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 .
Response to Amendment
The Amendment filed 03/04/2026 has been entered. Claims 1-20 remain pending in this application.
Information Disclosure Statement
The information disclosure statements submitted on 03/05/2026 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
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 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 of this title, 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.
Claim 1-2, 8, 10-12 and 18-19 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang et al. (US 11586849 B2 hereinafter Zhang) in view of GUERET et al. (US 20220138632 A1 hereinafter Gueret) and Bhide et al. (US 20200184350 A1 hereinafter Bhide)
As to independent claim 1, Zhang teaches a computer-implemented method comprising:
receiving, by one or more processors, an aggregate bias correction function for a machine learning model;[receives settings that optimize for bias (correction function) including SPD and WAOD Col. 4 ln. 42-46 "optimize a machine learning model via the implementation of one or more user defined fairness policies that can mitigate statistical bias" Col. 13 ln. 4-16 "SPD and/or WAOD to measure one or more fairness criteria"]
generating, by the processors and using the aggregate bias correction function, an individualized threshold corresponding to an input data object for the machine learning model, [generates a threshold based on metrics (attributes) Col. 2 ln. 4-24 "determining, by a system operatively coupled to a processor, a threshold setting for a machine learning model based on preferential weight values assigned to a fairness metric and a utility metric of the machine learning model"], [protected attributes Col. 7 ln. 61-67]
generating, by the one or more processors and using the machine learning model, a predictive output for the input data object based at least in part on the plurality of contextual attributes; [Y prediction output Col. 7-8 ln. 61-17 " “Y” can be a machine learning classifier's prediction"]
generating, by the one or more processors, a bias adjusted output for the input data object based at least in part on a comparison between the individualized threshold and the predictive output, [output a visualization based on thresholds applied (bias based adjustments Col. 9 ln. 42-54 "visualization component 202 can evaluate a plurality of threshold settings for the machine learning model to generate one or more sample sets comprising various versions of the machine learning model"]
providing, by the one or more processors, data indicative of the bias adjusted output [visual reflects changes Col. 9-10 ln. 42-11"one or more visualizations generated by the visualization component 202 can depict how changing the thresholds (e.g., classification thresholds) can impact SPD, WAPD, and/or utility of the machine learning model"]
Zhang does not specifically teach wherein the individualized threshold is based at least in part on (i) a plurality of contextual attributes of the input data object and (ii) a plurality of protected attributes of the input data object; wherein the bias adjusted output comprises a reduced bias relative to the predictive output without using a randomized component for the machine learning model;
However, Gueret teaches wherein the individualized threshold is based at least in part on (i) a plurality of contextual attributes of the input data object and (ii) a plurality of protected attributes of the input data object; [calibration defined threshold based on context attributes (income, occupation) and protected attributes (females, southerners) ¶24, ¶17-19 "characteristics of the user, such as age, gender, race, ethnicity, income, occupation, geographical location, and/or other distinguishable characteristics."…"rules are provided for four example groups: Youth (specifically 20 to 30 year old individuals), Females, Northerners, and Southerners. An example rule, as described herein, includes a definition and an objective"]
wherein the bias adjusted output comprises a reduced bias relative to the predictive output without using a randomized component for the machine learning model; [rule based calibration (non-random) that avoids bias ¶43 " rules associated with the one or more groups. Accordingly, based on the rules, the calibrated qualification model may avoid developing a bias"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the bias mitigation disclosed by Zhang by incorporating the wherein the individualized threshold is based at least in part on (i) a plurality of contextual attributes of the input data object and (ii) a plurality of protected attributes of the input data object; wherein the bias adjusted output comprises a reduced bias relative to the predictive output without using a randomized component for the machine learning model disclosed by Gueret because both techniques address the same field of machine learning and by incorporating Gueret into Zhang alleviates waste of computing resources an provides more desired results from models [Gueret ¶13]
Zhang and Gueret do not specifically teach generating, by the one or more processors and using the aggregate bias correction function, an individualized threshold tailored to an individual attribute corresponding to an input data object for the machine learning model.
