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 .
Examiner’s Note
Under consideration as a judicial exceptions to 35 U.S.C. 101 as outlined by the process in MPEP 2016, Examiner notes that independent claims 1, 8, and 15 recite mathematical concepts, namely, the computation of a “bi-causal transport distance between the first joint distribution and the second joint distribution.” However, Examiner finds that using this computation to regularize a machine learning model (recited in claim 1 as “computing, by the at least one processor based on the bi-causal transport distance, a regularizer that reduces the conditional demographic disparity; and applying, by the at least one processor, the regularizer to the model”) constitutes an improvement which integrates the exception into a practical application.
Drawings
The drawings are objected to because Fig. 5 appears to be a reproduction of color text, which has resulted in text that is not fully legible, especially but not limited to subscripts and superscripts. Corrected drawing sheets in compliance with 37 CFR 1.121(d) are required in reply to the Office action to avoid abandonment of the application. Any amended replacement drawing sheet should include all of the figures appearing on the immediate prior version of the sheet, even if only one figure is being amended. The figure or figure number of an amended drawing should not be labeled as “amended.” If a drawing figure is to be canceled, the appropriate figure must be removed from the replacement sheet, and where necessary, the remaining figures must be renumbered and appropriate changes made to the brief description of the several views of the drawings for consistency. Additional replacement sheets may be necessary to show the renumbering of the remaining figures. Each drawing sheet submitted after the filing date of an application must be labeled in the top margin as either “Replacement Sheet” or “New Sheet” pursuant to 37 CFR 1.121(d). If the changes are not accepted by the examiner, the applicant will be notified and informed of any required corrective action in the next Office action. The objection to the drawings will not be held in abeyance.
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 1-18 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 recites the limitation “determining, by the at least one processor, a first joint distribution of model outputs and a second feature from among the at least one feature based on a first level of a first feature from among the at least one feature and a second joint distribution of model outputs and the second feature based on a second level of the first feature.” Examiner finds the word “level” in the context of “a first level of a first feature” and “a second level of the first feature” to be ambiguous. The term “level” is not explicitly defined in the disclosure. The “level” could reasonably be interpreted as a particular value of a categorical feature (e.g., “dog” versus “cat” for a “species” feature), a grouping for a numerical feature (e.g., grouping a numerical value for a household income feature into a quantile), a position in a feature hierarchy (e.g., “dog” versus “mammal”), a layer in a multi-layer feature extraction mechanism of a neural network model, among others.
A person having ordinary skill would therefore be unable to determine the metes and bounds of the claim.
Claims 8 and 15 recite the same or analogous language, and therefore are rejected by the same argument.
Claims 2-7, 9-14, and 16-18 depend upon claim 1, 8, and 15, and therefore are rejected by the same argument.
In further examination below, the “level” of a feature will be interpreted to include any attribute of the data that differentiates the distributions.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-2 and 5-7 rejected under 35 U.S.C. 103 over Backhoff et al., “Causal transport in discrete time and applications,” 2017, arXiv:1606.04062v2 (hereafter Backhoff) in view of Xu et al., “Algorithmic Decision Making with Conditional Fairness,” 2021, arXiv:2006.10483v5 (hereafter Xu).
Regarding claim 1:
Backhoff teaches:
“identifying, by the at least one processor, at least one feature associated with data that is inputted into the model; determining, by the at least one processor, a first joint distribution of model outputs and a second feature from among the at least one feature based on a first level of a first feature from among the at least one feature and a second joint distribution of model outputs and the second feature based on a second level of the first feature”: Backhoff, section 1, paragraph 1, “In this article we consider the optimal transport problem between two discrete time stochastic processes [a first joint distribution of model outputs … and a second joint distribution of model outputs] under the so-called causality constraint, highlighted recently by the work of Lassalle in [Las15] in a more general setting. A transport plan between two processes is said to be causal if, from an observed trajectory of the first process, the ‘mass’ can be split at each moment of time [identifying, by the at least one processor, at least one feature associated with data that is inputted into the model] into the second process only based on the information available up to that time [based on a first level of a first feature from among the at least one feature … based on a second level of the first feature, interpreted as including an identified attribute of the data that differentiates the distributions].”
