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
It is noted that applicant has not filed a certified copy of the IN202311079233 application as required by 37 CFR 1.55.
Note regarding reference characters in claims 5 and 10
Reference characters corresponding to elements recited in the detailed description and the drawings may be used in conjunction with the recitation of the same element or group of elements in the claims. The reference characters, however, should be enclosed within parentheses so as to avoid confusion with other numbers or characters which may appear in the claims. Generally, the presence or absence of such reference characters does not affect the scope of a claim.
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 10-18 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter. The claim(s) does/do not fall within at least one of the four categories of patent eligible subject matter because claim 10 is considered to be drawn to software per se, which is not patent eligible subject matter. The claim recites a “system (102) for transferring data from one storage to another storage… comprising a transfer AI agent.” The presence of reference characters does not affect the scope of a claim, and “storage” may be a signal per se since the instant specification merely provides examples of storage unit 204 and memory 112 and does not exclude a transitory signal per se (e.g., [0038]-[0039] of the instant specification). The AI agent as configured may likewise be interpreted as software code.
The dependent claims do not rectify this issue and are therefore rejected with parent claim 10.
Claims 1-19 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. Note that the courts do not distinguish between mental processes that are performed entirely in the human mind and mental processes that require a human to use a physical aid (e.g., pen and paper or a slide rule) to perform the claim limitation (refer to MPEP 2106.04(a)(2)).
Example independent claim 19 recites the following abstract idea limitations: a system for transferring data from one storage to another storage (transferring information as part of certain methods of organizing activity—e.g., Alice transfers written documents to Bob for collaboration; the storage may be, e.g., a filing cabinet or mental memory), [caused] to perform instructions that cause the system to perform operations comprising:
identifying, by [an entity], one or more Machine Learning (ML)/ Artificial Intelligence (AI) models and data associated with the one or more ML/AI models to be transferred to the other storage selected by the [entity] (observation as part of a mental process—e.g., Alice looks through a set of written information to identify particular aspects);
organizing, by the [entity], the one or more ML/AI models and data to be transferred to the other storage (evaluation as part of a mental process—e.g., Alice prepares her the identified information by taking notes or memorizing the identified information; Alice can further take memorizing or take notes on how she intends to perform the transfer);
abstracting, by the [entity], relevant information from the one or more ML/AI models and the data (evaluation and judgement as part of a mental process—e.g., Alice picks out particular parts of the identified data which she will send, depending on considerations such as sensitivity or usefulness to the collaboration), wherein the relevant information is encrypted (evaluation as part of a mental process—e.g., Alice encrypts the picked out parts using an algorithm such as a Caesar cipher); and
applying, by the [entity], one or more obfuscation techniques on the encrypted relevant information (evaluation as part of a mental process—e.g., Alice obfuscates the encrypted information using an algorithm such as that of redaction or adding random noise values), wherein the encrypted relevant information is transferred to the other storage (certain methods of organizing human activity—e.g., Alice writes up a final obfuscated and encrypted document and gives it to Bob).
Example independent claim 19 recites the following limitations which may comprise additional elements sufficient to amount to significantly more than the judicial exception: “A non-transitory machine-readable medium including data, which when used by” the system causes it to perform the instructions of the abstract idea; the entity further comprising “a transfer AI agent.”
With respect to step 2A, this judicial exception is not integrated into a practical application because adding the words "apply it" (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely using a computer as a tool to perform an abstract idea are not considered to be sufficient—e.g., see MPEP 2106.05(f). In this case, the claim is drawn to looking through information concerning models and associated data, finding particular relevant information to send, and encrypting and obfuscating the relevant information before sending it. This is performed at a high level of generality such that it may be implemented under the human Alice-and-Bob framework discussed above. While the claim does include the instructions on a non-transitory machine-readable medium, it is noted that: (1) storing instructions on memory is a base level implementation by any computer which can perform the abstract idea; (2) the non-transitory machine-readable medium can be interpreted as a paper with instructions.
The claim does not recite any particular machine for performing the abstract idea, and the “transfer AI agent” is specified at a level of generality where its actions may be performed by a human. As such, the claimed invention is addressing a problem that transcends computing (picking out information to send and securing it during transmission using any kind of obfuscation and encryption) rather than improving the functioning of a computer, or an improvement to other technology or a technical field.
