Prosecution Insights
Last updated: April 19, 2026
Application No. 18/374,955

TRAINING FOUNDATION MODELS ON TABULAR DATA

Non-Final OA §101§112
Filed
Sep 29, 2023
Examiner
SHECHTMAN, CHERYL MARIA
Art Unit
2164
Tech Center
2100 — Computer Architecture & Software
Assignee
International Business Machines Corporation
OA Round
3 (Non-Final)
72%
Grant Probability
Favorable
3-4
OA Rounds
3y 2m
To Grant
99%
With Interview

Examiner Intelligence

Grants 72% — above average
72%
Career Allow Rate
215 granted / 300 resolved
+16.7% vs TC avg
Strong +28% interview lift
Without
With
+28.1%
Interview Lift
resolved cases with interview
Typical timeline
3y 2m
Avg Prosecution
21 currently pending
Career history
321
Total Applications
across all art units

Statute-Specific Performance

§101
22.2%
-17.8% vs TC avg
§103
34.8%
-5.2% vs TC avg
§102
18.4%
-21.6% vs TC avg
§112
19.6%
-20.4% vs TC avg
Black line = Tech Center average estimate • Based on career data from 300 resolved cases

Office Action

§101 §112
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 . A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on February 20, 2026 has been entered. Claims 1-20 are pending. Claims 1, 5, 9, 13, 17 and 19 are amended. Response to Arguments Referring to the 35 USC 101 rejection of claims 1-20, Applicant’s amendments to the claims and arguments have been considered but are not found persuasive. Applicant argues that the claims, as amended, specifically, the limitations “training, by the processor set, a tabular foundation model that performs the predictive data management task in a data fabric application, based on the masked subset of the identified informative features within the first cluster of the plurality of clusters of the received plurality of tabular data records” and “optimizing, by the processor set, the trained tabular foundation model using self-supervision techniques” do not recite mental steps and integrates the judicial exception into a practical application. However Examiner respectfully disagrees. Examiner submits that the act of training a tabular foundation model based on the masked subset of features is a mental step because a user can perform the masking mentally by ignoring certain types of features from the set of records in the table in order to perform a predictive data management task. Furthermore, the tabular foundation model performing the predictive data management task in a data fabric application does not provide integration into a practical application because the performing of a prediction task within a data fabric application is described at a high level of generality and does not provide specific details on the training set or how the predictive task is achieved aside from abstract idea of masking the set of informative features. This limitation therefore provides nothing more than mere instructions to implement the abstract idea on a generic computer and within a generic computer application. With respect to the optimization limitation, a user can decide how to improve the trained model for future use in the human mind. Furthermore, the optimization being performed using self-supervision techniques does not integrate the judicial exception into a practical application because the self-supervision techniques are merely considered linking the mental step of optimizing the model to the technical field of self-supervision, which does not add significantly more to the judicial exception, see Parker v. Flook, 437 U.S. 584, 198 USPQ 193 (1978), in MPEP 2106.05(h). Applicant argues that the claims are directed to an improved method of training a foundation model with tabular data, however, Examiner submits that this improvement is only achieved through the implementation of the mental steps of generating clusters, identifying informative features and masking steps. There are no further details claimed that would lead one to realize the technical improvement in training the tabular data-rather the training step is recited at a high level of generality. As such, the claims do not provide a practical application for the judicial exception or reflect a technical improvement. Claims 1-20 remain rejected under 35 USC 101 for the reasons stated above and further as addressed in this Office action. Claim Remarks Claims 9-16 are not interpreted by the Examiner as containing transitory signals. The instant specification in para 41 expressly disavows the term ‘computer readable storage media’ from including transitory signals per se. 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. Claims 1, 9 and 17 recite receiving, by a processor set, a plurality of tabular data records; generating, by the processor set, a plurality of clusters within the received plurality of tabular data records, wherein each cluster is associated with a specific real-world entity; identifying, by the processor set, informative features within a first cluster of the plurality of clusters of the received plurality of tabular data records by identifying a random subset of tabular records for each cluster; masking, by the processor set, a subset of the identified informative features within the first cluster of the plurality of clusters of the received plurality of tabular data records, the subset of the identified informative features including maximum information associated with one or more source rows of the first cluster for a predictive data management task; training, by the processor set, a tabular foundation model that performs the predictive data management task in a data fabric application, based on the masked subset of the identified informative features within the first cluster of the plurality of clusters of the received plurality of tabular data records; and optimizing, by the processor set, the trained tabular foundation model using self-supervision techniques. Step 1: The claims as a whole fall within one or more statutory categories. Step 2A prong 1: At least claims 1, 9 and 17 recite limitations that are abstract ideas. The limitation “generating a plurality of clusters within the received plurality of tabular data records, wherein each cluster is associated with a specific real-world entity” is a mental step. One can mentally create groupings of data records in the human mind. Thus, the claimed limitation can be performed by the