“computing, by the at least one processor, a bi-causal transport distance between the first joint distribution and the second joint distribution” and (bold only) “computing, by the at least one processor based on the bi-causal transport distance, a regularizer that reduces the conditional demographic disparity”: Backhoff, section 1, paragraph 2, ““In this article we will also link these objects to the notion of nested distance, whose systematic investigation was initiated by Pflu [Pfl09] and Pflug–Pichler [PP12, PP14, PP15], and had a precursor in the ‘Markov-constructions’ studied by Ruschendorf [R¨us85]. Roughly, the nested distance is defined through a problem of optimal transport over plans which are bicausal, this notion being the symmetrized analogue of causality [computing, by the at least one processor, a bi-causal transport distance between the first joint distribution and the second joint distribution][based on the bi-causal transport distance]. Interestingly, [R¨us85] and [PP14] established a recursive formulation for the problem, and [PP12, PP14] further obtained a dual formulation for the nested distance. Moreover, Pflug–Pichler [PP12] applied these considerations to the practical problem of reducing the complexity of multistage stochastic programs, by showing that the difference between the optimal value of a program w.r.t. two different noise distributions is dominated by the nested distance between them.”
Backhoff does not explicitly teach:
“A method for optimizing conditional demographic parity in a machine learning model, the method being implemented by at least one processor, the method comprising”
(bold only) “computing, by the at least one processor based on the bi-causal transport distance, a regularizer that reduces the conditional demographic disparity”
“applying, by the at least one processor, the regularizer to the model”
Xu teaches:
“A method for optimizing conditional demographic parity in a machine learning model, the method being implemented by at least one processor, the method comprising”: Xu, page 3, paragraph 1, “We propose a novel Derivable Conditional Fairness Regularizer (DCFR) to optimize conditional fairness by learning conditional independent representation [method for optimizing conditional demographic parity in a machine learning model]. Our DCFR can be easily integrated into decision-making models to improve fairness”; Xu, page 4, paragraph 2, “Our method is based on an equivalent relation of conditional independence [Daudin, 1980] and is tractable in common machine learning [the method being implemented by at least one processor] algorithms.”
(bold only) “computing, by the at least one processor based on the bi-causal transport distance, a regularizer that reduces the conditional demographic disparity”: Xu, page 3, paragraph 1, “We propose a novel Derivable Conditional Fairness Regularizer (DCFR) to optimize conditional fairness by learning conditional independent representation. Our DCFR can be easily integrated into decision-making models [applying, by the at least one processor, the regularizer to the model] to improve fairness”
“applying, by the at least one processor, the regularizer to the model”: Xu, page 3, paragraph 1, “We propose a novel Derivable Conditional Fairness Regularizer (DCFR) to optimize conditional fairness by learning conditional independent representation. Our DCFR can be easily integrated into decision-making models [applying, by the at least one processor, the regularizer to the model] to improve fairness.”
Xu and Backhoff are analogous arts as they are both related to distribution analysis. It would have been obvious to a person having ordinary skill in the art prior to the effective filing date of the claimed invention to have applied the regularization mechanisms of Xu to the teachings of Backhoff to arrive at the present invention, in order to optimize conditional fairness, as stated in Xu, page 3, paragraph 1, “We propose a novel Derivable Conditional Fairness Regularizer (DCFR) to optimize conditional fairness by learning conditional independent representation. Our DCFR can be easily integrated into decision-making models to improve fairness.”
Regarding claim 2:
Backhoff as modified by Xu teaches “The method of claim 1.”
