DETAILED ACTION
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 § 112
The following is a quotation of 35 U.S.C. 112(b):
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.
Claim 6, 23 – 25, 28, 32, 35, 36, 41, 43 – 46, 49, 53, 56, 57, 62 and 64 is 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 pre-AIA the applicant regards as the invention.
Claim 1 line 7 and claim 6 line 1, claim 23 lines 1 and 4, claim 28 line 1, claim 1 lines 1 and 9 and claim 49 line 1 refer to the limitation “a machine learning model” making it impossible to be certain if the later instances intend to refer to the initial instance or introduce a new instance. For claim interpretation purposes the later instances are read as “the machine … .”
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, 6, 19, 22 – 25, 28, 41, 44 – 46, 49 and 64 is/are rejected under 35 U.S.C. 103 as being unpatentable over Wang et al. (hereinafter Wang, US 2020/0257652) in view of Gupta et al., (hereinafter Gupta, US 2021/0263781).
Regarding claim 1, Wang discloses:
a method for providing computing resources, the method comprising, using a main processor (see at least Fig. 2, Fig. 5 and Fig. 11 that illustrate a computing device with a central process that provides various computing resources to perform the method and all the features and functions disclosed throughout the reference),
a computing device having available computing power and comprising a readymade runtime environment that enables interface with a computer network (see at least Fig. 10 and 11 that discloses such computer systems with central processing unit(s) and network interface connections)
dividing a translated task into one or more portions, wherein the translated task comprises a task gist extracted from a task of training of a machine learning model or inference using the machine learning model (see at least ph. [0064] - [0065] for the compression (changed into a different digital form, i.e. translated) of the instructions sets (providing a sense of what instructions to perform, i.e. a gist) of the task(s) and the task(s) being split as part of the processing and at least ph. [0009] discloses this as part of a deep learning / machine learning process);
transferring one portion of the one or more portions to each of the selected computing devices to be executed by the readymade runtime environment of each of the respective computing devices (see at least ph. [0086] for transferring data as a part of the learning instructions and as such would as appropriate transfer tasks/split tasks of the instructions sets as per at least ph. [0065]); and
obtaining results of the execution of the one or more portions of the translated task from the selected computing devices (see at least Fig. 5 for the actual execution of the task that will then produce the result of said execution, and this is as done from the compressed instruction set(s) as described in at least ph. [0064] – [0065]).
Wang does not expressly disclose, however, Gupta discloses:
a method for providing decentralized computing resources (see at least Fig. 1 and 2 (Fig. 2 showing a storage area network)), the method comprising, using a main processor (see at least Fig. 2 for processor(s) ):
selecting, from a plurality of computing devices, one or more computing devices, each of which having available computing power and comprising a readymade runtime environment that enables interface with a computer network (devices as shown in Fig. 1 and/or Fig. 2 where these systems and their resources perform functions that are available as executing functions/runtime);
It would have been obvious for a person of ordinary skill in the art before the effective filing date of the invention to modify the teachings of Wang by the teachings of Gupta in order to allow for additional computing devices to assist in the completion of a task, thereby improving the total time required to perform a task.
Claim 23 is an alternate method version of claim 1 and is rejected in like manner, where Wang discloses the training of machine learning model or inference in at least ph. [0009] and [0045] – [0046] and Gupta discloses the use of decentralized computing resources that would allow for the training to be performed over multiple other computing devices.
Claim 44 is system version of claim 1 and is rejected in like manner, where Wang discloses such a system with memory and a processor in at least Fig. 11.
Regarding claims 2, 24 and 45, the rejections of claims 1, 23 and 44 are incorporated and Wang discloses:
the task is training the machine learning model, and wherein the task gist comprises a model structure, model weights, hyperparameters, and datasets (see at least ph. [0059] that discloses that the model weight is loaded as part of the instruction set (needing additional internal memory as opposed to other data for the model, ph. [0061] discloses that a split model structure is disclosed as part of the deep learning instruction set, based on parameter information of the deep learning model associated with the instruction set (note, these are set as part of the deep learning model architecture then, as opposed to parameter values that change as the model runs and *learns*, making these parameters hyperparameters) and see at least ph. [0046] for complex data structures (data sets) transmitted into an artificial intelligence neural network for the deep learning model of the instruction set described above).
Regarding claims 3, 25 and 46 the rejections of claims 1, 23 and 44 are incorporated and Wang discloses:
the task is inference using the machine learning model, and wherein the task gist comprises model structure, model weights, and input data (see at least ph. [0059] that discloses that the model weight is loaded as part of the instruction set (needing additional internal memory as opposed to other data for the model, ph. [0061] discloses that a split model structure is disclosed as part of the deep learning instruction set, based on parameter information of the deep learning model associated with the instruction set and at least ph. [0055] discloses inputted data used by the deep learning model).
Regarding claims 6, 28 and 49 the rejections of claims 1, 23 and 44 are incorporated and Wang discloses:
the task of training of a machine learning model or inference using the machine learning model comprises the model structure, model weights, hyperparameters and datasets (see at least ph. [0059] that discloses that the model weight is loaded as part of the instruction set (needing additional internal memory as opposed to other data for the model, ph. [0061] discloses that a split model structure is disclosed as part of the deep learning instruction set, based on parameter information of the deep learning model associated with the instruction set (note, these are set as part of the deep learning model architecture then, as opposed to parameter values that change as the model runs and *learns*, making these parameters hyperparameters) and see at least ph. [0046] for complex data structures (data sets) transmitted into an artificial intelligence neural network for the deep learning model of the instruction set described above).
Regarding claims 19 and 41 the rejections of claims 1 and 23 are incorporated and Wang discloses:
the portion of the translated task comprises a portion of the training dataset (see at least ph. [0045] – [0046] for the training utilizing the data structures in the artificial intelligence as part of the deep learning model, where these are part of the compressed and later decompressed/translated instruction sets disclosed in at least ph. [0064] – [0065]), the method comprising:
integrating the results of execution of the portions of the translated task to obtain a trained model (see at least ph. [0045] – [0046] for the trained model that was derived from tasks that were split as per at least ph. [0009] and compressed and decompressed as per at least ph. [0064] – [0065] to get the final model disclosed).
Regarding claim 22 and 64 the rejections of claims 1 and 44 are incorporated and Wang discloses:
extracting the task gist from the task of training of the machine learning model or inference using the machine learning model, to generate the translated task (see at least Fig. 5 for the decoding of the instruction set which is different than the task that includes the training disclosed in at least ph. [0045] – [0046] and these instruction sets are decoded and compressed/decompressed as per at least Fig. 5 and ph. [0009]).
Allowable Subject Matter
Claim(s) 10, 13, 14, 21, 32, 35, 36 43, 53, 56, 57 and 62 would be allowable if rewritten to overcome the rejection(s) under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), 2nd paragraph, set forth in this Office action and to include all of the limitations of the base claim and any intervening claims.
Other References Cited Not Relied Upon
Chen (US 2020/0117505) discloses computing tasks to processing units according to a job queue with individual task queues.
Wang et al. (US 2021/0073169) discloses a deep neural network learning while utilizing tasks for task queues.
Lu et al. (US 2026/0169791) discloses task processing for multi-chip and subtasks with task queues.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to CRAIG C DORAIS whose telephone number is (571)270-3371. The examiner can normally be reached M-F 9:00 am - 6:00pm.
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, Pierre Vital can be reached at 5712724215. 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.
/CRAIG C DORAIS/Primary Examiner, Art Unit 2198