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 .
Claim Rejections - 35 USC § 101
Claims 1, 16, 17, and 20-22 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more.
Claim 1
Step 1 This part of the eligibility analysis evaluates whether the claim falls within any statutory category. See MPEP 2106.03. The claim recites “An information processing method executed by one or more processors, the method comprising: via the one or more processors, (a) extracting a learning subset... (b) extracting a first evaluation subset... (c) extracting a second evaluation subset... (d) obtaining a value of evaluation result... (e) obtaining a value of evaluation result... (f) obtaining a value of domain generalization..., repeating steps of (a)-(f)... and determining a learning subset candidate...” Therefore, the claim is a process, which is a statutory category of invention under 35 U.S.C. § 101.
Step 2A Prong 1 This part of the eligibility analysis evaluates whether the claim recites a judicial exception. As explained in MPEP 2106.04, subsection II, a claim “recites” a judicial exception when the judicial exception is “set forth” or “described” in the claim.
Limitation “(a) extracting a learning subset from a dataset having one domain under a designated condition and training a learning model by using the learning subset, wherein the dataset includes a plurality of subsets having different domains” is classified as Mathematical Concept. The underlying operational logic is a judicial exception because machine learning model training involves mathematical optimization algorithms that constitute abstract mathematical concepts, including when expressed entirely in plain text or prose representations without formal algebraic symbols.
Limitation “(b) extracting a first evaluation subset from the dataset having the one domain under the designated condition and training the learning model by using the first evaluation subset” is classified as Mathematical Concept. The underlying operational logic is a judicial exception because machine learning model training involves mathematical optimization algorithms that constitute abstract mathematical concepts, including when expressed entirely in plain text or prose representations without formal symbols.
Limitation “(c) extracting a second evaluation subset from the dataset having an other domain different from the one domain under the designated condition and training the learning model by using the second evaluation subset” is classified as Mathematical Concept. The underlying operational logic is a judicial exception because machine learning model training involves mathematical optimization algorithms that constitute abstract mathematical concepts, including when expressed entirely in plain text or prose representations without formal symbols.
Limitation “(d) obtaining a value of evaluation result of the trained learning model by using the first evaluation subset” is classified as Mental Process. The underlying operational logic is a judicial exception because this recites the activity of reading/retrieving numerical performance metrics from stored data, which is an information gathering and retrieval practice practically performable in the human mind or by a human using pen and paper.
Limitation “(e) obtaining a value of evaluation result of the trained learning model by using the second evaluation subset” is classified as Mental Process. The underlying operational logic is a judicial exception because this recites the activity of reading/retrieving numerical performance metrics from stored data, which is an information gathering and retrieval practice practically performable in the human mind or by a human using pen and paper.
Limitation “(f) obtaining a value of domain generalization for the learning set, by comparing the value (ii) and the value (i)” is classified as Mental Process. The underlying operational logic is a judicial exception because this recites the activity of performing numerical comparison operations to derive derived metrics, which represents a logical comparison practically performable in the human mind or by a human using pen and paper.
Limitation “repeating steps of (a)-(f) and obtaining values of domain generalization for a plurality of learning subsets” is classified as Mental Process. The underlying operational logic is a judicial exception because this recites repeatedly performing data retrieval, comparison, and calculation operations across multiple iterations, which represents an iterative computation workflow practically performable in the human mind or by a human using pen and paper.
Limitation “determining a learning subset candidate, which produces a highest value of domain generalization among the extracted and trained learning subsets” is classified as Mental Process. The underlying operational logic is a judicial exception because this recites the activity of identifying/selecting an optimal option based on numerical criteria (argmax selection), which is a logical comparison and selection process practically performable in the human mind or by a human using pen and paper.
“Unless it is clear that a claim recites distinct exceptions, such as a law of nature and an abstract idea, care should be taken not to parse the claim into multiple exceptions, particularly in claims involving abstract ideas.” MPEP 2106.04, subsection II.B. However, if possible, the examiner should consider the limitations together as a single abstract idea rather than as a plurality of separate abstract ideas to be analyzed individually. “For example, in a claim that includes a series of steps that recite mental steps as well as a mathematical calculation, an examiner should identify the claim as reciting both a mental process and a mathematical concept for Step 2A, Prong One to make the analysis clear on the record.” MPEP 2106.04, subsection II.B. Under such circumstances, however, the Supreme Court has treated such claims in the same manner as claims reciting a single judicial exception. Id. (discussing Bilski v. Kappos, 561 U.S. 593 (2010)). The limitations mentioned are considered together as a single abstract idea for further analysis. (Step 2A, Prong One: YES).
