DETAILED ACTION
This action is in response to the reply received 06/25/2026. After consideration of applicant's amendments and/or remarks:
Claims 1-17 are rejected under 35 USC § 103.
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 § 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, 3, 10, and 12 are rejected under 35 U.S.C. 103 as being unpatentable over Swan et al., U.S. PG-Publication No. 2020/0311617 A1 (hereinafter SWAN), in view of Mehta et al., U.S. PG-Publication No. 2022/0108035 A1 (hereinafter MEHTA), further in view of Li et al., U.S. PG-Publication No. 2018/0300360 A1 (hereinafter LI).
Claim 1
SWAN discloses a machine learning management system. ¶ 0074: The model training system, model hosting system, and associated data stored collectively operate as a “machine learning service.”
SWAN discloses one or more processors configured to execute code to cause the one or more processors to: receive a selected model and associated training data for the selected model through [a] communications interface from the customer. ¶ 0048: A processing unit executes computer program instructions implementing the disclosed functions. ¶ 0026: A user device submits a training request containing a container image and an indication of the associated training data location. ¶ 0027: The container’s executable instructions represent an algorithm that “defines a machine learning model.” ¶ 0029: The container image may have been “previously created/uploaded by the user.” ¶ 0036: The training system retrieves or streams training data from the location identified in the training request. ¶ 0075: The system provides application programming interfaces for submitting training, deployment, and execution requests.
SWAN discloses store the selected model and the associated training. ¶ 0071: Container data store 170 stores images used to form training and scoring containers. ¶ 0069: Training data store 160 stores the data used to train models. ¶ 0074: The illustrated systems and data stores may be combined to perform their disclosed functions.
SWAN does not expressly disclose a repository having one or more partitions, the one or more partitions being separate from others of the one or more partitions, store the selected model and the associated training data in the partition dedicated to the customer; and manage the one or more partitions to ensure that the customer can only access the partition dedicated to the customer.
MEHTA discloses a repository having one or more partitions, the one or more partitions being separate from others of the one or more partitions. ¶ 0024: Entity store 203 includes customer databases and “multiple separate distributed and/or replicated data stores.”
MEHTA discloses store the selected model and the associated training data in the partition dedicated to the customer. ¶ 0021: Database 123 supplies customer training data and can contain trained ML models available to that customer.
MEHTA discloses manage the one or more partitions to ensure that the customer can only access the partition dedicated to the customer. ¶ 0021: Database 123 “is only accessible by a single customer;” each additional customer “accesses its own database.”
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the model and training data storage of SWAN to incorporate the separate customer database arrangement taught by MEHTA. One of ordinary skill in the art would be motivated to integrate that arrangement into SWAN, with a reasonable expectation of success, in order to protect customer data while supporting multiple customers by maintaining data in separate customer stores, as taught by MEHTA at ¶ 0015.
SWAN-MEHTA does not expressly disclose each partition having a data store with a communications interface to allow a customer to access a partition dedicated to the customer.
LI discloses each partition having a data store with a communications interface to allow a customer to access a partition dedicated to the customer. ¶ 0045: Describes “an API corresponding to the storage space (including the exchange partition and the private partition)” and a dedicated service system interface with that API. ¶ 0045: Data uploaded through a dedicated service system “can only first enter the private partition.” ¶ 0055: An upload table is created when an organization uploads data to its private partition, and the data is synchronized to that table.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to modify the customer data transfer arrangement of SWAN-MEHTA to incorporate the storage specific interface arrangement and private partition upload rule taught by LI, implementing the existing model and training data intake through the interface of each customer store. One of ordinary skill in the art would be motivated to integrate that arrangement into SWAN-MEHTA, with a reasonable expectation of success, in order to reduce unintended exposure of incoming customer models and training data by directing them into the customer’s private storage before subsequent use of sharing, as supported by LI at ¶¶ 0027 and 0045.
Claim 3
SWAN discloses wherein the selected model is a trained model. ¶ 0054: A user submits a deployment request identifying “one or more trained machine learning models,” including locations of their model data files. ¶ 0056: The hosting system retrieves the identified trained model’s data files and inserts them into the scoring container as part of its code. ¶ 0051: The scoring container contains model execution code and model data; execution generates predicted outputs.
Claims 10 and 12
Claims 10 and 12 are rejected utilizing the rationale for claims 1 and 3; the claims are directed to a method performed by the system.
