Detailed Action
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1-20 are pending for examination. Claims 1, 9, and 17 are independent.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1
According to the first part of the analysis, in the instant case, claims 1-20 are directed to the four statutory categories. Thus, each of the claims falls within one of the four statutory categories (i.e., process, machine, manufacture, or composition of matter).
Regarding Claim 1:
2A Prong 1:
calculate category information corresponding to samples based on a plurality of models; (This step for calculating category information is practically performable in the human mind and is understood to be a recitation of a mental process (i.e., evaluation).)
calculate task loss values associated with respective ones of the samples and respective ones of the plurality of models based on product category information; (This step for calculating task loss values is understood to be a recitation of mathematical calculations (See spec para 0033 and equation 1) or mental process (i.e., evaluation).)
calculate gating loss values for a model gate based on category frequency information; (This step for calculating gating loss values is understood to be a recitation of mathematical calculations (See spec para 0034 and equation 1) or mental process (i.e., evaluation).) and
train the model gate based on a sum of the task loss and the gating loss, the training to derive weights corresponding to respective ones of the plurality of models. (This step for training the model is understood to be a recitation of mathematical calculations (See spec para 0034-0035 and equation 4) or mental process (i.e., evaluation).)
2A Prong 2: This judicial exception is not integrated into a practical application.
Additional elements:
An apparatus to train machine learning models to reduce categorization bias, the apparatus comprising: interface circuitry; machine readable instructions; and programmable circuitry to at least one of instantiate or execute the machine readable instructions to: (The apparatus and comprising circuitry are understood to be generic computer elements - See MPEP 2106.05(f).)
The additional elements as disclosed above alone or in combination do not integrate the judicial exception into practical application as they are generic computer functions that are implemented to perform the disclosed abstract idea above.
2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Additional elements:
An apparatus to train machine learning models to reduce categorization bias, the apparatus comprising: interface circuitry; machine readable instructions; and programmable circuitry to at least one of instantiate or execute the machine readable instructions to: (The apparatus and comprising instructions and circuitry are understood to be generic computer elements - See MPEP 2106.05(f).)
The additional elements as disclosed above in combination of the abstract idea
are not sufficient to amount to significantly more than the judicial exception as they are
generic computer functions that are implemented to perform the disclosed abstract idea above.
Regarding Claim 9
2A Prong 1:
expert circuitry to evaluate category information corresponding to data points based on a plurality of models; (This step for evaluating category information is practically performable in the human mind and is understood to be a recitation of a mental process (i.e., evaluation).)
task loss circuitry to determine task loss values associated with respective ones of the data points and respective ones of the plurality of models based on product category information; (This step for task loss values is understood to be a recitation of mathematical calculations (See spec para 0033 and equation 1) or mental process (i.e., evaluation).)
gating loss circuitry determine gating loss values for a model gate based on category frequency information; (This step for calculating gating loss values is understood to be a recitation of mathematical calculations (See spec para 0034 and equation 1) or mental process (i.e., evaluation).) and
summation circuitry to train the model gate based on a sum of the task loss and the gating loss, the training to derive weights corresponding to respective ones of the plurality of models. (This step for training the model is understood to be a recitation of mathematical calculations (See spec para 0034-0035 and equation 4) or mental process (i.e., evaluation).)
2A Prong 2: This judicial exception is not integrated into a practical application.
Additional elements:
An apparatus to reduce categorization bias, the apparatus comprising: interface circuitry to retrieve data; computer readable instructions; and programmable circuitry to instantiate: expert circuitry; task loss circuitry; gating loss circuitry; summation circuitry; (The apparatus and comprising circuitry are understood to be generic computer elements - See MPEP 2106.05(f).)
The additional elements as disclosed above alone or in combination do not integrate the judicial exception into practical application as they are generic computer functions that are implemented to perform the disclosed abstract idea above.
2B: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Additional elements:
An apparatus to reduce categorization bias, the apparatus comprising: interface circuitry to retrieve data; computer readable instructions; and programmable circuitry to instantiate: expert circuitry; task loss circuitry; gating loss circuitry; summation circuitry; (The apparatus and comprising circuitry and instructions are understood to be generic computer elements - See MPEP 2106.05(f).)
