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 responsive to the following communication: RCE filed Feb. 18, 2026.
Claims 1, 4-11, 14-24 are pending in the case. Claims 1and 11 are independent claims.
Claim Rejections - 35 USC § 103
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.
Claims 1, 4-7, 10, 11, 4-17 and 20-24 are rejected under 35 U.S.C. 103 as being unpatentable over Rajamoni et al. (hereinafter Rajamoni) U.S. Patent Publication No. 2021/0304062 in view of Vivona et al. (hereinafter Vivona) U.S. Patent Publication No. 2022/0129706 and in further view of Haraldson et al. (hereinafter Haraldson) U.S. Patent Publication No. 2023/0325711.
With respect to independent claim 1, Rajamoni teaches a method comprising:
identifying, by a controller for a federated learning system, a first dataset available to a first node of the federated learning system, wherein the first dataset comprises first features indicating information maintained by the first node and common to a group of nodes of the federated learning system,(see e.g., Para [5][6][18]-[21][62]-[75] –“Each customized learning request issued to a data party triggers the data party to locally train a model based on training data owned by the data party and one or more model parameters stored in the shared namespace in the object store, and upload a local model resulting from the local training to a corresponding namespace in the object store the data party is assigned with.”)
configuring, by the controller, the first node to train a first model using the first dataset (see e.g., Para [5] – “aggregating the at least one local model retrieved from the object store to obtain a shared model. The shared model is uploaded to the shared namespace in the object store. Each data party is notified of the shared model uploaded to the shared namespace in the object store. “);
causing, by the controller, formation of a global model in the federated learning system that aggregates the first model from the first node with models from the group of nodes of the federated learning system (see e.g., Para [5][18]-[21]);
configuring, by the controller, the first node to train a second model using the second dataset (see e.g., Para [5][6][19][20][62]-[75]).
Rajamoni does not expressly show the first dataset being determined based at least in part on an intersection of datasets features between the group of nodes, wherein the intersection of features is associated with an entity type. However, Vivona teaches similar feature (see e.g. Abstract and Para [67]-[82][110]-[120] – “The first and second training datasets have first and second sample sets that share one or more shared sample features. The shared sample features are common between the first and second sample sets.” “The technology disclosed can generate input to the classifier 730 which is a fusion of the embeddings (or spatial representation) produces by the two encoders 713 and 715. These outputs are produced from real data row 711 accessible to the computing device and synthetic data row 723 which is generated by the trained generator. The generator generates the data using shared features. The shared features are common across datasets of multiple computing devices. Therefore, given a real sample, we use the imported generator from the other client to synthesize a row from the other client conditioned on the real row's shared features.”). Both Rajamoni and Vivona are directed to federated learning. Accordingly, it would have been obvious to the skilled artisan before the effective filing date of the claimed invention having Rajamoni and Vivona in front of them to modify the system of Rajamoni to include the above feature. The motivation to combine Rajamoni and Vivona comes from Vivona. Vivona discloses the motivation to improve training performance by utilizing different and heterogenous data sources (see e.g. Vivona para [45][46]).
Rajamoni-Vivona does not expressly show determining, by the controller for the federated learning system, a second dataset available to the first node of the federated learning system, wherein the second dataset comprises second features indicating information maintained by the first node and common to the first node as a subset of the group of nodes, the second features being associated with a geographical proximity.
However, Rajamoni teaches local learning (see e.g., Para [5][6][19][20][62]-[75] –“Each customized learning request issued to a data party triggers the data party to locally train a model based on training data owned by the data party and one or more model parameters stored in the shared namespace in the object store, and upload a local model resulting from the local training to a corresponding namespace in the object store the data party is assigned with.”) Further, Haraldson teaches that each group (subset of nodes) can train its own machine learning model and the grouping can be based on geographical similarity (see e.g., Para [22][23][139][172] - “Due to the heterogeneous nature of the mobile network with various different configurations, geographical locations, various user behavior, however, a single model may not be an appropriate fit for all sites. The standard solution to this problem is to collect more data and introduce more features. Another solution is to group sites having similar patterns and train one ML model per each group.”) Both Rajamoni and Haraldson are directed to federated learning. Accordingly, it would have been obvious to the skilled artisan before the effective filing date of the claimed invention having Rajamoni and Haraldson in front of them to further modify the modified system of Rajamoni to include the above feature. The motivation to combine Rajamoni and Haraldson comes from Haraldson. Haraldson discloses the motivation to improve model fitness by grouping the nodes for federated learning based on nodes similarity such as geographical similarity (see e.g. Haraldson Para [22][23][139][172]).