Claims 2, 4, 9, 11, and 13 are rejected under 35 U.S.C. 103 as being unpatentable over SWAN, in view of MEHTA, further in view of LI, further in view of Mueller et al., U.S. PG-Publication No. 2021/0326717 A1 (hereinafter MUELLER).
Claim 2
MUELLER discloses train the selected model using the associated training data prior to storing the selected model and the associated training data. ¶ 0105: The virtual machine executes the selected container’s instructions to train the model, “applying the training data retrieved by the model training system” as inputs. ¶ 0107: The system packages model files “once training is complete” and stores them; model files may be stored after completion of training. ¶ 0052: Model artifacts may be written to the same storage location as the user supplied datasets, or to a separate output location.
MUELLER discloses send the selected model and the associated training data to the customer after the training. ¶ 0054: After the training exploration concludes and model artifacts are stored, the interface allows the user to download those artifacts. ¶ 0077: The completed trial interface permits downloading “each set of data resulting from the split of the dataset” and model artifacts including “model code/weights.”
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to modify the training completion and customer storage operations of SWAN-MEHTA-LI to incorporate the model and dataset retention and download capabilities taught by MUELLER, storing the trained model and retained associated training data after training and returning both to the customer. One of ordinary kill in the art would be motivated to integrate those capabilities into SWAN-MEHTA-LI, with a reasonable expectation of success, in order to facilitate inspection and refinement of the completed model by preserving and supplying the training data associated with its results, as supported by MUELLER at ¶¶ 0020 and 0077.
Claim 4
MUELLER discloses receive a request for a trained model stored in the partition dedicated to the customer. ¶ 0054: The user selects an interface icon to download model artifacts previously stored in an object storage location.
MUELLER discloses access the partition dedicated to the customer to retrieve the trained model and the training data associated with the trained model. ¶ 0052: Stored exploration artifact models, and their storage location may be the same location holding the user supplied datasets. ¶ 0077: Selecting a particular trial exposes that trial’s downloadable dataset portions, preprocessing information, and model artifacts.
MUELLER discloses send the trained model and the training data associated with the trained model to the customer. ¶ 0077: The user may download the trial’s dataset portions and its model code and weights.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to modify the customer model retrieval operation of SWAN-MEHTA-LI to incorporate the associated model and training data retrieval and download capabilities taught by MUELLER, returning both in response to a request for the stored trained model. One of ordinary kill in the art would be motivated to integrate those capabilities into SWAN-MEHTA-LI, with a reasonable expectation of success, in order to enable the customer to examine the training basis of the requested model by supplying its associated data alongside the model artifacts, as supported by MUELLER at ¶¶ 0020 and 0077.
Claim 9
MUELLER discloses wherein the partition in the repository dedicated to the customer contains one or more partitions, each of the one or more partitions comprising sub-repositories dedicated to one or more of different models, different versions of a same model, and different components. ¶ 0033: Customer object storage locations may be “folders, buckets, etc.” that store files or objects. ¶ 0052: The system may create multiple object storage locations for datasets and generated artifacts, including models and pipeline code. ¶ 0076: Separate model training trials are individually identified and the user can select a particular trial to inspect its model algorithm, hyperparameters, and results. ¶ 0077: The selected trial has corresponding datasets, preprocessing materials, and model code and weights available for inspection and download.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to modify the internal organization of the customer partition of SWAN-MEHTA-LI to incorporate the separate storage locations and model trial organization taught by MUELLER, arranging subordinate repositories dedicated to different models within that customer partition. One of ordinary kill in the art would be motivated to integrate those capabilities into SWAN-MEHTA-LI, with a reasonable expectation of success, in order facilitate inspection and retrieval of the correct model’s materials by keeping each model’s artifacts and associated data in an identifiable subordinate collection, as supported by MUELLER at ¶¶ 0076 and 0077.
Claims 11 and 13
Claims 11 and 13 are rejected utilizing the rationale for claims 2 and 4; the claims are directed to a method performed by the system.
Claims 5-6, 8, 14-15, and 17 are rejected under 35 U.S.C. 103 as being unpatentable over SWAN, in view of MEHTA, further in view of LI, further in view of Bhattacharjee et al., U.S. PG-Publication No. 2020/0334567 A1 (hereinafter BHATTACHARJEE).
