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 .
Priority
Acknowledgment is made of applicant's claim for foreign priority based on an application filed in Korea on January 3, 2023. It is noted, however, that applicant has not filed a certified copy of the KR10-2023-0000664 application as required by 37 CFR 1.55. Under the priority document exchange, USPTO attempted to electronically receive the foreign application to which priority is claimed and failed on June 3, 2024.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on April 24, 2024, March 19, 2025, and February 13, 2026 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-15 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an
abstract idea (mental process) without significantly more.
Claim 1:
Regarding claim 1, in step 1 of the 101-analysis set forth in MPEP 2106, the claim recites “An electronic apparatus comprising: a communication interface; a memory configured to store at least one instruction; and at least one processor”, and an apparatus or machine is one of the four statutory categories of invention.
In step 2A prong 1 of the 101-analysis set forth in the MPEP 2106, the examiner has determined that the following limitations recite a process that, under the broadest reasonable interpretation, covers a mental process but for recitation of generic computer components:
“determine whether to transmit information regarding the trained global neural network model to the server based on a result of the evaluating” (this is a mental process, a person could mentally determine a whether to transmit information or no in regards to a result of evaluating, see MPEP § 2106.04(a)(2)(III)),
“evaluate the trained global neural network model” (this is a mental process, a person could mentally evaluate a neural network model, see MPEP § 2106.04(a)(2)(III)),
If claim limitations, under the broadest reasonable interpretation, covers performance of the limitations as a mental process but for the recitation of generic computer components, then it falls within the mental process grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea.
In step 2A prong 2 of the 101-analysis set forth in MPEP 2106, the examiner has determined that the following additional elements do not integrate this judicial exception into a practical application:
“An electronic apparatus comprising:” (Using an electronic apparatus is considered generic computer component being used as tool to perform functions of the judicial exception – see MPEP § 2106.05(f)),
“a communication interface;” (a communication interface is considered mere instructions to apply an exception using generic computer – see MPEP § 2106.05(f)),
“a memory configured to store at least one instruction;” (Using a memory to store instructions is considered generic computer component being used as tool to perform functions of the judicial exception – see MPEP § 2106.05(f)),
“at least one processor;” (Using a processor is considered generic computer component being used as tool to perform functions of the judicial exception – see MPEP § 2106.05(f)),
“receive information regarding a global neural network model and information regarding evaluation data from a server using the communication interface;” (receiving information from a server using a communication interface is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)),
“obtain a data set for training the global neural network model;” (obtaining a data set is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)),.
“train the global neural network model based on the data set;” (training a neural network model based on a data set is considered mere instructions to apply an exception using generic computer – see MPEP § 2106.05(f)),
“by inputting the evaluation data to the trained global neural network model;” (inputting data to a global neural network model is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)),
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is “directed” to an abstract idea.
In step 2B of the 101-analysis set forth in the 2019 PEG, the examiner has determined that the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
As discussed above, additional elements iii, v, and vi recites generic computer component being used as tool to perform functions of the judicial exception, additional elements vii, viii, and x recites insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362, and additional element iv and ix recites mere instructions to apply an exception using generic computer, which is not indicative of significantly more.
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claim 2:
Regarding claim 2, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis to claim 1. Further, claim 2 recites the following additional elements:
“evaluate the trained global neural network model by comparing the first accuracy level with the second accuracy level.” (this is a mental process, a person could mentally evaluate a neural network model by comparing two accuracy levels, see MPEP § 2106.04(a)(2)(III)),
If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process but for the recitation of generic computer components, then it falls within the mental process grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea.
“The electronic apparatus as claimed in claim 1, wherein the at least one processor is further configured to: obtain a first accuracy level regarding a result value output by inputting the evaluation data to the global neural network model;” (In step 2A, prong 2, this is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)). (In step 2B, this is also considered insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362).
“obtain a second accuracy level regarding a result value output by inputting the evaluation data to the trained global neural network model;” (In step 2A, prong 2, this is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)). (In step 2B, this is also considered insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 3:
Regarding claim 3, it is dependent upon claim 2, and thereby incorporates the limitations of, and corresponding analysis to claim 2. Further, claim 3 recites the following additional elements:
“based on determining that the second accuracy level is higher than the first accuracy level, determine whether to transmit the information regarding the trained global neural network model to the server.” (this is a mental process, a person could mentally evaluate determining if a second accuracy level is higher than the first and determining whether to transmit data based on the determination, see MPEP § 2106.04(a)(2)(III)),
If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process but for the recitation of generic compute components, then it falls within the mental process grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea.
“The electronic apparatus as claimed in claim 2, wherein the at least one processor is further configured to” (In step 2A, prong 2, this is considered generic computer component being used as tool to perform functions of the judicial exception, see MPEP § 2106.05(f)). (In step 2B, this is considered generic computer component being used as tool to perform functions of the judicial exception - see MPEP § 2106.05(f)).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 4:
Regarding claim 4, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis to claim 1. Further, claim 4 recites the following additional elements:
“compare version information corresponding to a local neural network model stored in the electronic apparatus with the version information corresponding to the global neural network model;” (this is a mental process, a person could mentally evaluate version and address information regarding a global neural network and comparing version information of a global network model with version information of a local model , see MPEP § 2106.04(a)(2)(III)),
“based on determining that a version of the global neural network model is higher than a version of the local neural network model,” (this is a mental process, a person could mentally evaluate determining if a version of a global neural network model is higher than a local neural network model, see MPEP § 2106.04(a)(2)(III)),
If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process but for the recitation of generic compute components, then it falls within the mental process grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea.
“The electronic apparatus as claimed in claim 1, wherein the information regarding the global neural network model comprises version information corresponding to the global neural network model and address information indicating an address from which the global neural network model is downloadable;” (In step 2A, prong 2, this is considered as mere field of use or technological environment in which to apply a judicial exception, see MPEP § 2106.05(h)). (In step 2B, this is also considered mere field of use or technological environment in which to apply a judicial exception - see MPEP § 2106.05(f)).
“and wherein the at least one processor is further configured to:” (In step 2A, prong 2, this is considered generic computer component being used as tool to perform functions of the judicial exception, see MPEP § 2106.05(f)). (In step 2B, this is considered generic computer component being used as tool to perform functions of the judicial exception - see MPEP § 2106.05(f)).
“download the global neural network model using the communication interface based on the address information.” (In step 2A, prong 2, this is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)). (In step 2B, this is also considered insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 5:
Regarding claim 5, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis to claim 1. Further, claim 5 recites the following additional elements:
“The electronic apparatus as claimed in claim 1, wherein the at least one processor is further configured to: receive a version file comprising the information regarding the global neural network model and the information regarding the evaluation data from the server using the communication interface;” (In step 2A, prong 2, this is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)). (In step 2B, this is also considered insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362).
“obtain address information regarding a data set pre-stored in the electronic apparatus;” (In step 2A, prong 2, this is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)). (In step 2B, this is also considered insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362).
“add the obtained address information regarding the data set to the version file.” (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer, see MPEP § 2106.05(f)). (In step 2B, this is also considered mere instructions to apply an exception using generic computer - see MPEP § 2106.05(f)).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 6:
Regarding claim 6, it is dependent upon claim 5, and thereby incorporates the limitations of, and corresponding analysis to claim 5. Further, claim 6 recites the following additional elements:
“The electronic apparatus as claimed in claim 5, wherein the at least one processor is further configured to: update the version file to include parameter information regarding the trained global network model based on the result of the evaluating;” (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer, see MPEP § 2106.05(f)). (In step 2B, this is also considered mere instructions to apply an exception using generic computer - see MPEP § 2106.05(f)).
