7 of the present application corresponds to claims 5 and 6 of the copending application, as both recite the same invention. As such this claim is rejected as nonstatutory double patenting.
Examiner notes the differences between the claims are highlighted below by bolding all limitations that differ, italicizing additional limitations, and underlining limitations that will be addressed below.
The recitations of “wherein the neural network l/J is trained prior… wherein the neural network l/J is fixed…” of claims 5 and 6 in the co-pending application reads on claim 7 of the instant application.
Instant Application 18/506,017
Co-pending Application 18/449,426
7. The method of claim 1,
wherein the fourth neural network is fixed after it is initialized.
5. The method of claim 1,
wherein the neural network l/J is trained prior to the training of the neural network 0 and the neural network rJ using the total loss, and wherein the neural network l/J is trained using a traditional supervised
learning method.
6. The method of claim 5,
wherein the neural network l/J is fixed during the training of the neural network 0 and the neural network n using the total loss.
Claims 14 (machine) is rejected on the same grounds under nonstatutory double patenting as claim 7 as they are substantially similar, respectively, Mutatis mutandis.
Claims 6, 13 and 20 are rejected on the ground of nonstatutory obviousness-type double patenting as being unpatentable over claims 11 of co-pending application no. 18/449,426 in view of Bertran, Martin, et al. "Adversarially learned representations for information obfuscation and inference." International Conference on Machine Learning. PMLR, 2019. (“Bertran”). The claims of the instant application and the claims of the reference patent are compared in the table below.
Regarding claim 6 and analogous claims 13 and 20, which depends upon claim 1:
Examiner notes the differences between the claims are highlighted below by bolding all limitations that differ, italicizing additional limitations, and underlining limitations that will be addressed below.
Instant Application 18/506,017
Co-pending Application 18/449,426
6. The method of claim 1,
wherein the generic feature suppression loss is an estimation of an upper bound of mutual information between the generic feature and the transformed dataset.
11. The method of claim 1,
wherein the unannotated useful attribute preservation loss is an estimation of mutual information between the unannotated useful attribute F and the transformed dataset.
Claim 11 of the co-pending application recites all of the limitations of claim 6 of the instant application except “an upper bound of mutual information.” However, Bertran teaches an upper bound of mutual information (Bertran, Section 2.1, “Lemma 2.1, presented next, shows that we can bound [wherein the generic feature suppression loss is an estimation of an upper bound of mutual information] the solution of Eq.(2) by considering mappings that go directly from the latent variables U and S to the obfuscated variable Y . This simplifies the analysis since ∣U ×S∣ ≪∣X∣ for many problems of interest.
PNG
media_image1.png
366
396
media_image1.png
Greyscale
”)
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to provide a loss as an estimation of an upper bound of mutual information between a feature and a dataset, as disclosed in Bertran, as doing so simplifies the analysis for many problems of interest (Bertran, Section 2.1, “This simplifies the analysis since ∣U ×S∣ ≪∣X∣ for many problems of interest.”)
Claims 13 (machine) and 20 (manufacture) are rejected on the same grounds under nonstatutory double patenting as claim 6 as they are substantially similar, respectively, Mutatis mutandis.
Claim Objections
Claim 20 is objected to because of the following informalities: “between the generic feature 'and the transformed dataset” should read as “between the generic feature [[']]and the transformed dataset.” Appropriate correction is 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 1-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 1 and analogous claims 8 and 15 recite “wherein the data transformation module accepts an original dataset as input to a first neural network and a second neural network and outputs a transformed dataset;” it is unclear if the first and second neural networks are a part of the data transformation module or if the data transformation module is merely transmitting the original dataset to the first and second neural networks. Further, it is unclear if the data transformation module or the first and second neural networks are outputting the transformed dataset. Similarly, the claim and analogous claims recite “wherein the sensitive attribute suppression module accepts the transformed dataset as input to a third neural network and calculates… wherein the annotated useful attribute preservation module accepts the transformed dataset as input to a fourth neural network, and calculates…” It is unclear if the respective neural networks are a part of the respective modules or if the modules are merely transmitting the data to the respective neural networks. Further, it is unclear if the modules or the neural networks are performing the calculating steps.
Claims 2-7 are further rejected on virtue of their dependencies to claim 1.
Claims 9-14 are further rejected on virtue of their dependencies to claim 8.
Claims 17-20 are further rejected on virtue of their dependencies to claim 20.