Claim 5
BHARRACHARJEE discloses wherein the one or more processors are further configured to execute code to allow the one or more processors to access one or more computing devices located at a customer location. ¶ 0014: Client devices include laptops, desktops, and workstations running a training application that interacts with model training manager 140. ¶ 0020: The model training manager instructs client devices to execute training processes and return results. ¶ 0036: The system sends training data and model parameters to the selected computing device for execution. ¶¶ 0022: Training jobs may be restricted to devices belonging to specified domains or network, with authentication credentials required.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to modify the model training architecture of SWAM-MEHTA-LI to incorporate the client training application and remote task distribution taught by BHATACHARJEE, enabling the management system to access customer computing devices and direct training of untrained models on those devices while permitting customers to select and perform training locally. One of ordinary skill n the art would be motivated to integrate those capabilities into SWAN-MEHTA-LI, with a reasonable expectation of success, in order to use otherwise idle computing resources and avoid requiring a dedicated training center by distributing training work to participating customer devices, as taught by BHATTACHARJEE at ¶ 0011.
Claim 6
BHATTACHARJEE discloses wherein the code that causes the one or more processors to access the one or more computing devices comprises code to allow the one or more processors to train untrained models on the one or more computing devices remotely. ¶ 0020: The manager remotely assigns training tasks, instructs client devices to perform model training, and combines their returned results to generate a trained model. ¶ 0014: The client training application includes model implementations used with supplied training data and model parameters to train a model. ¶ 0017: The client applies received model parameters, obtains training data, feeds that data into the model, and returns results that may include model weights.
Claim 8
BHARRACHARJEE discloses wherein the one or more processors are further configured to execute code that causes the one or more processors to allow the customer to train models locally at a customer location. ¶ 0012: Users download a training application or library, register their client device, obtain training jobs and data, and perform model training on those devices. ¶ 0016: The user may select a training job through the training application’s user interface. ¶ 0017: The client application feeds training data into the model locally and returns the resulting model information to the manager.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to modify the model training architecture of SWAM-MEHTA-LI to incorporate the client training application and remote task distribution taught by BHATACHARJEE, enabling the management system to access customer computing devices and direct training of untrained models on those devices while permitting customers to select and perform training locally. One of ordinary skill in the art would be motivated to integrate those capabilities into SWAN-MEHTA-LI, with a reasonable expectation of success, in order to use otherwise idle computing resources and avoid requiring a dedicated training center by distributing training work to participating customer devices, as taught by BHATTACHARJEE at ¶ 0011.
Claims 14-15 and 17
Claims 14-15 and 17 are rejected utilizing the rationale for claims 5-6 and 8; the claims are directed to a method performed by the system.
Claims 7 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over SWAN, in view of MEHTA, further in view of LI, further in view of Cai et al., U.S. PG-Publication No. 2014/0366004 A1 (hereinafter LI).
Claim 7
LI discloses wherein the code that causes the one or more processors to access the one or more computing devices causes the one or more processors to debug trained models on the one or more computing devices remotely. ¶ 0017: Describes “remote debugging of an application in a cloud system” while the system is in production mode. ¶ 0043: The server switches from production mode to a debugging mode, with an isolation level adapted to the debugging request, and establishes connections to debugging initiators. ¶ 0045: A connection between the debugging initiator and the server node is established “to perform a remote debugging process.”
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to modify the execution of trained models on customer computing devices of SWAN-MEHTA-LI to incorporate the remote debugging connections and isolated debugging modes taught by CAI. One of ordinary skill in the art would be motivated to integrate those debugging connections and modes into SWAM-MEHTA-LI, with a reasonable expectation of success, in order to minimize disruption to production operations while diagnosing the deployed model by isolating the node undergoing remote debugging from production processing, as taught by CAI at ¶ 0020.
Claim 16
Claim 16 is rejected utilizing the rationale for claim 7; the claim is directed to a method performed by the system.
Response to Arguments
Applicant’s arguments with respect to claim(s) 1-17 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument. See Lee et al., U.S. PG-Publication No. 2018/0300360 A1.
Conclusion
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 FRANK D MILLS whose telephone number is (571)270-3194. The examiner can normally be reached M-F 9-5:30 CT.
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, KEVIN YOUNG can be reached at (571)270-3180. 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.
/FRANK D MILLS/Primary Examiner, Art Unit 2194 September 30, 2026