With respect to step 2B, the claim(s) does/do not include additional elements that are sufficient to amount to significantly more than the judicial exception because adding the words "apply it" (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely using a computer as a tool to perform an abstract idea are not considered to be sufficient-e.g., see MPEP 2106.05(f). In this case, a machine-readable medium is a base level computer component of any computer that would be capable of performing the abstract idea. And the “transfer AI agent” is not defined beyond the human-performable actions which it is taking in the claim. As such, it may be any software for adding the words “apply it” concerning automation of the abstract idea.
Independent claims 1 and 10 recite substantially similar claim language, and are therefore rejected under the same analysis.
Regarding dependent claim 2, it recites the following abstract idea limitations: wherein the one or more obfuscation techniques comprises intentionally making the data unintelligible, preventing third parties from one or more of generating sensitive information and deducing sensitive information (merely further specifying the obfuscation algorithm at a high level of generality—e.g., Alice redacts information with black ink or adds random noise to the document). Since claim 2 merely further specifies the abstract idea, it is likewise rejected under the same analysis.
Regarding dependent claim 3, it recites the following abstract idea limitations: maintaining, by the transfer Al agent, one or more detailed logs corresponding to a transfer of the one or more ML/AI models and the data for monitoring the one or more ML/AI models and the data, auditing the one or more ML/AI models and the data, and troubleshooting (recording information as part of a mental process—e.g., Alice either writes down or memorizes information about each transfer). Since claim 3 merely further specifies the abstract idea, it is likewise rejected under the same analysis.
Regarding 4, it merely further specifies the form of data being looked over, and is therefore likewise rejected under the same analysis.
Regarding dependent claim 5, it recites the following abstract idea limitations: wherein organizing the one or more ML/AI models and the data comprises: categorizing the one or more ML/AI models and the data into a plurality of categories, further wherein the plurality of categories comprises one or more facts having immutable data points representing specific events or user attributes, one or more inferences derived by one or more Al agents based on factual data and an observed behaviour, one or more patterns and tendencies observed from one or more user interactions with one or more system (102)s or the one or more Al agents (evaluation as part of a mental process—e.g., Alice categorizes the data according to a given algorithm or requirement); and structuring the one or more ML/AI models and the data based on the plurality of categories (fitting values into a model as part of a mental process—e.g., Alice structures the models using categories as values). Since claim 5 merely further specifies the abstract idea, it is likewise rejected under the same analysis.
Regarding claims 6-8, they are rejected for substantially the same reasons as claims 3-5 above (i.e., they merely further specify the abstract idea concerning data collection, interpretation and categorization, and updating).
Regarding claim 9, it recites the following abstract idea limitations: performing, by the transfer Al agent, a validation check on the one or more ML/AI models and the data to ensure that the one or more ML/AI models and the data is not corrupted (observation, evaluation, and judgement as part of a mental process—e.g., Alice double checks her data to make sure it is correct), and the one or more ML/AI models and the data is meeting a predefined standard prior to the transfer (i.e., further specifying that there is a predefined standard to double check against; that Alice performs the double checking before the transfer to Bob); and packaging the one or more ML/AI models and the data with metadata/scripts facilitating one or more of an immediate fine-tuning, a subsequent fine-tuning, and a training to be done during the transfer of the data (appending data as part of certain methods of organizing human activity—e.g., Alice adds metadata notes to the transferred information). Since claim 9 merely further specifies the abstract idea, it is likewise rejected under the same analysis.
