DETAILED ACTION
1. This office action is in response to the Application No. 18778094 filed on 04/27/2026. Claims 1-20 are presented for examination and are currently pending. Applicant’s arguments have been carefully and respectfully considered.
Response of Arguments
2. The Applicants argument regarding the prior art have been considered and the Examiner is withdrawing the rejections in the previous Office action because Applicant’s amendment necessitated new grounds of rejection presented in this Office Action.
It is noted that arguments regarding independent claims 1 and 11 have been considered but are moot because a new reference has now been used to remap the independent claims 1 and 11.
The dependent claims 2-10 and 12-20 which depend directly or indirectly from independent claims 1 and 11 are not patentable because the instant claims are still obvious over the prior art of record.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
3. Claims 1-3, 6, 8, 11-13, 16 and 18 are rejected under 35 U.S.C. 103 as
being unpatentable over Jain et al. (US12111747 filed 05/10/2024) in view of Chen et
al. (US20250077792 filed 08/31/2023) and further in view of Rollwage et al. (US20240404514 filed 08/30/2023)
Regarding claim 1, Jain teaches an apparatus (FIG. 2 shows a block diagram showing some of the components typically incorporated in at least some of the computer systems and other devices on which the disclosed system operates in accordance with some implementations of the present technology, col. 1, ln 50-54) for training an excitation model (For example, an LLM is configured or trained using reinforcement learning from human feedback (RLHF), instruction tuning (col. 21, ln 51-54); For example, the data generation platform can redirect a prompt to a second LLM (e.g., distinct from the first LLM (col. 29, ln 18-20); Based on particular model architectures and training data used to generate or tune LLMs (col. 2, ln 58-59). The Examiner notes the second LLM is an excitation model),
wherein the apparatus comprises: at least a processor; memory communicatively connected to the at least a processor (processor(s) 208, including a CPU for executing computer programs, a GPU for executing computer graphic programs and handling computing graphical elements; storage(s) 210, including at least one computer memory for storing programs (e.g., application(s) 212, model (s) 214, and other programs) and data while they are being used, col. 12, ln 14-20),
wherein the memory contains instructions configuring the at least a processor to: instantiate a representation generator (generation instances associated with the system, col. 30, ln 4-5);
collect a first dataset from a system (For example, the data generation platform 102 obtains, from a first database, a plurality of training prompts and respective
performance metric values associated with providing respective training prompts to the first LLM, (col. 36, ln 21-24); For example, the data generation platform 102 receives inputs such as unstructured data, including text data, such as a prompt (col. 9, ln 37-40); An event database can include data associated with events
relating to the data generation platform 102, col. 9, ln 10-11);
generate, using the representation generator and the first dataset, a first representation dataset (The data generation platform 102 can generate a first vector
representation for the expected test output (col. 49, ln 31-32); the data generation platform can provide the prompt to the selected model (e.g., LLM) for generation of the requested output, (col. 6, 7-9); output generation request with a first performance criterion associated with the first LLM of a plurality of LLMs (col. 26, ln 8-9); machine learning models (e.g., LLMs), col. 31, ln 8. The Examiner notes the output is the generated first representation dataset and the first LLM is a representation generator which is a machine learning model),
wherein the first representation dataset comprises a plurality of evaluation metrics (The data generation platform can determine estimated performance metric values associated with generating the output, col. 6, ln 14-16)
output one or more excitation elements from the excitation model (the data generation platform can determine another model (e.g., a second LLM) for generation of the output, col. 6, ln 12-14. The Examiner notes the second LLM is the excitation model and the output generated is an excitation element) using the first representation dataset (For example, the data generation platform can redirect a prompt to a second LLM (col. 29, ln 18-20); natural language output e.g., prompts …, col. 9, ln 13-14),
transmit the excitation element to the system (the deployment database can include a server system (e.g., physical or virtual) that stores validated outputs or results from one or more LLMs (col. 10, ln 7-9); In response to validating the generated output, the data generation platform can transmit this information to an associated data store or deployment system, col. 4, ln 46-49);
collect a second dataset from the system (For example, the data generation platform 102 obtains, from a first database (col. 36, ln 21-22); A deployment database can include data associated with deploying, using, or viewing results associated with the data generation platform 102, col. 10, ln 5-7);
generate an error signal (The data generation platform 102 can provide the indication of the validation error, col. 51, ln 31-33) as a function of the second dataset and the first representation dataset (For example, the data generation platform 102 can generate a second code sample using a generative model that cures determined validation errors (e.g., by including an indication of the appropriate validation errors within a prompt of the LLM, col. 45, ln 66-67 to col. 46 ln 1-4);