Step 2A Prong 2 This part of the eligibility analysis evaluates whether the claim as a whole integrates the recited judicial exception into a practical application of the exception or whether the claim is “directed to” the judicial exception. This evaluation is performed by (1) identifying whether there are any additional elements recited in the claim beyond the judicial exception, and (2) evaluating those additional elements individually and in combination to determine whether the claim as a whole integrates the exception into a practical application. See MPEP 2106.04(d).
The additional elements:
one or more processors
extracting subsets from datasets
training learning models using extracted subsets
obtaining evaluation results
comparing numerical values
iterating through multiple subsets
determining an optimal subset based on metric comparison
MPEP § 2106.05(a) “Improvements to the Functioning of a Computer or to Any Other Technology or Technical Field
An eligibility evaluation of one or more processors, extracting subsets from datasets, training learning models using extracted subsets, obtaining evaluation results, comparing numerical values, iterating through multiple subsets, and determining optimal subset based on metric comparison confirms that the claim lacks improvements to functioning of computer or technical field under MPEP § 2106.05(a). The claim language defines broad operational goals at a high level of functional generality rather than an explicit technical mechanism that improves underlying computer processing speeds, memory limits, or network architecture hardware throughput. As detailed in specification paragraphs [0073]-[0083], these operations rely on conventional data structures and routine computer functions, leading to the conclusion that these elements do not integrate the exception into a practical application via an improvement to a computer or technical field.
MPEP § 2106.05(b) Particular Machine. The judicial exception does not apply to any particular machine.
An eligibility evaluation of one or more processors, extracting subsets from datasets, training learning models using extracted subsets, obtaining evaluation results, comparing numerical values, iterating through multiple subsets, and determining optimal subset based on metric comparison confirms that the recited elements do not constitute a specialized or non-generic structural machine configuration under MPEP § 2106.05(b). While the claim may recite hardware circuits, the specification defines these components broadly enough to encompass general-purpose processors, memory boundaries, or standard state machines. As detailed in specification paragraphs [0068]-[0072], the operating environment utilizes ordinary processing units and basic storage structures, leading to the conclusion that these elements do not constitute a particular machine.
MPEP § 2106.05(c) Particular Transformation.
An eligibility evaluation of one or more processors, extracting subsets from datasets, training learning models using extracted subsets, obtaining evaluation results, comparing numerical values, iterating through multiple subsets, and determining optimal subset based on metric comparison confirms these elements do not describe a permanent change of state for tangible items because they involve purely electronic manipulation of digital records rather than structural modifications to physical matter. As detailed in specification paragraphs [0070]-[0072], the system manipulates digital bits rather than tangible items, leading to the conclusion that these elements do not constitute a particular transformation.
MPEP § 2106.05(e) Other Meaningful Limitations.
An eligibility evaluation of one or more processors, extracting subsets from datasets, training learning models using extracted subsets, obtaining evaluation results, comparing numerical values, iterating through multiple subsets, and determining optimal subset based on metric comparison confirms these elements describe a processing framework at a high level of functional generality because they define broad operational goals-such as basic data ingestion, storage, or interface delivery-without imposing specific technical constraints. As detailed in specification paragraphs [0079]-[0083], the boundaries are left functionally wide open, leading to the conclusion that these elements do not introduce meaningful limitations.
MPEP § 2106.05(f) Mere Instructions Apply to Exception
An eligibility evaluation of one or more processors, extracting subsets from datasets, training learning models using extracted subsets, obtaining evaluation results, comparing numerical values, iterating through multiple subsets, and determining optimal subset based on metric comparison confirms these elements recite a desired operational outcome at a high level of functional generality because they state what broad results are accomplished without specifying concrete technical mechanisms or specialized physical parameters to achieve them. As detailed in specification paragraphs [0073]-[0083], the descriptions focus entirely on high-level goals rather than specific architectural boundaries, leading to the conclusion that these elements represent a mere instruction to apply the exception.