Xu further teaches “further comprising calculating a conditional demographic disparity between the first joint distribution and the second joint distribution with respect to the first feature”: Xu, page 2, paragraph 3, “In practice, there frequently exist a certain set of variables we term as fair variables, which are pre-decision covariates such as the department choice in Berkeley’s graduate admission problem. The effects of fair variables are irrelevant in assessing the fairness of the decision support algorithm. We thus define conditional fairness as a more sound fairness metric by conditioning on the fairness variables [calculating a conditional demographic disparity between the first joint distribution and the second joint distribution with respect to the first feature]. In detail, outcome variables should be independent of sensitive attributes conditional on these fair variables.”
Xu and Backhoff are combinable for the rationale given under claim 1.
Regarding claim 5:
Backhoff as modified by Xu teaches “The method of claim 1.”
Xu further teaches “wherein the model is configured to use an artificial intelligence technique for making a decision based on input data that relates to a person, and wherein the decision relates to at least one from among a consumer finance question, a health insurance question, and a hiring question”: Xu, page 1, paragraph 1, “Nowadays fairness issues have raised great concerns in decision-making systems such as loan applications [a consumer finance question] [Mukerjee et al., 2002], hiring processes [a hiring question] [Rivera, 2012], and criminal justice [Larson et al., 2016]. Poorly designed algorithms tend to amplify the bias existed in data, resulting in discriminations towards specific groups of individuals based on their inherent characteristics, which are often named as sensitive attributes in fairness problems. For example, race is a sensitive attribute for crime judgment. ProPublica [Larson et al., 2016] found it is unfair that African Americans were more likely to be incorrectly labeled as higher risk compared with Caucasians in the COMPAS system. However, what is fair and how to develop fair algorithms for algorithmic decision making [use an artificial intelligence technique for making a decision] are of paramount importance for both academic research and practical applications.”
Xu and Backhoff are analogous arts as they are both related to distribution analysis. It would have been obvious to a person having ordinary skill in the art prior to the effective filing date of the claimed invention to have applied the decision-making considerations of Xu to the teachings of Backhoff to arrive at the present invention, in order to counter discrimination towards specific groups, as stated in Xu, page 1, paragraph 1, “Nowadays fairness issues have raised great concerns in decision-making systems such as loan applications [Mukerjee et al., 2002], hiring processes [Rivera, 2012], and criminal justice [Larson et al., 2016]. Poorly designed algorithms tend to amplify the bias existed in data, resulting in discriminations towards specific groups of individuals based on their inherent characteristics, which are often named as sensitive attributes in fairness problems.”
Regarding claim 6:
Backhoff as modified by Xu teaches “The method of claim 1.”
Xu further teaches “wherein the first feature includes at least one from among race, gender, national origin, and disability”: Xu, page 1, paragraph 1, “Nowadays fairness issues have raised great concerns in decision-making systems such as loan applications [Mukerjee et al., 2002], hiring processes [Rivera, 2012], and criminal justice [Larson et al., 2016]. Poorly designed algorithms tend to amplify the bias existed in data, resulting in discriminations towards specific groups of individuals based on their inherent characteristics, which are often named as sensitive attributes in fairness problems. For example, race is a sensitive attribute for crime judgment [includes at least one from among race, gender, national origin, and disability]. ProPublica [Larson et al., 2016] found it is unfair that African Americans were more likely to be incorrectly labeled as higher risk compared with Caucasians in the COMPAS system. However, what is fair and how to develop fair algorithms for algorithmic decision making are of paramount importance for both academic research and practical applications.”
Xu and Backhoff are analogous arts as they are both related to distribution analysis. It would have been obvious to a person having ordinary skill in the art prior to the effective filing date of the claimed invention to have applied the group consideration of Xu to the teachings of Backhoff to arrive at the present invention, in order to counter discrimination towards specific groups, as stated in Xu, page 1, paragraph 1, “Nowadays fairness issues have raised great concerns in decision-making systems such as loan applications [Mukerjee et al., 2002], hiring processes [Rivera, 2012], and criminal justice [Larson et al., 2016]. Poorly designed algorithms tend to amplify the bias existed in data, resulting in discriminations towards specific groups of individuals based on their inherent characteristics, which are often named as sensitive attributes in fairness problems.”
