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 .
Priority
The present application has a provisional application No. 63/470,408 filed on June 01, 2023.
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 1-23 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding Claim 1
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 1 is a method claim thus it falls into one of the four categories of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding independent claim 1, following limitations recite a judicial exception:
“inferring, from an input that contains a value of a feature that has a plurality of multipliers, a probability of a class”
[Mathematical Calculations] – inferring a probability or calculating the probability of the class uses mathematical computation which recites to an abstract idea.
“selecting based on the value of the feature in the input, from the plurality of multipliers of the feature, a multiplier that is specific to both of the feature and the value of the feature”
[Mental Process] – selecting one over another based on some kind of indicator like the value of the feature simply requires an action of comparison between the values which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
“Classifying the input based on a multiplicative product of the probability of the class and the multiplier that is specific to both of the feature and the value of the feature”
[Mental Process] – classifying using some kind of indicator such as a number simply requires an action of matching between the values and the classes which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
[Mathematical Process] – using multiplicative product of the probability of the class purely uses mathematical computations which recites to an abstract idea.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 1, the claim recites additional elements of
“wherein the method is performed by one or more computers”
Performing the method using a computer component is recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional element does no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer. (see MPEP 2106.05(f)) This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 2
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 2 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 2, following limitations recite a judicial exception:
“inferring, from a second input that contains a second value of the feature, a second probability of the class”
[Mathematical Calculations] – inferring a probability or calculating the probability of the class uses mathematical computation which recites to an abstract idea.
“selecting based on the second value of the feature in the second input, from the plurality of multipliers of the feature, a second multiplier that is specific to both of the feature and the second value of the feature”
[Mental Process] – selecting one over another based on some kind of indicator like the value of the feature simply requires an action of comparison between the values which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
“Classifying the second input based on the second multiplier that is specific to both of the feature and the second value of the feature and not based on the first value-specific multiplier”
[Mental Process] – classifying using some kind of indicator such as a number or a multiplier simply requires an action of matching between the values and the classes which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 2 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 3
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 3 is a dependent claim of 2, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 3, following limitations recite a judicial exception:
“inferring, from a third input that contains the second value of the feature, a third probability of the class”
[Mathematical Calculations] – inferring a probability or calculating the probability of the class uses mathematical computation which recites to an abstract idea.
“inferring, from a fourth input that contains the second value of the feature, a fourth probability of the class”
[Mathematical Calculations] – inferring a probability or calculating the probability of the class uses mathematical computation which recites to an abstract idea.
“detecting, not based on the first probability of the class, a minimum probability for the second value of the feature and a maximum probability for the second value of the feature based on: the second probability of the class, the third probability of the class, and the fourth probability of the class”
[Mental Process] – detecting the minimum and the maximum among many probabilities simply requires an action of comparison of each number to another which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 3 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 4
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 4 is a dependent claim of 3, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 4, following limitations recite a judicial exception:
“generating the second multiplier that is specific to both of the feature and the second value of the feature based on the minimum probability for the second value of the feature and the maximum probability for the second value of the feature”
[Mathematical Calculations] – generating or calculating a multiplier based on the minimum and maximum probabilities requires mathematical computations which recites to an abstract idea.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 4 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 5
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 5 is a dependent claim of 4, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 5, following limitations recite a judicial exception:
“generating the second multiplier that is specific to both of the feature and the second value of the feature is based on a first ratio of the minimum probability for the second value of the feature over the maximum probability for the second value of the feature and a second ratio of the maximum probability for the second value of the feature over the minimum probability for the second value of the feature”
[Mathematical Calculations] – generating the multiplier based on ratios of both max/min and min/max simply uses mathematical computation which recites to an abstract idea.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 5 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 6
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 6 is a dependent claim of 5, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 6, following limitations recite a judicial exception:
“the second multiplier that is specific to both of the feature and the second value of the feature is generated in a range from the first ration to the second ratio”
[Mathematical Calculations] – generating the multiplier based the range of the two ratios requires mathematical comparisons and computations which recites to an abstract idea.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 6 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 7
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 7 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 7, following limitations recite a judicial exception:
“generating, by a bi-objective optimizer, the plurality of multipliers of the feature”
[Mathematical Calculations] – generating multiple multipliers using the optimizer simply requires mathematical computations to optimize and find the best multipliers accordingly, thus it recites to an abstract idea.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 7 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 8
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 8 is a dependent claim of 7, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 8, following limitations recite a judicial exception:
“receiving, by the bi-objective optimizer, two validation scores that are based on the plurality of multipliers of the feature”
[Mathematical Calculations] – calculating validation scores based on the multipliers uses mathematical computations which recites to an abstract idea.
[Mental Process] – receiving the scores by the optimizer which can be considered as receiving the scores for optimization based on that scores which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 8 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 9
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 9 is a dependent claim of 8, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 9, following limitations recite a judicial exception:
“the two validation scores that are based on the plurality of multipliers of the feature are a fitness score and a fairness score”
[Mental Process] – these fitness and fairness scores measured based on the multipliers simply requires an action of comparing the multipliers and deciding the scores based on those multipliers which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 9 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 10
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 10 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 10, following limitations recite a judicial exception:
“said inferring is performed by a classifier that was trained”
[Mathematical Calculations] – inferring a probability or calculating the probability of the class uses mathematical computation which recites to an abstract idea
[Mental Process] – a trained classifier is simply an action of comparing things to classify them to a specific class with a higher accuracy which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
“the method further comprises without retraining the classifier: adjusting the multiplier that is specific to both of the feature and the value of the feature; reclassifying the input”
[Mental Process] – adjusting the multipliers can simply mean change the number according to the needs and reclassifying according to the changed multiplier is again a simple action of comparing and matching which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 10 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 11
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 11 is a dependent claim of 10, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 11, following limitations recite a judicial exception:
“said adjusting and said reclassifying do not use the classifier”
[Mental Process] – adjusting and reclassifying are actions of modification and reclassification according to the modification. Also, the classifier is just steps of actions of comparing and matching thus all of them involve observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 11 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 12
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 12 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 12, following limitations recite a judicial exception:
“generating multiple pluralities of multipliers of the feature”
[Mathematical Calculations] – generating the multipliers of the feature based on some kind of indicators requires mathematical computations which recites to an abstract idea.
“detecting a subset of the multiple pluralities of multipliers of the feature that are on a bi-objective Pareto frontier”
[Mental Process] – detecting the subset of the multipliers that are on the bi-objective Pareto frontier requires drawing a curve with these multipliers and detecting those multipliers on the curve simply requires comparing and matching of multipliers to the curve which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 12 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 13
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 13 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 13, following limitations recite a judicial exception:
“inferring, from the input that contains the value of the feature, a second probability of a second class and a third probability of a third class”
[Mathematical Calculations] – inferring a probability or calculating the probability of the class uses mathematical computation which recites to an abstract idea.
“rescaling, based on said multiplicative product of the probability of the class and the multiplier that is specific to both of the feature and the value of the feature, the second probability of the second class and the third probability of the third class”
[Mathematical Calculation] – rescaling or normalizing the probability after the multiplicative product requires mathematical computations which recites to an abstract idea.
“said classifying the input is not based on said rescaling the second probability of the second class”
[Mental Process] – classifying using some kind of indicator such as a number simply requires an action of matching between the values and the classes which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 13 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 14
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 14 is a dependent claim of 1, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 14, following limitations recite a judicial exception:
“inferring a second probability of a second class that is less than the probability of the first class and a third probability of a third class”
[Mathematical Calculations] – inferring a probability or calculating the probability of the class uses mathematical computation which recites to an abstract idea.
“said classifying the input comprises classifying the input as the second class”
[Mental Process] – classifying using some kind of indicator such as a number simply requires an action of matching between the values and the classes which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 14 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 15
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 15 is a method claim thus it falls into one of the four categories of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding independent claim 15, following limitations recite a judicial exception:
“generating, by a bi-objective optimizer, a first multiplier for a first class of a plurality of mutually exclusive classes and a second multiplier for a second class of the plurality of mutually exclusive classes”
[Mathematical Calculations] – generating multiple multipliers using the optimizer simply requires mathematical computations to optimize and find the best multipliers accordingly, thus it recites to an abstract idea.
“generating, from an input, an inference that contains a probability of the first class and a probability of the second class”
[Mathematical Calculations] – generating an inference containing probability or calculating the probability of the classes uses mathematical computation which recites to an abstract idea.
“Classifying the input based on a first multiplicative product of the probability of the first class and the first multiplier, and a second multiplicative product of the probability of the second class and the second multiplier”
[Mental Process] – classifying using some kind of indicator such as a number simply requires an action of matching between the values and the classes which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
[Mathematical Process] – using multiplicative product of the probability of the class purely uses mathematical computations which recites to an abstract idea.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 15, the claim recites additional elements of
“wherein the method is performed by one or more computers”
Performing the method using a computer component is recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional element does no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer. (see MPEP 2106.05(f)) This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 16
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 16 is a dependent claim of 15, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 16, following limitations recite a judicial exception:
“said classifying the input uses a plurality of multipliers that contains the first multiplier and the second multiplier”
[Mental Process] – classifying using some kind of indicator such as a number or a multiplier simply requires an action of matching between the values and the classes which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
[Mathematical Calculations] – using multipliers indicates applying these numbers to other numbers which involves mathematical computations and thus it recites to an abstract idea.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 16 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 17
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 17 is a non-transitory computer readable medium claim thus it falls into one of the four categories of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding independent claim 17, following limitations recite a judicial exception:
“inferring, from an input that contains a value of a feature that has a plurality of multipliers, a probability of a class”
[Mathematical Calculations] – inferring a probability or calculating the probability of the class uses mathematical computation which recites to an abstract idea.