MPEP § 2106.05(g) Insignificant Extra-Solution Activity
An eligibility evaluation of one or more processors, extracting subsets from datasets, training learning models using extracted subsets, obtaining evaluation results, comparing numerical values, iterating through multiple subsets, and determining optimal subset based on metric comparison confirms these elements represent routine preprocessing or post-processing because they describe basic data acquisition or display operations that occur prior to or following the primary algorithmic exception logic. They are merely incidental to the primary process since they serve as conventional input/output mechanisms that facilitate standard interaction without adding an inventive or technical step to the actual core solution.
MPEP § 2106.05(h) Field of Use and Technological Environment
An eligibility evaluation of one or more processors, extracting subsets from datasets, training learning models using extracted subsets, obtaining evaluation results, comparing numerical values, iterating through multiple subsets, and determining optimal subset based on metric comparison confirms these elements do not integrate into a field of use. Instead, they represent a conventional field-of-use limitation without technological improvement. As detailed in specification paragraphs [0001]- [0009], the environment is described as generic backdrop, leading to the conclusion that these elements are not integrated into a meaningful field of use.
Because the additional elements represent a mere instruction to apply the exceptions using generic computer components at a high level of functional generality, the judicial exceptions are NOT integrated into a practical application. Therefore, the claim is directed to abstract ideas, necessitating an analysis under Step 2B.
Step 2B The ordered combination of additional elements fails to add an inventive concept because they utilize routine, well-understood, and conventional processing activities executed on generic hardware. As detailed across specification paragraphs [0068]-[0072] and [0073]-[0083], these additional elements describe a general software/system process running on conventional hardware rather than a specific, non-generic technological solution to a technical problem inherent to the machine itself. The “domain generalization” metric is treated as a nominal label given to a purely mathematical result, which does not alter the physical or computational nature of the hardware. Because the ordered combination of additional elements only utilizes standard computer components and conventional steps to execute the abstract mathematical optimization and mental process concepts, the independent claim fails to provide an inventive concept and is patent ineligible under 35 U.S.C. § 101.
Claims 16 and 17 are similar to claim 1. The claims are rejected based on the same reasons.
Claim 17 incorporates all limitations of claim 1 and adds a non-transitory computer-readable tangible recording medium on which a program is recorded for causing processors to execute the method. This feature describes an algorithmic outcome at a high level of functional generality, citing specification paragraphs [0068]-[0072] that describe the hardware configuration. The broad claim language fails to capture the narrow non-generic technical structures or physical hardware architecture configurations disclosed in those paragraphs. By claiming the storage medium at a high level of functional generality without incorporating the specification's structural limitations, the text leaves standard industry operations or generic computer automation to accomplish the goal. Because the broad text fails to incorporate specific technological structures from the specification, this limitation fails to provide a patent-eligible restriction and the claim is ineligible under 35 U.S.C. 101.
Claim 19 recites “wherein the value of the evaluation result is a value of classification accuracy of the trained learning model, or a degree of association between a feature amount and a response variable for each of the domains” This feature describes an algorithmic outcome at a high level of functional generality, citing specification paragraphs [0057]-[0064] that describe performance metrics and evaluation results. Because the broad text fails to incorporate specific technological structures from the specification, the claim is ineligible under 35 U.S.C. 101.
Claim 20 recites “wherein the highest value of domain generalization is a value when the value obtained by subtracting the value (ii) from the value (i) produces a minimum value among the extracted and trained learning subsets” This feature describes an outcome at a high level of functional generality, citing specification paragraphs [0100] that describe evaluating domain generalization by using differences between evaluation results. Because the broad text fails to incorporate specific technological structures from the specification, this limitation fails to provide a patent-eligible restriction and the claim is ineligible under 35 U.S.C. 101.
Claim 21 recites “wherein the highest value of domain generalization is a value when the value (ii) by using the second evaluation subset produces a maximum value among the extracted and trained learning subsets” This feature describes an outcome at a high level of functional generality, citing specification paragraphs [0057]-[0064] that describe performance metrics and evaluation results for different domains. The broad claim language fails to capture the narrow non-generic technical structures or physical hardware architecture configurations disclosed in those paragraphs. Because the broad text fails to incorporate specific technological structures from the specification, the claim is ineligible under 35 U.S.C. 101.