Regarding claim 7:
Backhoff as modified by Xu teaches “The method of claim 1.”
Xu further teaches “wherein the second feature includes one from among a level of education, a grade point average (GPA), and a level of income”: Xu, page 11, paragraph 5, “Adult: The goal of the Adult dataset is to predict whether a person makes more than $50k per year or not. Each instance contains 112 attributes including sex, gender, education level [includes one from among a level of education, a grade point average (GPA), and a level of income], occupation, etc. In our experiments, we set gender as the sensitive attribute, and consider occupation (with 14 possible categorical values) as the fair variable. The target variable (income) is binary and we set ‘$50k per year’ as the favored outcome.”
Xu and Backhoff are analogous arts as they are both related to distribution analysis. It would have been obvious to a person having ordinary skill in the art prior to the effective filing date of the claimed invention to have applied the group consideration of Xu to the teachings of Backhoff to arrive at the present invention, in order to counter discrimination towards specific groups, as stated in Xu, page 1, paragraph 1, “Nowadays fairness issues have raised great concerns in decision-making systems such as loan applications [Mukerjee et al., 2002], hiring processes [Rivera, 2012], and criminal justice [Larson et al., 2016]. Poorly designed algorithms tend to amplify the bias existed in data, resulting in discriminations towards specific groups of individuals based on their inherent characteristics, which are often named as sensitive attributes in fairness problems.”
Claim 3 rejected under 35 U.S.C. 103 over Backhoff as modified by Xu in view of Hardt et al., “Equality of Opportunity in Supervised Learning,” 2016, arXiv:1610.02413v1 (hereafter Hardt).
Backhoff as modified by Xu teaches “The method of claim 2.”
Backhoff as modified by Xu does not explicitly teach “wherein the calculating of the conditional demographic disparity comprises calculating a Kolmogorov distance between the first joint distribution and the second joint distribution.”
Hardt teaches “wherein the calculating of the conditional demographic disparity comprises calculating a Kolmogorov distance between the first joint distribution and the second joint distribution”: Hardt, section 5.1, paragraph 6, “We can now show that an equalized odds predictor [calculating of the conditional demographic disparity] derived from a nearly optimal regressor is still nearly optimal among all equal odds predictors, while quantifying the loss in terms of the conditional Kolmogorov distance [calculating a Kolmogorov distance between the first joint distribution and the second joint distribution].”
Hardt and Backhoff as modified by Xu are analogous arts as they are both related to disparity metrics. It would have been obvious to a person having ordinary skill in the art prior to the effective filing date of the claimed invention to have combined the Kolmogorov distance of Hardt with the teachings of Backhoff as modified by Xu to arrive at the present invention, in order to employ a near-optimal disparity measurement, as stated in Hardt, section 5.1, paragraph 1, “We can furthermore show that if we can approximate the (unconstrained) Bayes optimal regressor well enough, then we can also construct a nearly optimal non-discriminating classifier.”
Claims 4, 8-9, 11-16, and 18 rejected under 35 U.S.C. 103 over Backhoff as modified by Xu in view of Chen et al., US Pre-Grant Publication No. 2024/0419939 (hereafter Chen).
Regarding claim 4:
Backhoff as modified by Xu teaches “The method of claim 1.”
Backhoff further teaches (bold only) “wherein the computing of the bi-causal transport distance comprises estimating a nested Wasserstein distance between the first joint distribution and the second joint distribution by applying a Sinkhorn divergence algorithm to a set of samples of each of the first joint distribution and the second joint distribution”: Backhoff, section 6, paragraph 1, “We start however with a discussion about geometric properties of the space of probability measures endowed with a bicausal Wasserstein distance (equiv. nested distance) [a nested Wasserstein distance between the first joint distribution and the second joint distribution].”