“selecting based on the value of the feature in the input, from the plurality of multipliers of the feature, a multiplier that is specific to both of the feature and the value of the feature”
[Mental Process] – selecting one over another based on some kind of indicator like the value of the feature simply requires an action of comparison between the values which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
“Classifying the input based on a multiplicative product of the probability of the class and the multiplier that is specific to both of the feature and the value of the feature”
[Mental Process] – classifying using some kind of indicator such as a number simply requires an action of matching between the values and the classes which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
[Mathematical Process] – using multiplicative product of the probability of the class purely uses mathematical computations which recites to an abstract idea.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 17, the claim recites additional elements of
“executed by one or more processors”
Executing the instructions using a computer component is recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional element does no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer. (see MPEP 2106.05(f)) This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
Regarding Claim 18
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 18 is a dependent claim of 17, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 18, following limitations recite a judicial exception:
“inferring, from a second input that contains a second value of the feature, a second probability of the class”
[Mathematical Calculations] – inferring a probability or calculating the probability of the class uses mathematical computation which recites to an abstract idea.
“selecting based on the second value of the feature in the second input, from the plurality of multipliers of the feature, a second multiplier that is specific to both of the feature and the second value of the feature”
[Mental Process] – selecting one over another based on some kind of indicator like the value of the feature simply requires an action of comparison between the values which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
“Classifying the second input based on the second multiplier that is specific to both of the feature and the second value of the feature and not based on the first value-specific multiplier”
[Mental Process] – classifying using some kind of indicator such as a number or a multiplier simply requires an action of matching between the values and the classes which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 18 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 19
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 19 is a dependent claim of 17, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 19, following limitations recite a judicial exception:
“generating, by a bi-objective optimizer, the plurality of multipliers of the feature”
[Mathematical Calculations] – generating multiple multipliers using the optimizer simply requires mathematical computations to optimize and find the best multipliers accordingly, thus it recites to an abstract idea.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 19 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 20
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 20 is a dependent claim of 17, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 20, following limitations recite a judicial exception:
“said inferring is performed by a classifier that was trained”
[Mathematical Calculations] – inferring a probability or calculating the probability of the class uses mathematical computation which recites to an abstract idea
[Mental Process] – a trained classifier is simply an action of comparing things to classify them to a specific class with a higher accuracy which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
“the instructions further cause without retraining the classifier: adjusting the multiplier that is specific to both of the feature and the value of the feature; reclassifying the input”
[Mental Process] – adjusting the multipliers can simply mean change the number according to the needs and reclassifying according to the changed multiplier is again a simple action of comparing and matching which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 20 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 21
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 21 is a dependent claim of 17, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 21, following limitations recite a judicial exception:
“generating multiple pluralities of multipliers of the feature”
[Mathematical Calculations] – generating the multipliers of the feature based on some kind of indicators requires mathematical computations which recites to an abstract idea.
“detecting a subset of the multiple pluralities of multipliers of the feature that are on a bi-objective Pareto frontier”
[Mental Process] – detecting the subset of the multipliers that are on the bi-objective Pareto frontier requires drawing a curve with these multipliers and detecting those multipliers on the curve simply requires comparing and matching of multipliers to the curve which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 21 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 22
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 22 is a dependent claim of 17, thus it falls within the same category of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding dependent claim 22, following limitations recite a judicial exception:
“inferring a second probability of a second class that is less than the probability of the first class and a third probability of a third class”
[Mathematical Calculations] – inferring a probability or calculating the probability of the class uses mathematical computation which recites to an abstract idea.
“said classifying the input comprises classifying the input as the second class”
[Mental Process] – classifying using some kind of indicator such as a number simply requires an action of matching between the values and the classes which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
The claim 22 does not recite any additional elements other than abstract ideas, so it does not integrate into a practical application. Thus, this claim is directed to the abstract idea.
Regarding Claim 23
Step 1 – whether the claim falls within any statutory category. See MPEP 2016.03
Claim 23 is a non-transitory computer readable medium claim thus it falls into one of the four categories of statutory subject matter.
Step 2A Prong 1 – whether the claim recites a judicial exception. See MPEP 2106.04, subsection II.
Regarding independent claim 23, following limitations recite a judicial exception:
“generating, by a bi-objective optimizer, a first multiplier for a first class of a plurality of mutually exclusive classes and a second multiplier for a second class of the plurality of mutually exclusive classes”
[Mathematical Calculations] – generating multiple multipliers using the optimizer simply requires mathematical computations to optimize and find the best multipliers accordingly, thus it recites to an abstract idea.
“generating, from an input, an inference that contains a probability of the first class and a probability of the second class”
[Mathematical Calculations] – generating an inference containing probability or calculating the probability of the classes uses mathematical computation which recites to an abstract idea.
“Classifying the input based on a first multiplicative product of the probability of the first class and the first multiplier, and a second multiplicative product of the probability of the second class and the second multiplier”
[Mental Process] – classifying using some kind of indicator such as a number simply requires an action of matching between the values and the classes which involves observations, evaluations, judgments, and opinions that is capable of being performed in the human mind with the assistance of paper and pen.
[Mathematical Process] – using multiplicative product of the probability of the class purely uses mathematical computations which recites to an abstract idea.
Step 2A Prong 2 – whether the claim recites additional elements that integrate the exception into a practical application of the exception?
Regarding Claim 23, the claim recites additional elements of
“executed by one or more processors”
Performing the method using a computer component is recited at a high level of generality and is merely adding words “apply it” to the judicial exception. (see MPEP 2106.05(f))
[Even when viewed in combination, the additional element does no more than automate the mental processes that a person could perform, using computer components as a tool, thus the claim as a whole does not integrate into a practical application.]
Step 2B – whether the claim as a whole amount to significantly more than the judicial exception? I.e. Are there any additional elements (features/limitations/step) recited in the claim beyond the abstract idea?
The claim does not provide an inventive concept (significantly more than the abstract idea). The claim is ineligible.
As explained above, the additional element [1] is considered a mere instruction to apply an exception to the generic computer. (see MPEP 2106.05(f)) This limitation remains a mere instruction to apply an exception even upon reconsideration. Even when considered in combination, the additional element represents a mere instruction to apply an exception, which cannot provide an inventive concept.
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.
Claims 1, 2, 10, 11, 17, 18, 20 are rejected under 35 U.S.C. 102 as being anticipated over Bhaila et al. (Bhaila), Non-Patent Literature, “Fair Collective Classification in Networked Data”, published on 2022, Pages: 10.
As to independent Claim 1,
Bhaila teaches a method comprising:
Inferring, from an input that contains a value of a feature that has a plurality of multipliers, a probability of a class(Bhaila, Pg1416-1417, Right Column, Subsection A. Collective Classification, Lines4-9, “The input is an unweighted, undirected, and attributed graph G = (V,E,X,Y) where V is a set of N nodes, E is a set of edges that connect node pairs in V, X represents the node feature matrix, and Y denotes node labels. Each node vi is defined by its feature vector xi which includes a sensitive attribute si”, Pg1417, Left Column, Lines6-8, “The task then is to simultaneously infer the values yi for vi or a probability distribution over those label values” and Pg1419, Left Column, Paragraph1, Lines14-16, “In our setup with binary s and binary y, we derive these weights for four combinations of s and y”, wherein each node vi corresponds to each input of the claimed invention such that each input vi has a sensitive feature s and its value si. Then it infers probability value yi or probability distribution over the label values of the input, vi, which corresponds to the probability of a class or a label, meaning each input will be inferred with multiple yi according to the number of labels. There is the plurality of weights according to the combination or at least four, which corresponds to the plurality of multipliers, thus it is equivalent to the claimed invention.)
selecting based on the value of the feature in the input, from the plurality of multipliers of the feature, a multiplier that is specific to both of the feature and the value of the feature (Bhaila, Pg1419, Right Column, Paragraph1, Lines4-5, “We also apply the weight based on the node’s own sensitive attribute and predicted node label”, Pg1419, Left Column, Paragraph1, Lines14-16, “In our setup with binary s and binary y, we derive these weights for four combinations of s and y” and Pg1419, Right Column, Paragraph1, Lines9-10, “wc,si is the weight due to node vi’s own sensitive attribute and the label to be estimated”, wherein as mentioned above there are multiple multipliers to the sensitive feature s such that the multiplier or the weight is specific to the feature value si so using the weight wc,si , is equivalent to the claimed invention)
classifying the input based on a multiplicative product of the probability of the class and the multiplier that is specific to both of the feature and the value of the feature (Bhaila, Pg1419, Left Column, Paragraph1, Lines1-4, “We propose to incorporate this reweighting technique in the collective classification model during each iterative update to assign weights to nodes for each possible combination of s and y”, Pg1419, Algorithm2 Fair Collective Classification via Node Reweighting, Line6,
PNG
media_image1.png
24
203
media_image1.png
Greyscale
and Pg1419, Equation8,
PNG
media_image2.png
46
350
media_image2.png
Greyscale
, wherein the summation part in the equation 8 shows that the feature specific weight is multiplied to the probability of a class c, which corresponds to the multiplicative product. Also, this multiplication is incorporated into the classification model such that the algorithm 2 shows that c^i(t+1) is the value obtained from the multiplication to classify the input with the final label y^i which is equivalent to the claimed invention)
wherein the method is performed by one or more computers (Bhaila, Pg1421, Left Column, Subsection B. Compared Methods, Lines14-15, “All methods evaluated in this section are implemented on the basis of AIF360”, wherein AIF360 is an open-source AI fairness toolkit that runs on a computer with the code, which is equivalent to the claimed invention.)
