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 .
DETAILED ACTION
This action is in response to the amendments filed 11 March 2026. Claims 1, 8, and 15 are amended. Claims 1-20 are pending and have been examined.
Response to Arguments
Applicant' s arguments, see pages 8-9, filed 11 March 2026, with respect to the interpretation of Claims 1, 3, 4, and 6 under 35 U.S.C. 112(f) have been fully considered and are persuasive.
APPLICANT'S ARGUMENT: Applicant argues (page 9, paragraphs 1-2) that "Assignee's Representative does not concede that these limitations invoke § 112(f), as the terms denote structural software modules executed by a processor and are understood by those of ordinary skill in the art as referring to specific algorithmic components of a machine learning training system. ¶ Nevertheless, even if §112(f) were deemed applicable, the specification clearly discloses corresponding structure for performing the recited functions, including algorithmic processes for executing the random unidimensional augmentation algorithm, generating the one-dimensional augmentation policy search space, defining augmentation parameter mappings, and executing search-based optimization of the global augmentation parameter during training."
EXAMINER'S RESPONSE: Examiner agrees that the cited components of amended Claims 1, 3, 4, and 6 recite sufficient structure to perform the claimed function when interpreted in light of the specification. The cited components, "a data augmentation component" (Claim 1), "a global parameter component" (Claim 3), "a parameter component" (Claim 4), and "a search component" (Claim 6), are no longer being interpreted under 35 U.S.C. 112(f) in light of Applicant's arguments.
Applicant's arguments, see pages 9-12, filed 11 March 2026, with respect to the rejections of Claims 1-20 under 35 U.S.C. 101 have been fully considered but they are not persuasive.
APPLICANT'S ARGUMENT: Applicant argues (page 10, paragraph 2) that "These limitations describe a specific technological process implemented by a computer to control augmentation of training data during machine-learning model training. The claims therefore recite a concrete computational technique rather than a mental process. In particular, the claims require ... operations that cannot practically be performed in the human mind."
Applicant argues (page 10, paragraph 3) that "the claims recite a specific augmentation policy architecture defined by a single scalar global augmentation parameter that simultaneously controls distortion magnitude and the number of augmentation operations and that defines a one-dimensional augmentation policy search space. These limitations describe a concrete technical solution for controlling machine-learning training processes. ... ¶ Because the claims recite a specific technological process for training machine-learning models using dynamically controlled augmentation policies, the claims do not recite a judicial exception and therefore satisfy Step 2A."
EXAMINER'S RESPONSE: Examiner respectfully disagrees. As currently claimed, the random unidimensional augmentation algorithm of amended Claim 1 appears to recite several mental process steps and appears to represent an improvement in the recited mental processes rather than an improvement in a computer or computing technology. The additional elements of amended Claim 1 that recite training of a machine learning model appear either to recite use of computing machinery as a tool to perform the mental process steps or appear merely to link the mental process steps to a particular field of use. In the absence of a recited additional element that integrates the mental process steps into a practical application or provide an inventive concept, amended Claim 1 is directed to the mental process.
APPLICANT'S ARGUMENT: Applicant argues (page 11, paragraph 1) that "Even assuming arguendo that the claims could be characterized as involving mathematical operations or data analysis, the claims integrate any such concept into a practical application. The claims do not merely analyze information; rather, they actively modify training data and control the training process of a machine-learning model."
Applicant argues (page 11, paragraph 2) that "These limitations integrate the claimed algorithm into a concrete machine-learning training pipeline that transforms the training dataset and controls operation of the model during training."
Applicant argues (page 11, paragraph 3) that "the present claims recite specific rules governing augmentation of training data during machine-learning training."
EXAMINER'S RESPONSE: Examiner respectfully disagrees. As currently recited, the additional elements of amended Claim 1 do not integrate the mental process steps into a practical application, but appear either to recite use of computing machinery as a tool to perform the mental process steps or appear merely to link the mental process steps to a particular field of use.
APPLICANT'S ARGUMENT: Applicant argues (page 12, paragraph 3) that "These elements collectively define a specific improvement to machine-learning training techniques, namely reducing augmentation policy search complexity while dynamically optimizing augmentation during model training."
Applicant argues (page 12, paragraph 4) that "The claims therefore recite far more than merely implementing an abstract idea on a generic computer. Instead, they define a specific technological architecture that improves the operation of machine-learning training systems. Such claims satisfy the 'significantly more' requirement of Alice Step 2."
EXAMINER'S RESPONSE: Examiner respectfully disagrees. As currently recited, the additional elements of amended Claim 1 do not provide significantly more, but appear either to recite use of computing machinery as a tool to perform the mental process steps or appear merely to link the mental process steps to a particular field of use.