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 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
In regards to claim 1,
Step 1: This part of the eligibility analysis evaluates whether the claim(s) falls within any statutory category. See MPEP 2106.03:
The claim directs to a statutory category – process.
Step 2A – Prong 1: Judicial Exception Recited?
MPEP 2106.04(a)(2)(II) “The mathematical concepts grouping is defined as mathematical relationships, mathematical formulas or equations, and mathematical calculations.”
Further, the MPEP recites “It is important to note that a mathematical concept need not be expressed in mathematical symbols, because "[w]ords used in a claim operating on data to solve a problem can serve the same purpose as a formula." In re Grams, 888 F.2d 835, 837 and n.1, 12 USPQ2d 1824, 1826 and n.1 (Fed. Cir. 1989). See, e.g., SAP America, Inc. v. InvestPic, LLC, 898 F.3d 1161, 1163, 127 USPQ2d 1597, 1599 (Fed. Cir. 2018) (holding that claims to a ‘‘series of mathematical calculations based on selected information’’ are directed to abstract ideas); Digitech Image Techs., LLC v. Elecs. for Imaging, Inc., 758 F.3d 1344, 1350, 111 USPQ2d 1717, 1721 (Fed. Cir. 2014) (holding that claims to a ‘‘process of organizing information through mathematical correlations’’ are directed to an abstract idea); and Bancorp Servs., LLC v. Sun Life Assurance Co. of Can. (U.S.), 687 F.3d 1266, 1280, 103 USPQ2d 1425, 1434 (Fed. Cir. 2012) (identifying the concept of ‘‘managing a stable value protected life insurance policy by performing calculations and manipulating the results’’ as an abstract idea).”
Yes, the claim recites a mathematical concept, specifically:
and calculates, for each attribute of a plurality of annotated sensitive attributes, a sensitive attribute suppression loss;
and calculates, for each attribute of a plurality of annotated useful attributes, a useful attribute preservation loss;
and calculates, for an unannotated generic attribute, a generic feature suppression loss;
combining the sensitive attribute suppression loss, the useful attribute preservation loss, and the generic feature suppression loss into a total loss;
These limitations encompass calculating loss values for respective modules wherein each loss value will subsequently be summed.
Therefore, the claim recites a mathematical concept.
Step 2A – Prong 2: Integrated into a Practical Solution?
MPEP 2106.05(f) Mere Instructions To Apply An Exception has found simply adding a general purpose computer or computer components after the fact to an abstract idea (e.g., a fundamental economic practice or mathematical equation) does not integrate a judicial exception into a practical application or provide significantly more. The following steps are mere instructions to apply:
executing a machine learning model on at least one computer comprising a processor and a memory; (mere instructions to implement the abstract idea at a high level of generality using a generic computer)
providing a data transformation module of the machine learning model…
providing a sensitive attribute suppression module of the machine learning model…
providing an annotated useful attribute preservation module of the machine learning model…
providing a generic feature suppression module of the machine learning model… (BRI of the respective modules is software executed on the generic computer)
a first neural network and a second neural network…
a third neural network
a fourth neural network
and training the first neural network and the second neural network with the total loss
MPEP 2106.05(g) Insignificant Extra-Solution Activity has found mere data gathering to be insignificant extra-solution activity. The following steps are insignificant extra-solution activities:
Mere Data Gathering:
wherein the data transformation module accepts an original dataset as input to a first neural network and a second neural network and outputs a transformed dataset; (This limitation encompasses providing a software (ie data transformation module) wherein the software obtains a dataset, provides the dataset to a first and second neural network and outputs a transformed dataset)
wherein the sensitive attribute suppression module accepts the transformed dataset as input to a third neural network…
wherein the annotated useful attribute preservation module accepts the transformed dataset as input to a fourth neural network… (These limitations encompass providing a software wherein the software obtains a transformed dataset, provides the transformed dataset to a neural network wherein the neural network does not provide any outputs)
a generic feature suppression module of the machine learning model that accepts parameters of a distribution of a latent variable from the first neural network (This limitation encompasses software that obtains data from the first neural network)
The additional elements have been considered both individually and as an ordered combination in to determine whether they integrate the exception into a practical application.
The full context of the claim encompasses executing a machine learning model on a generic computer wherein the machine learning model comprises of software that obtains a dataset to input or manipulate for transmitting to a plurality of neural networks, calculating a plurality of loss values using the software to subsequently calculate a total loss, and training particular neural networks with the total loss wherein the training is recited at the highest level of generality.