“control the communication interface to transmit the updated version file to the server.” (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer, see MPEP § 2106.05(f)). (In step 2B, this is also considered mere instructions to apply an exception using generic computer - see MPEP § 2106.05(f)).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 7:
Regarding claim 7, it is dependent upon claim 6, and thereby incorporates the limitations of, and corresponding analysis to claim 6. Further, claim 7 recites the following additional elements:
“The electronic apparatus as claimed in claim 6, wherein the at least one processor is further configured to control the communication interface to delete the address information regarding the data set from the updated version file before the updated version file is transmitted to the server.” (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer, see MPEP § 2106.05(f)). (In step 2B, this is also considered mere instructions to apply an exception using generic computer - see MPEP § 2106.05(f)).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 8:
Regarding claim 8, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis to claim 1. Further, claim 8 recites the following additional elements:
“The electronic apparatus as claimed in claim 1, wherein a new version of the global neural network model is generated by the server based on the information regarding the trained global neural network model received from the electronic apparatus.” (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer, see MPEP § 2106.05(f)). (In step 2B, this is also considered mere instructions to apply an exception using generic computer - see MPEP § 2106.05(f)).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 9:
Regarding claim 9, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis to claim 1. Further, claim 9 recites the following additional elements:
“The electronic apparatus as claimed in claim 1, wherein the at least one processor is further configured to: perform Secured Sockets Layer/Transport Layer Security (SSL/TLS) encoding on the information regarding the trained global neural network model;” (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer, see MPEP § 2106.05(f)). (In step 2B, this is also considered mere instructions to apply an exception using generic computer - see MPEP § 2106.05(f)).
“control the communication interface to transmit the encoded trained global neural network model to the server.” (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer, see MPEP § 2106.05(f)). (In step 2B, this is also considered mere instructions to apply an exception using generic computer - see MPEP § 2106.05(f)).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 10:
Regarding claim 10, it is dependent upon claim 1, and thereby incorporates the limitations of, and corresponding analysis to claim 1. Further, claim 10 recites the following additional elements:
“The electronic apparatus as claimed in claim 1, wherein the at least one processor is further configured to store the trained global neural network model in the memory as a local neural network model.” (In step 2A, prong 2, this is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)). (In step 2B, this is also considered insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see Storing and retrieving information in memory, Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015)).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 11:
Regarding claim 11, in step 1 of the 101-analysis set forth in MPEP 2106, the claim recites “A controlling method of an electronic apparatus”, and a process or method is one of the four statutory categories of invention.
In step 2A prong 1 of the 101-analysis set forth in the MPEP 2106, the examiner has determined that the following limitations recite a process that, under the broadest reasonable interpretation, covers a mental process but for recitation of generic computer components:
“determining whether to transmit information regarding the trained global neural network model to the server based on a result of the evaluating” (this is a mental process, a person could mentally determine a whether to transmit information or no in regards to a result of evaluating, see MPEP § 2106.04(a)(2)(III)),
“evaluating the trained global neural network model” (this is a mental process, a person could mentally evaluate a neural network model, see MPEP § 2106.04(a)(2)(III)),
If claim limitations, under the broadest reasonable interpretation, covers performance of the limitations as a mental process but for the recitation of generic computer components, then it falls within the mental process grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea.
In step 2A prong 2 of the 101-analysis set forth in MPEP 2106, the examiner has determined that the following additional elements do not integrate this judicial exception into a practical application:
“A controlling method of an electronic apparatus, the method comprising:” (Using an electronic apparatus is considered generic computer component being used as tool to perform functions of the judicial exception – see MPEP § 2106.05(f)),
“receiving information regarding a global neural network model and information regarding evaluation data from a server;” (receiving information from a server using a communication interface is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)),
“obtaining a data set for training the global neural network model;” (obtaining a data set is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)),.
“training the global neural network model based on the data set;” (training a neural network model based on a data set is considered mere instructions to apply an exception using generic computer – see MPEP § 2106.05(f)),
“by inputting the evaluation data to the trained global neural network model;” (inputting data to a global neural network model is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)),
Since the claim as a whole, looking at the additional elements individually and in combination, does not contain any other additional elements that are indicative of integration into a practical application, the claim is “directed” to an abstract idea.
In step 2B of the 101-analysis set forth in the 2019 PEG, the examiner has determined that the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
As discussed above, additional element iii recites generic computer component being used as tool to perform functions of the judicial exception, additional elements iv, v, and vii recites insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362, and additional element vi recites mere instructions to apply an exception using generic computer, which is not indicative of significantly more.
Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Claim 12:
Regarding claim 12, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis to claim 11. Further, claim 12 recites the following additional elements:
“evaluating the trained global neural network model by comparing the first accuracy level with the second accuracy level.” (this is a mental process, a person could mentally evaluate a neural network model by comparing two accuracy levels, see MPEP § 2106.04(a)(2)(III)),
If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process but for the recitation of generic computer components, then it falls within the mental process grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea.
“The method as claimed in claim 11, wherein the evaluating comprises: obtaining a first accuracy level regarding a result value output by inputting the evaluation data to the global neural network model;” (In step 2A, prong 2, this is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)). (In step 2B, this is also considered insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362).
“obtain a second accuracy level regarding a result value output by inputting the evaluation data to the trained global neural network model;” (In step 2A, prong 2, this is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)). (In step 2B, this is also considered insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 13:
Regarding claim 13, it is dependent upon claim 12, and thereby incorporates the limitations of, and corresponding analysis to claim 12. Further, claim 13 recites the following additional elements:
“The method as claimed in claim 12, wherein the determining comprises: based on determining that the second accuracy level is higher than the first accuracy level, determining whether to transmit the information regarding the trained global neural network model to the server.” (this is a mental process, a person could mentally evaluate determining if a second accuracy level is higher than the first and determining whether to transmit data based on the determination, see MPEP § 2106.04(a)(2)(III)),
If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process but for the recitation of generic compute components, then it falls within the mental process grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea.
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 14:
Regarding claim 14, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis to claim 11. Further, claim 14 recites the following additional elements:
“comparing version information corresponding to a local neural network model stored in the electronic apparatus with the version information corresponding to the global neural network model;” (this is a mental process, a person could mentally evaluate version and address information regarding a global neural network and comparing version information of a global network model with version information of a local model , see MPEP § 2106.04(a)(2)(III)),
“based on determining that a version of the global neural network model is higher than a version of the local neural network model,” (this is a mental process, a person could mentally evaluate determining if a version of a global neural network model is higher than a local neural network model, see MPEP § 2106.04(a)(2)(III)),
If claim limitations, under their broadest reasonable interpretation, covers performance of the limitations as a mental process but for the recitation of generic compute components, then it falls within the mental process grouping of abstract ideas. Accordingly, the claim “recites” an abstract idea.
“The method as claimed in claim 11, wherein the information regarding the global neural network model comprises version information corresponding to the global neural network model and address information indicating an address from which the global neural network model is downloadable;” (In step 2A, prong 2, this is considered as mere field of use or technological environment in which to apply a judicial exception, see MPEP § 2106.05(h)). (In step 2B, this is also considered mere field of use or technological environment in which to apply a judicial exception - see MPEP § 2106.05(f)).
“downloading the global neural network model based on the address information.” (In step 2A, prong 2, this is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)). (In step 2B, this is also considered insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim 15:
Regarding claim 15, it is dependent upon claim 11, and thereby incorporates the limitations of, and corresponding analysis to claim 11. Further, claim 15 recites the following additional elements:
“The method as claimed in claim 11, wherein the receiving comprises: receiving a version file comprising the information regarding the global neural network model and the information regarding the evaluation data from the server;” (In step 2A, prong 2, this is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)). (In step 2B, this is also considered insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362).
“wherein the obtaining comprises: obtaining address information regarding a data set pre-stored in the electronic apparatus;” (In step 2A, prong 2, this is considered insignificant extra-solution activity of mere data gathering – see MPEP § 2106.05(g)). (In step 2B, this is also considered insignificant extra-solution activity of mere data gathering, which is a well understood routine and conventional activity, see receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362).