modify the representation generator using the error signal (Based on such a determination, the data generation platform 102 can transmit the validation indicator 1222 (e.g., including indications of associated validation errors) to an LLM (e.g., the first model) for modification (col. 45, ln 62-66); For example, the access control engine 114 modifies or changes the LLM for execution of the prompt associated with the output generation request based on the user identifier, the attribute, and/or the performance evaluation 408, col. 16, ln 43-47),
wherein modifying the representation generator is configured to generate at least a modified evaluation metric (In some implementations, the data generation platform 102 can generate the modified output by providing an indication of a validation error (e.g., associated with the validation indicator) to an LLM (col. 51, ln 23-26); The data generation platform can determine estimated performance metric values associated with generating the output, col. 6, ln 14-16) comprising
a reprioritization of the one or more excitation elements (Accordingly, the data
generation platform 102 enables the prioritization of relevant performance metrics (e.g., cost) over other metrics (e.g., memory usage) according to system requirements (col. 31, ln 43-46); the data generation platform 102 can determine the composite performance metric value based on weights that correspond to
the order of priority (col. 21, ln 24-27); modify the output (e.g., by resubmitting the
output to the LLM) to modify the sentiment associated with the output, col. 27, ln 5-7);
output a second representation dataset (The data generation platform 102 can generate … a second vector representation for the test output (col. 49 ln 31-34); The LLM can generate an output based on the query and the retrieved documents col. 21, ln 66-7 to col. 22 ln 1) using the modified representation generator (In some implementations, the model (e.g., LLM) includes augmented or modified LLMs, such as retrieval-augmented generation (RAG) algorithms, col. 21, ln 57-59); and
output a second excitation element from the excitation model (the process 1100 can generate a second output by providing the prompt to the second model (col. 37, ln 54-55); the platform can generate the output using the second LLM and transmit the output to a computing system, col. 6 ln 27-30) using the second representation dataset (The data generation platform 102 can generate … a second vector representation for the test output (col. 49 ln 31-34)).
Jain does not explicitly teach wherein the representation generator identifies executive functions to be a focus area and prioritizes the focus area comprising emotional regulation; wherein the first representation dataset comprises a plurality of evaluation metrics which includes at least a quantitative measurement used to assess a cognitive function; wherein the first representation dataset comprises a plurality of embeddings, wherein each embedding represents a quantitative measurement of a cognitive function across at least two domains, and wherein the representation generator computes a cosine similarity score between embeddings to identify the focus area for prioritization;
wherein an excitation element comprises a system recommendation comprising at least a neurocognitive exercise to increase performance in a cognitive or behavioral domain and a recommendation in a work or career domain; tune the representation generator iteratively using the error signal to correct a deficiency of the system;
Chen teaches tune the representation generator iteratively using the error signal to correct a deficiency of the system (When fine-tuning the pretrained machine learning model 408 using the adaptation component 420, during each training iteration (or fine-tuning iteration), the set of low-rank matrices of the adaptation component 420 are updated based on the error signal 412 determined from the predicted output 406 and the pseudo label 418 [0074]);
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Jain to incorporate the teachings of Chen for the benefit of fine-tuning a machine learning model [0011] to perform one or more domain-specific tasks (Chen [0025])
Jain and Chen does not explicitly teach wherein the representation generator identifies executive functions to be a focus area and prioritizes the focus area comprising emotional regulation; wherein the first representation dataset comprises a plurality of evaluation metrics which includes at least a quantitative measurement used to assess a cognitive function; wherein the first representation dataset comprises a plurality of embeddings, wherein each embedding represents a quantitative measurement of a cognitive function across at least two domains, and wherein the representation generator computes a cosine similarity score between embeddings to identify the focus area for prioritization; wherein an excitation element comprises a system recommendation comprising at least a neurocognitive exercise to increase performance in a cognitive or behavioral domain and a recommendation in a work or career domain;
Rollwage teaches instantiate a representation generator (instantiate under specific (stressful) situations ... a key component of a CBT based recommender system 32 is to understand the individual subcomponents of this qualitative disorder model in a more quantitative manner [0462]),
wherein the representation generator identifies executive functions to be a focus area (For example, one or more interventions are identified by the first module 30 as further described below and included in the third system prompt [0664]; In one example, the task of this recommender system 32 is to choose from a number of interventions defined by CBT in order to maximise patient outcomes [0409]; the recommender module 32 may comprise one or more machine learning models to identify interventions [0745]) and