With respect to dependent claim 4, the modified Rajamoni teaches the subset of nodes of the federated learning system comprises only the first node (see e.g., Para [62]-[75] – there is no limit regarding the number of the subset of nodes).
With respect to dependent claim 5, the modified Rajamoni teaches causing, by the controller, formation of a sub-aggregated model that aggregates the second model from the first node models from the subset of the group of nodes that are trained using the second features (see e.g., Para [62]-[75]).
With respect to dependent claim 6, the modified Rajamoni teaches identifying the first dataset and the second dataset available to the first node of the federated learning system comprises: receiving, at the controller, a manifest of data classes available to the first node (see e.g., Para [5][6][19][20] [62]-[75]).
With respect to dependent claim 7, the modified Rajamoni teaches identifying the first dataset and the second dataset available to the first node of the federated learning system comprises: causing, by the controller, the subset of the group of nodes of the federated learning system to employ a private set intersection protocol, to identify the second features (see e.g., Para [5][6][19][20] [62]-[75]).
With respect to dependent claim 10, the modified Rajamoni teaches causing, by the controller, the global model to be sent to the first node for use (see e.g., Para [18]-[21]).
Claim 11 is rejected for the similar reasons discussed above with respect to claim 1.
Claim 14 is rejected for the similar reasons discussed above with respect to claim 4.
Claim 15 is rejected for the similar reasons discussed above with respect to claim 5.
Claim 16 is rejected for the similar reasons discussed above with respect to claim 6.
Claim 17 is rejected for the similar reasons discussed above with respect to claim 7.
Claim 20 is rejected for the similar reasons discussed above with respect to claim 1.
With respect to dependent claim 21, the modified Rajamoni teaches the intersection of features is a first intersection of features, the subset of the group of nodes further comprises a second node, and wherein determining the second dataset further comprises: determining a second intersection of features between the first node and the second node, wherein the second intersection of features is associated with the geographical proximity (see e.g. Vivona Para [120] and Haraldson Para [22][23][139][172] – the examiner notes that the intersection can be based on any similarity such as geographic proximity).
With respect to dependent claim 22, the modified Rajamoni teaches the geographical proximity is a first geographical proximity, and wherein the intersection of features is associated with a second geographical proximity (see e.g. Vivona Para [120] and Haraldson Para [22][23][139][172] – The examiner notes it is not clear whether the first geographical proximity is same as the second geographical proximity. Furthermore, the intersection of features can be based on similarity of geographical proximity. Depends on how the similarity is defined, the second geographical proximity can be different from the first geographical proximity. For example, vehicles within a zip code area has a first geographical proximity and vehicles with the same zip code and parked in a parking lot have a second geographical proximity).
Claim 23 is rejected for the similar reasons discussed above with respect to claim 21.
Claim 24 is rejected for the similar reasons discussed above with respect to claim 22.
Claims 8-9 and 18-19 are rejected under 35 U.S.C. 103 as being unpatentable over Rajamoni in view of Vivona, Haraldson and further in view of Gopalakrishnan (hereinafter Gopalakrishnan) U.S. Patent Publication No. 2022/0067645.
With respect to dependent claim 8, Rajamoni does not expressly show the features are represented as a hash value in the federated learning system. However, Gopalakrishnan teaches the above feature (see e.g. para [18]). Both Rajamoni and Gopalakrishnan are directed to federated learning. Accordingly, it would have been obvious to the skilled artisan before the effective filing date of the claimed invention having Rajamoni and Gopalakrishnan in front of them to modify the system of Rajamoni to include the above feature. The motivation to combine Rajamoni and Gopalakrishnan comes from Gopalakrishnan. Gopalakrishnan discloses the motivation to use hash value to label data elements (see e.g. Gopalakrishnan para [18]).
With respect to dependent claim 9, the modified Rajamoni teaches the hash value is used as a group label as part of a command to aggregate models among the subset of the group of nodes that are based on the second features (see e.g. Gopalakrishnan para [18]).
Claim 18 is rejected for the similar reasons discussed above with respect to claim 8.
Claim 19 is rejected for the similar reasons discussed above with respect to claim 9.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to PEIYONG WENG whose telephone number is (571)270-1660. The examiner can normally be reached on Mon.-Fri. 8 am to 5 pm.
If attempts to reach the examiner by telephone are unsuccessful, the examiner's supervisor, Matthew Ell, can be reached on (571) 270-3264. 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://portal.uspto.gov/external/portal. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free).
/PEI YONG WENG/Primary Examiner, Art Unit 2141