However, Bhide teaches generating, by the one or more processors and using the aggregate bias correction function, an individualized threshold tailored to an individual attribute corresponding to an input data object for the machine learning model. [uses an individual bias threshold and (attributes ¶29) ¶9 "training a bias detector that learns to detect a sample that has an individual bias greater than a predetermined individual bias threshold value with constraints on a group bias"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the input data disclosed by Zhang and Gueret by incorporating the generating, by the one or more processors and using the aggregate bias correction function, an individualized threshold tailored to an individual attribute corresponding to an input data object for the machine learning model disclosed by Bhide because all techniques address the same field of machine learning and by incorporating Bhide into Zhang and Gueret addresses individual fairness to better mitigate bias for trustworthy models [Bhide ¶1-2]
As to dependent claim 2, the rejection of claim 1 is incorporated, Zhang, Gueret and Bhide further teach a contextual bias correction function configured to output an individualized contextual threshold for the input data object based at least in part on the plurality of contextual attributes; and [Gueret context attributes (income, occupation) ¶17-19 used to calibrate a model ¶26 " determine groups of calibration data according to rules and feature attributes"]
a protected bias correction function configured to output an individualized protection threshold for the input data object based at least in part on the plurality of protected attributes. [Zhang threshold based on age (protected attribute) Col. 14-15 ln .62-6 "classification thresholds are set to the same for the age groups younger and older than 26"], [Gueret protected attributes (females, southerners) ¶17-19]
As to dependent claim 8, the rejection of claim 1 is incorporated, Zhang, Gueret and Bhide further teach wherein the predictive output comprises a classification probability corresponding to one or more classifications, and wherein the bias adjusted output comprises a predicted classification from the one or more classifications. [Gueret confidence score and classification (qualified or not) ¶17 "The qualification model may be a binary classification model that provides a binary output (e.g., “qualified” or “not qualified”). Additionally, or alternatively, the qualification model may determine and/or indicate a confidence score associated with a prediction of whether the user is qualified"]
As to dependent claim 10, the rejection of claim 1 is incorporated, Zhang, Gueret and Bhide further teach wherein the individualized threshold is a real number between zero and one. [Zhang 0.5 Col. 13 ln. 56-67]
As to independent claim 11, Zhang teaches a system comprising: one or more processors; and one or more memories storing processor-executable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising: [computer, processors, memory and instructions ¶88]
receiving an aggregate bias correction function for a machine learning model;[receives settings that optimize for bias (correction function) including SPD and WAOD Col. 4 ln. 42-46 "optimize a machine learning model via the implementation of one or more user defined fairness policies that can mitigate statistical bias" Col. 13 ln. 4-16 "SPD and/or WAOD to measure one or more fairness criteria"]
generating, using the aggregate bias correction function, an individualized threshold corresponding to an input data object for the machine learning model, [generates a threshold based on metrics (attributes) Col. 2 ln. 4-24 "determining, by a system operatively coupled to a processor, a threshold setting for a machine learning model based on preferential weight values assigned to a fairness metric and a utility metric of the machine learning model"], [protected attributes Col. 7 ln. 61-67]
generating, using the machine learning model, a predictive output for the input data object based at least in part on the plurality of contextual attributes; [Y prediction output Col. 7-8 ln. 61-17 " “Y” can be a machine learning classifier's prediction"]
generating a bias adjusted output for the input data object based at least in part on a comparison between the individualized threshold and the predictive output, [output a visualization based on thresholds applied (bias based adjustments Col. 9 ln. 42-54 "visualization component 202 can evaluate a plurality of threshold settings for the machine learning model to generate one or more sample sets comprising various versions of the machine learning model"]
providing data indicative of the bias adjusted output [visual reflects changes Col. 9-10 ln. 42-11"one or more visualizations generated by the visualization component 202 can depict how changing the thresholds (e.g., classification thresholds) can impact SPD, WAPD, and/or utility of the machine learning model"]
Zhang does not specifically teach wherein the individualized threshold is based at least in part on (i) a plurality of contextual attributes of the input data object and (ii) a plurality of protected attributes of the input data object; wherein the bias adjusted output comprises a reduced bias relative to the predictive output without using a randomized component for the machine learning model;
However, Gueret teaches wherein the individualized threshold is based at least in part on (i) a plurality of contextual attributes of the input data object and (ii) a plurality of protected attributes of the input data object; [calibration defined threshold based on context attributes (income, occupation) and protected attributes (females, southerners) ¶24, ¶17-19 "characteristics of the user, such as age, gender, race, ethnicity, income, occupation, geographical location, and/or other distinguishable characteristics."…"rules are provided for four example groups: Youth (specifically 20 to 30 year old individuals), Females, Northerners, and Southerners. An example rule, as described herein, includes a definition and an objective"]
wherein the bias adjusted output comprises a reduced bias relative to the predictive output without using a randomized component for the machine learning model; [rule based calibration (non-random) that avoids bias ¶43 " rules associated with the one or more groups. Accordingly, based on the rules, the calibrated qualification model may avoid developing a bias"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the bias mitigation disclosed by Zhang by incorporating the wherein the individualized threshold is based at least in part on (i) a plurality of contextual attributes of the input data object and (ii) a plurality of protected attributes of the input data object; wherein the bias adjusted output comprises a reduced bias relative to the predictive output without using a randomized component for the machine learning model disclosed by Gueret because both techniques address the same field of machine learning and by incorporating Gueret into Zhang alleviates waste of computing resources an provides more desired results from models [Gueret ¶13]
Zhang and Gueret do not specifically teach generating, by the one or more processors and using the aggregate bias correction function, an individualized threshold tailored to an individual attribute corresponding to an input data object for the machine learning model.