prioritizes the focus area comprising emotional regulation (The identified interventions may be included in the third system prompt in order of how useful they are determined to be. The suggested interventions may further comprise one or more interventions from a pre-set treatment plan ... This may include things like ... exercises related to one's feelings (e.g. emotion regulation techniques)[0664]. The Examiner notes the interventions from a pre-set treatment plan is the prioritized focus area);
collect a first dataset from a system (providing the input data to a first module [0057]);
generate, using the representation generator ( For example, the recommender module 32 may comprise a neural network that is trained to maximize one or more metrics [0745]) and the first dataset (The knowledge bank 33 comprises stored therapy-related knowledge. As has been described previously, the input text data from the user is provided to the first module 30 [0393]),
a first representation dataset, wherein the first representation dataset comprises a plurality of evaluation metrics which includes at least a quantitative measurement (In some examples, the recommender module 32 may be configured to identify interventions while maximizing and/or optimizing all three of the target metrics [0744]) used to assess a cognitive function (These models output subject profile information represent clinical concepts or diagnoses, such as cognitive distortion [0020]),
wherein the first representation dataset comprises a plurality of embeddings (and in some cases the first module 30, analyses the input data further to generate dynamically determined system responses [0571]; This is done by transforming the system response into an embedding [0804]),
wherein each embedding represents a quantitative measurement of a cognitive function across at least two domains (The distance between this embedding and the embeddings of all the intervention steps are then measured [0804]), and
wherein the representation generator computes a cosine similarity score between embeddings to identify the focus area for prioritization (The similarity measure may be generated by embedding the input text using a language model, embedding all reference entries in the knowledge bank 33 using the same language model, computing the cosine similarity (or some other similarity measure) between the input embedding and all reference embeddings [0393]. The Examiner notes the language model in the excitation model);
output one or more excitation elements from the excitation model using the first representation dataset (The output from the LLM may be a label (e.g. all-or-nothing thinking) and a probability [0520]),
wherein an excitation element comprises a system recommendation comprising at least a neurocognitive exercise to increase performance in a cognitive or behavioral domain (These models output subject profile information represent clinical concepts or diagnoses, such as cognitive distortion [0020]) and
a recommendation in a work or career domain; transmit the excitation element to the system (a system might be implemented in a mental health application that relies on cognitive behavioural therapy [0444]);
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Jain and Chen to incorporate the teachings of Rollwage for the benefit of chatbots or voice-based agents producing text and dialogue that is human-like that has long been a challenge in artificial intelligence (Rollwage [0003])
Regarding claim 2, Modified Jain teaches the apparatus of claim 1, Jain teaches wherein the representation generator comprises a machine learning model (the data generation platform can provide the prompt to the selected model (e.g., LLM) for generation of the requested output, (col. 6, 7-9); output generation request with a first performance criterion associated with the first LLM of a plurality of LLMs (col. 26, ln 8-9); machine learning models (e.g., LLMs), col. 31, ln 8. The Examiner notes the first LLM is a representation generator which is a machine learning model).
Regarding claim 3, Modified Jain teaches the apparatus of claim 2, Jain teaches wherein the machine learning model is configured to: identify the focus area for the system to optimize (In some implementations, the data generation platform 102 can receive user input that selects particular performance metrics (e.g., in an order of priority) for determination of corresponding values. For example, the data generation platform 102 can determine the composite performance metric value based on weights that correspond to the order of priority, col. 21, ln 21-27);
prioritize the focus area (Accordingly, the data generation platform 102 enables the prioritization of relevant performance metrics (e.g., cost) over other metrics (e.g., memory usage) according to system requirements (col. 31, ln 43-46); and
generate the representation dataset that further examines the focus area (modify the output (e.g., by resubmitting the output to the LLM) to modify the sentiment associated with the output, col. 27, ln 5-7); For example, the data generation platform 102 can determine the composite performance metric value based on weights that correspond to the order of priority, col. 21, ln 21-27)
Regarding claim 6, Modified Jain teaches the apparatus of claim 1, Jain teaches wherein the excitation model comprises a large language model (For example, the data generation platform can redirect a prompt to a second LLM (e.g., distinct from the first LLM (col. 29, ln 18-20)).