Applicant's arguments, see pages 12-17, filed 11 March 2026, with respect to the rejections of Claims 1-20 under 35 U.S.C. 103 have been fully considered and are persuasive. The rejections of Claims 1-20 under 35 U.S.C. 103 have been withdrawn.
APPLICANT'S ARGUMENT: Applicant argues (page 13, paragraph 2) that "The rejection identifies Cubuk as teaching augmentation policies and Benton as teaching learning invariances during training, but the Office Action does not identify any disclosure in either reference of a single scalar parameter that simultaneously determines both the magnitude of augmentation operations and the number of augmentation operations applied to each training sample."
Applicant argues (page 14, paragraph 3) that "Benton is directed to learning transformation invariances within neural network models and does not disclose or suggest the augmentation policy architecture recited in the present claims. ... Rather, Benton is concerned with learning invariances through neural network training and transformation distributions, not with defining or optimizing augmentation policies using a single scalar parameter that simultaneously governs augmentation magnitude and operation count."
Applicant argues (page 15, paragraph 3) that "The rejection does not identify any disclosure in either Cubuk or Benton of generating such a one-dimensional augmentation policy search space defined by a single scalar parameter that jointly governs augmentation magnitude and operation count. The Office Action instead appears to rely on general teachings regarding augmentation or invariance learning."
Applicant argues (page 16, paragraph 3) that "The rejection appears to reconstruct the claimed invention by combining isolated concepts from the references without identifying any teaching or suggestion that would have motivated a person of ordinary skill in the art to combine those references in the manner required by the claims. Such hindsight reconstruction is improper."
EXAMINER'S RESPONSE: Examiner notes that Applicant's arguments pertain to newly claimed matter. Examiner agrees that the combination of Cubuk and Benton does not teach the features of amended Claim 1. The rejection of amended Claims 1-20 has been withdrawn in light of arguments and/or amendments.
APPLICANT'S ARGUMENT: Applicant argues (page 16, paragraph 1) that "The claims therefore recite a closed-loop training process in which augmentation behavior is dynamically modified based on performance metrics generated during model training. ¶ ... The rejection fails to identify any teaching in either reference of dynamically updating such a parameter during model training in the manner claimed."
EXAMINER'S RESPONSE: Examiner notes that Applicant's arguments pertain to newly claimed matter. Examiner further notes that the features upon which applicant's argument relies (i.e., "a closed-loop training process") are not recited in the rejected claims. Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims.
Examiner agrees that the combination of Cubuk and Benton does not teach all features of amended Claim 1. The rejection of amended Claims 1-20 has been withdrawn in light of arguments and/or amendments.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding Claim 1
Step 1
Claim 1 recites a system, and thus the claimed machine falls within a statutory category of invention.
Step 2A Prong 1
The claim recites random unidimensional augmentation algorithm ... to augment a dataset via a plurality of augmentation operations, wherein the random unidimensional augmentation algorithm: employs a single scalar global augmentation parameter that jointly defines: a distortion magnitude associated with the plurality of augmentation operations, and a number of augmentation operations applied to each training sample from the dataset, such that both the distortion magnitude and the number of augmentation operations are derived from the single scalar global augmentation parameter, which is a mental process. The claim recites generates a one-dimensional augmentation policy search space corresponding to a range of values of the global augmentation parameter, which is a mental process. The claim recites selects candidate values of the global augmentation parameter by executing a search algorithm configured to identify an optimal augmentation configuration ..., wherein the augmentation operations are selected from a pool of possible augmentation operations based on uniform probability, which is a mental process. The claim recites the system ... augments the dataset ... using the selected augmentation operations in accordance with the candidate global augmentation parameter, wherein the data augmentation component iteratively adjusts the global augmentation parameter during successive training iterations ..., thereby improving efficiency of the training of the machine learning model, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The additional element comprising: a non-transitory computer readable memory that stores computer-executable components; and a processor, operably coupled to the memory, that executes the computer-executable components stored in the memory invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element a data augmentation component that executes invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element during training of a machine learning model invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element based on training performance metrics produced by the machine learning model during training does not amount to more than generally linking the use of a judicial exception to a particular field of use (see MPEP 2106.05(h), "limit the use of the abstract idea to a particular technological environment"). The additional element automatically ... during model training invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element based on feedback derived from the training performance metrics of the machine learning model does not amount to more than generally linking the use of a judicial exception to a particular field of use (see MPEP 2106.05(h), "limit the use of the abstract idea to a particular technological environment").
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 2
Step 1
Regarding Claim 2, the rejection of Claim 1 is incorporated.