human mind. Furthermore, the limitation “identifying informative features within a first cluster of the plurality of clusters of the received plurality of tabular data records by identifying a random subset of tabular records for each cluster” is also a mental step. A user can randomly select a set of records from a table and identify selected features within the set of records. Thus, the claimed limitation can be performed by the human mind. Furthermore, the limitation “masking a subset of the identified informative features within the first cluster of the plurality of clusters of the received plurality of tabular data records, the subset of the identified informative features including maximum information associated with one or more source rows of the first cluster” is also a mental step. A user can perform the masking mentally by ignoring certain types of features from the set of records in the table in order to perform a predictive data management task. Thus, the claimed limitation can be performed by the human mind. Furthermore, the limitation “training a tabular foundation model that performs the predictive data management task in a data fabric application, based on the masked subset of the identified informative features within the first cluster of the plurality of clusters of the received plurality of tabular data records” is also a mental step. A user can mentally perform the training based on the user’s prior mental step of deciding which subset of data to mask or ignore in order to then input into the model as seed data. Thus, the claimed limitation can be performed by the human mind. Finally, the limitation “optimizing the trained tabular foundation model” is a mental step. A user can decide how to improve the trained model for future use. Thus, the claimed limitation can be performed by the human mind. Step 2A prong 2: Claims 1, 9 and 17 recite the limitation “receiving a plurality of tabular data records”. This limitation is an additional element and is insignificant extra-solution activity as retrieval/receiving of data (i.e. mere data gathering) such as 'obtaining information' as identified in MPEP 2106.05(g) and does not provide integration into a practical application. Claims 1, 9 and 17 recite that the masking of the subset of identified informative features is for a predictive data management task and that the tabular foundation model performs the predictive data management task in a data fabric application. These limitations are additional elements and do not provide integration into a practical application because the masking for a predictive task and the performing of the prediction task within a data fabric application are described at a high level of generality and do not provide specific details on the training set or how the predictive task is achieved aside from abstract idea of masking the set of informative features. These limitations therefore provide nothing more than mere instructions to implement the abstract idea on a generic computer and within a generic computer application. Claims 1, 9 and 17 also recite that the optimizing of the model is done “using self-supervision techniques”. This limitation is an additional element and does not provide integration into a practical application because the self-supervision techniques are merely considered linking the mental step of optimizing the model to the technical field of self-supervision, which does not add significantly more to the judicial exception, see Parker v. Flook, 437 U.S. 584, 198 USPQ 193 (1978), in MPEP 2106.05(h). Furthermore, Claims 1, 9 and 17 recite the following additional elements “a processor set”, “computer readable storage media having program instructions”, and “a system”, note that these recited additional elements are a high-level recitation of generic computer components to perform the mental process and applied on a computer as in MPEP 2106.05(f), which does not provide integration into a practical application. Step 2B: the conclusions for the additional elements representing mere implementation using a computer are carried over and do not provide significantly more. With respect to the "receiving” limitation identified as insignificant extra-solution activity above when re-evaluated this element is well-understood, routine, and conventional as evidenced by the court cases in MPEP 2106.05(d)(II), "i. Receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information); … OIP Techs., Inc., v. Amazon.com, Inc., 788 F.3d 1359, 1363, 115 USPQ2d 1090, 1093 (Fed. Cir. 2015) (sending messages over a network); buySAFE, Inc. v. Google, Inc., 765 F.3d 1350, 1355, 112 USPQ2d 1093, 1096 (Fed. Cir. 2014) (computer receives and sends information over a network);" and thus remains insignificant extra-solution activity that does not provide significantly more. Therefore, the claims as a whole do not change this conclusion and the claims are ineligible. Claims 2 and 10 depend from claims 1 and 9 and thus include all the limitations of claims 1 and 9, therefore claims 2 and 10 recite the same abstract ideas of "mental processes". Claims 2 and 10 furthermore recite exporting at least a portion of the trained tabular foundation model to a database or a user device. Step 1: Claims 2 and 10 as a whole fall within one or more statutory categories. Step 2A prong 1: Claims 2 and 10, depending from claims 1 and 9, also recite the same abstract ideas of "mental processes" as the parent claims. Step 2A prong 2: Claims 2 and 10 recite the limitation: “exporting at least a portion of the trained tabular foundation model to a database or a user device”. The exporting step is an additional element and is insignificant extra-solution activity, specifically post-solution activity as storing data to a database or storage within a device, such as 'the outputting step' as identified in MPEP 2106.05(g) and does not provide integration into a practical application. Step 2B: With respect to the "exporting” limitation identified as insignificant extra-solution activity above, when re-evaluated this element is well-understood, routine, and conventional as evidenced by the court cases in MPEP 2106.05(d)(II), "i iv. Storing and retrieving information in memory, Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015); OIP Techs., 788 F.3d at 1363, 115 