Backhoff as modified by Xu does not explicitly teach (bold only) “wherein the computing of the bi-causal transport distance comprises estimating a nested Wasserstein distance between the first joint distribution and the second joint distribution by applying a Sinkhorn divergence algorithm to a set of samples of each of the first joint distribution and the second joint distribution.”
Chen teaches (bold only) “wherein the computing of the bi-causal transport distance comprises estimating a nested Wasserstein distance between the first joint distribution and the second joint distribution by applying a Sinkhorn divergence algorithm to a set of samples of each of the first joint distribution and the second joint distribution”: Chen, paragraph 0113, “Optimal Transport (OT) theory is introduced to study the problem of resource allocation with lowest transport cost. OT is a mathematical model that defines distances or similarities between objects such as probability distributions, either continuous or discrete, as the cost of an optimal transport plan from one to the other. By regarding the ‘cost’ as distance, Wasserstein Distance is a commonly used metric for matching two distributions in OT [transport distance comprises … a … Wasserstein distance]”; Chen, paragraph 0114, “The Wasserstein distance between µ and v may then be defined according to the following Equation (7)”; Chen, paragraph 0116, “Roughly speaking, OT is equivalent to constrained k-Means clustering. In order to parameterize k-Means using the optimal transport plan in Equation (7), GNN model processor 104 may employ a novel k-Means objective for clustering users according to the following Equation (8)”; Chen, paragraph 0118, “Equation (8) can be solved by Linear Programming (LP), yet with computational burden. This cost can be largely mitigated by introducing a strictly convex entropy regularized OT via fast Sinkhorn’s algorithms [by applying a Sinkhorn divergence algorithm to a set of samples of each of the first joint distribution and the second joint distribution].”
Chen and Backhoff are analogous arts as they are both related to optimal transport computation. It would have been obvious to a person having ordinary skill in the art prior to the effective filing date of the claimed invention to have combined the Sinkhorn algorithm of Chen with the teachings of Backhoff to arrive at the present invention, in order to relieve computational burden, as stated in Chen, paragraph 0118, “Equation (8) can be solved by Linear Programming (LP), yet with computational burden. This cost can be largely mitigated by introducing a strictly convex entropy regularized OT via fast Sinkhorn’s algorithms.”
Regarding claims 8-9 and 12-14:
These claims are analogous to claims 1-2 and 5-7 respectively, which are taught by Backhoff as modified by Xu, as shown above, except for the performance of the claimed invention by (bold only) “A computing apparatus for optimizing conditional demographic parity in a machine learning model, the computing apparatus comprising: a processor; a memory; and a communication interface coupled to each of the processor and the memory, wherein the processor is configured to,” which is not explicitly taught by Backhoff as modified by Xu.
Chen teaches (bold only) “A computing apparatus for optimizing conditional demographic parity in a machine learning model, the computing apparatus comprising: a processor; a memory; and a communication interface coupled to each of the processor and the memory, wherein the processor is configured to”: Chen, paragraph 0087, “As shown in FIG. 3, device 300 may include a bus 302 [a communication interface coupled to each of the processor and the memory], a processor [processor] 304, memory [memory] 306, a storage component 308, an input component 310, an output component 312, and a communication interface 314. Bus 302 may include a component that permits communication among the components of device 300.”
Chen and Backhoff are analogous arts as they are both related to optimal transport computation. It would have been obvious to a person having ordinary skill in the art prior to the effective filing date of the claimed invention to have combined the use of a computing system of Chen with the teachings of Backhoff to arrive at the present invention, in order to solve the computations using a computer implementation, as stated in Chen, paragraph 0087, “As shown in FIG. 3, device 300 may include a bus 302, a processor 304, memory [memory] 306, a storage component 308, an input component 310, an output component 312, and a communication interface 314. Bus 302 may include a component that permits communication among the components of device 300.”