Dependent claims 11-18 are substantially similar to claims 2-9 above, and are therefore likewise rejected.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 2 and 11 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the enablement requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to enable one skilled in the art to which it pertains, or with which it is most nearly connected, to make and/or use the invention. Each of claims 2 and 11 recites “preventing third parties from one or more of generating sensitive information and deducing sensitive information,” which renders the claim as failing to comply with the enablement requirement. The “sensitive information” is unbounded and is not previously defined in the claims. As such, the broadest reasonable interpretation is that the claimed “sensitive information” refers substantially to any and all sensitive information. This would include sensitive information beyond the scope of the claimed data transferring and transfer AI agent (e.g., state secrets, personal secrets between any given people, PII on any computer in the world, and so forth). Since the claimed invention concerns implementing and securing a specific data transfer (“relevant information” concerning models and their associated data), it clearly does not disclose how to prevent third parties from generating or deducing the broad class of any and all sensitive information that exists. While known prior art techniques include many different kinds of encryption and obfuscation techniques, these are applied to specific use cases (e.g., federated learning) rather than for absolute coverage of all sensitive data that exists. The instant specification likewise does not disclose preventing generating or deducing sensitive information at that broad level of generality. For instance, it does not disclose how to prevent a given person from “generating” sensitive information in their mind; nor does it disclose how to prevent, e.g., a spy from overhearing state secrets. As such, one of ordinary skill in the art would require undue experimentation to arrive at a solution for preventing third parties from generating/deducing sensitive information at that level of generality.
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-19 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. Independent claim 1 recites multiple limitations which render it indefinite. It is first noted that claim 1 is drawn to a “method for transferring data” while also reciting “data associated with the one or more ML/AI models,” which renders it indefinite because it is note clear whether “data associated with the one or more ML/AI models” is the same as the data in the preamble.
Additionally, it is not clear how to interpret “one or more Machine Learning (ML)/ Artificial Intelligence (Al) models and data associated with the one or more ML/AI models to be transferred to the other storage selected by the transfer Al agent” concerning whether only the “data associated” is to be transferred, or whether the models are to be transferred as well.
Claim 1 further recites “applying, by the transfer Al agent, one or more obfuscation techniques on the encrypted relevant information, wherein the encrypted relevant information is transferred to the other storage,” which renders the claim indefinite because it is not clear whether “wherein the encrypted relevant information is transferred to the other storage” refers to the obfuscated encrypted relevant information or to transferring the encrypted relevant information before it is obfuscated. For the purpose of applying prior art, this limitation has been interpreted as referring to transferring the obfuscated encrypted relevant information.
Finally, claim 1 recites “the other storage selected by the transfer AI agent,” which renders the claim indefinite because the selection is in past tense and may have taken place outside of the claim scope. As such, a person of ordinary skill in the art could not interpret the metes and bounds of the claim so as to understand how to avoid infringement concerning storage selection.
Independent claims 10 and 19 recite substantially similar claim language, and are rejected under the same analysis as above.
The dependent claims do not rectify the above-addressed issues and are therefore likewise rejected with their respective parent claims.
Claims 3 and 12 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. Each of claims 3 and 12 recite “maintaining, by the transfer Al agent, one or more detailed logs corresponding to a transfer of the one or more ML/AI models and the data,” which renders these claims indefinite because it is not clear whether “a transfer of the one or more ML/AI models and the data” refers to the step of “wherein the encrypted relevant information is transferred” from the respective parent claims, or to a different transfer. Additionally, it is not clear whether “a transfer of the one or more ML/AI models and the data” is actually part of the claim scope because it could refer to a transfer which takes place beyond the claim scope (e.g., a previously recorded transfer). The claim merely requires maintaining the logs, which may have been obtained without performing the actual logging operation. As such, a person of ordinary skill in the art could not interpret the metes and bounds of the claim so as to understand how to avoid infringement concerning logging data transfers.
Also, the term “detailed” in “detailed logs” is a relative term which renders the claim indefinite. The term “detailed” is not defined by the claim, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention.
Claims 4 and 13 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. Each of claims 4 and 13 recites a list of claim elements under “one or more of” and “and /or” combinations where it is unclear how these claim elements are delineated. For instance, “wherein the one or more ML/AI models is one or more of a language model, a 3-Dimensional (3D) model, an image model, 3D mannerisms, a voice model including tonal voices and the data comprises one or more documents,” where it is not clear if “data comprises one or more documents” is only part of the “voice model including tonal voices” portion, or if it is a separate limitation after the claim language defining the ML/AL models is finished. The claims appear to be missing a semicolon and line break before “and the data comprises one or more documents.”