“adding the obtained address information regarding the data set to the version file.” (In step 2A, prong 2, this is considered mere instructions to apply an exception using generic computer, see MPEP § 2106.05(f)). (In step 2B, this is also considered mere instructions to apply an exception using generic computer - see MPEP § 2106.05(f)).
Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Claim Rejections - 35 USC § 102
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 the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claim(s) 1-3, 8, and 10-13 are rejected under 35 U.S.C. 102(a)(1) and 102(a)(2) as being anticipated by Singh et al, (European Patent Application EP 4075348 A1) effectively filed on April 14, 2021, (hereafter Singh).
Claim 1:
Regarding claim 1, Singh teaches “An electronic apparatus comprising: a communication interface; a memory configured to store at least one instruction; and at least one processor configured to: receive information regarding a global neural network model and information regarding evaluation data from a server using the communication interface;”
See Singh in paragraph [0017] describing, “Within the context of embodiments of the invention, a "module" can be understood to mean for example a processor and/or a memory unit for storing program instructions. By way of example, the processor is configured specifically to execute the program instructions such that the processor performs functions to implement or perform the method according to embodiments of the invention or a step of the method according to embodiments of the invention. The applicable modules can comprise further elements. These elements are for example one or more interfaces (e.g. database interfaces, communication interfaces - e.g. network interface, WLAN interface) and/or a memory unit.” Here, Singh establishes a processor, a memory configured to store instructions and communication interface as part of the embodiment described. Further, see Singh in paragraph [0086] describing, “The interface 106 is configured to communicate with the other nodes to perform a consensus method for jointly evaluating the check results. Therefore, the interface can be configured to receive check results CR' provided by other nodes, which are assigned to the second modified machine learning model mML2. The evaluation result EVAL of the joint evaluation of the check results CR', CR is transmitted to the replacement module 107. For example, the evaluation result EVAL indicates an improvement of the modified machine learning model if a majority of check results CR' indicate an improvement.” Here, Singh establishes a receiving of nodes assigned to a model done by an interface, and establishes the information regarding evaluation data being transmitted or received using the nodes. As known in decentralized federated learning any node can act as a server. Further, see Singh in paragraph [0015] describing, “Unless indicated otherwise in the description below, the terms "perform", "calculate", "computer-aided", "compute", "replace", "generate", "configure", "reconstruct" and the like preferably relate to actions and/or processes and/or processing steps that alter and/or produce data and/or that convert data into other data, the data being able to be presented or available as physical variables, in particular, for example as electrical impulses. In particular, the expression "computer" should be interpreted as broadly as possible in order to cover in particular all electronic devices having data processing properties. Computers can therefore be for example personal computers, servers, programmable logic controllers (PLCs), handheld computer systems, Pocket PC devices, mobile radios and other communication devices that can process data in computer-aided fashion, processors and other electronic devices for data processing.” Here, Singh establishes a server that can be used to process data in computer-aided fashion for processes described. Further, see Singh in paragraph [0062] describing, “In the first step S1 the global machine learning model is received or read in by a first node, e.g., it is received from a first node's storage module.” Here, Singh establishes a receiving of a global neural network using a node assigned to it which implies the global neural network received being from an interface established to be considered a communication interface.
Further, Singh teaches “obtain a data set for training the global neural network model;”
See Singh in paragraph [0063] describing, “In the next step S2 the first node modifies the global machine learning model, i.e., the global machine learning model is trained by the first node using a locally stored dataset of said first node generating a modified machine learning model.”
Further, Singh teaches “train the global neural network model based on the data set;”
See Singh in paragraph [0064] describing, “The global machine learning model can for example be an artificial neural network (NN). Training the NN using a local dataset results in a modified machine learning model/NN having adjusted weights.”
Further, Singh teaches “evaluate the trained global neural network model by inputting the evaluation data to the trained global neural network model;”
See Singh in paragraph [0067] describing, “The modified/local machine learning model is preferably only sent to these other nodes if it meets a predefined quality criterion which is set by the first node. The quality criterion can for example define a level of improvement of the modified machine learning model compared to the global machine learning model. The first node can therefore preferably perform first (i.e., before sending) a verification if this quality criterion is met. This verification can for example comprise checking the modified machine learning model against the local dataset of the first node. The output of the modified machine learning model can then be compared to the output of the global machine learning model for determining an improvement.” Here, Singh establishes an evaluating of a global machine learning model and local model is input if it meets a quality criterion. Further, see Singh in paragraph [0031] describing, “A quality criterion can for example be determined by evaluating the machine learning model using a given dataset.” Here, Singh establishes the quality criterion is evaluated using an evaluation of a given dataset for the machine learning model.
Further, Singh teaches “and determine whether to transmit information regarding the trained global neural network model to the server based on a result of the evaluating”
See Singh in paragraph [0067] describing, “The output of the modified machine learning model can then be compared to the output of the global machine learning model for determining an improvement. If the model modification meets the given quality criterion, the modified machine learning model can be broadcasted to the other nodes. If the quality criterion is not met, the modified machine learning model is preferably not shared with the other nodes. This enables a first quality control of the modification of the machine learning model locally performed on the first node.” Here Singh establishes a determination of transmitting information regarding a modified or trained global machine learning model to nodes based on evaluating, the nodes in this instance can be seen as the server.
Claim 2:
Regarding claim 2, Singh teaches the limitations of claim 1.
Further, Singh teaches “The electronic apparatus as claimed in claim 1, wherein the at least one processor is further configured to: obtain a first accuracy level regarding a result value output by inputting the evaluation data to the global neural network model;”
See Singh in paragraph [0067] describing, “The modified/local machine learning model is preferably only sent to these other nodes if it meets a predefined quality criterion which is set by the first node. The quality criterion can for example define a level of improvement of the modified machine learning model compared to the global machine learning model. The first node can therefore preferably perform first (i.e., before sending) a verification if this quality criterion is met. This verification can for example comprise checking the modified machine learning model against the local dataset of the first node.” Here, Singh establishes obtaining a first accuracy level with the quality criterion of the first node using a comparison of a global neural network and a modified machine learning model, a dataset is also used to check for this model ultimately describing evaluation of data of the model.
Further, Singh teaches “obtain a second accuracy level regarding a result value output by inputting the evaluation data to the trained global neural network model;”
See Singh in paragraph [0067] describing, “The first node can therefore preferably perform first (i.e., before sending) a verification if this quality criterion is met. This verification can for example comprise checking the modified machine learning model against the local dataset of the first node. The output of the modified machine learning model can then be compared to the output of the global machine learning model for determining an improvement.”
Further, Singh teaches “evaluate the trained global neural network model by comparing the first accuracy level with the second accuracy level.”
See Singh in paragraph [0067] describing, “The output of the modified machine learning model can then be compared to the output of the global machine learning model for determining an improvement. If the model modification meets the given quality criterion, the modified machine learning model can be broadcasted to the other nodes. If the quality criterion is not met, the modified machine learning model is preferably not shared with the other nodes. This enables a first quality control of the modification of the machine learning model locally performed on the first node.” Further, see Singh in paragraph [0031] describing, “Within the context of embodiments of the invention, a "quality criterion" of a modified machine learning model can be understood to mean for example a predefined performance value, e.g., with respect to computing speed, and/or model accuracy. A quality criterion can for example be determined by evaluating the machine learning model using a given dataset. For example, an improvement of a modified machine learning model over the original machine learning model can be determined by comparing the respective model outputs when a given dataset is applied to the models (i.e., when the models are run on a given dataset).”
Claim 3:
Regarding claim 3, Singh teaches the limitations of claim 2.