Therefore, no meaningful limits are imposed on practicing the abstract idea.
The claim is directed to the abstract idea.
Step 2B: Claim provides an Inventive Concept?
No, as discussed with respect to Step 2A, the additional limitation is mere data gathering Insignificant Extra-Solution Activity) and a generic device do not impose any meaningful limits on practicing the abstract idea and therefore the claim does not provide an inventive concept in Step 2B.
The claim recites receiving data by generic device.
This has been determined to be insignificant extra-solution activity as found in MPEP § 2106.05(d)(II)(i): Receiving or transmitting data over a network, e.g., using the Internet to
gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary
computer to forward information); TLI Communications LLC v. AV Auto. LLC, 823 F.3d 607,
610, 118 USPQ2d 1744, 1745 (Fed. Cir. 2016) (using a telephone for image transmission); OIP
Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015)
(sending messages over a network); buy SAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112
USPQ2d 1093, 1096 (Fed. Cir. 2014) (computer receives and sends information over a network);
but see DDR Holdings, LLC v. Hotels.com, L.P., 773 F.3d 1245, 1258, 113 USPQ2d 1097, 1106
(Fed. Cir. 2014) ("Unlike the claims in Ultramercial, the claims at issue here specify how
interactions with the Internet are manipulated to yield a desired result‐‐a result that overrides
the routine and conventional sequence of events ordinarily triggered by the click of a hyperlink."
(emphasis added)).
The additional elements have been considered both individually and as an ordered
combination in the significantly more consideration.
The claim is ineligible.
In regards to claim 2,
Step 1: This part of the eligibility analysis evaluates whether the claim(s) falls within any statutory category. See MPEP 2106.03:
The claim directs to a statutory category – process.
Step 2A Prong 1: The claim recites the following abstract ideas:
The abstract idea(s) in the parent claim(s).
Step 2A Prong 2: The claim recites the following additional elements which, considered individually and as an ordered combination, do not integrate the abstract idea into a practical application:
The additional element(s) in the parent claim(s).
wherein the third neural network is trained jointly with the training of the first neural network and the second neural network using the sensitive attribute suppression loss, and wherein the third neural network is trained using supervised learning
This limitation merely indicates a field of use or technological environment in which the judicial exception is performed. This type of limitation merely confines the use of the abstract idea to a particular technological environment and thus fails to add an inventive concept to the claims. See MPEP § 2106.05(h)
Step 2B: The claim recites the following additional elements which, considered individually and as an ordered combination, do not amount to significantly more than the abstract idea:
The additional element(s) in the parent claim(s).
wherein the third neural network is trained jointly with the training of the first neural network and the second neural network using the sensitive attribute suppression loss, and wherein the third neural network is trained using supervised learning
This limitation merely indicates a field of use or technological environment in which the judicial exception is performed. This type of limitation merely confines the use of the abstract idea to a particular technological environment and thus fails to add an inventive concept to the claims. See MPEP § 2106.05(h)
In regards to claim 3,
Step 1: This part of the eligibility analysis evaluates whether the claim(s) falls within any statutory category. See MPEP 2106.03:
The claim directs to a statutory category – process.
Step 2A Prong 1: The claim recites the following abstract ideas:
The abstract idea(s) in the parent claim(s).
wherein the first neural network is trained using gradient descent
This limitation directs to a mathematical calculation. See MPEP 2106.04(a)(2)(I)(C.)
Step 2A Prong 2: The claim recites the following additional elements which, considered individually and as an ordered combination, do not integrate the abstract idea into a practical application:
The additional element(s) in the parent claim(s).
Step 2B: The claim recites the following additional elements which, considered individually and as an ordered combination, do not amount to significantly more than the abstract idea:
The additional element(s) in the parent claim(s).
In regards to claim 4,
Step 1: This part of the eligibility analysis evaluates whether the claim(s) falls within any statutory category. See MPEP 2106.03:
The claim directs to a statutory category – process.
Step 2A Prong 1: The claim recites the following abstract ideas:
The abstract idea(s) in the parent claim(s).
wherein the sensitive attribute suppression loss is a constraint to an estimation of mutual information between each attribute of the plurality of annotated sensitive attributes and the transformed dataset
This limitation directs to a mathematical relationship. See MPEP 2106.04(a)(2)(I)(A.) i. a relationship between reaction rate and temperature, which relationship can be expressed in the form of a formula called the Arrhenius equation, Diamond v. Diehr; 450 U.S. at 178 n. 2, 179 n.5, 191-92, 209 USPQ at 4-5 (1981);
Step 2A Prong 2: The claim recites the following additional elements which, considered individually and as an ordered combination, do not integrate the abstract idea into a practical application:
The additional element(s) in the parent claim(s).