The additional elements as disclosed above in combination of the abstract idea
are not sufficient to amount to significantly more than the judicial exception as they are
generic computer functions that are implemented to perform the disclosed abstract idea above.
Regarding Claim 17: see the rejection of claim 1 above. Same rationale applies.
2A Prong 2 & 2B: The claim recites another additional element “the method comprising: […] by executing instructions with at least one processor” (mere instructions to apply the exception using a generic computer component - see MPEP 2106.05(f))
Regarding Claims 2, 10, and 18
2A Prong 1:
wherein a first one of the plurality of models is a softmax cross-entropy loss function. (This step is understood to be a recitation of mathematical calculations or mental process (i.e., evaluation).)
2A Prong 2 & 2B: The claim does not recite any additional elements.
Regarding Claims 3, 11, and 19
2A Prong 1:
wherein a second one of the plurality of models is a balanced softmax loss function. (This step is understood to be a recitation of mathematical calculations or mental process (i.e., evaluation).)
2A Prong 2 & 2B: The claim does not recite any additional elements.
Regarding Claims 4, 12, and 20
2A Prong 1:
wherein a third one of the plurality of models is an inverted softmax loss function. (This step is understood to be a recitation of mathematical calculations or mental process (i.e., evaluation).)
2A Prong 2 & 2B: The claim does not recite any additional elements.
Regarding Claims 5 and 13
2A Prong 1:
wherein a first category frequency corresponds to a first threshold quantity of the samples and a second category frequency corresponds to a second threshold quantity of the samples, the first threshold quantity greater than the second threshold quantity. (This step is understood to be a recitation of mental process (i.e., evaluation).)
2A Prong 2 & 2B: The claim does not recite any additional elements.
Regarding Claims 6 and 14
2A Prong 1:
wherein a third category frequency corresponds to a third threshold quantity of the samples, the second threshold quantity of the samples greater than the third threshold quantity of the samples. (This step is understood to be a recitation of mental process (i.e., evaluation).)
2A Prong 2 & 2B: The claim does not recite any additional elements.
Regarding Claims 7 and 15
2A Prong 1:
wherein the gating loss is multiplied by a multiplication factor to modulate a contribution of the gating loss. (This step is understood to be a recitation of mathematical calculations or mental process (i.e., evaluation).)
2A Prong 2 & 2B: The claim does not recite any additional elements.
Regarding Claims 8 and 16
2A Prong 1:
wherein the task loss values are multiplied by a multiplication factor to modulate contributions of the task loss values. (This step is understood to be a recitation of mathematical calculations or mental process (i.e., evaluation).)
2A Prong 2 & 2B: The claim does not recite any additional elements.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 1-3, 5-11, and 13-19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Wang et al. ("Long-Tailed Recognition By Routing Diverse Distribution-Aware Experts", hereinafter "Wang") in view of Liu et al. (US 20220261551 A1, hereinafter "Liu").
Regarding Claim 1
Wang discloses: An apparatus to train machine learning models to reduce categorization bias ([Abstract and Section 3] describes training expert models to reduce model bias. [Page 14 Appendix A.1] discloses a GPU for implementation (i.e. an apparatus).),
calculate category information corresponding to samples based on a plurality of models ([Section 3, Fig 2, and Fig 8] describes multi-expert model with experts (i.e. plurality of models) that calculate classification probabilities (i.e. calculate category information) corresponding to inputs.);
calculate task loss values associated with respective ones of the samples and respective ones of the plurality of models based on ([Section 3, equations 8-9, Fig 2, and Fig 8] describes calculating individual loss (i.e. task loss) associated with samples (i.e. x) of respective experts (i.e. models based on category information).);
calculate gating loss values for a model gate based on category frequency information ([Section 3, equations 10-14, Fig 2, and Fig 8] describes calculating a diversity loss (i.e. gating loss) with KL divergence which analyzes differences between probability distributions and class-wise temperature (i.e. category frequency information).); and
train the model gate based on a sum of the task loss and the gating loss, the training to derive weights corresponding to respective ones of the plurality of models ([Section 3, equations 15-17, Fig 2, and Fig 8] describes optimizing experts (i.e. train model) with a total loss including a sum of the individual loss (i.e. task loss) and diversity loss (i.e. gating loss).).