Further Singh teaches, “The electronic apparatus as claimed in claim 2, wherein the at least one processor is further configured to, based on determining that the second accuracy level is higher than the first accuracy level, determine whether to transmit the information regarding the trained global neural network model to the server”
See Singh in paragraph [0031] describing, “Within the context of embodiments of the invention, a "quality criterion" of a modified machine learning model can be understood to mean for example a predefined performance value, e.g., with respect to computing speed, and/or model accuracy. A quality criterion can for example be determined by evaluating the machine learning model using a given dataset. For example, an improvement of a modified machine learning model over the original machine learning model can be determined by comparing the respective model outputs when a given dataset is applied to the models (i.e., when the models are run on a given dataset).” Here Singh establishes a determination of a second accuracy level being higher than the first with the check of improvement of an output of a first model compared to another model. Further, see Singh in paragraph [0049] describing, “According to a further embodiment of the method a model change value can be determined by comparing the modified machine learning model with the machine learning model, and only the model change value is sent to the other nodes for checking the modified machine learning model against the respective local datasets.” Here Singh establishes determining transmitting information regarding the modified or trained model with the model change value to other nodes, which as known in decentralized federating learning any node can act as a server at any given time, this is also based on the comparison of a first model and a modified one.
Claim 8:
Regarding claim 8, Singh teaches the limitations of claim 5.
Further, Singh teaches “The electronic apparatus as claimed in claim 1, wherein a new version of the global neural network model is generated by the server based on the information regarding the trained global neural network model received from the electronic apparatus.”
See Singh in paragraph [0084] describing, “The receiver 104 is configured to receive a second modified machine learning model mML2, which has been generated (based on the global machine learning model ML) by another node of the decentralized distributed database.” Here, as known in decentralized federated learning a node can act as a server, Singh establishes a node generating a new version of the global model that is modified or trained and is received from a receiver. Further, see Singh in paragraph [0079] describing, “The quality control system 100 comprises a storage module 101, a model generator 102, a sender 103, a receiver 104, a checking module 105, an interface 106, and a replacement module 107. The quality control system 100 is preferably configured to perform the method steps as exemplary described in Figure 1.” Here, Singh establishes the quality control system which can be the electronic apparatus comprising of the receiver mentioned.
Claim 10:
Regarding claim 10, Singh teaches the limitations of claim 1.
Further, Singh teaches “The electronic apparatus as claimed in claim 1, wherein the at least one processor is further configured to store the trained global neural network model in the memory as a local neural network model.”
See Singh in paragraph [0061] describing, “The machine learning model, also referred to as global machine learning model or global model, is stored on all nodes which participate in the federated learning process. The machine learning model is collectively trained on all participating nodes. Hence, such a participating node comprises a machine learning module/model generator configured to train a machine learning model using a locally stored dataset. The nodes therefore can apply their local learnings to the global model to run their specific machine learning models applications. Datasets of the execution of the global model or a local machine learning model can be continuously stored in the local storage of each node and further be used for training the machine learning model.”
Claim 11:
Regarding claim 11, Singh teaches “A controlling method of an electronic apparatus, the method comprising: receiving information regarding a global neural network model and information regarding evaluation data from a server;”
See Singh in paragraph [0017] describing, “Within the context of embodiments of the invention, a "module" can be understood to mean for example a processor and/or a memory unit for storing program instructions. By way of example, the processor is configured specifically to execute the program instructions such that the processor performs functions to implement or perform the method according to embodiments of the invention or a step of the method according to embodiments of the invention. The applicable modules can comprise further elements. These elements are for example one or more interfaces (e.g. database interfaces, communication interfaces - e.g. network interface, WLAN interface) and/or a memory unit.” Here, Singh establishes a processor, a memory configured to store instructions and communication interface as part of the embodiment described. Further, see Singh in paragraph [0086] describing, “The interface 106 is configured to communicate with the other nodes to perform a consensus method for jointly evaluating the check results. Therefore, the interface can be configured to receive check results CR' provided by other nodes, which are assigned to the second modified machine learning model mML2. The evaluation result EVAL of the joint evaluation of the check results CR', CR is transmitted to the replacement module 107. For example, the evaluation result EVAL indicates an improvement of the modified machine learning model if a majority of check results CR' indicate an improvement.” Here, Singh establishes a receiving of nodes assigned to a model done by an interface, and establishes the information regarding evaluation data being transmitted or received using the nodes. As known in decentralized federated learning any node can act as a server. Further, see Singh in paragraph [0015] describing, “Unless indicated otherwise in the description below, the terms "perform", "calculate", "computer-aided", "compute", "replace", "generate", "configure", "reconstruct" and the like preferably relate to actions and/or processes and/or processing steps that alter and/or produce data and/or that convert data into other data, the data being able to be presented or available as physical variables, in particular, for example as electrical impulses. In particular, the expression "computer" should be interpreted as broadly as possible in order to cover in particular all electronic devices having data processing properties. Computers can therefore be for example personal computers, servers, programmable logic controllers (PLCs), handheld computer systems, Pocket PC devices, mobile radios and other communication devices that can process data in computer-aided fashion, processors and other electronic devices for data processing.” Here, Singh establishes a server that can be used to process data in computer-aided fashion for processes described. Further, see Singh in paragraph [0062] describing, “In the first step S1 the global machine learning model is received or read in by a first node, e.g., it is received from a first node's storage module.” Here, Singh establishes a receiving of a global neural network using a node assigned to it which implies the global neural network received being from an interface established to be considered a communication interface.
Further, Singh teaches “obtaining a data set for training the global neural network model;”
See Singh in paragraph [0063] describing, “In the next step S2 the first node modifies the global machine learning model, i.e., the global machine learning model is trained by the first node using a locally stored dataset of said first node generating a modified machine learning model.”
Further, Singh teaches “training the global neural network model based on the data set;”
See Singh in paragraph [0064] describing, “The global machine learning model can for example be an artificial neural network (NN). Training the NN using a local dataset results in a modified machine learning model/NN having adjusted weights.”
Further, Singh teaches “evaluating the trained global neural network model by inputting the evaluation data to the trained global neural network model;”
See Singh in paragraph [0067] describing, “The modified/local machine learning model is preferably only sent to these other nodes if it meets a predefined quality criterion which is set by the first node. The quality criterion can for example define a level of improvement of the modified machine learning model compared to the global machine learning model. The first node can therefore preferably perform first (i.e., before sending) a verification if this quality criterion is met. This verification can for example comprise checking the modified machine learning model against the local dataset of the first node. The output of the modified machine learning model can then be compared to the output of the global machine learning model for determining an improvement.” Here, Singh establishes an evaluating of a global machine learning model and local model is input if it meets a quality criterion. Further, see Singh in paragraph [0031] describing, “A quality criterion can for example be determined by evaluating the machine learning model using a given dataset.” Here, Singh establishes the quality criterion is evaluated using an evaluation of a given dataset for the machine learning model.
Further, Singh teaches “and determining whether to transmit information regarding the trained global neural network model to the server based on a result of the evaluating;”
See Singh in paragraph [0067] describing, “The output of the modified machine learning model can then be compared to the output of the global machine learning model for determining an improvement. If the model modification meets the given quality criterion, the modified machine learning model can be broadcasted to the other nodes. If the quality criterion is not met, the modified machine learning model is preferably not shared with the other nodes. This enables a first quality control of the modification of the machine learning model locally performed on the first node.” Here Singh establishes a determination of transmitting information regarding a modified or trained global machine learning model to nodes based on evaluating, the nodes in this instance can be seen as the server.
Claim 12:
Regarding claim 12, Singh teaches the limitations of claim 11.
Further, Singh teaches “The method as claimed in claim 11, wherein the evaluating comprises: obtaining a first accuracy level regarding a result value output by inputting the evaluation data to the global neural network model;”
See Singh in paragraph [0067] describing, “The modified/local machine learning model is preferably only sent to these other nodes if it meets a predefined quality criterion which is set by the first node. The quality criterion can for example define a level of improvement of the modified machine learning model compared to the global machine learning model. The first node can therefore preferably perform first (i.e., before sending) a verification if this quality criterion is met. This verification can for example comprise checking the modified machine learning model against the local dataset of the first node.” Here, Singh establishes obtaining a first accuracy level with the quality criterion of the first node using a comparison of a global neural network and a modified machine learning model, a dataset is also used to check for this model ultimately describing evaluation of data of the model.