Regarding claim 11:
This claim is analogous to claim 4, which is taught by Backhoff as modified by Xu and Chen, as shown above, except for the performance of the claimed invention by (bold only) “A computing apparatus for optimizing conditional demographic parity in a machine learning model, the computing apparatus comprising: a processor; a memory; and a communication interface coupled to each of the processor and the memory, wherein the processor is configured to,” which is taught by Chen, paragraph 0087, “As shown in FIG. 3, device 300 may include a bus 302 [a communication interface coupled to each of the processor and the memory], a processor [processor] 304, memory [memory] 306, a storage component 308, an input component 310, an output component 312, and a communication interface 314. Bus 302 may include a component that permits communication among the components of device 300.”
Chen and Backhoff are analogous arts as they are both related to optimal transport computation. It would have been obvious to a person having ordinary skill in the art prior to the effective filing date of the claimed invention to have combined the use of a computing system of Chen with the teachings of Backhoff to arrive at the present invention, in order to solve the computations using a computer implementation, as stated in Chen, paragraph 0087, “As shown in FIG. 3, device 300 may include a bus 302, a processor 304, memory [memory] 306, a storage component 308, an input component 310, an output component 312, and a communication interface 314. Bus 302 may include a component that permits communication among the components of device 300.”
Regarding claims 15-16:
These claims are analogous to claims 1-2, which are taught by Backhoff as modified by Xu, as shown above, except for the performance of the claimed invention by (bold only) “A non-transitory computer readable storage medium storing instructions for optimizing conditional demographic parity in a machine learning model, the storage medium comprising executable code which, when executed by a processor, causes the processor to,” which is not explicitly taught by Backhoff as modified by Xu.
Chen teaches (bold only) “A non-transitory computer readable storage medium storing instructions for optimizing conditional demographic parity in a machine learning model, the storage medium comprising executable code which, when executed by a processor, causes the processor to”: Chen, paragraph 0008, “According to non-limiting embodiments or aspects, provided is a computer program product for determining long-range dependencies using a non-local graph neural network including at least one non-transitory computer-readable medium including program instructions [A non-transitory computer readable storage medium storing instructions] that, when executed by at least one processor, cause the at least one processor to [which, when executed by a processor, causes the processor to].”
Chen and Backhoff are analogous arts as they are both related to optimal transport computation. It would have been obvious to a person having ordinary skill in the art prior to the effective filing date of the claimed invention to have combined the use of a computing system of Chen with the teachings of Backhoff to arrive at the present invention, in order to solve the computations using a computer implementation, as stated in Chen, paragraph 0087, “As shown in FIG. 3, device 300 may include a bus 302, a processor 304, memory [memory] 306, a storage component 308, an input component 310, an output component 312, and a communication interface 314. Bus 302 may include a component that permits communication among the components of device 300.”
Regarding claim 18:
This claim is analogous to claim 4, which is taught by Backhoff as modified by Xu and Chen, as shown above, except for the performance of the claimed invention by (bold only) “A non-transitory computer readable storage medium storing instructions for optimizing conditional demographic parity in a machine learning model, the storage medium comprising executable code which, when executed by a processor, causes the processor to,” which is taught by Chen, paragraph 0008, “According to non-limiting embodiments or aspects, provided is a computer program product for determining long-range dependencies using a non-local graph neural network including at least one non-transitory computer-readable medium including program instructions [A non-transitory computer readable storage medium storing instructions] that, when executed by at least one processor, cause the at least one processor to [which, when executed by a processor, causes the processor to].”
Chen and Backhoff are analogous arts as they are both related to optimal transport computation. It would have been obvious to a person having ordinary skill in the art prior to the effective filing date of the claimed invention to have combined the use of a computing system of Chen with the teachings of Backhoff to arrive at the present invention, in order to solve the computations using a computer implementation, as stated in Chen, paragraph 0087, “As shown in FIG. 3, device 300 may include a bus 302, a processor 304, memory [memory] 306, a storage component 308, an input component 310, an output component 312, and a communication interface 314. Bus 302 may include a component that permits communication among the components of device 300.”
