DETAILED ACTION
This action is in response to the claims filed 12/12/2023 for Application number 18/536,977. Claims 1-20 are currently pending.
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 .
Information Disclosure Statement
The information disclosure statements (IDS) submitted on 12/12/2023, 03/26/2025, and 05/14/2026 are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statements are being considered by the examiner.
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 7 and 15 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.
The term “top matching” copies of the independently trained neural networks in claim 7 and 15 is a relative term which renders the claim indefinite. The term “top” is not defined by the claim, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention. The specification only discloses selecting top matching copies in para 0020, while not defining what is considered to be “top matching”. Since top matching is a subjective definition which differs from person to person, the metes and bounds of the claim is not made clear and one of ordinary skill in the art would not be able to properly avoid infringing upon a claim when no definition of “top matching” has been made.
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 1-5, 9-13, and 17-19 are rejected under 35 U.S.C. 103 as being unpatentable over Martins ("WO 2021144037 A1", cited by Applicant in the IDS filed 03/26/2025, hereinafter "Martins") in view of Chhibber et al. ("US 20230040721 A1", hereinafter "Chhibber").
Regarding claim 1, Martins teaches A method for deploying a neural network to a new edge server in an edge computing environment (“FIG. 1 illustrates a system 100 of machine learning according to an embodiment. As shown, a central server node or computing device 102 is in communication with one or more local client nodes or computing devices 104.” [¶0040]), comprising:
deploying a plurality of copies of a centralized neural network respectively to a corresponding plurality of edge servers (“As shown, each local client node 104 may communicate model updates to central server node 102, and central server node 102 may send the updated central model to the local client nodes 104” [¶0041]), wherein each of the copies of the centralized neural network is independently operated and trained at a respective one of the edge servers based on inputs received at that edge server to create independently trained neural networks (“For example, the central server node 102 may transmit to the local client nodes 104 a central model (e.g., newly initialized or partially trained through previous rounds of federated learning). The local client nodes 104 may train their individual models locally with their own data. (corresponds to “independently operated”)” [¶0042]);
and
generating a new edge neural network for deployment to the new edge server, including performing neural network breeding based on the independently trained neural networks and the stored edge devices information (“…It is relevant to notice that, given locality-preserving hashing functions, the similarity measure can be simple and still useful to identify the proper central models in the model pool.” [¶0075; identifying and selecting proper models would correspond to “performing neural network breeding”]), and based on anticipated edge devices information for edge devices expected to access the new edge server (“The deployment of a model to a new edge node whose characteristics and settings (scenario) is known (“expected edge device information”). Since it is a new deployment, local data is not available. Embodiments allow the node to read a model from the model pool to match the target scenario (partial key).” [¶0070]).
However fails to explicitly teach storing, by each of the edge servers, edge devices information including a physical location of each edge device that accessed the independently trained neural network operating at the edge server
Chhibber teaches storing, by each of the edge servers, edge devices information including a physical location of each edge device that accessed the independently trained neural network operating at the edge server (“In some embodiments, user device 120 is operable to collect (and to analyze using transaction request analysis module 122) user behavior information. As used herein “user behavior information” refers to any information about how one or more users 104 has used a particular user device 120… user behavior information includes but is not limited to: location information (e.g., a geolocation of user device 120)” [¶0041])
It would have been obvious to one of ordinary skill in the art before the effective filing date to modify Martins’ teachings by storing edge device information including a physical location as taught by Chhibber. One would have been motivated to make this modification as collecting sensitive device data, user behavior data and user information with this kind of hybrid machine learning implementation not only helps increase the security of the computer system using machine learning models but also reduces risk around privacy and sensitive information. [¶0029, Chhibber]
Regarding claim 2, Martins/Chhibber teaches The method of claim 1, and further comprising:
Martins teaches sending, from each of the edge servers to the cloud-based data center (¶0040), a copy of the independently trained neural network operating at the edge server and the edge devices information stored by the edge server, wherein the cloud-based data center generates the new edge neural network in response to the request based on the copies of the independently trained neural networks sent from the edge servers (“The reading operator acts on the similarity vector W and the models Q in the model pool to produce an aggregated model Θ.sub.READ . That action is called a reading operation. The reading operation is performed in response to a model request from a local client node 104.” [¶0052-¶0053]); and
deploying, from the cloud-based data center to the new edge server, the new edge neural network (See ¶0071, new edge node, new deployment).
Chhibber teaches receiving, at a cloud-based data center (¶0031, cloud), a request for the new edge server including a physical location of the new edge server (“Similarly, headstart models may be distributed to newly installed (or reset) edge servers 130 and 140.” [¶0092; physical location would be included as Chhibber’s FL system collect user device information]);
Same motivation to combine the teachings of Martins/Chhibber as claim 1.