Claim 22 recites “wherein at least one of the learning subset, the first evaluation subset, and the second evaluation subset involves searching for a subset with higher value of domain generalization by repeating operations of adding or deleting samples with respect to a starting subset” This feature describes an outcome at a high level of functional generality, citing specification paragraphs [0169]-[0173] that describe the iterative search algorithm. Because the broad text fails to incorporate specific technological structures from the specification, the claim is ineligible under 35 U.S.C. 101.
Response to Arguments
Section Rejection under 35 U.S.C – pg. 5-6
Applicant argues that “Applicant has amended claims 1 and 16 to clarify the scope of the present invention and canceled claims 2-12, 14, 15 and 18 to simplify the issues in view of the Examiner's rejection, without prejudice or disclaimer to the subject matter included therein. Claim 1, as amended, is directed to one of the embodiments described in the present specification, which selects/determines a learning model, among a plurality of learning models, providing the performance (classification accuracy) with the smallest performance degradation even with the domain shift. Applicant submits that the claimed method provides sufficiently specific improvements in the algorithms over prior art methods, and provides significantly more than a mere replacement of a computer with pen and paper, which is an improved method for selecting a model with high classification accuracy even when there are domain shifts among the models. Applicant respectfully submits that the claim 1, as amended, and for the same reasons claim 16, as amended, comply with the requirements of 35 U.S.C. § 101. Accordingly, reconsideration and withdrawal of this rejection are respectfully requested.”
The Applicant's arguments rely heavily on technical benefits disclosed exclusively in the specification. However, features, advantages, or specific embodiments described in the specification cannot be read into the claims to establish eligibility or patentability if they are not explicitly recited in the claim body. Because claims 1 and 16 fails to incorporate the specific structural and technical parameters described in the specification, the Applicant's arguments regarding the technical nature of the invention are legally unpersuasive as applied to the pending claim boundaries.
The applicant argues that claim 1 is directed to an embodiment described in the specification that selects/determines a learning model providing performance with the smallest performance degradation during a domain shift. The Examiner disagrees.
While specification paragraphs [0009]-[0011] discuss evaluating domain generalization performance to select a model candidate, the actual text of claim 1 does not recite any specific structural limitations, narrow computational rules, or specialized data structure constraints. Rather, the claim language uses broad, high-level functional concepts such as “extracting a learning subset,” “training a learning model,” and “obtaining a value of evaluation result.” Because these steps describe broad operational categories rather than a specific technical implementation, the basic character of the claim remains directed to unintegrated judicial exceptions.
The applicant submits that the claimed method provides sufficiently specific algorithmic improvements over prior art methods. Under MPEP § 2106.05(a), a claim must describe an explicit technical mechanism that improves underlying computer processing speeds, memory utilization, or hardware throughput to establish a patent-eligible improvement. The additional elements - processors, extracting data subsets, training generic models, and retrieving performance values-represent well-understood, routine, and conventional activities in the field of machine learning and data processing. The claim merely automates a standard data handling workflow on generic computing hardware, which fails to improve the underlying functioning of the computer or the technical field beyond generic computation.
The applicant argues that the claim provides “significantly more” than a mere replacement of a computer with pen and paper. The Examiner disagrees. While the claim recites physical hardware nouns like “one or more processors,” these elements act as a mere generic environment and vehicle for executing software instructions.
The claim fails to recite a specialized machine configuration or a particular physical transformation of tangible matter. Evaluating performance metrics and tracking loop repetitions are mental concepts that remain performable by a human using basic physical aids. The generic recitation of standard computer hardware to execute an abstract workflow does not provide an inventive concept under Step 2B.
The applicant states that the claim defines an improved method for selecting a model with high classification accuracy during a domain shift. The Examiner disagrees. The claim concludes with the broad goal of “determining a learning subset candidate, which produces a highest value of domain generalization.” This language recites a desired operational outcome at a high level of functional generality without specifying the concrete mechanics or technical parameters required to achieve it.