However, Bhide teaches generating, by the one or more processors and using the aggregate bias correction function, an individualized threshold tailored to an individual attribute corresponding to an input data object for the machine learning model. [uses an individual bias threshold and (attributes ¶29) ¶9 "training a bias detector that learns to detect a sample that has an individual bias greater than a predetermined individual bias threshold value with constraints on a group bias"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the input data disclosed by Zhang and Gueret by incorporating the generating, by the one or more processors and using the aggregate bias correction function, an individualized threshold tailored to an individual attribute corresponding to an input data object for the machine learning model disclosed by Bhide because all techniques address the same field of machine learning and by incorporating Bhide into Zhang and Gueret addresses individual fairness to better mitigate bias for trustworthy models [Bhide ¶1-2]
As to dependent claim 12, the rejection of claim 11 is incorporated, Zhang, Gueret and Bhide further teach a contextual bias correction function configured to output an individualized contextual threshold for the input data object based at least in part on the plurality of contextual attributes; and [Gueret context attributes (income, occupation) ¶17-19 used to calibrate a model ¶26 " determine groups of calibration data according to rules and feature attributes"]
a protected bias correction function configured to output an individualized protection threshold for the input data object based at least in part on the plurality of protected attributes. [Zhang threshold based on age (protected attribute) Col. 14-15 ln .62-6 "classification thresholds are set to the same for the age groups younger and older than 26"], [Gueret protected attributes (females, southerners) ¶17-19]
As to independent claim 18, Zhang teaches One or more non-transitory computer-readable media storing processor-executable instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising: [processors, memory and instructions ¶88]
receiving an aggregate bias correction function for a machine learning model; [receives settings that optimize for bias (correction function) including SPD and WAOD Col. 4 ln. 42-46 "optimize a machine learning model via the implementation of one or more user defined fairness policies that can mitigate statistical bias" Col. 13 ln. 4-16 "SPD and/or WAOD to measure one or more fairness criteria"]
generating, using the aggregate bias correction function, an individualized threshold corresponding to an input data object for the machine learning model, [generates a threshold based on metrics (attributes) Col. 2 ln. 4-24 "determining, by a system operatively coupled to a processor, a threshold setting for a machine learning model based on preferential weight values assigned to a fairness metric and a utility metric of the machine learning model"], [protected attributes Col. 7 ln. 61-67]
generating, using the machine learning model, a predictive output for the input data object based at least in part on the plurality of contextual attributes; [Y prediction output Col. 7-8 ln. 61-17 " “Y” can be a machine learning classifier's prediction"]
generating a bias adjusted output for the input data object based at least in part on a comparison between the individualized threshold and the predictive output, [output a visualization based on thresholds applied (bias based adjustments Col. 9 ln. 42-54 "visualization component 202 can evaluate a plurality of threshold settings for the machine learning model to generate one or more sample sets comprising various versions of the machine learning model"]
providing data indicative of the bias adjusted output [visual reflects changes Col. 9-10 ln. 42-11"one or more visualizations generated by the visualization component 202 can depict how changing the thresholds (e.g., classification thresholds) can impact SPD, WAPD, and/or utility of the machine learning model"]
Zhang does not specifically teach wherein the individualized threshold is based at least in part on (i) a plurality of contextual attributes of the input data object and (ii) a plurality of protected attributes of the input data object; wherein the bias adjusted output comprises a reduced bias relative to the predictive output without using a randomized component for the machine learning model;