Step 2B: The claim recites the following additional elements which, considered individually and as an ordered combination, do not amount to significantly more than the abstract idea:
The additional element(s) in the parent claim(s).
In regards to claim 5,
Step 1: This part of the eligibility analysis evaluates whether the claim(s) falls within any statutory category. See MPEP 2106.03:
The claim directs to a statutory category – process.
Step 2A Prong 1: The claim recites the following abstract ideas:
The abstract idea(s) in the parent claim(s).
wherein the useful attribute preservation loss is a constraint to an estimation of mutual information between each attribute of a plurality of annotated useful attributes and the transformed dataset
This limitation directs to a mathematical relationship. See MPEP 2106.04(a)(2)(I)(A.) i. a relationship between reaction rate and temperature, which relationship can be expressed in the form of a formula called the Arrhenius equation, Diamond v. Diehr; 450 U.S. at 178 n. 2, 179 n.5, 191-92, 209 USPQ at 4-5 (1981);
Step 2A Prong 2: The claim recites the following additional elements which, considered individually and as an ordered combination, do not integrate the abstract idea into a practical application:
The additional element(s) in the parent claim(s).
Step 2B: The claim recites the following additional elements which, considered individually and as an ordered combination, do not amount to significantly more than the abstract idea:
The additional element(s) in the parent claim(s).
In regards to claim 6,
Step 1: This part of the eligibility analysis evaluates whether the claim(s) falls within any statutory category. See MPEP 2106.03:
The claim directs to a statutory category – process.
Step 2A Prong 1: The claim recites the following abstract ideas:
The abstract idea(s) in the parent claim(s).
wherein the generic feature suppression loss is an estimation of an upper bound of mutual information between the generic feature and the transformed dataset
This limitation directs to a mathematical relationship. See MPEP 2106.04(a)(2)(I)(A.) i. a relationship between reaction rate and temperature, which relationship can be expressed in the form of a formula called the Arrhenius equation, Diamond v. Diehr; 450 U.S. at 178 n. 2, 179 n.5, 191-92, 209 USPQ at 4-5 (1981);
Step 2A Prong 2: The claim recites the following additional elements which, considered individually and as an ordered combination, do not integrate the abstract idea into a practical application:
The additional element(s) in the parent claim(s).
Step 2B: The claim recites the following additional elements which, considered individually and as an ordered combination, do not amount to significantly more than the abstract idea:
The additional element(s) in the parent claim(s).
In regards to claim 7,
Step 1: This part of the eligibility analysis evaluates whether the claim(s) falls within any statutory category. See MPEP 2106.03:
The claim directs to a statutory category – process.
Step 2A Prong 1: The claim recites the following abstract ideas:
The abstract idea(s) in the parent claim(s).
Step 2A Prong 2: The claim recites the following additional elements which, considered individually and as an ordered combination, do not integrate the abstract idea into a practical application:
The additional element(s) in the parent claim(s).
wherein the fourth neural network is fixed after it is initialized
This limitation directs to merely applying (or equivalent) an abstract idea, or implementing an abstract idea on a computer, or using a computer as a tool to perform an abstract idea. See MPEP 2106.05(f)
Step 2B: The claim recites the following additional elements which, considered individually and as an ordered combination, do not amount to significantly more than the abstract idea:
The additional element(s) in the parent claim(s).
wherein the fourth neural network is fixed after it is initialized
This limitation directs to merely applying (or equivalent) an abstract idea, or implementing an abstract idea on a computer, or using a computer as a tool to perform an abstract idea. See MPEP 2106.05(f)
Claims 8 (machine) and 15 (manufacture) are rejected on the same grounds under 35 U.S.C. 101 as claim 1 as they are substantially similar, respectively, Mutatis mutandis.
Claims 9 (machine) and 16 (manufacture) are rejected on the same grounds under 35 U.S.C. 101 as claim 2 as they are substantially similar, respectively, Mutatis mutandis.
Claims 10 (machine) and 17 (manufacture) are rejected on the same grounds under 35 U.S.C. 101 as claim 3 as they are substantially similar, respectively, Mutatis mutandis.