Claims 10 and 17 rejected under 35 U.S.C. 103 over Backhoff as modified by Xu and Hardt in view of Chen.
Regarding claim 10:
This claim is analogous to claim 3, which is taught by Backhoff as modified by Xu and Hardt, as shown above, except for the performance of the claimed invention by (bold only) “A computing apparatus for optimizing conditional demographic parity in a machine learning model, the computing apparatus comprising: a processor; a memory; and a communication interface coupled to each of the processor and the memory, wherein the processor is configured to,” which is not explicitly taught by Backhoff as modified by Xu and Hardt.
Chen teaches (bold only) “A computing apparatus for optimizing conditional demographic parity in a machine learning model, the computing apparatus comprising: a processor; a memory; and a communication interface coupled to each of the processor and the memory, wherein the processor is configured to”: Chen, paragraph 0087, “As shown in FIG. 3, device 300 may include a bus 302 [a communication interface coupled to each of the processor and the memory], a processor [processor] 304, memory [memory] 306, a storage component 308, an input component 310, an output component 312, and a communication interface 314. Bus 302 may include a component that permits communication among the components of device 300.”
Chen and Backhoff are analogous arts as they are both related to optimal transport computation. It would have been obvious to a person having ordinary skill in the art prior to the effective filing date of the claimed invention to have combined the use of a computing system of Chen with the teachings of Backhoff to arrive at the present invention, in order to solve the computations using a computer implementation, as stated in Chen, paragraph 0087, “As shown in FIG. 3, device 300 may include a bus 302, a processor 304, memory [memory] 306, a storage component 308, an input component 310, an output component 312, and a communication interface 314. Bus 302 may include a component that permits communication among the components of device 300.”
Regarding claim 17:
This claim is analogous to claim 3, which is taught by Backhoff as modified by Xu and Hardt, as shown above, except for the performance of the claimed invention by (bold only) “A non-transitory computer readable storage medium storing instructions for optimizing conditional demographic parity in a machine learning model, the storage medium comprising executable code which, when executed by a processor, causes the processor to,” which is not explicitly taught by Backhoff as modified by Xu and Hardt.
Chen teaches (bold only) “A non-transitory computer readable storage medium storing instructions for optimizing conditional demographic parity in a machine learning model, the storage medium comprising executable code which, when executed by a processor, causes the processor to”: Chen, paragraph 0008, “According to non-limiting embodiments or aspects, provided is a computer program product for determining long-range dependencies using a non-local graph neural network including at least one non-transitory computer-readable medium including program instructions [A non-transitory computer readable storage medium storing instructions] that, when executed by at least one processor, cause the at least one processor to [which, when executed by a processor, causes the processor to].”
Chen and Backhoff are analogous arts as they are both related to optimal transport computation. It would have been obvious to a person having ordinary skill in the art prior to the effective filing date of the claimed invention to have combined the use of a computing system of Chen with the teachings of Backhoff to arrive at the present invention, in order to solve the computations using a computer implementation, as stated in Chen, paragraph 0087, “As shown in FIG. 3, device 300 may include a bus 302, a processor 304, memory [memory] 306, a storage component 308, an input component 310, an output component 312, and a communication interface 314. Bus 302 may include a component that permits communication among the components of device 300.”
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Yeng et al., “Decision-making with Side Information: A Causal Transport Robust Approach,” 2022, https://optimization-online.org/?p=20639, discloses methods for making decisions over joint distributions using a causal transport distance metric.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to VINCENT SPRAUL whose telephone number is (703) 756-1511. The examiner can normally be reached M-F 9:00 am - 5:00 pm.
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.
/VAS/Examiner, Art Unit 2129
/MICHAEL J HUNTLEY/Supervisory Patent Examiner, Art Unit 2129