USPQ2d at 1092-93" and thus remains insignificant extra-solution activity that does not provide significantly more. Therefore, claims 2 and 10 as a whole do not change this conclusion and the claims are ineligible. Claims 3, 4, 7, 8, 11, 12, 15, 16, 18 and 20 depend from claims 1, 9 and 17 and thus include all the limitations of claims 1, 9 and 17, therefore claims 3, 4, 7, 8, 11, 12, 15, 16, 18 and 20 recite the same abstract idea of "mental process". Claims 3, 4, 7, 8, 11, 12, 15, 16, 18 and 20 furthermore recite: bucketing, by the processor set, the received plurality of tabular data records into at least one bucket by analyzing the plurality of tabular data records and assigning each of the tabular data records to a bucket of the at least one bucket based on at least one of the identified informative features; and generating, by the processor set, a plurality of transitive links between data records having a transitive relationship (claims 3, 4, 11, 12, and 18); wherein the tabular foundation model is trained to predict a representative row that captures information from at least one source row (claims 7 and 15); and wherein identifying the informative features within the first cluster further comprises using an explainable entity matching technique to identify a column a column in the first cluster containing the informative features (claims 8, 16 and 20). Step 1: Claims 3, 4, 7, 8, 11, 12, 15, 16, 18 and 20 as a whole fall within one or more statutory categories. Step 2A prong 1: Claims 3, 4, 7, 8, 11, 12, 15, 16, 18 and 20 recite limitations that are abstract ideas. The limitations “bucketing, by the processor set, the received plurality of tabular data records into at least one bucket by analyzing the plurality of tabular data records and assigning each of the tabular data records to a bucket of the at least one bucket based on at least one of the identified informative features; and generating, by the processor set, a plurality of transitive links between data records having a transitive relationship” are mental steps. One can mentally analyze records and determine a category or bucket to which the records belong. Furthermore, one can mentally generate links between records based on determining that the records are related. Thus, the claimed limitations can be performed by the human mind. Step 2A prong 2: Claims 3, 4, 7, 8, 11, 12, 15, 16, 18 and 20 do not recite any additional elements that would integrate the judicial exception into a practical application. Step 2B: Claims 3, 4, 7, 8, 11, 12, 15, 16, 18 and 20 do not recite any additional elements that would provide significantly more than the judicial exception. Therefore, claims 3, 4, 7, 8, 11, 12, 15, 16, 18 and 20 as a whole are ineligible. Claims 5, 6, 13, 14 and 19 depend from claims 1, 9 and 17 and thus include all the limitations of claims 1, 9 and 17, therefore claims 5, 6, 13, 14 and 19 recite the same abstract idea of "mental process". Claims 5, 6, 13, 14 and 19 recite: “wherein the at least three loss functions comprise a representative row generation function, a masked cell modeling function, and an entity matching function; and wherein the representative row generation function measures a categorical cross entropy loss, the masked cell modeling function measures a cross entropy loss, and the entity matching function measures a contrastive learning score”. Step 1: Claims 5, 6, 13, 14 and 19 as a whole fall within one or more statutory categories. Step 2A prong 1: Claims 5, 6, 13, 14 and 19 recite limitations that are abstract ideas. The limitations “wherein the at least three loss functions comprise a representative row generation function, a masked cell modeling function, and an entity matching function; and wherein the representative row generation function measures a categorical cross entropy loss, the masked cell modeling function measures a cross entropy loss, and the entity matching function measures a contrastive learning score” are considered mathematical calculations. Step 2A prong 2: Claims 5, 6, 13, 14 and 19 do not recite any additional elements that would integrate the judicial exception into a practical application. Step 2B: Claims 5, 6, 13, 14 and 19 do not recite any additional elements that would provide significantly more than the judicial exception. Therefore, claims 5, 6, 13, 14 and 19 as a whole are ineligible. To expedite a complete examination of the instant application, the claims rejected under 35 U.S.C. 101 (nonstatutory) above are further rejected as set forth below in anticipation of applicant amending these claims to place them within the four statutory categories of the invention. Claim Rejections - 35 USC § 112 The following is a quotation of 35 U.S.C. 112(b): (b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention. The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph: The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention. Claims 1-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. Regarding claims 1, 9 and 17, the phrase "maximum information" in lines 10, 12, and 13 of the respective claims renders the claims indefinite because it is unclear as to what is meant by “maximum information”. The specification does not describe a standard for ascertaining what constitutes a maximum amount of information. For purposes of examination, Examiner will assume that the limitation reads “information”. All claims depending from the aforenoted claims are also rejected by virtue of their dependencies. Due to the 35 USC 112 rejections, the claims have been examined as best understood by the Examiner. Novel and/or Non-obvious Subject Matter Claims 1-20 were found to be novel for the reasons indicated in the Final action dated September 25, 2025. Conclusion Any inquiry concerning this communication or earlier communications from the examiner should be directed to CHERYL M SHECHTMAN whose telephone number is (571)272-4018. The examiner can normally be reached on Mon-Fri: 8am-4pm. 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, Amy Ng can be reached on 571-270-1698. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300. Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000. CHERYL M SHECHTMANPatent Examiner Art Unit 2164 /C.M.S/ /AMY NG/Supervisory Patent Examiner, Art Unit 2164
Read full office action