Regarding claim 3, Martins/Chhibber teaches The method of claim 1, Martins teaches wherein, for each of the edge servers, the stored edge devices information further includes inputs provided by edge devices to the independently trained neural network operating at the edge server. (“For example, such significant differences in distribution of local training data could arise from intrinsic local characteristics of the different local client nodes (e.g., hardware, software, geographical location, system type). Regardless of the reason for differing distributions, if such differences are significant it could affect the ability to maintain a well-performing centralized model that represents each of the local client nodes, e.g. because updates from local client nodes having a different distribution may result in decreased performance or even prevent the global model from converging.” [¶0008])
Regarding claim 4, Martins/Chhibber teaches The method of claim 1, Martins teaches wherein the anticipated edge devices information includes inputs expected to be provided by edge devices to the new edge server. (“Target data distribution refers to the distribution of the training data that a model is targeted for, and target scenario refers to non-training data related characteristics of the deployment environment. For example, a target scenario for a mobile phone could include manufacturer information, network standard (e.g., 4G, 5G), device resource information (e.g., memory, processing capabilities, and so on). The target scenario information may, in some embodiments, depend on the problem domain.” [¶0065])
Regarding claim 5, Martins/Chhibber teaches The method of claim 1, and further comprising:
Martins teaches comparing, for each of the edge servers, the edge devices information from the edge server to the anticipated edge devices information for the new edge server to determine a comparison percentage corresponding to the independently trained neural network operating on the edge server. (“In some embodiments, updating the model pool further comprises computing a similarity score comparing the key corresponding to the local client node to the keys collectively corresponding to each of the plurality of central models,” [¶00104; similarity score corresponds to “comparison percentage”)
Regarding claims 9-13, they are substantially similar to claim 1-5 respectively, and are rejected in the same manner, the same art, and reasoning applying.
Regarding claims 17-19, they are substantially similar to claims 1, 2, and 5 respectively, and are rejected in the same manner, the same art, and reasoning applying.
Claims 6, 7, 14, and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Martins in view of Chhibber and further in view of Andoni et al. ("WO 2021021546 A1", hereinafter "Andoni").
Regarding claim 6, Martins/Chhibber teaches The method of claim 5, however fails to explicitly teach wherein the neural network breeding includes discarding any copies of the independently trained neural networks that have a corresponding comparison percentage below a threshold comparison percentage.
Andoni teaches wherein the neural network breeding includes discarding any copies of the independently trained neural networks that have a corresponding comparison percentage below a threshold comparison percentage. (“Alternatively, or in addition, a candidate neural network with lower estimated relative fitness can become extinct and discarded from consideration such that neural networks in subsequent epochs do not inherit traits of the extinct neural network. Discarding neural networks that have lower estimated relative fitness can be used to prune an evolutionary possibility space to remove evolutionary branches that are unlikely to lead to a reliable and high-performing neural network.” [¶0022])
It would have been obvious to one of ordinary skill in the art before the effective filing date to modify Martins’/Chhibber’s teachings in order to implement the automated neural network generation method of Andoni. One would have been motivated to make this modification in order to determine which neural networks of a population would be used or selected based on a fitness evaluation. [¶0013, Andoni]
Regarding claim 7, Martins/Chhibber teaches The method of claim 5, however fails to explicitly teach wherein the neural network breeding includes selecting top matching copies of the independently trained neural networks to be used for the neural network breeding based on the comparison percentage corresponding to each of the copies of the independently trained neural networks.
Andoni teaches wherein the neural network breeding includes selecting top matching copies of the independently trained neural networks to be used for the neural network breeding based on the comparison percentage corresponding to each of the copies of the independently trained neural networks. (“In this example, the ranking values 314 can indicate the relative ranking (in terms of expected fitness) of each of the 1000 neural networks, or the ranking values 314 can identify or indicate a subset of the population 302 that is expected to be fittest (e.g., the top 50% in terms of fitness).” [¶0074])
Same motivation to combine the teachings of Martins/Chhibber/Andoni as claim 6.
Regarding claims 14-15, they are substantially similar to claims 6 and 7 respectively, and are rejected in the same manner, the same art, and reasoning applying.
Claims 8, 16, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Martins in view of Chhibber and further in view of Gallego-Duran et al. ("Experiments on Neuroevolution and Online Weight Adaptation in Complex Environments", hereinafter "Gallego-Duran").
Regarding claim 8, Martins/Chhibber teaches The method of claim 5, however fails to explicitly teach wherein the neural network breeding includes performing a hyperNEAT calculation that includes determining a weighted average, based on the comparison percentages, of parameters extracted from neurons in one or more of the copies of the independently trained neural networks.
Gallego-Duran teaches wherein the neural network breeding includes performing a hyperNEAT calculation that includes determining a weighted average, based on the comparison percentages, of parameters extracted from neurons in one or more of the copies of the independently trained neural networks. (“Therefore, produced ANNs have a sort of fixed, hardcoded behaviour that will not change even if it is required. In order to address this issue, weight adaptation has been added to HyperNEAT, encoded as a pattern of local rules that modulate each weight.” [pg. 134, 2.4, ¶1; See further §3 discloses “All this functions will be created by the CPPN as a weighted average of a predefined subset of the functions that the CPPN uses as internal nodes”])
It would have been obvious to one of ordinary skill in the art before the effective filing date to modify Martins’/Chhibber’s teachings in order to perform a hyperNEAT calculation as taught by Gallego-Duran. One would have been motivated ot make this modification as HyperNEAT has shown a great potential for evolving large scale neural networks, by discovering geometric regularities, thus being suitable for evolving complex controllers. [Abstract, Gallego-Duran]
Regarding claims 16 and 20, they are substantially similar to claim 8 respectively, and are rejected in the same manner, the same art, and reasoning applying.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MICHAEL H HOANG whose telephone number is (571)272-8491. The examiner can normally be reached Mon-Fri 8:30AM-4:30PM.
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, Kakali Chaki can be reached at (571) 272-3719. 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.
/MICHAEL H HOANG/PRIMARY EXAMINER, Art Unit 2122
Prosecution Insights