Claims 11 (machine) and 18 (manufacture) are rejected on the same grounds under 35 U.S.C. 101 as claim 4 as they are substantially similar, respectively, Mutatis mutandis.
Claims 12 (machine) and 19 (manufacture) are rejected on the same grounds under 35 U.S.C. 101 as claim 5 as they are substantially similar, respectively, Mutatis mutandis.
Claims 13 (machine) and 20 (manufacture) are rejected on the same grounds under 35 U.S.C. 101 as claim 6 as they are substantially similar, respectively, Mutatis mutandis.
Claim 14 (machine) is rejected on the same grounds under 35 U.S.C. 101 as claim 7 as they are substantially similar, respectively, Mutatis mutandis.
Claim Rejections - 35 USC § 102
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 (i.e., changing from AIA to pre-AIA ) 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.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claim(s) 1-3, 8-10, 7 and 14-17 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by Chen, Chun-Fu, et al. "Mass: Multi-attribute selective suppression." arXiv preprint arXiv:2210.09904 (24 Oct 2022). (“Chen”)
In regards to claim 1 and analogous claims 8 and 15,
Chen teaches A method comprising: executing a machine learning model on at least one computer comprising a processor and a memory;
(Chen, Table 7 teaches performing model training with GPUs
PNG
media_image2.png
227
942
media_image2.png
Greyscale
)
Chen teaches providing a data transformation module of the machine learning model, wherein the data transformation module accepts an original dataset as input to a first neural network and a second neural network and outputs a transformed dataset;
(Chen, Section 3.2, “We propose a Generative Adversarial Network (GAN)-based solution in tackling the data transformation problem. Our framework consists of three major components: a Data Modifier [providing a data transformation module ie Data Modifier of the machine learning model], a Suppression Branch, and a Preservation Branch, as depicted in Fig. 2. The data modifier 𝐺 is the generator while both branches are the discriminators in the GAN framework. The data modifier tries to generate new data such that the similarity between original data and modified data are maximized and minimized via the suppression branch and the preservation branch, respectively [wherein the data transformation module accepts an original dataset ie Original data X as input to a first neural network and a second neural network ie models in the suppression branch (See Chen Section 3.2.2 and fig. 2 for the suppression branch receiving both the original data and the modified data) and outputs a transformed dataset ie Modified Data X’; see annotated figure 2]. In a nutshell, the data modifier learns a transformation that is to be applied to the original data vectors, where the learned transformation is jointly regularized by both the suppression and preservation branches to ensure all targeted attributes in 𝑆 are indeed suppressed in the transform while all other attributes, explicitly specified or not, are left intact as much as possible.”
PNG
media_image3.png
459
1407
media_image3.png
Greyscale
)
Chen teaches providing a sensitive attribute suppression module of the machine learning model, wherein the sensitive attribute suppression module accepts the transformed dataset as input to a third neural network and calculates, for each attribute of a plurality of annotated sensitive attributes, a sensitive attribute suppression loss;
(Chen, Section 3.2.2, “The suppression branch [providing a sensitive attribute suppression module of the machine learning model] is designed to make the targeted attributes in 𝑆 as unrecognizable as possible. It utilizes the corresponding set of inference models pretrained on the original dataset X. Each pretrained model corresponds to a specific attribute 𝑠 ∈ 𝑆 and is composed of a feature extractor 𝐹𝑠, which converts raw input data to a feature vector z = 𝐹𝑠 (x), and a projector 𝑃𝑠, which maps the features to the attribute’s label p = 𝑃𝑠 z , where z is the feature representation and p is the prediction logit.