As to dependent Claim 2,
Bhaila teaches all the limitations of Claim 1 and further teaches the method of Claim 1 wherein:
said probability of the class is a first probability of the class (Bhaila, Pg1416-1417, Right Column, Subsection A. Collective Classification, Lines4-9, “The input is an unweighted, undirected, and attributed graph G = (V,E,X,Y) where V is a set of N nodes, E is a set of edges that connect node pairs in V, X represents the node feature matrix, and Y denotes node labels. Each node vi is defined by its feature vector xi which includes a sensitive attribute si” and Pg1417, Left Column, Lines6-8, “The task then is to simultaneously infer the values yi for vi or a probability distribution over those label values”, wherein as mentioned above in Claim1, the probability yi is one of the label values of the input vi such that it inherently encompasses first probability of the label or the class for the first input)
said multiplier that is specific to both of the feature and the value of the feature is a first value-specific multiplier (Bhaila, Pg1419, Right Column, Paragraph1, Lines9-10, “wc,si is the weight due to node vi’s own sensitive attribute and the label to be estimated”, wherein as mentioned in Claim 1 that the multiplier or the weight wc,si is specific to the feature and the value of the feature as ‘si’ represents the feature value of the feature s, meaning if si is the first value of the feature, then it is equivalent to the claimed invention)
the method further comprises:
inferring, from a second input that contains a second value of the feature, a second probability of the class (Bhaila, Pg1416-1417, Right Column, Subsection A. Collective Classification, Lines4-9, “The input is an unweighted, undirected, and attributed graph G = (V,E,X,Y) where V is a set of N nodes, E is a set of edges that connect node pairs in V, X represents the node feature matrix, and Y denotes node labels. Each node vi is defined by its feature vector xi which includes a sensitive attribute si”, Pg1417, Left Column, Lines6-8, “The task then is to simultaneously infer the values yi for vi or a probability distribution over those label values” and Pg1419, Left Column, Paragraph1, Lines14-16, “In our setup with binary s and binary y, we derive these weights for four combinations of s and y”, wherein as mentioned in Claim 1, the input vi is not a single input but it comprises multiple inputs such as v1, v2, … and so on. Thus, each input vi will be inferred with its own unique probability of the class, yi or y1 and y2 respect to the input, which is equivalent to the claimed invention)
selecting based on the second value of the feature in the second input, from the plurality of multipliers of the feature, a second multiplier that is specific to both of the feature and the second value of the feature (Bhaila, Pg1419, Right Column, Paragraph1, Lines4-5, “We also apply the weight based on the node’s own sensitive attribute and predicted node label”, Pg1419, Left Column, Paragraph1, Lines14-16, “In our setup with binary s and binary y, we derive these weights for four combinations of s and y” and Pg1419, Right Column, Paragraph1, Lines9-10, “wc,si is the weight due to node vi’s own sensitive attribute and the label to be estimated”, wherein as mentioned in Claim 1, each weight or multiplier is the sensitive feature-specific and there is the plurality of multipliers which the feature s comprises, thus it inherently means there is a second multiplier or a multiplier that is specific to the second value of the feature for the second input)
classifying the second input based on the second multiplier that is specific to both of the feature and the second value of the feature and not based on the first value-specific multiplier (Bhaila, Pg1419, Left Column, Paragraph1, Lines1-4, “We propose to incorporate this reweighting technique in the collective classification model during each iterative update to assign weights to nodes for each possible combination of s and y”, Pg1419, Algorithm2 Fair Collective Classification via Node Reweighting, Line6,
PNG
media_image1.png
24
203
media_image1.png
Greyscale
and Pg1419, Equation8,
PNG
media_image2.png
46
350
media_image2.png
Greyscale
, wherein as mentioned above, the final classification is according to the weight or multiplier that is specific to the sensitive feature si from the plurality of multipliers of the feature which inherently means this si could be either the first value or the second value of the feature not both, which is equivalent to the claimed invention.)
As to dependent Claim 10,
Bhaila teaches, as mentioned above, all the limitations of Claim 1. Bhaila teaches about classifying an input, which has a feature consists of a plurality of multipliers, based on the inferred probability of the class by multiplying the probability to the multiplier. Bhaila further teaches the method of claim 1 wherein:
said inferring is performed by a classifier that was trained (Bhaila, Pg1420, Right Column, Subsection b) Label Flipping (LF), Paragraph2, Lines1-3, “Both of these postprocessing techniques can be applied over final CC predictions without modifying the local classifier, the relational classifier, and the inference method”, wherein as mentioned in Claim 1 Bhaila teaches about the inference method of each input vi to yield the class probability, yi , in which the classifier that infers the probability does not required to be retrained as it uses post-processing techniques which inherently uses trained classifiers, thus it is equivalent to the claimed invention);
the method further comprising without retraining the classifier:
adjusting the multiplier that is specific to both of the feature and the value of the feature (Bhaila, Pg1419, Left Column, Algorithm2, Line4 or Equation7,
PNG
media_image3.png
73
671
media_image3.png
Greyscale
and Pg1419, Left Column, Paragraph2, Lines1-4, “Then for each iteration of inference, we derive weights using known labels and predictions obtained from fR in the previous iteration as shown in Algorithm 2 until the predictions converge to a fair and stable state and Pg1419, Equation 6,
PNG
media_image4.png
101
571
media_image4.png
Greyscale
, wherein as mentioned above, the classifiers does not further retrain and this equation 7 shows that the algorithm iteratively derives or adjusts the weight, w(t)c,a to w(t+1)c,a that is specific to the feature s and the value of the feature si or here uses ‘a’ which is inherently equivalent to si shown in the equation 6 or s=a, which is functionally equivalent to the claimed invention);
reclassifying the input (Bhaila, Pg1419, Left Column, Paragraph1, Lines1-4, “We propose to incorporate this reweighting technique in the collective classification model during each iterative update to assign weights to nodes for each possible combination of s and y”, Pg1419, Algorithm2 Fair Collective Classification via Node Reweighting, Line6,
PNG
media_image5.png
55
426
media_image5.png
Greyscale
and Pg1417, Table1, Notation,
PNG
media_image6.png
41
801
media_image6.png
Greyscale
PNG
media_image7.png
34
648
media_image7.png
Greyscale
, wherein as mentioned in Claim1, this classification is done using the trained classifier and iteratively adjusting the weights such that by calculating
PNG
media_image8.png
55
85
media_image8.png
Greyscale
which is the new probability distribution from applying the new weight mentioned above to yield the
PNG
media_image9.png
53
103
media_image9.png
Greyscale
which is the new value of the label to reclassify the input using this value, thus it is functionally equivalent to the claimed invention.)
As to dependent Claim 11,
Bhaila teaches, as mentioned above, all the limitations of Claim 10. Bhaila teaches about reclassifying an input by adjusting the multiplier that is specific to the feature and the value of the feature without retraining the classifier. Bhaila further teaches the method of claim 10 wherein said adjusting and said reclassifying do not use the classifier (Bhaila, Pg1417, Table1, Notation,
PNG
media_image10.png
35
470
media_image10.png
Greyscale
, Pg1417, Left Column, 1) Local Classifier, Lines 1-2, “The local classifier fL estimates label probabilities P(yi|xi) using only node attributes xi” and Pg1419, Algorithm 2, wherein the algorithm shows that the classifier fL which is the trained classifier that infers the probabilities of the class in the beginning like a black-box. According to the algorithm, this classifier is no longer used in adjusting and reclassification process as it is used only in the line 1 to infer the probabilities of the class, thus it is functionally equivalent to the claimed invention.
PNG
media_image11.png
621
644
media_image11.png
Greyscale
As to dependent Claim 17,
it is a non-transitory computer-readable media claim that contains similar limitations of Claim 1 and thus rejected under the same rationale.
As to dependent Claim 18,
it is a non-transitory computer-readable media claim that contains similar limitations of Claim 2 and thus rejected under the same rationale.
As to dependent Claim 20,
it is a non-transitory computer-readable media claim that contains similar limitations of Claim 10 and thus rejected under the same rationale.
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 3 is rejected under 35 U.S.C. 103 as being unpatentable over Bhaila as mentioned in Claims 1 and 2 in view of Pranto et al. (Pranto), Non-Patent Literature, “Performance Analysis of Ensemble Based Approaches to Mitigate Class Imbalance Problem After Applying Normalization”, published on July 2021, Pages: 5.
As to dependent Claim 3,
Bhaila teaches all the limitations of Claim 2 and further teaches the method of Claim 2 further comprising:
Inferring, from a third input that contains the second value of the feature, a third probability of the class (Bhaila, Pg1416-1417, Right Column, Subsection A. Collective Classification, Lines4-9, “The input is an unweighted, undirected, and attributed graph G = (V,E,X,Y) where V is a set of N nodes, E is a set of edges that connect node pairs in V, X represents the node feature matrix, and Y denotes node labels. Each node vi is defined by its feature vector xi which includes a sensitive attribute si”, Pg1417, Left Column, Lines6-8, “The task then is to simultaneously infer the values yi for vi or a probability distribution over those label values” and Pg1419, Left Column, Paragraph1, Lines14-16, “In our setup with binary s and binary y, we derive these weights for four combinations of s and y”, wherein as mentioned in Claim 2, the input vi is not a single input but it comprises multiple inputs such as v1, v2, … and so on. Thus, each input vi will be inferred with its own unique probability of the class, yi or y1 and y2 respect to the input, which is equivalent to the claimed invention)
Inferring, from a fourth input that contains the second value of the feature, a fourth probability of the class (Bhaila, Pg1416-1417, Right Column, Subsection A. Collective Classification, Lines4-9, “The input is an unweighted, undirected, and attributed graph G = (V,E,X,Y) where V is a set of N nodes, E is a set of edges that connect node pairs in V, X represents the node feature matrix, and Y denotes node labels. Each node vi is defined by its feature vector xi which includes a sensitive attribute si”, Pg1417, Left Column, Lines6-8, “The task then is to simultaneously infer the values yi for vi or a probability distribution over those label values” and Pg1419, Left Column, Paragraph1, Lines14-16, “In our setup with binary s and binary y, we derive these weights for four combinations of s and y”, wherein as mentioned above, the input vi is not a single input but it comprises multiple inputs such as v1, v2, … and so on. Thus, each input vi will be inferred with its own unique probability of the class, yi or y1 and y2 respect to the input, which is equivalent to the claimed invention)
Bhaila teaches about each input vi or v1, v2, …, vn has its own unique probability, yi, for the class and the three inputs (second, third and fourth) have the second value of the feature. However, Bhaila fails to teach the following limitation, but from the same field of endeavor, Pranto teaches detecting, not based on the first probability of the class, a minimum probability for the second value of the feature and a maximum probability for the second value of the feature based on: the second probability of the class, the third probability of the class, and the fourth probability of the class (Pranto, Pg2, Section III, normalization technique 1, "1) min-max: For the minimum value min(xi) and the maximum value max(xi) of the data set, min-max normalization can be represented as Eqn. 1.