Further, Singh teaches “obtaining a second accuracy level regarding a result value output by inputting the evaluation data to the trained global neural network model;”
See Singh in paragraph [0067] describing, “The first node can therefore preferably perform first (i.e., before sending) a verification if this quality criterion is met. This verification can for example comprise checking the modified machine learning model against the local dataset of the first node. The output of the modified machine learning model can then be compared to the output of the global machine learning model for determining an improvement.”
Further, Singh teaches “evaluating the trained global neural network model by comparing the first accuracy level with the second accuracy level.”
See Singh in paragraph [0067] describing, “The output of the modified machine learning model can then be compared to the output of the global machine learning model for determining an improvement. If the model modification meets the given quality criterion, the modified machine learning model can be broadcasted to the other nodes. If the quality criterion is not met, the modified machine learning model is preferably not shared with the other nodes. This enables a first quality control of the modification of the machine learning model locally performed on the first node.” Further, see Singh in paragraph [0031] describing, “Within the context of embodiments of the invention, a "quality criterion" of a modified machine learning model can be understood to mean for example a predefined performance value, e.g., with respect to computing speed, and/or model accuracy. A quality criterion can for example be determined by evaluating the machine learning model using a given dataset. For example, an improvement of a modified machine learning model over the original machine learning model can be determined by comparing the respective model outputs when a given dataset is applied to the models (i.e., when the models are run on a given dataset).” Here, Singh shows an evaluation of an output of a global machine learning model being compared to an output of another model which is being interpreted as the first and second accuracy level. An evaluation is then explicitly made from the comparisons.
Claim 13:
Regarding claim 13, Singh teaches the limitations of claim 12.
Further Singh teaches, “The method as claimed in claim 12, wherein the determining comprises: based on determining that the second accuracy level is higher than the first accuracy level, determining whether to transmit the information regarding the trained global neural network model to the server.”
See Singh in paragraph [0031] describing, “Within the context of embodiments of the invention, a "quality criterion" of a modified machine learning model can be understood to mean for example a predefined performance value, e.g., with respect to computing speed, and/or model accuracy. A quality criterion can for example be determined by evaluating the machine learning model using a given dataset. For example, an improvement of a modified machine learning model over the original machine learning model can be determined by comparing the respective model outputs when a given dataset is applied to the models (i.e., when the models are run on a given dataset).” Here Singh establishes a determination of a second accuracy level being higher than the first with the check of improvement of an output of a first model compared to another model. Further, see Singh in paragraph [0049] describing, “According to a further embodiment of the method a model change value can be determined by comparing the modified machine learning model with the machine learning model, and only the model change value is sent to the other nodes for checking the modified machine learning model against the respective local datasets.” Here Singh establishes determining transmitting information regarding the modified or trained model with the model change value to other nodes, which as known in decentralized federating learning any node can act as a server at any given time, this is also based on the comparison of a first model and a modified one.
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.
Claim(s) 4 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Singh et al., (European Patent Application EP 4075348 A1) effectively filed on April 14, 2021, (hereafter Singh), in view of Kim J. et al, (US. Patent Application Publication 20230308903 A1) effectively filed on November 8, 2022, (hereafter Kim).
Claim 4:
Regarding claim 4, Singh teaches the limitations of claim 1.
Further, Singh teaches “The electronic apparatus as claimed in claim 1, wherein the information regarding the global neural network model comprises version information corresponding to the global neural network model”
See Singh in paragraph [0080] describing, “The storage module 101 is configured to store a global machine learning model ML which can be understood as the current/present version of the global model.” Here, Singh establishes information regarding version information of a global machine learning model being stored.
Further, Singh teaches “wherein the at least one processor is further configured to: compare version information corresponding to a local neural network model stored in the electronic apparatus with the version information corresponding to the global neural network model;”
See Singh in paragraph [0081] describing, “The model generator 102 is configured to generate a first modified machine learning model mML1 based on the global machine learning model ML and using the first local dataset LD. The global machine learning model ML is trained using the local dataset LD resulting in the first modified machine learning model mML1.” Here, Singh establishes two different versions of a model with the model using the local dataset or the local neural network model and the global model, the modified version based on both implies a comparison of the two being done.
Singh does not appear to explicitly teach “and address information indicating an address from which the global neural network model is downloadable, based on determining that a version of the global neural network model is higher than a version of the local neural network model, download the global neural network model using the communication interface based on the address information”,
However in the same field of art, Kim teaches, “and address information indicating an address from which the global neural network model is downloadable,”
See Kim in paragraph [0034] describing, “In the attached drawings, user equipments (UEs) are shown for example.” Here Kim establishes that UE means user equipments. Further, see Kim in paragraph [0185] describing, “Online model deployment (i.e., downloading new models) may be required, through which AI/ML models can be deployed from NW endpoints to devices as needed to adapt to changed AI/ML tasks and environments. For this, it may be necessary to continuously monitor the model performance at the UE.” Here, Kim establishes NW, which is network in this embodiment, endpoints which can be seen as an address and the address information that indicates an address from where a model can be downloaded, and mentions model performance being monitored by the UE. Further, see Kim in paragraph [0190] describing, “The cloud server can train the global model by aggregating the partially trained local model on each end device. Within each training iteration, the UE can perform training based on the model downloaded from the AI server, using local training data.” Here Kim establishes a global model being downloadable using a UE. Further, see paragraph [0227] describing, “UE List (including GPSIs, External Group ID, or IP addresses) ”. This further establishes the UE comprising different address information.
Further, Kim teaches, “based on determining that a version of the global neural network model is higher than a version of the local neural network model, download the global neural network model using the communication interface based on the address information.”
See Kim in paragraph [0190] describing, “The cloud server can train the global model by aggregating the partially trained local model on each end device. Within each training iteration, the UE can perform training based on the model downloaded from the AI server, using local training data. Then, the UE may report the intermediate training result to the cloud server through the 5G UL channel. The server may aggregate the intermediate training results of the UE and update the global model. Then the updated global model is deployed back to the UE and the UE can perform training for the next iteration.” Here Kim establishes performing training of a version of a global model with a partially trained local model, meaning a determination was made to then perform training based on a model downloaded which is the downloading of a global model from an AI server. Further, see Kim in paragraph [0053] describing, “The wireless devices 100a to 100f may be connected to the network 300 via the BSs 200. An AI technology may be applied to the wireless devices 100a to 100f and the wireless devices 100a to 100f may be connected to the AI server 400 via the network 300.” Here, establishes wireless devices being connected to an AI server. Further, see Kim in paragraph [0086] describing “wireless devices 100 and 200 may correspond to the wireless devices 100 and 200 of FIG. 2 and may be configured by various elements, components, units/portions, and/or modules. For example, each of the wireless devices 100 and 200 may include a communication unit 110, a control unit 120, a memory unit 130, and additional components 140.” Here, Kim establishes these wireless devices including a communication, control, and memory unit. Further, see Kim in paragraph [0086] describing “The control unit 120 may transmit the information stored in the memory unit 130 to the exterior (e.g., other communication devices) via the communication unit 110 through a wireless/wired interface or store, in the memory unit 130, information received through the wireless/wired interface from the exterior (e.g., other communication devices) via the communication unit 110.” Here, Kim establishes the communication unit consisting of an interface which since linked to the AI server can have the downloaded global model be done using a communication interface. In the previous limitations, it was established that models were downloaded based on address information.
It would have been obvious to one of ordinary skill in the art before the effective filing date of
the claimed invention to combine the base reference of Singh with the teachings of Kim by using Singh’s teachings of receiving information of a global neural network model, and training, evaluating, and determining to transmit information regarding the global neural network model, and incorporate with Kim’s teachings of downloadable global models based on address information.