As another example, it is not clear whether “and/or one or more machine learning (ML) models trained for making predictions tailored to individual users or specific use cases” is meant to be part of the definition of what the “data comprises” or another of the “one or more ML/AI models.” It is further not clear whether this limitation refers back to the “ML/AI models” of the previous claim language and respective parent claim, or to newly introduced models.
As such, the claims are indefinite because it is not clear how to interpret nor identify what is being defined and how it relates to the previous claim elements.
Claims 7 and 16 rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. The term “holistic view” of “to form a holistic view of the data” is a relative term which renders the respective claims indefinite. The term “holistic” is not defined by the claim, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention. In this case, [0054] of the instant specification merely recites the same language as in the claims.
Claims 9 and 18 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. Each of claims 9 and 18 recites “packaging the one or more ML/AI models and the data with metadata/scripts facilitating one or more of an immediate fine-tuning, a subsequent fine-tuning, and a training to be done during the transfer of the data,” which renders the respective claims indefinite because it is not clear how to interpret “to be done during the transfer of the data.” For instance, “during a transfer of the data” may mean during the time the time data is being transferred as packets on a network. It is not clear how “an immediate fine-tuning, a subsequent fine-tuning, and a training” are to be performed on the data while it is in transit.
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.
Claim(s) 1-2, 4-8, 10-11, 13-17, and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Liu (“MULTI-DIMENSIONAL FEDERATED LEARNING IN RECOMMENDER SYSTEMS”) in view of Chen (US 2025/0132901 A1).
Regarding claim 1, Liu discloses: A method for transferring data from one storage to another storage, comprising:
identifying, by a transfer Al agent (e.g., 3.3.2 of Liu stating that “users’ personal spaces should serve as intermediate agents that coordinate an indirect transfer framework;” FIG. 3.4 of Liu stating that “personal space Eu serves as the intermediate transfer agent”), one or more Machine Learning (ML)/ Artificial Intelligence (Al) models and data associated with the one or more ML/AI models to be transferred to the other storage selected by the transfer Al agent (e.g., “each domain separately owns and maintains its CF model fd… fd along with U(d) are pretrained by a domain-specific objective L(CF) d and then serve as part of the training environment for the cross-domain transfer model” in 3.3.1 of Liu, where U(d) are user embeddings);
Refer to at least 3.4.1 of Liu concerning the “objective of the system is to learn a set of decentralized zu for each user that contains ‘complete user information’ from all domains, so that each domain-specific DECd(·) can extract accurate U(d) u that contains sufficient information for later recommendation services.”
organizing, by the transfer Al agent, the one or more ML/AI models and data to be transferred to the other storage;
Refer to at least FIG. 3.3-3.4 and 3.5(b) of Liu with respect to organizing data transfer of model data and user information.
abstracting, by the transfer Al agent, relevant information from the one or more ML/AI models and the data; and
Refer to at least 3.4.1-3.4.2 of Liu, wherein “the only auxiliary information for each domain is the encoded zu that is maintained by Eu from all interacted domains,” “each domain now has its own encoder and decoder, and the only information that is shared across domains is the latent encoding zu maintained on each user’s personal space Eu…”
applying, by the transfer Al agent, one or more obfuscation techniques on the encrypted relevant information, wherein the relevant information is transferred to the other storage.
Refer to at least page 22 of Liu stating that “if the client wants to avoid information transfer from one dimension to another, the client can simply mask or disguise the corresponding information in the shared space before sending it to the service.” Alternatively, refer to at least pages 19, 52, 55, and 62 of Liu concerning applying differential privacy methods.
Although Liu discusses using encryption and obfuscation (masking / DP) generally, it does not specify: wherein the relevant information is encrypted; apply the one or more obfuscation techniques on encrypted relevant information; transferring encrypted relevant information after the obfuscation. However, Liu in view of Chen discloses: wherein the relevant information is encrypted; apply the one or more obfuscation techniques on encrypted relevant information; transferring encrypted relevant information after the obfuscation.
Refer to at least FIG. 1 and [0054] of Chen with respect to encrypting then masking information sent as part of federated learning.