PNG
media_image12.png
200
400
media_image12.png
Greyscale
", wherein as mentioned above, there are 3 inputs where they share the feature or the second value of the feature and applying the min-max normalization technique of choosing the maximum and the minimum values, or the probabilities, which is functionally equivalent to the claimed invention.)
Bhaila and Pranto are analogous to the claimed invention as they are from the same endeavor of machine learning-based data classification and model parameter optimization. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine classification using sensitive feature weights of Bhaila with the min/max bounding margins of statistical scaling of Pranto. The motivation is recited by Pranto (Pranto, Pg1, Right Column, Lines7-9, “But these methods are hindered by several drawbacks such as sampling methods may lose information or cause over-fitting”), such that over-fitting is a serious problem that hinders the overall performance as the computational requirement increases significantly, thus building a boundary to limit the computation requirements is necessary to avoid the over-fitting situations.
Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over Bhaila and Pranto mentioned in Claim 3 in view of Gidaris et al. (Gidaris), Non-Patent Literature, “Generating Classification Weights with GNN Denoising Autoencoders for Few-Shot Learning”, published on July 2019, Pages:10.
As to dependent Claim 4,
The combination of Bhaila and Pranto teaches, as mentioned above, all the limitations of Claim 3. Bhaila teaches about the inferring probabilities of the class for the three inputs that each input has the second value of the feature and Pranto teaches about finding the maximum and the minimum probabilities among the three for effective pruning range. However, the combination does not teach the following limitation, but from the same field of endeavor, Gidaris teaches the method of Claim 3 further comprising generating the second multiplier that is specific to both of the feature and the second value of the feature based on the minimum probability for the second value of the feature and the maximum probability for the second value of the feature (Gidaris, Pg22, Subsection Our approach, Lines20-29, “In this context, in order to be able to recognize novel classes one must be able to generate classification weight vectors for them. So, the goal of our work is to learn a meta model that fulfills exactly this task: i.e., given a set of novel classes with few training examples for each of them, as well as the classification weights of the base classes, it learns to output a new set of classification weight vectors (both for the base and novel classes) that can then be used from the feature classifier in order to classify in a unified way both types of classes”, wherein the combination of the three will functionally match the limitation of generating a weight that is specific to the second value of the feature (corresponding second multiplier) between the range of the minimum probability and the maximum probability for efficient pruning procedure.)
Bhaila, Pranto and Gidaris are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based data classification and model parameter optimization. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine adjusting classification outputs by applying class-specific multiplier weights of Bhaila, calculating the minimum and maximum values of a data distribution to establish boundaries of Pranto with dynamically generating classification weight vectors of Gidaris. The motivation is as recited by Gidaris (Gidaris, Pg22, Left Column, Subsection. DAE based model parameters generation, “Learning to perform such a meta-learning task, i.e., inferring the classification weights of a set of classes, is a difficult meta-problem that requires plenty of training data in order to be reliably solved. However, having access to such a large pool of data is not always possible; or otherwise stated the training data available for learning such meta-tasks might never be enough”) such that dynamically generating weights with limited data is not so reliable, thus combining the solid pruning range created from the actual probabilities of the class to efficiently handle the weight or multiplier generation specific to the sensitive feature (the second multiplier corresponding to the second value of the feature).
Claims 5 and 6 are rejected under 35 U.S.C. 103 as being unpatentable over Bhaila, Pranto and Gidaris as mentioned in Claim 4 in view of Pleiss et al. (Pleiss), Non-Patent Literature listed in IDS filed on 02/09/2024, “On Fairness and Calibration”, published on Nov 2017, Pages: 19.
As to dependent Claim 5,
The combination of Bhaila, Pranto and Gidaris teaches, as mentioned above, all the limitations of Claim 4. It teaches about generating a second multiplier that is specific to both of the feature and the second value of the feature. However, the combination does not teach the following limitation, but from the same field of endeavor, Pleiss teaches the method of Claim 4 wherein said generating the second multiplier that is specific to both of the feature and the second value of the feature is
based on a first ratio of the minimum probability for the second value of the feature over the maximum probability for the second value of the feature and a second ratio of the maximum probability for the second value of the feature over the minimum probability for the second value of the feature (Pleiss, Pg6, Paragraph2, Lines6-8, "In Lemma 4 in Section S2, we show that if
PNG
media_image13.png
200
400
media_image13.png
Greyscale
then the cost of h~_2 is a linear interpolation between the costs of h_2 and h^u2 (Figure 2d). More formally, we have that g2(h~_2) = (1-alpha)g2(h2) + alpha g2 (h^u2)), and thus setting
PNG
media_image14.png
200
400
media_image14.png
Greyscale
ensures that g2(g2(h~_2) = g1(h1) as desired” and Pg14, Lemma 4,
PNG
media_image15.png
200
400
media_image15.png
Greyscale
, wherein the two ratios are simply any continuous number between 0 and 1 which acts as an upper bound and a lower bound of the multiplier's selection range as mentioned in the claimed invention's specification paragraph [0036], "Multiplier 161 is a positive real number that , if less that one, generates a multiplicative product that is less than probability M1", thus these ratios must be a number that is also between 0 and 1. The alpha introduced by Pleiss is an interpolation parameter which operates as a functional mathematical equivalent to a multiplier's selection fractional range. Mathematically, a multiplier is defined as any scaling factor or coefficient that is multiplied by a variable to alter its magnitude or weight. The parameter alpha is explicitly utilized in a combinatorial formulation to determine the operational cost of the post-processed classifier, expressed in g2(h~_2) equation. In this algebraic structure, alpha acts directly as a multiplicate weight applied to the upper bound constraint, while (1-alpha) acts as the corresponding multiplier for the lower bound constraint. Therefore, generating a multiplier as mentioned in Claim 4 using the ratios is functionally equivalent to the claimed invention.)
Bhaila, Pranto, Girandis and Pleiss are analogous to the claimed invention as they are all from the same field of endeavor of machine learning-based data classification and model parameter optimization. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine adjusting classification outputs by applying class-specific multiplier weights of Bhaila, calculating the minimum and maximum values of a data distribution to establish boundaries of Pranto, dynamically generating classification weight vectors of Gidaris with the error-cost balancing using an interpolation parameter under calibration constraints of Pleiss. The motivation is recited by Pleiss (Pleiss, Pg1, Abstract, Lines4-5, “In this paper, we investigate the tension between minimizing error disparity across different population groups while maintaining calibrated probability estimates”), such that by adapting the fairness multiplier to the computational procedure in a low-latency environment without demolishing the accuracy performance such that without a proper pruning range, the computational burden will be infinitely high.
As to dependent Claim 6,
The combination of Bhaila, Pranto, Gidaris and Pleiss teaches, as mentioned above, all the limitations of Claim 5. It teaches about generating the second multiplier that is specific to the second value of the feature based on the ratios. The combination of Bhaila, Pranto and Gidaris does not teach the following limitations but Pleiss further teaches the method of Claim 5 wherein the second multiplier that is specific to both of the feature and the second value of the feature is generated in a range from the first ratio to the second ratio (Pleiss, Pg6, Paragraph2, Lines6-8, "In Lemma 4 in Section S2, we show that if
PNG
media_image13.png
200
400
media_image13.png
Greyscale
then the cost of h~_2 is a linear interpolation between the costs of h_2 and h^u2 (Figure 2d). More formally, we have that g2(h~_2) = (1-alpha)g2(h2) + alpha g2 (h^u2)), and thus setting
PNG
media_image14.png
200
400
media_image14.png
Greyscale
ensures that g2(g2(h~_2) = g1(h1) as desired” and Pg14, Lemma 4,
PNG
media_image15.png
200
400
media_image15.png
Greyscale
, wherein the two ratios are simply any continuous number between 0 and 1 which acts as an upper bound and a lower bound of the multiplier's selection range as mentioned in the claimed invention's specification paragraph [0036], "Multiplier 161 is a positive real number that , if less that one, generates a multiplicative product that is less than probability M1", thus these ratios must be a number that is also between 0 and 1. The alpha introduced by Pleiss is an interpolation parameter which operates as a functional mathematical equivalent to a multiplier's selection fractional range. Mathematically, a multiplier is defined as any scaling factor or coefficient that is multiplied by a variable to alter its magnitude or weight. The parameter alpha is explicitly utilized in a combinatorial formulation to determine the operational cost of the post-processed classifier, expressed in g2(h~_2) equation. In this algebraic structure, alpha acts directly as a multiplicate weight applied to the upper bound constraint, while (1-alpha) acts as the corresponding multiplier for the lower bound constraint. Therefore, generating a multiplier as mentioned in Claim 5 using the ratios is functionally equivalent to the claimed invention.).
Claims 7 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Bhaila as mentioned in Claim 1 in view of Deb et al. (Deb), Non-Patent Literature listed in IDS filed on 02/09/2024, “A Fast and Elitist Multiobjective Genetic Algorithm: NSGA-II Fairness and Calibration”, published on Apr 2002, Pages: 16 and in further view of Gidaris.