However, Gueret teaches wherein the individualized threshold is based at least in part on (i) a plurality of contextual attributes of the input data object and (ii) a plurality of protected attributes of the input data object; [calibration defined threshold based on context attributes (income, occupation) and protected attributes (females, southerners) ¶24, ¶17-19 "characteristics of the user, such as age, gender, race, ethnicity, income, occupation, geographical location, and/or other distinguishable characteristics."…"rules are provided for four example groups: Youth (specifically 20 to 30 year old individuals), Females, Northerners, and Southerners. An example rule, as described herein, includes a definition and an objective"]
wherein the bias adjusted output comprises a reduced bias relative to the predictive output without using a randomized component for the machine learning model; [rule based calibration (non-random) that avoids bias ¶43 " rules associated with the one or more groups. Accordingly, based on the rules, the calibrated qualification model may avoid developing a bias"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the bias mitigation disclosed by Zhang by incorporating the wherein the individualized threshold is based at least in part on (i) a plurality of contextual attributes of the input data object and (ii) a plurality of protected attributes of the input data object; wherein the bias adjusted output comprises a reduced bias relative to the predictive output without using a randomized component for the machine learning model disclosed by Gueret because both techniques address the same field of machine learning and by incorporating Gueret into Zhang alleviates waste of computing resources an provides more desired results from models [Gueret ¶13]
Zhang and Gueret do not specifically teach generating, by the one or more processors and using the aggregate bias correction function, an individualized threshold tailored to an individual attribute corresponding to an input data object for the machine learning model.
However, Bhide teaches generating, by the one or more processors and using the aggregate bias correction function, an individualized threshold tailored to an individual attribute corresponding to an input data object for the machine learning model. [uses an individual bias threshold and (attributes ¶29) ¶9 "training a bias detector that learns to detect a sample that has an individual bias greater than a predetermined individual bias threshold value with constraints on a group bias"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the input data disclosed by Zhang and Gueret by incorporating the generating, by the one or more processors and using the aggregate bias correction function, an individualized threshold tailored to an individual attribute corresponding to an input data object for the machine learning model disclosed by Bhide because all techniques address the same field of machine learning and by incorporating Bhide into Zhang and Gueret addresses individual fairness to better mitigate bias for trustworthy models [Bhide ¶1-2]
As to dependent claim 19, the rejection of claim 18 is incorporated, Zhang, Gueret and Bhide further teach wherein the predictive output comprises a classification probability corresponding to one or more classifications, and wherein the bias adjusted output comprises a predicted classification from the one or more classifications. [Gueret confidence score and classification (qualified or not) ¶17 "The qualification model may be a binary classification model that provides a binary output (e.g., “qualified” or “not qualified”). Additionally, or alternatively, the qualification model may determine and/or indicate a confidence score associated with a prediction of whether the user is qualified"]
Claims 3-4 and 13-14 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang in view of Gueret and Bhide as applied to the rejection of claim 2 and 12 above, and further in view of Murugesan et al. (US 20220114225 A1 hereinafter Murugesan)
As to dependent claim 3, the combination of Zhang, Gueret and Bhide teach all the limitations of claim 2 that is incorporated.
Zhang, Gueret and Bhide do not specifically teach wherein the input data object is associated with a contextual tensor comprising the plurality of contextual attributes.
