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 .
Continued Examination Under 37 CFR 1.114
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 9/26/2025 has been entered.
Remarks
Claims 1-18 and 24-27 are pending.
Allowable Subject Matter
Claims 2, 4, 6, 8, 14 and 26 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1, 9 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over John et al. (‘John’ hereinafter) (Publication Number 20190378016) in view of Parker et al. (‘Parker’ hereinafter) (Publication Number 20170300253).
As per claim 1, John teaches
An electronic device, comprising: (see abstract and background)
a plurality of nodes, each node comprising: a processor with circuitry configured to process input data through a model; and a local memory configured to store a portion of model data associated with the model; and a controller comprising circuitry configured to: (paragraphs [0029]-[0038])
copy a first subset of the portion of the model data from the local memory of a first node to the local memory of a second node, subset of portion of model; note that a subset can be the set itself, so the “first subset of the portion of the model data” is interpreted as “the portion of the model data” itself)
and cause a processor of the second node to use the copied first subset of the portion of the model data to perform at least one operation that processes the input data through the model to generate output data,
John does not explicitly indicate “responsive to data that identifies the first subset as being frequently accessed or associated with a latency condition”, “wherein the copying makes the first subset locally available to the second node without remote memory access”.
However, Parker discloses “responsive to data that identifies the first subset as being frequently accessed or associated with a latency condition” (more frequently access portion of the data set is apportioned for replication, paragraph [0017]; identify subsections of the data set that is more frequently accessed than the others and apportions as much of those subsections as it can for replication, paragraph [0022]), “wherein the copying makes the first subset locally available to the second node without remote memory access” (replicate to device’s local memory to minimize the delays caused by the remote access where portions is distributed selectively across the devices, paragraphs [0017],[0022], where one of skill in the art would know that replicating to minimize delays would mean that the replication can make the subset available without remote access as claimed).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine John and Parker because using the steps claimed would have given those skilled in the art the tools to improve the invention by providing a multi-device computer system that utilizes both local and remote accesses to exploit the available memory capacity to the full extent while efficiently running/executing an application (see Parker, background). This gives the user the advantage of more efficient use of expensive resources.
As per claim 9,
This claim is rejected on grounds corresponding to the reasons given above for rejected claim 1 and is similarly rejected.
As per claim 16,
This claim is rejected on grounds corresponding to the reasons given above for rejected claim 1 and is similarly rejected.
Claims 10 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over John et al. (‘John’ hereinafter) (Publication Number 20190378016) in view of Parker et al. (‘Parker’ hereinafter) (Publication Number 20170300253) and further in view of Murthy (Publication Number 20070276835).
As per claim 10,
Neither John nor Parker explicitly indicates “the first subset has a smaller size than the portion of the model data, and the method further comprising copying, by the controller to the local memory of the second node, the first subset of the portion of the model data, while maintaining a remainder of the portion at the local memory of the first node”.
However, Murthy discloses “the first subset has a smaller size than the portion of the model data, and the method further comprising copying, by the controller to the local memory of the second node, the first subset of the portion of the model data, while maintaining a remainder of the portion at the local memory of the first node” (multiple ACL system tables where one ACL table can be replicated to hold a subset of entries, paragraph [0053], where one ACL table is a portion of a database that hold multiple tables; note the John teaches model data and nodes as shown previously).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine John, Parker and Murthy because using the steps claimed would have given those skilled in the art the tools to improve the invention by providing for more efficient queries against smaller tables for more efficient use of resources. This gives the user the advantage of faster access to desired information.
As per claim 17,
This claim is rejected on grounds corresponding to the reasons given above for rejected claim 10 and is similarly rejected.
Claims 11 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over John et al. (‘John’ hereinafter) (Publication Number 20190378016) in view of Parker et al. (‘Parker’ hereinafter) (Publication Number 20170300253) and further in view of Hardin (Publication Number 20160171068).
As per claim 11, John teaches
at one or more times after copying the first subset of the portion of the model data from the local memory of the first node to the local memory of the second node: copying, by the controller to the local memory of the second node, a second subset of the portion of the model data stored at the local memory of the first node that satisfies the one or more conditions, (first node can store a portion of the first layer of the deep learning model and another portion of the first layer can be stored on a different host node, and further separating the input data amongst the host nodes, paragraph [0057], see also paragraphs [0062], [0067]-[0077]; deep learning model distributed based on distribution protocol such as distribution by layers or operations, paragraph [0057], where distribution protocol such as layers read on the one or more conditions for copying subset of portion of model).
Neither John nor Parker explicitly indicates “including overwriting the first subset of the portion of the model data copied at the local memory of the second node with the second subset, prior to the second node using the copied first subset for performing the at least one operation”.
However, Hardin discloses “including overwriting the first subset of the portion of the model data copied at the local memory of the second node with the second subset, prior to the second node using the copied first subset for performing the at least one operation” (overwrite update records in context data, paragraph [0155])
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine John, Parker and Hardin because using the steps claimed would have given those skilled in the art the tools to improve the invention by allowing users the ability to access user group data without being required to transfer or search through large volumes of available information (see Hardin, paragraph [0001]). This gives the user the advantage of more efficient access to distributed information.
As per claim 18,
This claim is rejected on grounds corresponding to the reasons given above for rejected claim 11 and is similarly rejected.
Claim 12 and 24 are rejected under 35 U.S.C. 103 as being unpatentable over John et al. (‘John’ hereinafter) (Publication Number 20190378016) in view of Parker et al. (‘Parker’ hereinafter) (Publication Number 20170300253) and further in view Popov (Publication Number 20210342427).
As per claim 12,
Neither John nor Parker explicitly indicates “local memories in the plurality of nodes have insufficient storage capacity for simultaneously storing all of the model data associated with the model”.