The teachings of Chen likewise concern privacy for federated learning, and are considered to be within the same field of endeavor and combinable as such.
Therefore it would have been obvious to one of ordinary skill in the art before the filing date of Applicant’s invention to modify the teachings of Liu to implement both encryption and obfuscation because the particular known technique (obfuscation, encryption, and using both together) was recognized as part of the ordinary capabilities of one skilled in the art, and for at least the purpose of improving privacy (i.e., protecting sensitive data, especially where it is bound by regulations and service agreements).
Regarding claim 2, Liu-Chen discloses: The method according to claim 1, wherein the one or more obfuscation techniques comprises intentionally making the data unintelligible (i.e., adding noise as per the cited portions of Liu and Chen concerning masking), preventing third parties from one or more of generating sensitive information and deducing sensitive information.
Refer to at least 3.1, 4.3.2, and page 63 of Liu with respect to protecting sensitive information from third parties (e.g., via noise obfuscation).
Regarding claim 4, Liu-Chen discloses: The method according to claim 1, wherein the one or more ML/AI models is one or more of a language model, a 3-Dimensional (3D) model, an image model, 3D mannerisms, a voice model including tonal voices
Refer to at least 1.2.1 of Liu concerning example models such as various language models.
and the data comprises one or more documents, information associated with one or more artificial intelligence (AI) models and/or one or more machine learning (ML) models trained for making predictions tailored to individual users or specific use cases, one or more user interactions with the one or more ML/AI models, learned knowledge based on the one or more user interaction, and one or more databases.
Refer to at least 1.2.1, 3.1, and 3.3.1 of Liu with respect to user data for recommender models (e.g., recommending movies based on user profile data).
Regarding claim 5, Liu-Chen discloses: The method according to claim 1, wherein organizing the one or more ML/AI models and the data comprises: categorizing the one or more ML/AI models and the data into a plurality of categories, further wherein the plurality of categories comprises one or more facts having immutable data points representing specific events or user attributes, one or more inferences derived by one or more Al agents based on factual data and an observed behaviour, one or more patterns and tendencies observed from one or more user interactions with one or more system (102)s or the one or more Al agents; and
Refer to at least 3.2 and 3.3.1 of Liu with respect to categorizing domain data (e.g., user profile data concerning movies)
structuring (e.g., FIG. 3.4 of Liu) the one or more ML/AI models and the data based on the plurality of categories.
Refer to at least 3.3.1 on page 33 of Liu stating that “The ultimate goal is to provide recommendations for a cold-start user in the target domain dt ∈ D with the user’s embedding transferred from other source/auxiliary domains… this problem is a federated collaborative transfer learning task that requires both federation of multiple domain services and the collaboration between user spaces.”
Regarding claim 6, it is rejected for substantially the same reasons as claim 1 above (e.g., “each domain now has its own encoder and decoder, and the only information that is shared across domains is the latent encoding zu maintained on each user’s personal space Eu” on page 37 of Liu).
Regarding claim 7, Liu discloses: The method according to claim 1, further comprising: merging one or more common data points in the one or more ML/AI models and the data originating from a plurality of interactions between a user and one or more Al agents;
Refer to at least 1.2.3 of Liu stating that “multi-behavior modeling usually adopts a multi-task or multi-label learning paradigm that aims to improve user engagement for different behaviors (e.g. click, purchase, like, comment, and review). In other words, multi-behavior modeling is essentially a transfer learning task that aims to learn a user representation model for downstream tasks of all behaviors;” page 80 of Liu describing “federated learning of heterogeneous user behaviors across all services and a decomposed view that coordinates all participants.”
and aggregating behavioral data from one or more touchpoints to form a holistic view of the data prior to the transfer of the one or more ML/AI models and the data.
Refer to at least 3.4.1 and FIG. 3.4 of Liu with respect to aggregating user data (e.g., gd for a given user and domain)
Regarding claim 8, it is rejected for substantially the same reasons as claims 1, 5, and 7 above (e.g., FIG. 3.4 of Liu).
Regarding independent claim 10, it is substantially similar to independent claim 1 above, and is therefore likewise rejected.