Wang does not explicitly disclose: the apparatus comprising: interface circuitry; machine readable instructions; and programmable circuitry to at least one of instantiate or execute the machine readable instructions to: the plurality of models based on product category information;
However, Liu discloses in the same field of endeavor: the apparatus comprising: interface circuitry; machine readable instructions; and programmable circuitry to at least one of instantiate or execute the machine readable instructions to ([Para 0096-0102, 0110-0118, and Fig 9]): calculate task loss values associated with respective ones of the samples and respective ones of the plurality of models based on product category information ([Para 0003, 0019, 0029, 0037, 0069-0072, Fig 1-3, and Fig 5] describes a plurality of models based on product category information and computing a loss.);
It would have been obvious to a person of ordinary skill in art before the effective filling date of the invention to implement the apparatus of Product Representation Learning disclosed by Liu into the method of Long-Tailed Recognition disclosed by Wang to provide an apparatus for a plurality of models based on product category information. The modification would have been obvious because one of the ordinary skills of the art would be motivated to utilize the feature of Product Representation Learning disclosed by Liu as all the references are in the field of machine learning. A person of ordinary skill of the art would have been motivated to perform the combination for being able for learning product representation to categorize products.
Regarding Claim 9
Wang discloses: An apparatus to reduce categorization bias, the apparatus comprising: ([Abstract and Section 3] describes training expert models to reduce model bias. [Page 14 Appendix A.1] discloses a GPU for implementation (i.e. an apparatus).)
expert ([Section 3, Fig 2, and Fig 8] describes multi-expert model with experts (i.e. plurality of models) that calculate classification probabilities (i.e. calculate category information) corresponding to inputs.)
task loss ([Section 3, equations 8-9, Fig 2, and Fig 8] describes calculating individual loss (i.e. task loss) associated with samples (i.e. x) of respective experts (i.e. models based on category information).)
gating loss ([Section 3, equations 10-14, Fig 2, and Fig 8] describes calculating a diversity loss (i.e. gating loss) with KL divergence which analyzes differences between probability distributions and class-wise temperature (i.e. category frequency information).) and
summation ([Section 3, equations 15-17, Fig 2, and Fig 8] describes optimizing experts (i.e. train model) with a total loss including a sum of the individual loss (i.e. task loss) and diversity loss (i.e. gating loss).)
Wang does not explicitly disclose: interface circuitry to retrieve data; computer readable instructions; and programmable circuitry to instantiate: plurality of models based on product category information;
However, Liu discloses in the same field of endeavor: interface circuitry to retrieve data; computer readable instructions; and programmable circuitry to instantiate ([Para 0096-0102, 0110-0118, and Fig 9]): plurality of models based on product category information; ([Para 0003, 0019, 0029, 0037, 0069-0072, Fig 1-3, and Fig 5] describes a plurality of models based on product category information and computing a loss.);
It would have been obvious to a person of ordinary skill in art before the effective filling date of the invention to implement the apparatus of Product Representation Learning disclosed by Liu into the method of Long-Tailed Recognition disclosed by Wang to provide an apparatus for a plurality of models based on product category information. The modification would have been obvious because one of the ordinary skills of the art would be motivated to utilize the feature of Product Representation Learning disclosed by Liu as all the references are in the field of machine learning. A person of ordinary skill of the art would have been motivated to perform the combination for being able for learning product representation to categorize products.
Regarding Claim 17
Wang in view of Liu discloses: A method of reducing categorization bias in machine learning models ([Abstract and Section 3], Wang describes training expert models to reduce model bias. [Page 14 Appendix A.1] discloses a GPU for implementation (i.e. an apparatus).), the method comprising: […] by executing instructions with at least one processor ([Para 0096-0102, 0110-0118, and Fig 9], Liu), (Claim 17 is a method claim that corresponds to claim 1 and the rest of the limitations are rejected on the same ground)
Regarding Claim 2
Wang in view of Liu discloses: The apparatus of claim 1, wherein a first one of the plurality of models is a softmax cross-entropy loss function. ([Section 3, equations 7-17, Fig 2, and Fig 8] Wang discloses a softmax cross entropy loss.)
Regarding Claim 3
Wang in view of Liu discloses: The apparatus of claim 1, wherein a second one of the plurality of models is a balanced softmax loss function. ([Section 3, equations 7-17, Fig 2, and Fig 8] Wang discloses routing a balancing classes with loss.)