However, Popov discloses “local memories in the plurality of nodes have insufficient storage capacity for simultaneously storing all of the model data associated with the model” (paragraph [0177]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine John, Parker and Popov because using the steps claimed would have given those skilled in the art the tools to improve the invention by ensuring that nodes have suitable resources to performing their function. This gives the user the advantage of avoiding preventable failures of storage systems.
As per claim 24,
This claim is rejected on grounds corresponding to the reasons given above for rejected claim 12 and is similarly rejected.
Claims 3, 5, 13 and 25 are rejected under 35 U.S.C. 103 as being unpatentable over John et al. (‘John’ hereinafter) (Publication Number 20190378016) in view of Parker et al. (‘Parker’ hereinafter) (Publication Number 20170300253) and further in view of Ma et al. (‘Ma’ hereinafter) (Publication Number 20240037378).
As per claim 3,
Neither John nor Parker explicitly indicates “while processing the input data through the model, the processor acquires model data stored in its local memory and acquires additional model data from other nodes when that data is not available locally, and uses the acquired model data to perform the processing”.
However, Ma discloses “while processing the input data through the model, the processor acquires model data stored in its local memory and acquires additional model data from other nodes when that data is not available locally, and uses the acquired model data to perform the processing” (row in table that is needed by model instance is moved to local memory, paragraph [0055]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine John, Parker and Ma because using the steps claimed would have given those skilled in the art the tools to improve the invention by providing for more efficient use of resources by ensuring rows necessary for model processing are available locally. This gives the user the advantage of faster processing of critical information.
As per claim 5,
Neither John nor Parker explicitly indicates “the model data associated with the model comprises embedding table data used for training or inference”.
However, Ma discloses “the model data associated with the model comprises embedding table data used for training or inference” (row in table that is needed by model instance is moved to local memory, paragraph [0055]; embedding table and deep learning machine process, paragraphs [0016],[0039]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine John, Parker and Ma because using the steps claimed would have given those skilled in the art the tools to improve the invention by providing for more efficient use of resources by ensuring rows necessary for model processing are available locally. This gives the user the advantage of faster processing of critical information.
As per claim 13,
Neither John nor Parker explicitly indicates “the model is a computational model that comprises a deep learning recommendation model; the portion of the model data stored in the local memory in each node includes at least one embedding table comprising a plurality of rows of model data; and the first subset of the portion of the model data that satisfies the one or more conditions includes individual rows of model data in the embedding table”.
However, Ma discloses “the model is a computational model that comprises a deep learning recommendation model; the portion of the model data stored in the local memory in each node includes at least one embedding table comprising a plurality of rows of model data; and the first subset of the portion of the model data that satisfies the one or more conditions includes individual rows of model data in the embedding table” (row in table that is needed by model instance is moved to local memory, paragraph [0055]; embedding table and deep learning machine process, paragraphs [0016],[0039]).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine John, Parker and Ma because using the steps claimed would have given those skilled in the art the tools to improve the invention by providing for more efficient use of resources by ensuring rows necessary for model processing are available locally. This gives the user the advantage of faster processing of critical information.
As per claim 25,
This claim is rejected on grounds corresponding to the reasons given above for rejected claim 13 and is similarly rejected.
Claims 7, 15 and 27 are rejected under 35 U.S.C. 103 as being unpatentable over John et al. (‘John’ hereinafter) (Publication Number 20190378016) in view of Parker et al. (‘Parker’ hereinafter) (Publication Number 20170300253) and further in view of Swift et al. (‘Swift’ hereinafter) (Patent Number 9053167).
As per claim 7,
Neither John nor Parker explicitly indicates “an amount of data in the first subset is based at least in part on: an available capacity for storing model data satisfying the one or more conditions in the local memory of the second node; and an amount of communication between the first node and the second node for communicating the model data.”
However, Swift discloses “an amount of data in the first subset is based at least in part on: an available capacity for storing model data satisfying the one or more conditions in the local memory of the second node; and an amount of communication between the first node and the second node for communicating the model data” (parameters for storage, paragraph [0124], where parameters read on claimed conditions; distribute data based on free storage capacity, paragraph [0125]-[0128], where Stacey teaches copy model data and frequently accessed as shown previously).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine John, Parker and Swift because using the steps claimed would have given those skilled in the art the tools to improve the invention by allowing applications to determine the device on which data shall be stored (see Swift, paragraph [0003]-[0004]). This gives the user the advantage of improved data security, availability, or accessibility.
As per claim 15,
This claim is rejected on grounds corresponding to the reasons given above for rejected claim 7 and is similarly rejected.
As per claim 27,
This claim is rejected on grounds corresponding to the reasons given above for rejected claim 7 and is similarly rejected.
Response to Arguments
Applicant’s arguments with respect to the 35 USC 101 rejections have been fully considered and are persuasive. The 35 USC 101 rejections have been withdrawn.
Applicant’s remaining arguments with respect to the 35 USC 103 rejections 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. It is noted that the newly added Parker reference, in combination with previously cited references, teaches the amended claims as shown above.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JAY A MORRISON whose telephone number is (571)272-7112. The examiner can normally be reached on Monday - Friday, 8:00 am - 4:00 pm ET.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Trujillo K James, can be reached at telephone number (571)272-3677. 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 Patent Center and the Private Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from Patent Center or Private PAIR. Status information for unpublished applications is available through Patent Center and Private PAIR for authorized users only. Should you have questions about access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free).
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) Form at https://www.uspto.gov/patents/uspto-automated- interview-request-air-form.
/JAY A MORRISON/Primary Examiner, Art Unit 2151