Step 2A Prong 1
The claim recites wherein the distortion magnitude controls an amount of augmentation associated with augmentation operations from the plurality of augmentation operations, which is a mental process. The claim recites wherein the system improves training efficiency of the machine learning model by enabling autonomous search-based optimization of augmentation parameters without manual grid tuning or human intervention, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 3
Step 1
Regarding Claim 3, the rejection of Claim 2 is incorporated.
Step 2A Prong 1
The claim recites a global parameter component that generates a one-dimensional search space based on the global augmentation parameter and plurality of augmentation operations selected to facilitate computational efficiency, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 4
Step 1
Regarding Claim 4, the rejection of Claim 3 is incorporated.
Step 2A Prong 1
The claim recites a parameter component that generates parameter definitions for the pool of possible augmentation operations, wherein the parameter definitions align the possible augmentation operations with the global augmentation parameter such that the amount of augmentation directly correlates with the global augmentation parameter, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 5
Step 1
Regarding Claim 5, the rejection of Claim 4 is incorporated.
Step 2A Prong 1
The claim recites wherein the parameter component controls a density of the one-dimensional search space by introducing a random uniform distribution into the parameter definitions, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 6
Step 1
Regarding Claim 6, the rejection of Claim 5 is incorporated.
Step 2A Prong 1
The claim recites a search component that executes a search algorithm on the one-dimensional search space based on the global augmentation parameter to execute an automated search space reduction, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 7
Step 1
Regarding Claim 7, the rejection of Claim 6 is incorporated.
Step 2A Prong 1
The claim recites wherein the search algorithm executes the automated search space reduction based on a unimodal function that characterizes a performance metric of the machine learning model, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2, Step 2B
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Regarding Claim 8
Step 1
Claim 8 recites a computer-implemented method, and thus the claimed process falls within a statutory category of invention.
Step 2A Prong 1
The claim recites random unidimensional augmentation algorithm ... augmenting a dataset via a plurality of augmentation operations, wherein the random unidimensional augmentation algorithm: employs a single scalar global augmentation parameter that jointly defines: a distortion magnitude associated with the plurality of augmentation operations applied to training samples of the dataset, and a number of augmentation operations applied to each training sample, such that both the distortion magnitude and the number of augmentation operations are derived from the single scalar global augmentation parameter, which is a mental process. The claim recites generates ... a one-dimensional augmentation policy search space corresponding to a range of candidate values of the global augmentation parameter, which is a mental process. The claim recites selecting ... a candidate value of the global augmentation parameter from the one-dimensional search space using a search algorithm that evaluates the performance metrics, which is a mental process. The claim recites augmenting subsequent training samples using augmentation operations selected from a pool of augmentation operations based on uniform probability in accordance with the selected candidate value of the global augmentation parameter, which is a mental process. The claim recites iteratively updating the global augmentation parameter during successive training iterations ..., thereby improving efficiency of training of the machine learning model, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2
The additional element executing, by a system operably coupled to a processor invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element during training of a machine learning model invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element by the system invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element training the machine learning model using the augmented dataset produced according to the global augmentation parameter invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element collecting ... performance metrics produced by the machine learning model for the augmented dataset amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering"). The additional element based on feedback derived from the training performance metrics does not amount to more than generally linking the use of a judicial exception to a particular field of use (see MPEP 2106.05(h), "limit the use of the abstract idea to a particular technological environment").
Step 2B
The additional element executing, by a system operably coupled to a processor invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element during training of a machine learning model invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element by the system invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element training the machine learning model using the augmented dataset produced according to the global augmentation parameter invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element collecting ... performance metrics produced by the machine learning model for the augmented dataset is well-understood, routine, conventional activity (see MPEP 2106.05(d), "storing and retrieving information in memory"). The additional element based on feedback derived from the training performance metrics does not amount to more than generally linking the use of a judicial exception to a particular field of use (see MPEP 2106.05(h), "limit the use of the abstract idea to a particular technological environment").
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Claims 9-14, dependent on Claim 8, incorporate the rejection of Claim 8. Claims 9-14 incorporate substantively all the limitations of Claims 2-7, respectively, and are rejected under the same rationales.
Regarding Claim 15
Step 1
Claim 15 recites a computer program product for a data augmentation process, the computer program product comprising a non-transitory computer readable storage medium having program instructions embodied therewith, and thus the claimed machine falls within a statutory category of invention.