During the training of the data modifier, the suppression branch guides the data modifier to degrade these pretrained models’ recognition accuracies on the transformed data X’ [wherein the sensitive attribute suppression module accepts the transformed dataset as input to a third neural network]. It does so by either measuring the similarity of features from the pretrained models or by comparing the prediction results against the ground truth labels, and then adding a corresponding penalty. Specifically, for each targeted attribute 𝑠 ∈ 𝑆, the feature-similarity loss function is defined as [and calculates, for each attribute of a plurality of annotated sensitive attributes, a sensitive attribute suppression loss]
PNG
media_image4.png
137
1061
media_image4.png
Greyscale
”)
Chen teaches providing an annotated useful attribute preservation module of the machine learning model, wherein the annotated useful attribute preservation module accepts the transformed dataset as input to a fourth neural network, and calculates, for each attribute of a plurality of annotated useful attributes, a useful attribute preservation loss;
(Chen, Section 3.2.3, “While the suppression branch guides the data modifier to decrease the confidence of machine learning models on certain attributes 𝑆 from the data, the preservation branch [providing an annotated useful attribute preservation module of the machine learning model] is responsible for guarding the data against said suppression and erasure such that maximum utility can be preserved through the transformation, in the sense that all the attributes 𝑅 = 𝐴 n 𝑆 [calculates, for each attribute of a plurality of annotated useful attributes] not targeted by the suppression branch should remain recognizable from the transformed data X’ [wherein the annotated useful attribute preservation module accepts the transformed dataset as input to a fourth neural network], just as they were from the original data X. In the proposed method, we design two types of losses: one is attribute-agnostic in the preservation branch, which is applicable to any attribute as the loss is defined in an agnostic way; another one is attribute-specific, which if the downstream tasks have been defined and we know the set of 𝑅, then we embedded them into the loss [a useful attribute preservation loss].”)
Chen teaches providing a generic feature suppression module of the machine learning model that accepts parameters of a distribution of a latent variable from the first neural network and calculates, for an unannotated generic attribute, a generic feature suppression loss;
(Chen, Section 3.2.2, “The suppression branch [providing a generic feature suppression module of the machine learning model] is designed to make the targeted attributes in 𝑆 as unrecognizable as possible. It utilizes the corresponding set of inference models pretrained on the original dataset X. Each pretrained model corresponds to a specific attribute 𝑠 ∈ 𝑆 and is composed of a feature extractor 𝐹𝑠, which converts raw input data to a feature vector z = 𝐹𝑠 (x), and a projector 𝑃𝑠, which maps the features to the attribute’s label p = 𝑃𝑠 z , where z is the feature representation and p is the prediction logit [that accepts parameters of a distribution of a latent variable from the first neural network]…
Specifically, for each targeted attribute 𝑠 ∈ 𝑆 [for an unannotated generic attribute], the feature-similarity loss function is defined as [a generic feature suppression loss; see negative KL-divergence]
PNG
media_image4.png
137
1061
media_image4.png
Greyscale
For example, we can use the cosine similarity between feature vectors
PNG
media_image5.png
38
282
media_image5.png
Greyscale
or the negative KL-divergence between the original and the transformed logits
PNG
media_image6.png
51
325
media_image6.png
Greyscale
..”)
Chen teaches combining the sensitive attribute suppression loss, the useful attribute preservation loss, and the generic feature suppression loss into a total loss;
(Chen, Section 3.2.3, “Finally, collecting all the loss terms from the data modifier, the suppression branch, as well as the attribute-agnostic and attribute-specific components of the preservation branch, we have the overall optimization loss [combining the sensitive attribute suppression loss, the useful attribute preservation loss, and the generic feature suppression loss into a total loss]
PNG
media_image7.png
65
624
media_image7.png
Greyscale
.”)
Chen teaches and training the first neural network and the second neural network with the total loss.
(Chen, Section 3.2, “In a nutshell, the data modifier learns a transformation that is to be applied to the original data vectors, where the learned transformation is jointly regularized [training the first neural network and the second neural network with the total loss] by both the suppression and preservation branches to ensure all targeted attributes in 𝑆 are indeed suppressed in the transformed data, while all other attributes, explicitly specified or not, are left intact as much as possible.”)
In regards to claim 2 and analogous claims 9 and 16,
Chen teaches The method of claim 1,
Chen teaches wherein the third neural network is trained jointly with the training of the first neural network and the second neural network using the sensitive attribute suppression loss,
(Chen, Section 3.2, “In a nutshell, the data modifier learns a transformation that is to be applied to the original data vectors, where the learned transformation is jointly regularized [wherein the third neural network is trained jointly with the training of the first neural network and the second neural network using the sensitive attribute suppression loss; see suppression branch losses as taught by Chen in claim 1] by both the suppression and preservation branches to ensure all targeted attributes in 𝑆 are indeed suppressed in the transformed data, while all other attributes, explicitly specified or not, are left intact as much as possible.”)
Chen teaches and wherein the third neural network is trained using supervised learning.
(Chen, Section 3.2.2, discloses training “…by comparing the prediction results against the ground truth labels [using supervised learning], and then adding a corresponding penalty.”)
In regards to claim 3 and analogous claims 10 and 17,
Chen teaches The method of claim 1,
Chen teaches wherein the first neural network is trained using gradient descent.