As to dependent Claim 7,
Bhaila teaches, as mentioned above, all the limitations of Claim 1. Bhaila teaches about classifying an input, which has a feature consists of a plurality of multipliers, based on the inferred probability of the class by multiplying the probability to the multiplier. However, Bhaila does not teach about generating the plurality of multipliers of the feature using a bi-objective optimizer. In the same field of endeavor, Deb teaches about the bi-objective optimizer (Deb, Pg182, Abstract, Lines20-25, "Moreover, we modify the definition of dominance in order to solve constrained multiobjective problems efficiently. Simulation results of the constrained NSGA-II on a number of test problems, including a five-objective seven-constraint nonlinear problem, are compared with another constrained multiobjective optimizer and much better performance of NSGA-II is observed", Abstract, Lines11-14, "Also, a selection operator is presented that creates a mating pool by combining the parent and offspring populations and selecting the best (with respect to fitness and spread) N solutions" and Deb, Pg188, Subsection B, "Unlike in single-objective optimization, there are two goals in a multiobjective optimization: 1) convergence to the Pareto-optimal set and 2) maintenance of diversity in solutions of the Preto-optimal set", wherein NSGA-II which is specifically designed to solve multi-objective optimization problems which represents the application of the framework to a "two-objective problem" (which is technically identical and synonymous with the claimed "bi-objective" optimizer.) Also, NSGA-II shows that the optimizer functions by maintaining an entire population of candidate solutions, which a population of N Pareto-optimal solutions corresponds exactly to the claimed plurality of multipliers designed to represent distinct operational trade-offs, thus it is inherently equivalent to the claimed invention.)
Bhaila and Deb are analogous to the claimed invention as they are all from the same field of endeavor of constrained multi-objective optimization. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine adjusting classification outputs by applying class-specific multiplier weights of Bhaila with the evolutionary optimization engine designed to efficiently solve such multi-objective problems of Deb. The motivation combine is recited by Deb (Deb, Pg182, Abstract, Lines20-21, “Moreover, we modify the definition of dominance in order to solve constrained multiobjective problems efficiently”) such that as mentioned it is computationally complex when it comes to solve multi-objective problems, thus incorporating Deb’s engine which is designed to efficiently solve such problems for optimization is a compromising choice.
Bhaila teaches, as mentioned above, about classifying an input, which has a feature consists of a plurality of multipliers, based on the inferred probability of the class by multiplying the probability to the multiplier. However, Bhaila does not teach about generating the plurality of multipliers of the feature. In the same field of endeavor, Gidaris teaches this limitation (Gidaris, Pg22, Left Column, Subsection. Our approach, Lines25-29, “it learns to output a new set of classification weight vectors (both for the base and novel classes) that can then be used from the feature classifier in order to classify in a unified way both types of classes”, wherein combining the sensitive feature weights of Bhaila with the generation of the set of weights using the bi-objective optimizer is functionally equivalent to the claimed invention.)
Bhaila, Deb and Gidaris are analogous to the claimed invention as they are from the same field of endeavor of multi-objective optimization and Lachine learning-based data classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine adjusting classification outputs by applying class-specific multiplier weights of Bhaila, the evolutionary optimization engine designed to efficiently solve such multi-objective problems of Deb with dynamically generating classification weight vectors of Gidaris. The motivation is as recited by Gidaris (Gidaris, Pg22, Left Column, Subsection. DAE based model parameters generation, “Learning to perform such a meta-learning task, i.e., inferring the classification weights of a set of classes, is a difficult meta-problem that requires plenty of training data in order to be reliably solved. However, having access to such a large pool of data is not always possible; or otherwise stated the training data available for learning such meta-tasks might never be enough”) such that dynamically generating weights or multipliers of the feature using limited data is not reliable, thus creating a pool of choices or weights/multipliers to be selected make the dynamic generation much more efficient).
As to dependent Claim 19,
it is a non-transitory computer-readable media claim that contains similar limitations of Claim 7 and thus rejected under the same rationale.
Claims 8 and 9 are rejected under 35 U.S.C. 103 as being unpatentable over Bhaila, Deb and Gidaris as mentioned in Claim 7 in view of Cruz et al. (Cruz), Non-Patent Literature listed in IDS filed on 02/09/2024, “FairGBM: Gradient Boosting with Fairness Constraints”, published on March 2023, Pages: 23.
As to dependent Claim 8,
The combination of Bhaila, Deb and Gidaris teaches, as mentioned above, all the limitations of Claim 7. The combination teaches about generating a plurality of multipliers of the feature using the bi-objective optimizer. However, the combination does not teach about receiving two validation scores that are based on the plurality of multipliers of the feature. In the same field of endeavor, Cruz teaches this limitation (Cruz, Pg6, Section3, Paragraph3, Lines4-7, "Fairlearn GS (Agarwal et al., 2018) is a similar method that instead uses a grid search over the constraint multipliers, and outputs a single (deterministic) classifier that achieves the best fairness-performance trade-off. RS Reweighing is a variation of GS that instead casts the choice of multipliers as another model hyperparameter," Cruz, Pg6, Section3, Paragraph4, Lines9-10, "The best performing model (the maximizer of
PNG
media_image16.png
200
400
media_image16.png
Greyscale
) of each trial on validation data is then separately selected", wherein the optimizer mentioned in Claim7 uses the objective function value for its computation which inherently means this value must be received from somewhere according to its objective of the usage. FairGBM shows that the validation scores are mathematically and causally governed by the specific values of the constraint multipliers (Lagrange multiplier). Also, these alternative multiplier configurations are evaluated on validation data, whereby the maximizer equation. FairGBM also shows that it relies on the fairness constraints with no impact on predictive performance, which these two metrics represent the overall performance of FairGMB such that the scores are based on the specific values of the constraint multipliers. If these two metrics that heavily relies on the multipliers are used in the NGSA-II objective function values, then it is functionally equivalent to the claimed invention.)
Bhaila, Deb, Gidaris and Cruz are analogous to the claimed invention as they are from the same field of endeavor of multi-objective parameter optimization and machine learning-based data classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine adjusting classification outputs by applying class-specific multiplier weights of Bhaila, the evolutionary optimization engine designed to efficiently solve such multi-objective problems of Deb, dynamically generating classification weight vectors of Gidaris with fairness/accuracy trade-off using the validation scores of Cruz. The motivation is as recited by Cruz (Cruz, Pg1, Abstract Lines3-6, “While fairness in these domains is a foremost concern, existing in-processing Fair ML methods are either incompatible with GBDT, or incur in significant performance losses while taking considerably longer to train.”) such that combining multi-objective or constrained optimization techniques to solve the inherent trade-offs between predictive accuracy and fairness metrics efficiently.
As to dependent Claim 9,
The combination of Bhaila, Deb, Gidaris and Cruz teaches, as mentioned above, all the limitations of Claim 8. The combination teaches about using the bi-objective optimizer to receive the validation scores that are based on the plurality of multipliers of the feature to generate the plurality of multipliers of the feature.
However, the combination of Bhaila, Deb and Gidaris does not teach the following limitation below but from the same field of endeavor, Cruz further teaches the method of Claim 8 wherein the two validation scores that are based on the plurality of multipliers of the feature are a fitness score and a fairness score (Cruz, Pg6, Section3, Paragraph3, Lines4-7, "Fairlearn GS (Agarwal et al., 2018) is a similar method that instead uses a grid search over the constraint multipliers, and outputs a single (deterministic) classifier that achieves the best fairness-performance trade-off. RS Reweighing is a variation of GS that instead casts the choice of multipliers as another model hyperparameter," Cruz, Pg6, Section3, Paragraph4, Lines9-10, "The best performing model (the maximizer of
PNG
media_image16.png
200
400
media_image16.png
Greyscale
) of each trial on validation data is then separately selected”, wherein as mentioned in Claim 8, these validation scores are based on the multipliers. These scores are here described as a performance score and a fairness score such that the performance score indicates the accuracy which is functionally equivalent to the claimed invention.)
Bhaila, Deb, Gidaris and Cruz are analogous to the claimed invention as they are from the same field of endeavor of multi-objective parameter optimization and machine learning-based data classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine adjusting classification outputs by applying class-specific multiplier weights of Bhaila, the evolutionary optimization engine designed to efficiently solve such multi-objective problems of Deb, dynamically generating classification weight vectors of Gidaris with fairness/accuracy trade-off using the validation scores of Cruz. The motivation is as recited by Cruz (Cruz, Pg1, Abstract Lines3-6, “While fairness in these domains is a foremost concern, existing in-processing Fair ML methods are either incompatible with GBDT, or incur in significant performance losses while taking considerably longer to train.”) such that combining multi-objective or constrained optimization techniques to solve the inherent trade-offs between predictive accuracy and fairness metrics efficiently.
Claims 12 and 21 are rejected under 35 U.S.C. 103 as being unpatentable over Bhaila as mentioned in Claim 1 in view of Gidaris and in further view of Deb.
As to dependent Claim 12,
Bhaila teaches, as mentioned above, all the limitations of Claim 1. Bhaila teaches about classifying an input, which has a feature consists of a plurality of multipliers, based on the inferred probability of the class by multiplying the probability to the multiplier. However, Bhaila does not teach the method of Claim 1 further comprising:
generating multiple pluralities of multipliers of the feature. In the same field of endeavor, Gidaris teaches this limitation (Gidaris, Pg22, Left Column, Subsection. Our approach, Lines25-29, “it learns to output a new set of classification weight vectors (both for the base and novel classes) that can then be used from the feature classifier in order to classify in a unified way both types of classes”, wherein combining the sensitive feature weights of Bhaila as mentioned in Claim 1 with the generation of the set of weights is functionally equivalent to the claimed invention.)