Step 2A Prong 1
The claim recites random unidimensional augmentation algorithm to augment a dataset via a plurality of augmentation operations ... wherein the random unidimensional augmentation algorithm: employs a single scalar global augmentation parameter that jointly defines: a distortion magnitude associated with the plurality of augmentation operations, and a number of augmentation operations applied to each training sample from the dataset, such that both the distortion magnitude and the number of augmentation operations are derived from the single scalar global augmentation parameter, which is a mental process. The claim recites generate a one-dimensional augmentation policy search space comprising candidate values of the global augmentation parameter, which is a mental process. The claim recites select augmentation operations from a pool of possible augmentation operations based on uniform probability, which is a mental process. The claim recites augment the dataset ... using the augmentation operations configured according to a candidate value of the global augmentation parameter, which is a mental process. The claim recites iteratively adjust the global augmentation parameter during successive training iterations based on the monitored training performance metrics to improve efficiency of the training of the machine learning model, which is a mental process.
Thus, the claim recites an abstract idea.
Step 2A Prong 2
The additional element execute ... a random unidimensional augmentation algorithm invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element during training of a machine learning model invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element applied to training samples of the dataset does not amount to more than generally linking the use of a judicial exception to a particular field of use (see MPEP 2106.05(h), "limit the use of the abstract idea to a particular technological environment"). The additional element monitor training performance metrics generated by the machine learning model during training amounts to insignificant extra-solution activity (see MPEP 2106.05(g), "mere data gathering"). The additional element based on the monitored training performance metrics does not amount to more than generally linking the use of a judicial exception to a particular field of use (see MPEP 2106.05(h), "limit the use of the abstract idea to a particular technological environment").
Step 2B
The additional element execute ... a random unidimensional augmentation algorithm invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element during training of a machine learning model invokes a computer or other machinery merely as a tool to perform an existing process (see MPEP 2106.05(f), "apply it"). The additional element applied to training samples of the dataset does not amount to more than generally linking the use of a judicial exception to a particular field of use (see MPEP 2106.05(h), "limit the use of the abstract idea to a particular technological environment"). The additional element monitor training performance metrics generated by the machine learning model during training is well-understood, routine, conventional activity (see MPEP 2106.05(d), "storing and retrieving information in memory"). The additional element based on the monitored training performance metrics does not amount to more than generally linking the use of a judicial exception to a particular field of use (see MPEP 2106.05(h), "limit the use of the abstract idea to a particular technological environment").
The claim lacks additional elements that integrate it into a practical application or provide significantly more, so it is directed to an abstract idea and is ineligible.
Claims 16-20, dependent on Claim 15, incorporate the rejection of Claim 15. Claims 16-20 incorporate substantively all the limitations of Claims 9-14, respectively, and are rejected under the same rationales.
Claim Rejections - 35 USC § 103
The rejections of Claims 1-20 under 35 U.S.C. 103 are withdrawn in light of arguments and/or amendments.
Conclusion
Claims 1, 8, and 15 are rejected only under 35 U.S.C. 101 and are not rejected under 35 U.S.C. 102. A complete search of Claims 1, 8, and 15 did not uncover any prior art that teaches or fairly suggests:
1. A system, comprising:
a non-transitory computer readable memory that stores computer-executable components; and
a processor, operably coupled to the memory, that executes the computer-executable components stored in the memory, wherein the computer-executable components comprise:
a data augmentation component that executes a random unidimensional augmentation algorithm during training of a machine learning model to augment a dataset via a plurality of augmentation operations, wherein the random unidimensional augmentation algorithm:
employs a single scalar global augmentation parameter that jointly defines: a distortion magnitude associated with the plurality of augmentation operations, and a number of augmentation operations applied to each training sample from the dataset, such that both the distortion magnitude and the number of augmentation operations are derived from the single scalar global augmentation parameter;
generates a one-dimensional augmentation policy search space corresponding to a range of values of the global augmentation parameter; and
selects candidate values of the global augmentation parameter by executing a search algorithm configured to identify an optimal augmentation configuration based on training performance metrics produced by the machine learning model during training, wherein the augmentation operations are selected from a pool of possible augmentation operations based on uniform probability, and the system automatically augments the dataset during model training using the selected augmentation operations in accordance with the candidate global augmentation parameter, wherein the data augmentation component iteratively adjusts the global augmentation parameter during successive training iterations based on feedback derived from the training performance metrics of the machine learning model, thereby improving efficiency of the training of the machine learning model.
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Hendrycks, et al., "AUGMIX: A Simple Data Processing Method To Improve Robustness And Uncertainty," teach a method to improve the robustness and uncertainty estimates of image classifiers by means of an efficient data processing technique that stochastically samples and layers simple augmentation operations in concert with a consistency loss to enforce a consistent embedding by the classifier.
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ROBERT N DAY whose telephone number is (703)756-1519. The examiner can normally be reached M-F 9-5.
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, Kakali Chaki can be reached at (571) 272-3719. 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.
/R.N.D./Examiner, Art Unit 2122
/KAKALI CHAKI/Supervisory Patent Examiner, Art Unit 2122
Prosecution Insights