(Chen, Appendix B.1, “We train the model for 100 epochs with temperature 0.07 via the stochastic gradient decent (SGD) [using gradient descent] optimizer.”)
In regards to claim 7 and analogous claim 14,
Chen teaches The method of claim 1,
Chen teaches wherein the fourth neural network is fixed after it is initialized.
(Chen, Section 3.2.3, “In addition to the attribute-agnostic case discussed above, there could also be attribute-specific situations, where the set 𝑅 of attributes that needs to be preserved is explicitly defined in advance for the data transformation. Therefore, we can pretrain the inference models on the original dataset X [wherein the fourth neural network is… initialized ie pretrained] like in the suppression branch. To account for this extra information, 𝑅, we formulate the loss function similar to Eq. 2, except that here we want to maximize the similarities as opposed to minimizing them, as follows, [wherein the fourth neural network is fixed ie updated to maximize similarities]
PNG
media_image8.png
65
634
media_image8.png
Greyscale
”)
Claim Rejections - 35 USC § 103
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 (i.e., changing from AIA to pre-AIA ) 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.
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) 4-6, 11-13 and 18-20 are rejected under 35 U.S.C. 103 as being unpatentable over Chen in view of Bertran, Martin, et al. "Adversarially learned representations for information obfuscation and inference." International Conference on Machine Learning. PMLR, 2019. (“Bertran”).
In regards to claim 4 and analogous claims 11 and 18,
Chen teaches The method of claim 1,
Bertran teaches wherein the sensitive attribute suppression loss is a constraint to an estimation of mutual information between each attribute of the plurality of annotated sensitive attributes and the transformed dataset.
Examiner interprets the limitation in light of para. [0033] of the specification of the instant application “(1) convert annotated sensitive attributes suppression into constraining the mutual information between transformed data and the sensitive attributes, and may further convert it into constraining the cross entropy between predicted conditional distribution and ground truth of each sensitive attributes.”
(Bertran, Section 1.1, “In Section 2 we motivate the proposed framework as a distribution matching problem, and show that this can be formulated as a constrained optimization problem where both the objective function and the constraints are defined in terms of mutual information.”; Bertran discloses an information theoretic approach for an adversarial game in a data-driven manner (see Bertran Abstract))
(Bertran, Section 3., “Assume we have access to a labeled dataset {(xi,si,ui)}N i=1, where si and ui are the true values of the secret and utility variables for observation xi. Learning a parametric stochastic representation Y = qθ(x,z) that optimizes Eq.(3) requires estimating the posteriors: PS∣Y , PU∣Y , and PU∣X; these estimators are obtained through parametric neural networks pη(s ∣ y), pψ(u ∣ y), and pφ(u ∣ x) respectively. Under this setup qθ is obtained by simultaneously optimizing the following adversarial objectives:
PNG
media_image9.png
133
388
media_image9.png
Greyscale
The first three equations in Eq.(10) are cross-entropy loss terms to ensure the estimators pη(s ∣ qθ) [estimation of mutual information between each attribute of the plurality of annotated sensitive attributes and the transformed dataset], pψ(u ∣ qθ), and pφ(u ∣ x) are all good estimators to the true posterior distributions.”)
Chen is considered to be analogous to the claimed invention because they are in the same field of multi-attribute selective suppression. Bertran is considered to be analogous to the claimed invention because they are in the same field of information theoretic approaches to neural networks in a data-driven manner and minimizing sensitive information leakages. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Chen to incorporate the teachings of Bertran in order to provide an information theoretic approach in a data-driven manner as doing so enables the model to learn domain-preserving stochastic transformations that maintain performance on existing algorithms while minimizing sensitive information leakage. (Bertran, Abstract, “In this work, we take an information theoretic approach that is implemented as an unconstrained adversarial game between Deep Neural Networks in a principled, data-driven manner. This approach enables us to learn domain-preserving stochastic transformations that maintain performance on existing algorithms while minimizing sensitive information leakage.”)
In regards to claim 5 and analogous claims 12 and 19,
Chen teaches The method of claim 1,
Bertran teaches wherein the useful attribute preservation loss is a constraint to an estimation of mutual information between each attribute of a plurality of annotated useful attributes and the transformed dataset.
Examiner interprets the limitation in light of para. [0033] of the specification of the instant application “(2) convert annotated useful attributes preservation into constraining the mutual information between transformed data and the useful attributes, and may further convert it into constraining the cross entropy between predicted conditional distribution and ground truth of each useful attributes.”