Bhaila and Gidaris are analogous to the claimed invention as they are from the same field of endeavor of multi-objective optimization and Lachine learning-based data classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine adjusting classification outputs by applying class-specific multiplier weights of Bhaila with dynamically generating classification weight vectors of Gidaris. The motivation is as recited by Gidaris (Gidaris, Pg22, Left Column, Subsection. DAE based model parameters generation, “Learning to perform such a meta learning task, i.e., inferring the classification weights of a set of classes, is a difficult meta-problem that requires plenty of training data in order to be reliably solved. However, having access to such a large pool of data is not always possible; or otherwise stated the training data available for learning such meta-tasks might never be enough”) such that dynamically generating weights or multipliers of the feature using limited data is not reliable, thus relying on Gidaris’ model to generate multipliers that are specific to the feature is an efficient adaptation).
Bhaila does not teach the limitation of detecting a subset of the multiple pluralities of multipliers of the feature that are on a bi-objective Pareto frontier. From the same field of endeavor, Deb teaches this limitation (Deb, Pg182, Abstract, Lines20-25, "Moreover, we modify the definition of dominance in order to solve constrained multiobjective problems efficiently. Simulation results of the constrained NSGA-II on a number of test problems, including a five-objective seven-constraint nonlinear problem, are compared with another constrained multiobjective optimizer and much better performance of NSGA-II is observed", Abstract, Lines11-14, "Also, a selection operator is presented that creates a mating pool by combining the parent and offspring populations and selecting the best (with respect to fitness and spread) N solutions" and Deb, Pg188, Subsection B, "Unlike in single objective optimization, there are two goals in a multiobjective optimization: 1) convergence to the Pareto-optimal set and 2) maintenance of diversity in solutions of the Preto-optimal set" and Deb, Pg186, Paragraph1, Lines6-14, "Now, solutions belonging to the best nondominated set F1 are of best solutions in the combined population and must be emphasized more than any other solution in the combined population. If the size of F1 is smaller than N, we definitely choose all members of the set F1 for the new population P_t+1. The remaining members of the population P_t+1 are chosen from subsequent nondominated fronts in the order of their ranking. Thus, solutions from the set F2 are chosen next, followed by solutions from the set F3, and so on", wherein as the population is a set of candidate solutions, which a population of N Pareto-optimal solutions such that NGSA-II optimizer or the multi-objective optimizer will continuously selects the best subset of multipliers that fits the object thus where those subsets lie is the Pareto frontier which is functionally equivalent to the claimed invention.)
Bhaila, Gidaris and Deb are analogous to the claimed invention as they are from the same field of endeavor of multi-objective optimization and Lachine learning-based data classification. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine adjusting classification outputs by applying class-specific multiplier weights of Bhaila, dynamically generating classification weight vectors of Gidaris with the evolutionary optimization engine designed to efficiently solve such multi-objective problems of Deb. The motivation is as recited by Deb (Deb, Pg182, Abstract, Lines20 21, “Moreover, we modify the definition of dominance in order to solve constrained multiobjective problems efficiently”) such that as mentioned it is computationally complex when it comes to solve multi-objective problems, thus incorporating Deb’s engine which is designed to efficiently solve such problems for optimization is a compromising choice.
As to dependent Claim 21,
it is a non-transitory computer-readable media claim that contains similar limitations of Claim 12 and thus rejected under the same rationale.
Claim 13 is rejected under 35 U.S.C. 103 as being unpatentable over Bhaila as mentioned in Claim 1 in view of Grandini et al. (Grandini), Non-Patent Literature, “METRICS FOR MULTI-CLASS CLASSIFICATION: AN OVERVIEW”, published on Aug 2020, Pages: 17 in further view of Kriegman et al. (Kriegman), US-Patent, US-10,255,527-B2, Patented on Apr 9, 2019 and in further view of Cruz.
As to dependent Claim 13,
Bhaila teaches, as mentioned above, all the limitations of Claim 1. Bhaila teaches about classifying an input, which has a feature consists of a plurality of multipliers, based on the inferred probability of the class by multiplying the probability to the multiplier. Bhaila also teaches about inferring the probabilities of binary classes. However, Bhaila fails to teach the method of Claim 1 wherein: the method further comprises:
inferring, from the input that contains the value of the feature, a second probability of a second class and a third probability of a third class. In the same field of endeavor, Grandini teaches this limitation (Grandini, Pg1, Introduction, Paragraph6, Lines1-4, "A classification model gives us the probability of belonging to a specific class for each possible units. Starting from the probability assigned by the model, in the two-class classification problem a threshold is usually applied to decide which class has to be predicted for each unit. While in the multi-class case, there are various possibilities; among them, the highest probability value and the softmax are the most employed techniques", wherein each unit or input in the multi-class model, each class will be assigned with a unique probability, thus if there are three classes then each class of the input will be given a probability which is functionally equivalent to the claimed invention.)
Bhaila and Grandini are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based statistical data classification and algorithmic prediction of discrete categorical outcomes. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine the collective classification mechanism of Bhaila with the multi-class classification framework of Grandini. The motivation is as recited by Grandini (Grandini, Pg2 ,Paragraph1, Lines1-3, “There are many metrics that come in handy to test the ability of any multi-class classifier and they turn out to be useful for: i) comparing the performance of two different models, ii) analysing the behaviour of the same model by tuning different parameters”) such that analyzing the model using different parameters, including bias mitigation to control the fairness among the classes or the inputs is crucial in the classification process.
Bhaila teaches about multiplicative product of the probability of the class and the multiplier that is specific to both of the feature and the value of the feature. As mentioned above, Grandini teaches about multi-class classification. However, they do no teach
rescaling, based on said multiplicative product of the probability of the class and the multiplier that is specific to both of the feature and the value of the feature, the second probability of the second class and the third probability of the third class. However, from the same field of endeavor, Kriegman teaches about rescaling (Kriegman, Pg29, Column 32, Paragraph1, Lines1-13, “Notably, P(A|X) and P(I|X) have the same denominator, p(s). Moreover, the probabilities P(A|X), P(B|X), P(C|X) . . . P(N|X), P(I|X) that the Person X corresponds to person A, B, C, . . . N or an imposter sums to 1 (i.e., person X must either be an imposter or a tagged person). Accordingly, in one or more embodiments, digital classification system 100 ignores p(s) and simply scales calculated probabilities to 1 after calculating the numerator of the equations outlined above. For instance, one or more embodiments of digital classification system 100 utilizes the following equation to scale the probabilities: P(A|X)+P(B|X)+ . . . +P(N|X)+P(I|X)=1”, wherein rescaling process is a mere normalization such that it does not change the original proportions between the probabilities which is functionally equivalent to the claimed invention.)
Kriegman also teaches said classifying the input is not based on said rescaling the second probability of the second class and the third probability of the third class (as mentioned above, the rescaling process is a mere normalization such that it does not change the original proportions between the probabilities meaning it will not do anything during the classification process mentioned in Claim 1 by Bhaila regardless of the classes that will be mentioned below.)
Bhaila, Grandini and Kriegman are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based statistical data classification, predictive score processing, and the algorithm manipulation of probability distributions to assign target categories to input data object. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine the collective classification mechanism of Bhaila, the multi-class classification framework of Grandini with the normalized classification system of Kriegman. The motivation is as by Kriegman (Kriegman, Pg14, Column1, Lines58-61, “conventional classification systems often have difficulty accurately classifying digital objects with regard to multiple possible classifications and multiple corresponding classification scores”) such that ensures that all modified and competing classification metrics are mathematically calibrated to sum up to 100%, thereby resolving the inherent inaccuracies and difficulties of comparing raw scores or probabilities.
However, they do not disclose the limitation of based on said multiplicative product of the probability of the class and the multiplier that is specific to both of the feature and the value of the feature. From the same field of endeavor, Cruz teaches this limitation (Cruz, Pg5, Descent step,
PNG
media_image17.png
138
538
media_image17.png
Greyscale
, wherein the descent step is to minimize a proxy-Lagrangian loss by mathematically isolating the single group experiencing the highest systemic bias. The step explicitly anchors the group with the maximal proxy loss, j, and creates a conditional mathematical branch for calculating pseudo-residuals. According to the equation8, when an instance belongs to the primary disadvantaged group, (si= j), its loss optimization is scaled by aggregating the multipliers of all other remaining groups. Conversely, when an instance belongs to alternative classes (si =k), it is processed under a separate mathematical derivative utilizing only its individual multiplier. This formula establishes a clear engineering precedent where the primary target class, j, is intentionally excluded and decoupled from the individual scaling or multiplier adjustments, thus it is functionally equivalent to the claimed invention of where the multiplicative product is only applied to the second and the third class, which these classes are considered to be non-target groups.)
Bhaila, Grandini, Kriegman and Cruz are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based statistical data classification, predictive score processing, and the algorithmic optimization of probability distributions to balance accuracy and fairness across multiple target categories. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine the collective classification mechanism of Bhaila, the multi-class classification framework of Grandini, the normalized classification system of Kriegman with the gradient boosting with fairness constraints framework of Cruz. The motivation is as by Cruz (Cruz, Pg1 Abstract, Lines3-7, “While fairness in these domains is a foremost concern, existing in-processing Fair ML methods are either incompatible with GBDT, or incur in significant performance losses while taking considerably longer to train”) such that to address the issue of severe performance or training-time overhead to balance all classes simultaneously when adjusting model outputs with target multipliers for bias mitigation within a competing multi-class environment, Cruz demonstrates optimizing fairness under constraints can be achieved efficiently by anchoring the single most disadvantaged minority class.
Claims 14 and 22 are rejected under 35 U.S.C. 103 as being unpatentable over Bhaila as mentioned in Claim 1 in view of Pleiss in further view of Grandini and in further view of Cruz.