However, Murugesan teaches wherein the input data object is associated with a contextual tensor comprising the plurality of contextual attributes. [contextual tensor with user/product features ¶33 "a user context tensor X.sub.z.sup.xj that is formed from three feature vectors: user feature vectors of the i-th user and j-th user, and a product feature vector z"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the input data disclosed by Zhang, Gueret and Bhide by incorporating the wherein the input data object is associated with a contextual tensor comprising the plurality of contextual attributes disclosed by Murugesan because all techniques address the same field of machine learning and by incorporating Murugesan into Zhang, Gueret and Bhide better adapts models to user preferences and marketing strategies [Murugesan ¶33]
As to dependent claim 4, the rejection of claim 3 is incorporated, Zhang, Gueret, Bhide and Murugesan further teach wherein generating the individualized threshold for the input data object comprises: generating, by the processors, the individualized contextual threshold for the input data object by applying the contextual bias correction function to the contextual tensor; and [Gueret context attributes (income, occupation) ¶17-19 used to calibrate a model ¶26 " determine groups of calibration data according to rules and feature attributes"]
generating, by the processors, the individualized threshold for the input data object based at least in part on the individualized contextual threshold for the input data object. [Gueret calibration defined threshold based on context attributes (income, occupation) and protected attributes (females, southerners) ¶24 "calibration data further includes respective profile data associated with profiles of the units that are analyzed relative to a profile threshold with respect to meeting one or more qualifications"]
As to dependent claim 13, the combination of Zhang, Gueret and Bhide teach all the limitations of claim 12 that is incorporated.
Zhang, Gueret and Bhide do not specifically teach wherein the input data object is associated with a contextual tensor comprising the plurality of contextual attributes.
However, Murugesan teaches wherein the input data object is associated with a contextual tensor comprising the plurality of contextual attributes. [contextual tensor with user/product features ¶33 "a user context tensor X.sub.z.sup.xj that is formed from three feature vectors: user feature vectors of the i-th user and j-th user, and a product feature vector z"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the input data disclosed by Zhang, Gueret and Bhide by incorporating the wherein the input data object is associated with a contextual tensor comprising the plurality of contextual attributes disclosed by Murugesan because all techniques address the same field of machine learning and by incorporating Murugesan into Zhang, Gueret and Bhide better adapts models to user preferences and marketing strategies [Murugesan ¶33]
As to dependent claim 14, the rejection of claim 13 is incorporated, Zhang, Gueret, Bhide and Murugesan further teach wherein generating the individualized threshold for the input data object comprises: generating the individualized contextual threshold for the input data object by applying the contextual bias correction function to the contextual tensor; and [Gueret context attributes (income, occupation) ¶17-19 used to calibrate a model ¶26 " determine groups of calibration data according to rules and feature attributes"]
generating the individualized threshold for the input data object based at least in part on the individualized contextual threshold for the input data object. [Gueret calibration defined threshold based on context attributes (income, occupation) and protected attributes (females, southerners) ¶24 "calibration data further includes respective profile data associated with profiles of the units that are analyzed relative to a profile threshold with respect to meeting one or more qualifications"]
Claims 5-6 and 15-16 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang in view of Gueret, Bhide and Murugesan, as applied to the rejection of claim 4 and 14 above, and further in view of Alter (US 20180301223 A1)
As to dependent claim 5, the combination of Zhang, Gueret, Bhide and Murugesan teach all the limitations of claim 4 that is incorporated.
Zhang, Gueret, Bhide and Murugesan do not specifically teach wherein the input data object is associated with a protected tensor comprising the plurality of protected attributes.
However, Alter teaches wherein the input data object is associated with a protected tensor comprising the plurality of protected attributes. [tensor with demographic factors (protected attributes) ¶176 "Each tensor can represent or contain values for polling data"…"shared axes can include demographic factors (e.g., age, income, occupation, marital status, number of children, party membership, etc.)"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the input data disclosed by Zhang, Gueret, Bhide and Murugesan by incorporating the wherein the input data object is associated with a protected tensor comprising the plurality of protected attributes disclosed by Alter because all techniques address the same field of machine learning and by incorporating Alter into Zhang, Gueret, Bhide and Murugesan improve outcomes of predictions with more understanding of tensor inputs [Alter ¶78]
As to dependent claim 6, the rejection of claim 5 is incorporated, Zhang, Gueret, Bhide, Murugesan and Alter further teach wherein generating the individualized threshold for the input data object comprises: generating, by the processors, the individualized protection threshold for the input data object by applying the protected bias correction function to the protected tensor; and [Zhang threshold based on age (protected attribute) Col. 14-15 ln .62-6 "classification thresholds are set to the same for the age groups younger and older than 26"]
generating, by the processors, the individualized threshold for the input data object based at least in part on the individualized contextual threshold and the individualized protection threshold. [Zhang generates a threshold based on metrics (attributes) Col. 2 ln. 4-24 "determining, by a system operatively coupled to a processor, a threshold setting for a machine learning model based on preferential weight values assigned to a fairness metric and a utility metric of the machine learning model"], [protected attributes Col. 7 ln. 61-67]
As to dependent claim 15, the combination of Zhang, Gueret, Bhide and Murugesan teach all the limitations of claim 14 that is incorporated.