Prosecution Timeline

Sep 29, 2023
Application Filed
Mar 08, 2025
Non-Final Rejection — §101, §112
Jun 09, 2025
Examiner Interview Summary
Jun 09, 2025
Applicant Interview (Telephonic)
Jun 16, 2025
Response Filed
Sep 20, 2025
Final Rejection — §101, §112
Nov 07, 2025
Examiner Interview Summary
Nov 07, 2025
Applicant Interview (Telephonic)
Nov 20, 2025
Response after Non-Final Action
Feb 20, 2026
Request for Continued Examination
Mar 04, 2026
Response after Non-Final Action
Mar 09, 2026
Non-Final Rejection — §101, §112 (current)

Precedent Cases

Applications granted by this same examiner with similar technology

Patent 12554725
System and Method for Searching Electronic Records using Gestures
2y 5m to grant Granted Feb 17, 2026
Patent 12536201
SYSTEM AND METHODS FOR VARYING OPTIMIZATION SOLUTIONS USING CONSTRAINTS BASED ON AN ENDPOINT
2y 5m to grant Granted Jan 27, 2026
Patent 12530380
OBJECT DATABASE FOR BUSINESS MODELING WITH IMPROVED DATA SECURITY
2y 5m to grant Granted Jan 20, 2026
Patent 12524404
METHOD, DEVICE, AND SYSTEM WITH PROCESSING-IN-MEMORY (PIM)-BASED HASH QUERYING
2y 5m to grant Granted Jan 13, 2026
Patent 12493590
METHOD AND SYSTEM FOR DEDUPLICATING POINT OF INTEREST DATABASES
2y 5m to grant Granted Dec 09, 2025
Study what changed to get past this examiner. Based on 5 most recent grants.

AI Strategy Recommendation

Get an AI-powered prosecution strategy using examiner precedents, rejection analysis, and claim mapping.
Powered by AI — typically takes 5-10 seconds

Prosecution Projections

3-4
Expected OA Rounds
72%
Grant Probability
99%
With Interview (+28.1%)
3y 2m
Median Time to Grant
High
PTA Risk
Based on 300 resolved cases by this examiner. Grant probability derived from career allow rate.

Sign in with your work email

Enter your email to receive a magic link. No password needed.

Personal email addresses (Gmail, Yahoo, etc.) are not accepted.

Free tier: 3 strategy analyses per month