One of ordinary skill in the art would be motivated to do so because by integrating Kim’s
frameworks into the methods of Singh, which are all in relation to federated learning, one of ordinary skill in the art would bring “a method of providing 5GS assistance information to AF to support AI and ML services” (Kim, paragraph [0202]), and “5GS can effectively support the FL operation of AF (or AS)” (Kim, paragraph [0011]).
Claim 14:
Regarding claim 14, Singh teaches the limitations of claim 11.
Further, Singh teaches “The method as claimed in claim 11, wherein the information regarding the global neural network model comprises version information corresponding to the global neural network model”
See Singh in paragraph [0080] describing, “The storage module 101 is configured to store a global machine learning model ML which can be understood as the current/present version of the global model.” Here, Singh establishes information regarding version information of a global machine learning model being stored.
Further, Singh teaches “wherein the controlling method comprises: comparing version information corresponding to a local neural network model stored in the electronic apparatus with the version information corresponding to the global neural network model;”
See Singh in paragraph [0081] describing, “The model generator 102 is configured to generate a first modified machine learning model mML1 based on the global machine learning model ML and using the first local dataset LD. The global machine learning model ML is trained using the local dataset LD resulting in the first modified machine learning model mML1.” Here, Singh establishes two different versions of a model with the model using the local dataset or the local neural network model and the global model, the modified version based on both implies a comparison of the two being done.
Singh does not appear to explicitly teach “and address information indicating an address from which the global neural network model is downloadable, based on determining that a version of the global neural network model is higher than a version of the local neural network model, download the global neural network model using the communication interface based on the address information”,
However in the same field of art, Kim teaches, “and address information indicating an address from which the global neural network model is downloadable,”
See Kim in paragraph [0034] describing, “In the attached drawings, user equipments (UEs) are shown for example.” Here Kim establishes that UE means user equipments. Further, see Kim in paragraph [0185] describing, “Online model deployment (i.e., downloading new models) may be required, through which AI/ML models can be deployed from NW endpoints to devices as needed to adapt to changed AI/ML tasks and environments. For this, it may be necessary to continuously monitor the model performance at the UE.” Here, Kim establishes NW, which is network in this embodiment, endpoints which can be seen as an address and the address information that indicates an address from where a model can be downloaded, and mentions model performance being monitored by the UE. Further, see Kim in paragraph [0190] describing, “The cloud server can train the global model by aggregating the partially trained local model on each end device. Within each training iteration, the UE can perform training based on the model downloaded from the AI server, using local training data.” Here Kim establishes a global model being downloadable using a UE. Further, see paragraph [0227] describing, “UE List (including GPSIs, External Group ID, or IP addresses) ”. This further establishes the UE comprising different address information.
Further, Kim teaches, “based on determining that a version of the global neural network model is higher than a version of the local neural network model, downloading the global neural network model based on the address information.”
See Kim in paragraph [0190] describing, “The cloud server can train the global model by aggregating the partially trained local model on each end device. Within each training iteration, the UE can perform training based on the model downloaded from the AI server, using local training data. Then, the UE may report the intermediate training result to the cloud server through the 5G UL channel. The server may aggregate the intermediate training results of the UE and update the global model. Then the updated global model is deployed back to the UE and the UE can perform training for the next iteration.” Here Kim establishes performing training of a version of a global model with a partially trained local model, meaning a determination was made to then perform training based on a model downloaded which is the downloading of a global model from an AI server. Further, see Kim in paragraph [0053] describing, “The wireless devices 100a to 100f may be connected to the network 300 via the BSs 200. An AI technology may be applied to the wireless devices 100a to 100f and the wireless devices 100a to 100f may be connected to the AI server 400 via the network 300.” Here, establishes wireless devices being connected to an AI server. Further, see Kim in paragraph [0086] describing “wireless devices 100 and 200 may correspond to the wireless devices 100 and 200 of FIG. 2 and may be configured by various elements, components, units/portions, and/or modules. For example, each of the wireless devices 100 and 200 may include a communication unit 110, a control unit 120, a memory unit 130, and additional components 140.” Here, Kim establishes these wireless devices including a communication, control, and memory unit. Further, see Kim in paragraph [0086] describing “The control unit 120 may transmit the information stored in the memory unit 130 to the exterior (e.g., other communication devices) via the communication unit 110 through a wireless/wired interface or store, in the memory unit 130, information received through the wireless/wired interface from the exterior (e.g., other communication devices) via the communication unit 110.” Here, Kim establishes the communication unit consisting of an interface which since linked to the AI server can have the downloaded global model be done using a communication interface. In the previous limitations, it was established that models were downloaded based on address information.
It would have been obvious to one of ordinary skill in the art before the effective filing date of
the claimed invention to combine the base reference of Singh with the teachings of Kim by using Singh’s teachings of receiving information of a global neural network model, and training, evaluating, and determining to transmit information regarding the global neural network model, and incorporate with Kim’s teachings of downloadable global models based on address information.
One of ordinary skill in the art would be motivated to do so because by integrating Kim’s
frameworks into the methods of Singh, which are all in relation to federated learning, one of ordinary skill in the art would bring “a method of providing 5GS assistance information to AF to support AI and ML services” (Kim, paragraph [0202]), and “5GS can effectively support the FL operation of AF (or AS)” (Kim, paragraph [0011]).
Claim(s) 5-6, and 15 is rejected under 35 U.S.C. 103 as being unpatentable over Singh et al., in view of Schiatti L. et al, (US. Patent Application Publication 20210067339 A1) effectively filed on August 26, 2019, (hereafter Schiatti).
Claim 5:
Regarding claim 5, Singh teaches the limitations of claim 1.
Further, Singh teaches “The electronic apparatus as claimed in claim 1, wherein the at least one processor is further configured to: receive a version file comprising the information regarding the global neural network model and the information regarding the evaluation data from the server using the communication interface;”
See Singh in paragraph [0080] describing, “The storage module 101 is configured to store a global machine learning model ML which can be understood as the current/present version of the global model. The global machine learning model ML is collectively trained using the respective model generators 102 on the nodes of a decentralized distributed database. The storage module 101 is further configured to store a first local dataset LD. Furthermore, the storage module 101 can store previous versions of the global machine learning model ML and/or further local machine learning models of the node 200.” Here, Singh establishes receiving version information of a global model. Further, see Singh in paragraph [0086] describing, “The interface 106 is configured to communicate with the other nodes to perform a consensus method for jointly evaluating the check results. Therefore, the interface can be configured to receive check results CR' provided by other nodes, which are assigned to the second modified machine learning model mML2. The evaluation result EVAL of the joint evaluation of the check results CR', CR is transmitted to the replacement module 107.” This establishes the use of a communication interface with evaluation data. The communication with other nodes in this instance can act in place for the server or be a server as known in the art for decentralized federated learning.
Further, Singh teaches “obtain address information regarding a data set pre-stored in the electronic apparatus;”
See Singh in paragraph [0019] describing, “Within the context of embodiments of the invention, "assign", in particular in regard to data and/or information, can be understood to mean for example computer-aided assignment of data and/or information. By way of example, a second datum is assigned to a first datum in this regard by means of a memory address or a unique identifier (UID), e.g. by storing the first datum together with the memory address or the unique identifier of the second datum together in a data record.”
Singh does not appear to explicitly teach “add the obtained address information regarding the data set to the version file”,
However in the same field of art, Schiatti teaches “add the obtained address information regarding the data set to the version file.”