Zhang, Gueret, Bhide and Murugesan do not specifically teach wherein the input data object is associated with a protected tensor comprising the plurality of protected attributes.
However, Alter teaches wherein the input data object is associated with a protected tensor comprising the plurality of protected attributes. [tensor with demographic factors (protected attributes) ¶176 "Each tensor can represent or contain values for polling data"…"shared axes can include demographic factors (e.g., age, income, occupation, marital status, number of children, party membership, etc.)"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the input data disclosed by Zhang, Gueret, Bhide and Murugesan by incorporating the wherein the input data object is associated with a protected tensor comprising the plurality of protected attributes disclosed by Alter because all techniques address the same field of machine learning and by incorporating Alter into Zhang, Gueret, Bhide and Murugesan improve outcomes of predictions with more understanding of tensor inputs [Alter ¶78]
As to dependent claim 16, the rejection of claim 15 is incorporated, Zhang, Gueret, Bhide Murugesan and Alter further teach wherein generating the individualized threshold for the input data object comprises: generating the individualized protection threshold for the input data object by applying the protected bias correction function to the protected tensor; and [Zhang threshold based on age (protected attribute) Col. 14-15 ln .62-6 "classification thresholds are set to the same for the age groups younger and older than 26"]
generating the individualized threshold for the input data object based at least in part on the individualized contextual threshold and the individualized protection threshold. [Zhang generates a threshold based on metrics (attributes) Col. 2 ln. 4-24 "determining, by a system operatively coupled to a processor, a threshold setting for a machine learning model based on preferential weight values assigned to a fairness metric and a utility metric of the machine learning model"], [protected attributes Col. 7 ln. 61-67]
Claims 7 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang in view of Gueret, Bhide, Murugesan and Alter, as applied to the rejection of claim 6 and 16 above, and further in view of Everhart (US 20150052059 A1)
As to dependent claim 7, the combination of Zhang, Gueret, Bhide, Murugesan and Alter teach all the limitations of claim 6 that is incorporated.
Zhang, Gueret, Bhide, Murugesan and Alter do not specifically teach wherein the individualized threshold is an aggregate bias correction threshold that comprises a product of the individualized contextual threshold and the individualized protection threshold.
However, Everhart teaches wherein the individualized threshold is an aggregate bias correction threshold that comprises a product of the individualized contextual threshold and the individualized protection threshold. [aggregate threshold as a product ¶32 "aggregate threshold density"…"densities may be multiplied together to produce a product density. The product density may then be compared to the aggregate threshold density]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the thresholds disclosed by Zhang, Gueret, Bhide, Murugesan and Alter by incorporating the wherein the individualized threshold is an aggregate bias correction threshold that comprises a product of the individualized contextual threshold and the individualized protection threshold disclosed by Everhart because all techniques address the same field of machine learning and by incorporating Everhart into Zhang, Gueret, Bhide, Murugesan and Alter help protect sensitive information in models for a more secure system [Everhart ¶8]
As to dependent claim 17, the combination of Zhang, Gueret, Bhide, Murugesan and Alter teach all the limitations of claim 16 that is incorporated.
Zhang, Gueret, Bhide, Murugesan and Alter do not specifically teach wherein the individualized threshold is an aggregate bias correction threshold that comprises a product of the individualized contextual threshold and the individualized protection threshold.
However, Everhart teaches wherein the individualized threshold is an aggregate bias correction threshold that comprises a product of the individualized contextual threshold and the individualized protection threshold. [aggregate threshold as a product ¶32 "aggregate threshold density"…"densities may be multiplied together to produce a product density. The product density may then be compared to the aggregate threshold density]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the thresholds disclosed by Zhang, Gueret, Bhide, Murugesan and Alter by incorporating the wherein the individualized threshold is an aggregate bias correction threshold that comprises a product of the individualized contextual threshold and the individualized protection threshold disclosed by Everhart because all techniques address the same field of machine learning and by incorporating Everhart into Zhang, Gueret, Bhide, Murugesan and Alter help protect sensitive information in models for a more secure system [Everhart ¶8]
Claim 9 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang in view of Gueret and Bhide, as applied to the rejection of claim 8 above, and further in view of Renders et al. (US 20100014762 A1 hereinafter Renders)
As to dependent claim 9, the combination of Zhang, Gueret and Bhide teach all the limitations of claim 8 that is incorporated.