As to dependent Claim 14,
Bhaila teaches, as mentioned above, all the limitations of Claim 1. Bhaila teaches about classifying an input, which has a feature consists of a plurality of multipliers, based on the inferred probability of the class by multiplying the probability to the multiplier.
Bhaila further teaches said probability of the class is a probability of a first class (Bhaila, Pg1416-1417, Right Column, Subsection A. Collective Classification, Lines4-9, “The input is an unweighted, undirected, and attributed graph G = (V,E,X,Y) where V is a set of N nodes, E is a set of edges that connect node pairs in V, X represents the node feature matrix, and Y denotes node labels. Each node vi is defined by its feature vector xi which includes a sensitive attribute si”, Pg1417, Left Column, Lines6-8, “The task then is to simultaneously infer the values yi for vi or a probability distribution over those label values” and Pg1419, Left Column, Paragraph1, Lines14-16, “In our setup with binary s and binary y, we derive these weights for four combinations of s and y”, wherein as mentioned in Claim 1, the classes are in binary such that each input vi, will be given a unique probability for each class meaning one class is going to be inherently the probability of the first class.)
Bhaila also teaches about that the multiplier is specific to both of the feature and the value of the feature. However, Bhaila does not teach that the multiplier is less than one. From the same field of endeavor, Pleiss teaches this limitation (Pleiss, Pg6, Paragraph2, Lines6-8, "In Lemma 4 in Section S2, we show that if
PNG
media_image18.png
57
613
media_image18.png
Greyscale
then the cost of h~_2 is a linear interpolation between the costs of h_2 and h^u2 (Figure 2d). More formally, we have that g2(h~_2) = (1-alpha)g2(h2) + alpha g2 (h^u2)), and thus setting
PNG
media_image19.png
35
167
media_image19.png
Greyscale
ensures that g2(g2(h~_2) = g1(h1) as desired” and Pg14, Lemma 4,
PNG
media_image20.png
78
813
media_image20.png
Greyscale
, wherein the alpha here or the interpolation parameter that directly acts like a multiplier that is directly multiplied to the probabilities of the classes as mentioned in Claim 5 which is a number between 0 and 1. As a scaler factor, it tries to scale down the majority group’s values such that the minority can be selected which inherently means it must be a value between 0 and 1 which is equivalent to the claimed invention)
Bhaila and Pleiss are analogous to the claimed invention as they are all from the same field of endeavor of machine learning-based data classification and model parameter optimization. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine adjusting classification outputs by applying class-specific multiplier weights of Bhaila with the error-cost balancing using an interpolation parameter under calibration constraints of Pleiss. The motivation is recited by Pleiss (Pleiss, Pg1, Abstract, Lines4-5, “In this paper, we investigate the tension between minimizing error disparity across different population groups while maintaining calibrated probability estimates”) such that by adapting the fairness multiplier as a scaling factor to modify the value of the label in a low-latency environment without demolishing the accuracy performance.
Bhaila as mentioned above teaches about the binary class. However, Bhaila does not teach about multi-classes the third class and its probability. In the same field of endeavor, Grandini teaches there is a third class as mentioned in Claim 13 (Grandini, Pg1, Introduction, Paragraph6, Lines1-4, "A classification model gives us the probability of belonging to a specific class for each possible units. Starting from the probability assigned by the model, in the two-class classification problem a threshold is usually applied to decide which class has to be predicted for each unit. While in the multi-class case, there are various possibilities; among them, the highest probability value and the softmax are the most employed techniques”). Grandini further teaches the method further comprises from the input that contains the value of the feature, inferring a second probability of a second class that is less than the probability of the first class and a third probability of a third class (Grandini, Page4, Last paragraph, Lines 2-4, “This also means that Balanced Accuracy is insensitive to imbalanced class distribution and it gives more weight to the instances coming from minority classes. On the other hand, Accuracy treats all instances alike and usually favors the majority class” and Pg3, Section 2, Paragraph5, Lines-1-2, “When we think about classes instead of individuals, there will be classes with a high number of units and others with just few ones. In this situation, highly populated classes will have higher weight compared to the smallest ones”, wherein there will be imbalances between the classes such that one class might be bigger or smaller than the other classes in which is inherently equivalent to the claimed invention’s the second class has the smallest probability compared to the first and the third class.)
Bhaila, Pleiss and Grandini are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based data classification and model parameter optimization. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine the collective classification mechanism of Bhaila, the error-cost balancing using an interpolation parameter under calibration constraints of Pleiss with multi-class input classification using various parameters of Grandini. The motivation is as recited by Grandini (Grandini, Pg2 ,Paragraph1, Lines1-3, “There are many metrics that come in handy to test the ability of any multi-class classifier and they turn out to be useful for: i) comparing the performance of two different models, ii) analysing the behaviour of the same model by tuning different parameters”) such that analyzing the model using different parameters, including bias mitigation to control the fairness among multi classes or the inputs is crucial in the classification process.
Bhaila teaches about classifying the input, but does not teach about the classifying the input as the second class. In the same field of endeavor, Cruz teaches this limitation (Cruz, Pg5, Descent step,
PNG
media_image17.png
138
538
media_image17.png
Greyscale
, wherein as mentioned in Claim 13, the equation is used to isolate the target group experiencing the highest systemic bias. This formula establishes a clear engineering precedent where the primary target class, j, is intentionally excluded and decoupled from the individual scaling or multiplier adjustments, meaning if the first class and the third classes that has higher probabilities as mentioned above are scaled down with the multiplier, the second class will be likely to be selected, which is functionally equivalent to the claimed invention.)
Bhaila, Pleiss, Grandini and Cruz are analogous to the claimed invention as they are from the same field of endeavor of machine learning-based statistical data classification, predictive score processing, and the algorithmic optimization of probability distributions to balance accuracy and fairness across multiple target categories. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine the collective classification mechanism of Bhaila, the multi-class classification framework of Grandini, the error-cost balancing using an interpolation parameter under calibration constraints of Pleiss with the gradient boosting with fairness constraints framework of Cruz. The motivation is as by Cruz (Cruz, Pg1 Abstract, Lines3-7, “While fairness in these domains is a foremost concern, existing in-processing Fair ML methods are either incompatible with GBDT, or incur in significant performance losses while taking considerably longer to train”) such that to address the issue of severe performance or training-time overhead to balance all classes simultaneously when adjusting model outputs with target multipliers for bias mitigation within a competing multi-class environment, Cruz demonstrates optimizing fairness under constraints can be achieved efficiently by anchoring the single most disadvantaged minority class.
As to dependent Claim 22,
it is a non-transitory computer-readable media claim that contains similar limitations of Claim 14 and thus rejected under the same rationale.
Claims 15, 16 and 23 are rejected under 35 U.S.C. 103 as being unpatentable over Fernando et al. (Fernando), Non-Patent Literature, “Dynamically Weighted Balanced Loss: Class Imbalanced Learning and Confidence Calibration of Deep Neural Networks”, published on July 2022, Pages: 12 in view of Deb as mentioned in Claim 7 in further view of Lohia et al. (Lohia), Non-Patent Literature listed in IDS on 12/13/2024, “BIAS MITIGATION POST-PROCESSING FOR INDIVIDUAL AND GROUP FAIRNESS”, published on 05/12/2019, Pages: 5.
As to dependent Claim 15,
Fernando teaches about generating, from an input, an inference that contains a probability of the first class and a probability of the second class (Fernando, Pg2943, Equation7,
PNG
media_image21.png
73
384
media_image21.png
Greyscale
"where wj is the class weight of class j, yij is the jth element of one-hot encoded label of instance xi, and pij is the predicted probability of the class j of instance xi" and Fernando, Pg2942, Subsection A, Deep Neural Network, Paragraph1, Lines 2-4, "each input vector xi is associated with a corresponding class label (classification target) yi", wherein pij is a class ‘j’ specific probability(j-th element meaning there can be more than one) assigned to the input xi which is inherently means there are multiple probabilities associated with the input xi which is equivalent to the claimed invention of generating a probability for each class including the first and the second class).
Fernando also teaches about wherein the method is performed by one or more computers (Fernando, Pg2948, Right Column, Paragrah1, Model Architecture and Training, Lines 1-3, "We relied on established methods for computer vision and used the state-of-the-art CNN models for image classification.", wherein relying on these tools inherently means at least one computer has been used.)
Fernando further teaches about a multiplier for each class where the classes are mutually exclusive (Fernando, Pg2943, Equation7,
PNG
media_image21.png
73
384
media_image21.png
Greyscale
,
"where wj is the class weight of class j, yij is the jth element of one-hot encoded label of instance xi, and pij is the predicted probability of the class j of instance xi " and Fernando, Pg2942, Subsection A, Deep Neural Network, Paragraph1, Lines 2-4, "each input vector xi is associated with a corresponding class label (classification target) yi”, wherein each input is only associated with corresponding class label yi and each class yi has j-th element such that each yij is assigned a specific j-th weight, wj (corresponding multiplier), meaning each class is mutually exclusive which is equivalent to the claimed invention.)
However, Fernando does not teach that corresponding multipliers to these classes are generated by a bi-optimizer. In the same field of endeavor, Deb teaches this limitation (Deb, Pg182, Abstract, Lines20-25, "Moreover, we modify the definition of dominance in order to solve constrained multiobjective problems efficiently. Simulation results of the constrained NSGA-II on a number of test problems, including a five-objective seven-constraint nonlinear problem, are compared with another constrained multiobjective optimizer and much better performance of NSGA-II is observed", Deb, Abstract, Lines11-14, "Also, a selection operator is presented that creates a mating pool by combining the parent and offspring populations and selecting the best (with respect to fitness and spread) N solutions" and Deb, Pg188, Subsection B, "Unlike in single-objective optimization, there are two goals in a multiobjective optimization: 1) convergence to the Pareto-optimal set and 2) maintenance of diversity in solutions of the Pareto-optimal set", wherein multi-objective evolutionary optimization uses a selection operator functions to evaluate a population of candidate solutions within a mating pool, where each individual solution candidate standardly represents a discrete vector variation of the class weight (wj) for the mutually exclusive classes. By executing this evolutionary search to concurrently balance the competing target goals of convergence and solution diversity, the optimizer systemically sifts through these multiple probability-dependent weight candidates and selects the single optimal configuration. Assuming that j-th class is the first class, repeating this procedure for another class or the second class is functionally equivalent to the claimed invention.)