The applicant argues that the claim solves a technical problem by technical means, but fails to point to any specific, narrow structural limitations within the claim text that define a specialized computational rule. The arguments rely entirely on high-level functional concepts, and the rejection under 35 U.S.C. § 101 is therefore maintained.
Consideration
Claim 1
An information processing method executed by one or more processors, the method comprising:
via the one or more processors,
(a) extracting a learning subset from a dataset having one domain under a designated condition and training a learning model by using the learning subset, wherein the dataset includes a plurality of subsets having different domains,
(b) extracting a first evaluation subset from the dataset having the one domain under the designated condition and training the learning model by using the first evaluation subset;
(c) extracting a second evaluation subset from the dataset having an other domain different from the one domain under the designated condition and training the learning model by using the second evaluation subset;
(d) obtaining a value of evaluation result of the trained learning model by using the first evaluation subset (i),
(e) obtaining a value of evaluation result of the trained learning model by using the second evaluation subset (ii),
(f) obtaining a value of domain generalization for the learning set, by comparing the value (ii) and the value (i),
repeating steps of (a)-(f) and obtaining values of domain generalization for a plurality of learning subsets, and
determining a learning subset candidate, which produces a highest value of domain generalization among the extracted and trained learning subsets;
wherein extracting the learning subset in step (a) comprises:
evaluating a degree of association between each of a plurality of feature amounts of the dataset and a response variable for each of the domains;
identifying specific feature amounts that maintain a degree of association that meets or exceeds a designated threshold across a specified minimum number of domains as possessing domain generality; and
ranking the extracted learning subset as having a higher degree of domain generalization based on a higher volume of the identified feature amounts possessing domain generality; and
wherein the repeating and determining steps comprise searching for the learning subset candidate by:
iteratively executing sequential sample modifications on an active learning data subset starting from an initial baseline data state;
comparing domain generality characteristics between a forward-modified subset candidate and a backward-modified subset candidate; and
updating the active learning data subset with whichever candidate exhibits a higher volume of feature amounts possessing domain generality.
The current claim was ineligible under 35 U.S.C. § 101 because it merely recited a broad functional goal-finding a subset with high domain generalization-without describing the specific rules to achieve that result.
The underlined text resolves this abstract generality by restricting the invention to a concrete technological process. The process explicitly evaluates feature-to-response associations across domains, isolates specific features meeting a threshold across a minimum number of domains to establish "domain generality," and ranks the extracted subset based on the volume of those robust features. The repeating and determining steps are limited to a computerized search loop that modifies an active learning data subset G from an initial baseline state. The processor must execute a comparative optimization sequence: evaluating an expanded candidate matrix GF obtained by adding a sample against a truncated candidate matrix GB obtained by deleting a sample, and dynamically updating the active data subset G with whichever option exhibits a higher volume of domain-generic features.
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure
U.S. Pb 2025/0148365 - Padfield discloses systems and methods for continuous training of machine learning (ML) models on changing data. In particular, the present disclosure provides example approaches to model training that take advantage of constantly evolving data that may be available in various ancillary systems that contain large amounts of data, but which are not specific to or dedicated for model training.
U.S. Pub 2025/0053618 - Soman discloses a method performed by a node for handling drift in data. The node obtains a dataset including a plurality of datapoints corresponding to a plurality of values of one or more dependent variables for a plurality of first features over a time period. The node determines, using machine learning and explainability, in the absence of determining whether or not the plurality of datapoints has a drift, whether or not there has been a change in respective one or more characteristics of a subset of the plurality of first features having a largest contribution to a variability of the datapoints in the plurality of datapoints based on a threshold from a first time period to a second time period. The node then initiates application of a drift policy on the plurality of datapoints based on a result of the determination.
Conclusion
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.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any 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 HAU HAI HOANG whose telephone number is (571)270-5894. The examiner can normally be reached 1st biwk: Mon-Thurs 7:00 AM-5:00 PM; 2nd biwk: Mon-Thurs: 7:00 am-5:00pm, Fri: 7:00 am - 4:00pm.
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, Boris Gorney can be reached at 571-270-5626. 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.
HAU HAI. HOANG
Primary Examiner
Art Unit 2154
/HAU H HOANG/ Primary Examiner, Art Unit 2154