Zhang, Gueret and Bhide do not specifically teach wherein the individualized threshold comprises a modified classification threshold corresponding to the one or more classifications, and wherein the predicted classification is based at least in part on a comparison between the modified classification threshold and the classification probability.
However, Renders teaches wherein the individualized threshold comprises a modified classification threshold corresponding to the one or more classifications, and wherein the predicted classification is based at least in part on a comparison between the modified classification threshold and the classification probability. [Uses probabilities and a classification assignor to compare including thresholds ¶22 "rescaled probabilities 22 are input to a classification assignor 30 that assigns or associates with the input object 12 one or more class labels or identifications 32. In hard categorization"…"calibrated thresholds 34 that are used by the classification assignor 30 for selecting the class label or identification"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the thresholds disclosed by Zhang, Gueret and Bhide by incorporating the wherein the individualized threshold comprises a modified classification threshold corresponding to the one or more classifications, and wherein the predicted classification is based at least in part on a comparison between the modified classification threshold and the classification probability disclosed by Renders because all techniques address the same field of machine learning and by incorporating Renders into Zhang, Gueret and Bhide improves precision and performance of models in training [Renders ¶4]
As to dependent claim 20, the combination of Zhang, Gueret and Bhide teach all the limitations of claim 19 that is incorporated.
Zhang, Gueret and Bhide do not specifically teach wherein the individualized threshold comprises a modified classification threshold corresponding to the one or more classifications, and wherein the predicted classification is based at least in part on a comparison between the modified classification threshold and the classification probability.
However, Renders teaches wherein the individualized threshold comprises a modified classification threshold corresponding to the one or more classifications, and wherein the predicted classification is based at least in part on a comparison between the modified classification threshold and the classification probability. [Uses probabilities and a classification assignor to compare including thresholds ¶22 "rescaled probabilities 22 are input to a classification assignor 30 that assigns or associates with the input object 12 one or more class labels or identifications 32. In hard categorization"…"calibrated thresholds 34 that are used by the classification assignor 30 for selecting the class label or identification"]
Accordingly, it would have been obvious to a person of ordinary skill in the art before the effective filling date of the claimed invention to modify the thresholds disclosed by Zhang, Gueret and Bhide by incorporating the wherein the individualized threshold comprises a modified classification threshold corresponding to the one or more classifications, and wherein the predicted classification is based at least in part on a comparison between the modified classification threshold and the classification probability disclosed by Renders because all techniques address the same field of machine learning and by incorporating Renders into Zhang, Gueret and Bhide improves precision and performance of models in training [Renders ¶4]
Response to Arguments
Applicant's arguments filed 03/04/2026, with respect to 101, these rejections have been withdrawn.
Applicant's arguments filed 03/04/2026. In the remark, applicant argues that:
(1) Zhang and Gueret fails to teach “generating, by the one or more processors and using the aggregate bias correction function, an individualized threshold tailored to an individual attribute corresponding to an input data object for the machine learning model, wherein the individualized threshold is based at least in part on (i) a plurality of contextual attributes of the input data object and (ii) a plurality of protected attributes of the input data object; ” as recited by amended claim 1.
As to point (1), Applicant’s arguments with respect to claim 1 have been considered but are moot in view of a new ground of rejection as set forth above of Zhang in view of Gueret and Bhide. Further see Zhang’s example of 0.6 people threshold (see Col. 14 ln. 36-50). Also see Gueret’s individual characteristics such as age, gender, race and profile threshold (see ¶24).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Applicant is required under 37 C.F.R. § 1.111(c) to consider these references fully when responding to this action.
Wei et al. (US 20210374581 A1) teaches protected attributes and mitigating disparities among them (see ¶48-49)
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to BEAU SPRATT whose telephone number is (571)272-9919. The examiner can normally be reached M-F 8:30-5 PST.
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, Jennifer Welch can be reached on 5712127212. 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.
/BEAU D SPRATT/ Primary Examiner, Art Unit 2143
Prosecution Insights