Fernando and Deb are analogous to the claimed invention as they are from the same field of endeavor of multi-objective optimization targeting competing goals. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine concurrent optimization of class imbalanced learning and confidence calibration of Fernando with the evolutionary optimization engine designed to efficiently solve such multi-objective problems of Deb. The motivation is as recited by Deb (Deb, Pg182, Abstract, Lines20-21, “Moreover, we modify the definition of dominance in order to solve constrained multiobjective problems efficiently”) such that as mentioned it is computationally complex when it comes to solve multi-objective problems, thus using Deb’s engine which is designed to solve such problems is a compromising choice.
As mentioned above, the combination of Fernando and Deb teaches about multiplicative product of the probability of the first class and the first multiplier, and a second multiplicative product of the probability of the second class and the second multiplier (Fernando, Pg2943, Equation7,
PNG
media_image21.png
73
384
media_image21.png
Greyscale
"where wj is the class weight of class j, yij is the jth element of one-hot encoded label of instance xi, and pij is the predicted probability of the class j of instance xi" and Fernando, Pg2942, Subsection A, Deep Neural Network, Paragraph1, Lines 2-4, "each input vector xi is associated with a corresponding class label (classification target) yi" and Deb, Pg182, Abstract, Lines20-25, "Moreover, we modify the definition of dominance in order to solve constrained multiobjective problems efficiently. Simulation results of the constrained NSGA-II on a number of test problems, including a five-objective seven-constraint nonlinear problem, are compared with another constrained multiobjective optimizer and much better performance of NSGA-II is observed", Deb, Abstract, Lines11-14, "Also, a selection operator is presented that creates a mating pool by combining the parent and offspring populations and selecting the best (with respect to fitness and spread) N solutions" and Deb, Pg188, Subsection B, "Unlike in single-objective optimization, there are two goals in a multiobjective optimization: 1) convergence to the Pareto-optimal set and 2) maintenance of diversity in solutions of the Pareto-optimal set", wherein as mentioned above the multiplier or weight that is specific to the first class and the second class, wij, produced from the multi-objective optimizer can be used again in the Equation
PNG
media_image21.png
73
384
media_image21.png
Greyscale
by replacing wj to the new wij, which is equivalent to the claimed invention.)
However, the combination does not teach that the input will be classified according to the multiplicative product mentioned above. In the same field of endeavor, the combination of Lohia and Fernando teaches this limitation (Lohia, Pg1, Introduction, Paragraph5, Lines1-3,"The general methodology of post-processing algorithms is to take a subset of samples and change their predicted labels appropriately to meet a group fairness requirement" and Fernando, Pg2941, Right Column, Paragraph2, Lines13-14, “shifting the decision threshold based on their misclassification costs”, wherein extracting the probability-dependent multiplier framework (
PNG
media_image22.png
38
62
media_image22.png
Greyscale
multiplied by pij mentioned above) and integrate it into the post-processing reclassification routing to execute the final decision-making, thereby classifying the input based on the resulting multiplicative product to achieve the identical threshold-shifting effect during inference which is functionally equivalent to the claimed invention.)
Fernando, Deb and Lohia are analogous to the claimed invention as they are from the same field of endeavor of multi-objective cost-sensitive machine learning and algorithmic decision-making under competing constraints. Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filing date, to combine concurrent optimization of class imbalanced learning and confidence calibration of Fernando, the evolutionary optimization engine designed to efficiently solve such multi-objective problems of Deb with post-processing reclassification architecture of Lohia. The motivation is as recited by Lohia (Lohia, Pg2847, Introduction, Paragraph3, Right Column, Lines5-10, “Advantages of post-processing algorithms are that they do not require access to the training process and are thus suitable for run-time environments. Moreover, post-processing algorithms operate in a black-box fashion, meaning that they do not need access to the internals of models, their derivatives, etc., and are therefore applicable to any machine learning model”) such that full model retraining or concurrent in-processing weight optimization introduces significant computational latency and requires full access to model derivatives and thus, adapting the established post-processing heuristics to improve system efficiency would be obvious.
As to dependent Claim 16,
The combination of Fernando, Deb and Lohia teaches all the limitations of Claim 15. It teaches about generating, by the bi-objective optimizer, the first and the second multipliers for mutually exclusive classes. It also teaches about generating inference that contains probabilities of the first and the second classes and classifying the input based on the multiplication of the probabilities and its corresponding multipliers. The combination further teaches
said classifying the input uses a plurality of multipliers that contains the first multiplier and the second multiplier (Fernando, Pg2943, Equation7,
PNG
media_image21.png
73
384
media_image21.png
Greyscale
, "where wj is the class weight of class j, yij is the jth element of one-hot encoded label of instance xi, and pij is the predicted probability of the class j of instance xi" and Fernando, Pg2942, Subsection A, Deep Neural Network, Paragraph1, Lines 2-4, "each input vector xi is associated with a corresponding class label (classification target) yi” and Deb, Pg182, Abstract, Lines20-25, "Moreover, we modify the definition of dominance in order to solve constrained multiobjective problems efficiently. Simulation results of the constrained NSGA-II on a number of test problems, including a five-objective seven-constraint nonlinear problem, are compared with another constrained multiobjective optimizer and much better performance of NSGA-II is observed", Deb, Abstract, Lines11-14, "Also, a selection operator is presented that creates a mating pool by combining the parent and offspring populations and selecting the best (with respect to fitness and spread) N solutions" and Deb, Pg188, Subsection B, "Unlike in single-objective optimization, there are two goals in a multiobjective optimization: 1) convergence to the Pareto-optimal set and 2) maintenance of diversity in solutions of the Pareto-optimal set" and Lohia, Pg1, Introduction, Paragraph5, Lines1-3, "The general methodology of post-processing algorithms is to take a subset of samples and change their predicted labels appropriately to meet a group fairness requirement", wherein as mentioned above in Claim 15 where generating the first and the second multipliers for the corresponding classes, it uses the selection operator functions to evaluate a generated population of candidate solutions or weights to choose the optimal solution or the weight wj. As there are multiple solutions or the weights in the pool to be chosen and as mentioned above using Lohia’s post-processing reclassification architecture, it is functionally equivalent to the claimed such that there is a plurality of multipliers to be chosen from the pool which contains these corresponding multipliers.);
Fernando also teaches the inference contains a plurality of probabilities that contains the probability of the first class and the probability of the second class (Fernando, Pg2943, Equation7,
PNG
media_image21.png
73
384
media_image21.png
Greyscale
, "where wj is the class weight of class j, yij is the jth element of one-hot encoded label of instance xi, and pij is the predicted probability of the class j of instance xi " and Fernando, Pg2942, Subsection A, Deep Neural Network, Paragraph1, Lines 2-4, "each input vector xi is associated with a corresponding class label (classification target) yi”, wherein as mentioned above, pij represents the probability associated with j-th class of input x, meaning there are multiple classes from 1 to j and a multiple probabilities associated with them from pi1 to pij, which is functionally equivalent to the claimed invention.);
Fernando further teaches a count of the plurality of probabilities equals a count of the plurality of mutually exclusive classes (Fernando, Pg2943, Equation7,
PNG
media_image21.png
73
384
media_image21.png
Greyscale
, "where wj is the class weight of class j, yij is the jth element of one-hot encoded label of instance xi, and pij is the predicted probability of the class j of instance xi " and Fernando, Pg2942, Subsection A, Deep Neural Network, Paragraph1, Lines 2-4, "each input vector xi is associated with a corresponding class label (classification target) yi”, wherein as mentioned above, there are multiple classes from 1 to j and all these classes are mutually exclusive as they get to have its own unique probability and there are only j amount of probabilities meaning that the number of probabilities from 1 to j and the number of mutually exclusive classes from 1 to j must be the same which is equivalent to the claimed invention);
Fernando further teaches a count of the plurality of multipliers is less than the count of the plurality of probabilities (Fernando, Pg2943, Right Column, Equation 8, paragraph4, Lines1-3,
PNG
media_image23.png
59
392
media_image23.png
Greyscale
“As such, misclassification errors for a class j with classwise cost wj will have wj-times more penalty than misclassification errors for the majority class with weight equals 1” and Pg2944, Left Column, Paragraph2, Lines3-5, “the weighting factor for well-classified instance is close to 1, and hence, the loss is unaffected and equivalent to CE.”, wherein the equation shows that the majority class will be assigned a weight in such that it will always be 1 that is unaffected. This is functionally equivalent to the claimed invention of a class not being assigned with a weight or assigned a weight that is 1 which is identical as multiplying a number by 1 yields the same number, meaning there is actually one less multiplier compared to the number of the mutually exclusive classes.)
As to dependent Claim 23,
it is a non-transitory computer-readable media claim that contains similar limitations of Claim 15 and thus rejected under the same rationale.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DONG YOON JUNG whose telephone number is (571)270-0198. The examiner can normally be reached 8am-5pm.
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, Cesar Paula can be reached at (571) 272-4128. 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.
/DONG YOON JUNG/ Examiner, Art Unit 2145
/CESAR B PAULA/ Supervisory Patent Examiner, Art Unit 2145
Prosecution Insights