(Bertran, Section 1.1, “In Section 2 we motivate the proposed framework as a distribution matching problem, and show that this can be formulated as a constrained optimization problem where both the objective function and the constraints are defined in terms of mutual information.”; Bertran discloses an information theoretic approach for an adversarial game in a data-driven manner (see Bertran Abstract))
(Bertran, Section 3., “Assume we have access to a labeled dataset {(xi,si,ui)}N i=1, where si and ui are the true values of the secret and utility variables for observation xi. Learning a parametric stochastic representation Y = qθ(x,z) that optimizes Eq.(3) requires estimating the posteriors: PS∣Y , PU∣Y , and PU∣X [estimation of mutual information between each attribute of a plurality of annotated useful attributes and the transformed dataset]; these estimators are obtained through parametric neural networks pη(s ∣ y), pψ(u ∣ y), and pφ(u ∣ x) respectively. Under this setup qθ is obtained by simultaneously optimizing the following adversarial objectives:
PNG
media_image9.png
133
388
media_image9.png
Greyscale
The first three equations in Eq.(10) are cross-entropy loss terms to ensure the estimators pη(s ∣ qθ), pψ(u ∣ qθ), and pφ(u ∣ x) are all good estimators to the true posterior distributions.”)
In regards to claim 6 and analogous claims 13 and 20,
Chen teaches The method of claim 1,
Bertran teaches wherein the generic feature suppression loss is an estimation of an upper bound of mutual information between the generic feature and the transformed dataset.
Examiner interprets the limitation in light of para. [0033] of the specification of the instant application “(3) convert unannotated generic features suppression into minimizing the mutual information between it and transformed data, and convert it into minimizing the KL divergence between the conditional distribution of latent variables and a unit Gaussian distribution.”
(Bertran, Section 1.1, “In Section 2 we motivate the proposed framework as a distribution matching problem, and show that this can be formulated as a constrained optimization problem where both the objective function and the constraints are defined in terms of mutual information.”; Bertran discloses an information theoretic approach for an adversarial game in a data-driven manner (see Bertran Abstract))
(Bertran, Section 2., “In the proposed formulation, we measure distances between distributions using KL divergence, DKL(p(u ∣ x) ∣∣ p(u ∣ y)) and DKL(p(s ∣ y) ∣∣ p(s)) in particular. By taking the expectation of these metrics with respect to X and Y we recover the following mutual information:
PNG
media_image10.png
49
373
media_image10.png
Greyscale
Both quantities have intuitive interpretations, I(U;X ∣ Y ) is the amount of information on U we lose by observing the filtered data Y instead of the original data X, we call this quantity information loss (Geiger & Kubin, 2011). I(S;Y ) is the mutual information we disclose on variable S by observing variable Y , this is the information we fail to obfuscate. Under this setting, our objective is:
PNG
media_image11.png
45
334
media_image11.png
Greyscale
Here k > 0 is a constant controlling our tolerance on the amount of information on S disclosed via Y . Since U is conditionally independent on Y given X, I(U;X ∣ Y ) = I(U;X)−I(U;Y), which leads to the the equivalent objective minp(y∣x) I(U;X ∣ Y ) ∼ maxp(y∣x)I(U;Y ).”; Bertran discloses using KL divergence to recover mutual information)
(Bertran, Section 2.1, “Lemma 2.1, presented next, shows that we can bound [wherein the generic feature suppression loss is an estimation of an upper bound of mutual information] the solution of Eq.(2) by considering mappings that go directly from the latent variables U and S to the obfuscated variable Y . This simplifies the analysis since ∣U ×S∣ ≪∣X∣ for many problems of interest.
PNG
media_image1.png
366
396
media_image1.png
Greyscale
”)
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
NPL: Madras, David, et al. "Learning Adversarially Fair and Transferable Representations." arXiv e-prints (2018): arXiv-1802.
NPL: Wu, Zhenyu, et al. "Privacy-Preserving Deep Action Recognition: An Adversarial Learning Framework and A New Dataset." arXiv preprint arXiv:1906.05675 (2019).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JASMINE THAI whose telephone number is (703)756-5904. The examiner can normally be reached M-F 8-4.
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, Michael Huntley can be reached at (303) 297-4307. 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.
/J.T.T./Examiner, Art Unit 2129
/MICHAEL J HUNTLEY/Supervisory Patent Examiner, Art Unit 2129
Prosecution Insights