Regarding claims 11 and 13-17, they are substantially similar to claims 2 and 4-8 above, and are therefore likewise rejected.
Regarding independent claim 19, it is substantially similar to independent claim 1 above, and is therefore likewise rejected.
Claim(s) 3 and 12 is/are rejected under 35 U.S.C. 103 as being unpatentable over Liu-Chen as applied to claims 1-2, 4-8, 10-11, 13-17, and 19 above, and further in view of Keshavamurthy (US 2025/0371370 A1).
Regarding claim 3, Liu-Chen does not specify: maintaining, by the transfer Al agent, one or more detailed logs corresponding to a transfer of the one or more ML/AI models and the data for monitoring the one or more ML/AI models and the data, auditing the one or more ML/AI models and the data, and troubleshooting. However, Liu-Chen in view of Keshavamurthy discloses: maintaining, by the transfer Al agent, one or more detailed logs corresponding to a transfer of the one or more ML/AI models and the data for monitoring the one or more ML/AI models and the data, auditing the one or more ML/AI models and the data, and troubleshooting (the latter portion of the claim being an intended use of the logs).
Refer to at least [0210] of Keshavamurthy with respect to maintaining a log of transfers of model training between UEs.
The teachings of Keshavamurthy likewise concern federated learning, and are considered to be within the same field of endeavor and combinable as such.
Therefore it would have been obvious to one of ordinary skill in the art before the filing date of Applicant’s invention to modify the teachings of Liu-Chen to further implement logging transfers because particular known technique (logging data accesses and transfers) was recognized as part of the ordinary capabilities of one skilled in the art.
Regarding claim 12, it is substantially similar to claim 3 above, and is therefore likewise rejected.
Claim(s) 9 and 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Liu-Chen as applied to claims 1-2, 4-8, 10-11, 13-17, and 19 above, and further in view of Chai (US 2024/0070266 A1).
Regarding claim 9, Liu-Chen discusses validating convergence (e.g., 3.4.4 of Liu), but does not fully specify: performing, by the transfer Al agent, a validation check on the one or more ML/AI models and the data to ensure that the one or more ML/AI models and the data is not corrupted, and the one or more ML/AI models and the data is meeting a predefined standard prior to the transfer; and packaging the one or more ML/AI models and the data with metadata/scripts facilitating one or more of an immediate fine-tuning, a subsequent fine-tuning, and a training to be done during the transfer of the data. However, Liu-Chen in view of Chai discloses: performing, by the transfer Al agent, a validation check on the one or more ML/AI models and the data to ensure that the one or more ML/AI models and the data is not corrupted, and the one or more ML/AI models and the data is meeting a predefined standard prior to the transfer; and packaging the one or more ML/AI models and the data with metadata/scripts facilitating one or more of an immediate fine-tuning, a subsequent fine-tuning, and a training to be done during the transfer of the data.
Refer to at least [0040] and [0068] of Chai stating that “a cyclic redundancy check (CRC) code or other form of checksum is computed on the trained model and then inserted into the model's metadata” and that “an integrity check begins with reading a stored CRC code or checksum value from model metadata. In operation 422, a fresh CRC value is computed on current model parameters and, in operation 424, the stored and computed CRC values are compared. In operation 426, the results of the comparison are used to produce an output of the integrity check (e.g., pass or fail).”
The teachings of Chai likewise concern secured machine learning, and are considered to be within the same field of endeavor and combinable as such.
Therefore it would have been obvious to one of ordinary skill in the art before the filing date of Applicant’s invention to modify the teachings of Liu-Chen to further implement inserting and validating a checksum in the model’s metadata for at least the purpose of maintaining integrity and preventing attacks inserting malicious code (e.g., to steal sensitive data) or information (e.g., to hurt training).
Regarding claim 18, it is substantially similar to claim 9 above, and is therefore likewise rejected.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to VADIM SAVENKOV whose telephone number is (571)270-5751. The examiner can normally be reached 12PM-8PM.
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, Jeffrey L Nickerson can be reached at (469) 295-9235. 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.
/Jeffrey Nickerson/Supervisory Patent Examiner, Art Unit 2432
/V.S/ Examiner, Art Unit 2432
Prosecution Insights