See Schiatti in paragraph [0030] describing, “For example, the machine learning model may be stored as a file with one or more data structures, such as a matrix, to arrange the information.” Here, Schiatti establishes a model being stored as file that contains data. Further, see Schiatti in paragraph [0033] describing, “The participant node 102 may further include a file system 114. The file system may include a storage for information. For example, the file system may include a repository, a database, a set of memory locations, and/or any other type of storage. The file system may index information, such as files.” Here, Schiatti establishes a file system for storing of a file that may include memory locations which can be seen as an address and adding obtained address information to the file. Further, see Schiatti in paragraph [0038] describing, “The DFL smart contract 118 may include a model link 202. The model link 202 may include a link to access a global model for a current round. The global model may include a model designated for further training. For example, the global model may be stored on the file system 114 (FIG. 1).” Here, Schiatti further establishes adding an obtained address with the inclusion of a link to access a global model which can be seen as the address information, and the global model being able to include a model designated for further training implies different versions of the model with the inclusion of a link to this model being seen as adding an obtained address to a version file, as a model can be stored as a file as mentioned.
It would have been obvious to one of ordinary skill in the art before the effective filing date of
the claimed invention to combine the base reference of Singh with the teachings of Schiatti by using Singh’s teachings of receiving information of a global neural network model, and training, evaluating, and determining to transmit information regarding the global neural network model, and incorporate with Schiatti’s teachings of adding address and parameter information regarding data to a version file of a model, and transmitting the file to a server.
One of ordinary skill in the art would be motivated to do so because by integrating Schiatti’s
frameworks into the methods of Singh, which are both in relation to decentralized federated learning, one of ordinary skill in the art would bring “distributed ledger technology [that] may decentralize the coordination and performance of federated learning. In addition, peer-to-peer file sharing and/or network file storage may decentralize model sharing” (Schiatti, paragraph [0019]), “learning process [that] is democratized by way of smart contract” (Schiatti, paragraph [0020]), “blockchain [that] provides a traceability and immutability for a federated learning” (Schiatti, paragraph [0021]), and “participant nodes participating in a federated learning process [that] may maintain privacy of private training data and models” (Schiatti, paragraph [0022]).
Claim 6:
Regarding claim 6, Singh in view of Schiatti teaches the limitations of claim 5.
Singh does not appear to explicitly teach “The electronic apparatus as claimed in claim 5, wherein the at least one processor is further configured to: update the version file to include parameter information regarding the trained global network model based on the result of the evaluating; control the communication interface to transmit the updated version file to the server.”,
However in the same field of art, Schiatti teaches “The electronic apparatus as claimed in claim 5, wherein the at least one processor is further configured to: update the version file to include parameter information regarding the trained global network model based on the result of the evaluating;”
See Schiatti in paragraph [0039] describing, “The DFL smart contract 118 may include training instructions 204. The training instructions 204 may include logic, such as executable logic and/or machine-executable instructions. Alternatively or in addition, the training instructions 204 may include enumerations and/or parameters for machine learning. For example, the training instructions 204 may include parameters to perform machine learning (i.e. number of training interactions, machine learning type, training data schema, etc). A particular participant node may access the training instructions 204 from the DFL smart contract 118. The machine learning framework 110 (FIG. 1) may train or further train a global model based on the training instructions 204.” Here, Schiatti establishes training instructions to include parameters that can train or further train a global model which implies updating this version of the model to include parameter information, as established in the previous limitations a model can be stored as file. Further, see Schiatti in paragraph [0040] describing, “The DFL smart contract 118 may include performance criteria 206. The performance criteria 206 may include acceptance criteria that establishes tolerated performance of a model.” Here, Schiatti establishes the DFL smart contract which includes the training instructions also includes a performance criterion which can be seen as evaluation. Further, see Schiatti in paragraph [0065] describing “The participant node 102 may compare the performance metric with the performance criteria 206 included in the DFL smart contract 118. In response to satisfaction of the performance criteria 206, the participant node 102 may generate a vote token indicative of approval of the aggregated model. In response to the performance criteria 206 not being satisfied, the participant node 102 may generate a vote token indicative of disapproval of the aggregated model.” Here, Schiatti establishes evaluating a model using the criteria to approve or disapprove of the updated model.
Further, Schiatti teaches “control the communication interface to transmit the updated version file to the server.”
See Schiatti in paragraph [0053] describing, “The participant node 102 may receive federated learning parameters (402). The federated learning parameters may include parameters that specify how to perform federated learning. For example, the federated learning parameters may include a global model, a link to a global model, one or more participant node identifiers, training instructions, and/or any other information stored in DFL smart contract 118. In some examples, the DFL controller 112, or some server, may generate a graphical user interface (not shown) to receive the federated learning parameters. The graphical user interface may include graphical controllers, fields, and/or other graphical components that receive various parameters.” Here, Schiatti establishes a graphical user interface (GUI) which can be seen as the communication interface in an analogous system, which is generated to receive federated learning parameters which are linked to a global model. The server generates the GUI to receive these parameters which can be seen as the interface being controlled to transmit these parameters to the server. As mention in previous limitations a model can be stored as a file, and the parameters of a global model are in relation to an updated version of a global model, implying a transmission of an updated version file to the server using an interface.
It would have been obvious to one of ordinary skill in the art before the effective filing date of
the claimed invention to combine the base reference of Singh with the teachings of Schiatti by using Singh’s teachings of receiving information of a global neural network model, and training, evaluating, and determining to transmit information regarding the global neural network model, and incorporate with Schiatti’s teachings of adding address and parameter information regarding data to a version file of a model, and transmitting the file to a server.
One of ordinary skill in the art would be motivated to do so because by integrating Schiatti’s
frameworks into the methods of Singh, which are both in relation to decentralized federated learning, one of ordinary skill in the art would bring “distributed ledger technology [that] may decentralize the coordination and performance of federated learning. In addition, peer-to-peer file sharing and/or network file storage may decentralize model sharing” (Schiatti, paragraph [0019]), “learning process [that] is democratized by way of smart contract” (Schiatti, paragraph [0020]), “blockchain [that] provides a traceability and immutability for a federated learning” (Schiatti, paragraph [0021]), and “participant nodes participating in a federated learning process [that] may maintain privacy of private training data and models” (Schiatti, paragraph [0022]).
Claim 15:
Regarding claim 15, Singh teaches the limitations of claim 11.
Further, Singh teaches “The method as claimed in claim 11, wherein the receiving comprises: receiving a version file comprising the information regarding the global neural network model and the information regarding the evaluation data from the server;”
See Singh in paragraph [0080] describing, “The storage module 101 is configured to store a global machine learning model ML which can be understood as the current/present version of the global model. The global machine learning model ML is collectively trained using the respective model generators 102 on the nodes of a decentralized distributed database. The storage module 101 is further configured to store a first local dataset LD. Furthermore, the storage module 101 can store previous versions of the global machine learning model ML and/or further local machine learning models of the node 200.” Here, Singh establishes receiving version information of a global model. Further, see Singh in paragraph [0086] describing, “The interface 106 is configured to communicate with the other nodes to perform a consensus method for jointly evaluating the check results. Therefore, the interface can be configured to receive check results CR' provided by other nodes, which are assigned to the second modified machine learning model mML2. The evaluation result EVAL of the joint evaluation of the check results CR', CR is transmitted to the replacement module 107.” This establishes the use of a communication interface with evaluation data. The communication with other nodes in this instance can act in place for the server or be a server as known in the art for decentralized federated learning.
Further, Singh teaches “and wherein the obtaining comprises: obtaining address information regarding a data set pre-stored in the electronic apparatus;”
See Singh in paragraph [0019] describing, “Within the context of embodiments of the invention, "assign", in particular in regard to data and/or information, can be understood to mean for example computer-aided assignment of data and/or information. By way of example, a second datum is assigned to a first datum in this regard by means of a memory address or a unique identifier (UID), e.g. by storing the first datum together with the memory address or the unique identifier of the second datum together in a data record.”
Singh does not appear to explicitly teach “adding the obtained address information regarding the data set to the version file”,
However in the same field of art, Schiatti teaches “adding the obtained address information regarding the data set to the version file.”