Regarding Claim 5
Wang in view of Liu discloses: The apparatus of claim 1, wherein a first category frequency corresponds to a first threshold quantity of the samples and a second category frequency corresponds to a second threshold quantity of the samples, the first threshold quantity greater than the second threshold quantity. ([Section 3, equations 7-17, Fig 2, and Fig 8] Wang discloses routing and distribution of instances with classes reduced for later experts. Also discloses a head and tail class.)
Regarding Claim 6
Wang in view of Liu discloses: The apparatus of claim 5, wherein a third category frequency corresponds to a third threshold quantity of the samples, the second threshold quantity of the samples greater than the third threshold quantity of the samples. ([Section 3, equations 7-17, Fig 2, and Fig 8] Wang discloses routing and distribution of instances with classes reduced for later experts.)
Regarding Claim 7
Wang in view of Liu discloses: The apparatus of claim 1, wherein the gating loss is multiplied by a multiplication factor to modulate a contribution of the gating loss. ([Section 3, equations 7-17, Fig 2, and Fig 8] Wang discloses the loss weighted by a hyperparameter λ.)
Regarding Claim 8
Wang in view of Liu discloses: The apparatus of claim 1, wherein the task loss values are multiplied by a multiplication factor to modulate contributions of the task loss values. ([Section 3, equations 7-17, Fig 2, and Fig 8] Wang discloses the loss weighted by a hyperparameter λ.)
Regarding Claim 10
(Claim 10 recites analogous limitations to claim 2 and therefore is rejected on the same ground as claim 2.)
Regarding Claim 11
(Claim 11 recites analogous limitations to claim 3 and therefore is rejected on the same ground as claim 3.)
Regarding Claim 13
(Claim 13 recites analogous limitations to claim 5 and therefore is rejected on the same ground as claim 5.)
Regarding Claim 14
(Claim 14 recites analogous limitations to claim 6 and therefore is rejected on the same ground as claim 6.)
Regarding Claim 15
(Claim 15 recites analogous limitations to claim 7 and therefore is rejected on the same ground as claim 7.)
Regarding Claim 16
(Claim 16 recites analogous limitations to claim 8 and therefore is rejected on the same ground as claim 6.)
Regarding Claim 18
(Claim 18 recites analogous limitations to claim 2 and therefore is rejected on the same ground as claim 2.)
Regarding Claim 19
(Claim 19 recites analogous limitations to claim 3 and therefore is rejected on the same ground as claim 3.)
Claim(s) 4, 12, and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Wang in view of Liu and Kim et al.(US 20220156888 A1, hereinafter "Kim").
Regarding Claim 4
Wang in view of Liu discloses: The apparatus of claim 1, wherein a third one of the plurality of models is an ([Section 3, equations 7-17, Fig 2, and Fig 8] Wang discloses a softmax cross entropy loss.)
Wang does not explicitly disclose: an inverted softmax loss function
However, Kim discloses in the same field of endeavor: an inverted softmax loss function ([Para 0095-0097 and Fig 7] discloses inverted softmax loss.)
It would have been obvious to a person of ordinary skill in art before the effective filling date of the invention to implement the function of Recognition Training disclosed by Kim into the method of Wang in view of Liu to perform an inverted softmax loss. The modification would have been obvious because one of the ordinary skills of the art would be motivated to utilize the feature of Recognition Training disclosed by Kim as all the references are in the field of machine learning. A person of ordinary skill of the art would have been motivated to perform the combination for being able to train and evaluate inverted softmax models.
Regarding Claim 12
(Claim 12 recites analogous limitations to claim 4 and therefore is rejected on the same ground as claim 4.)
Regarding Claim 20
(Claim 20 recites analogous limitations to claim 4 and therefore is rejected on the same ground as claim 4.)
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Royer et al. (US 20230281510) describes gating expert models.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to TEWODROS E MENGISTU whose telephone number is (571)270-7714. The examiner can normally be reached Mon-Fri 9:30-5:30.
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, ABDULLAH KAWSAR can be reached at (571)270-3169. 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.
/TEWODROS E MENGISTU/ Examiner, Art Unit 2127