See Schiatti in paragraph [0030] describing, “For example, the machine learning model may be stored as a file with one or more data structures, such as a matrix, to arrange the information.” Here, Schiatti establishes a model being stored as file that contains data. Further, see Schiatti in paragraph [0033] describing, “The participant node 102 may further include a file system 114. The file system may include a storage for information. For example, the file system may include a repository, a database, a set of memory locations, and/or any other type of storage. The file system may index information, such as files.” Here, Schiatti establishes a file system for storing of a file that may include memory locations which can be seen as an address and adding obtained address information to the file. Further, see Schiatti in paragraph [0038] describing, “The DFL smart contract 118 may include a model link 202. The model link 202 may include a link to access a global model for a current round. The global model may include a model designated for further training. For example, the global model may be stored on the file system 114 (FIG. 1).” Here, Schiatti further establishes adding an obtained address with the inclusion of a link to access a global model which can be seen as the address information, and the global model being able to include a model designated for further training implies different versions of the model with the inclusion of a link to this model being seen as adding an obtained address to a version file, as a model can be stored as a file as mentioned.
It would have been obvious to one of ordinary skill in the art before the effective filing date of
the claimed invention to combine the base reference of Singh with the teachings of Schiatti by using Singh’s teachings of receiving information of a global neural network model, and training, evaluating, and determining to transmit information regarding the global neural network model, and incorporate with Schiatti’s teachings of adding address and parameter information regarding data to a version file of a model, and transmitting the file to a server.
One of ordinary skill in the art would be motivated to do so because by integrating Schiatti’s
frameworks into the methods of Singh, which are both in relation to decentralized federated learning, one of ordinary skill in the art would bring “distributed ledger technology [that] may decentralize the coordination and performance of federated learning. In addition, peer-to-peer file sharing and/or network file storage may decentralize model sharing” (Schiatti, paragraph [0019]), “learning process [that] is democratized by way of smart contract” (Schiatti, paragraph [0020]), “blockchain [that] provides a traceability and immutability for a federated learning” (Schiatti, paragraph [0021]), and “participant nodes participating in a federated learning process [that] may maintain privacy of private training data and models” (Schiatti, paragraph [0022]).
Claim(s) 7 are rejected under 35 U.S.C. 103 as being unpatentable over Singh et al., in view of Schiatti L. et al, and further in view of Kairouz et al., "Advances and Open Problems in Federated Learning", available at https://www.emerald.com/ftmal/article/14/1-2/1/1332154, published on March 9, 2021, (hereafter Kairouz).
Claim 7:
Regarding claim 7, Singh in view of Schiatti teaches the limitations of claim 6.
Neither Singh or Schaiatti appear to explicitly teach “The electronic apparatus as claimed in claim 6, wherein the at least one processor is further configured to control the communication interface to delete the address information regarding the data set from the updated version file before the updated version file is transmitted to the server”,
However in the same field of art, Kairouz teaches, “The electronic apparatus as claimed in claim 6, wherein the at least one processor is further configured to control the communication interface to delete the address information regarding the data set from the updated version file before the updated version file is transmitted to the server”
See Kairouz in Section 4 page 62 describing, “One of the primary attractions of the federated learning model is that it can provide a level of privacy to participating users through data minimization: the raw user data never leaves the device, and only updates to models (e.g., gradient updates) are sent to the central server. These model updates are more focused on the learning task at hand than is the raw data (i.e., they contain strictly no additional information about the user, and typically significantly less, compared to the raw data), and the individual updates only need to be held ephemerally by the server.” Here Kairouz establishes a transmission of data to a server regarding an updated model. The additional information not being passed in this transmission can be seen as it being deleted prior, raw data as known in the art comprises address information if it is stored and with it being stored in a device can be seen as deleting the address information regarding the data set from the updated version before transmitting to the server.
It would have been obvious to one of ordinary skill in the art before the effective filing date of
the claimed invention to combine the base reference of Singh with the teachings of Schaiatti, and further with the teachings of Kairouz by using Singh’s teachings of receiving information of a global neural network model, and training, evaluating, and determining to transmit information regarding the global neural network model, with Schaiatti’s teachings of receiving information regarding evaluation data and incorporate with Kairouz’s teachings of determining whether to transmit information regarding a global neural network model to a server.
One of ordinary skill in the art would be motivated to do so because by integrating Kairouz’s
frameworks into the methods of Schaiatti and Singh, which are all in relation to federated learning, one of ordinary skill in the art would bring “Performing local updates and communicating less frequently with the central server addresses the core challenges of respecting data locality constraints and of the limited communication capabilities of mobile device clients” (Kairouz, Section 3.2 Optimization Algorithms for Federated Learning pages 34-35).
Claim(s) 9 is rejected under 35 U.S.C. 103 as being unpatentable over Singh et al., in view of Jithish et al., "Distributed Anomaly Detection in Smart Grids: A Federated Learning-Based Approach", available at https://ieeexplore.ieee.org/document/10018378, effectively published on December 20, 2022, (hereafter Jithish).
Claim 9:
Regarding claim 9, Singh teaches the limitations of claim 1.
Singh does not appear to explicitly teach “The electronic apparatus as claimed in claim 1, wherein the at least one processor is further configured to: perform Secured Sockets Layer/Transport Layer Security (SSL/TLS) encoding on the information regarding the trained global neural network model; and control the communication interface to transmit the encoded trained global neural network model to the server”,
However in the same field of art, Jithish teaches, “The electronic apparatus as claimed in claim 1, wherein the at least one processor is further configured to: perform Secured Sockets Layer/Transport Layer Security (SSL/TLS) encoding on the information regarding the trained global neural network model;”
See Jithish in Abstract describing, “After local training, local model parameters are sent to the server to improve the global model. We secure the model parameter updates from adversaries using the SSL/TLS protocol.” Here Jithish establishes performing training using SSL/TLS regarding a global model. Further, see Jithish in Section VI Discussion describing, “Therefore, we use SSL/TLS protocol to secure the FL training process against cyberattacks, ensuring that client-server communication is encrypted.” Here, Jithish establishes encrypting which can be seen as encoding.
Further, Jithish teaches, “and control the communication interface to transmit the encoded trained global neural network model to the server.”
See Jithish in Abstract describing, “After local training, local model parameters are sent to the server to improve the global model. We secure the model parameter updates from adversaries using the SSL/TLS protocol.” Here Jithish establishes transmitting or sending information regarding the global model to a server. As mentioned in the previous limitation this information is encoded/encrypted in the SSL/TLS protocol. Further, see Jithish in Section V Federated Anomaly Detection for Smart Grids describing, “Communication Protocol: Raspberry Pi client devices and the laptop (server) communicate using a router through WiFi interface.” Here, Jithish a communication interface for communicating to the server.
It would have been obvious to one of ordinary skill in the art before the effective filing date of
the claimed invention to combine the base reference of Singh with the teachings of Jithish by using Singh’s teachings of receiving information of a global neural network model, and training, evaluating, and determining to transmit information regarding the global neural network model, and incorporate with Jithish’s teachings of performing SSL/TLS encoding on information regarding a global model.
One of ordinary skill in the art would be motivated to do so because by integrating Jithish’s
frameworks into the methods of Singh, which are all in relation to federated learning, one of ordinary skill in the art would bring “proposed FL-based models perform efficiently in terms of memory, CPU usage, bandwidth and power consumption at edge devices and are suitable for implementation in resource-constrained environments, such as smart meters, for anomaly detection” (Jithish, Abstract).
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to HASSAN R SESAY whose telephone number is (571)272-8493. The examiner can normally be reached Monday-Friday 8am-5pm.
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, Usmaan Saeed can be reached at (571) 272-4046. 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.
/HASSAN RAMADAN SESAY/Examiner, Art Unit 2146
/USMAAN SAEED/Supervisory Patent Examiner, Art Unit 2146
Prosecution Insights