Regarding claim 8, Modified Jain teaches the apparatus of claim 1, Jain teaches wherein the excitation model further comprises a neural network (An LLM can include an artificial neural network, col. 21 ln 48-49).
Regarding claim 11, claim 11 is similar to claim 1. It is rejected in the same manner and reasoning applying.
Regarding claim 12, claim 12 is similar to claim 2. It is rejected in the same manner and reasoning applying.
Regarding claim 13, claim 13 is similar to claim 3. It is rejected in the same manner and reasoning applying.
Regarding claim 16, claim 16 is similar to claim 6. It is rejected in the same manner and reasoning applying.
Regarding claim 18, claim 18 is similar to claim 8. It is rejected in the same manner and reasoning applying.
4. Claims 4, 5, 7, 9, 10, 14, 15, 17, 19 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Jain et al. (US12111747 filed 05/10/2024) in view of Chen et
al. (US20250077792 filed 08/31/2023) in view of Rollwage et al. (US20240404514 filed 08/30/2023) and further in view of Kelly et al. (US20240289560 filed 02/27/2024)
Regarding claim 4, Modified Jain teaches the apparatus of claim 2, Kelly teaches wherein the machine learning model is iteratively trained on a plurality of datasets as a function of the representation dataset (In some examples, the machine learning classifier model 410 may be trained, using one or more supervisory training techniques (e.g., backpropagation of errors, etc.) to assign a contextual classification based on a plurality of historical contextual classifications respectively assigned to a plurality of historical text documents [0119]. The Examiner notes backpropagation which involves iteratively tuning their internal parameters to minimize prediction error).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Jain, Chen and Rollwage to incorporate the teachings of Kelly for the benefit of the Large Language Model (LLM) which may be iteratively retrained to continuously and automatically improve without a human in the loop (Kelly [0091]).
Regarding claim 5, Modified Jain teaches the apparatus of claim 1, Kelly teaches wherein collecting the first dataset comprises receiving information from a generative data model (In some embodiments, an initial document subset 402 is identified, from a document data store 404, and for a generative text request 406 [0098]; For example, a generative text request may be generated and/or provided to a backend service, such as a generative service [0148]; In some examples, the generative service 606 may include a third-party service, such as OpenAI [0158]. The Examiner notes initial document subset 402 as first dataset receives generative text request 406, (Fig. 4) from generative data model such OpenAI).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Modified Jain to incorporate the teachings of Kelly for the benefit of the Large Language Model (LLM) which may be iteratively retrained to continuously and automatically improve without a human in the loop (Kelly [0091]).
Regarding claim 7, Modified Jain teaches the apparatus of claim 6, Kelly teaches wherein the large language model comprises a generative pretrained transformer (The LLM 420 may include any type of LLM, such as a generative pre-trained transformer, and/or the like [0133]).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Modified Jain to incorporate the teachings of Kelly for the benefit of the Large Language Model (LLM) which may be iteratively retrained to continuously and automatically improve without a human in the loop (Kelly [0091]).
Regarding claim 9, Modified Jain teaches the apparatus of claim 1, Kelly teaches wherein the excitation element is presented to the system through a graphical user interface, wherein the graphical user interface is configured to display a data structure to the system using a display device (For instance, the generative text output, using some of the techniques of the present disclosure, may trigger the performance of actions at a client device, such as the display, transmission, and/or the like of data reflective of generative text [0182]).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Modified Jain to incorporate the teachings of Kelly for the benefit of the Large Language Model (LLM) which may be iteratively retrained to continuously and automatically improve without a human in the loop (Kelly [0091]).
Regarding claim 10, Modified Jain teaches the apparatus of claim 9, Kelly teaches wherein the graphical user interface comprises a plurality of visual elements (the user interface may be a user application, browser, user interface, and/or similar words used herein interchangeably executing on and/or accessible via the client computing entity 102 to interact with and/or cause display of information/data from the computing entity 200 [0046]; The external computing entities 108, for example, may include and/or be associated with one or more entities that may be configured to receive, transmit, store, manage, and/or facilitate datasets, such as the document data store [0030])
associated with a plurality of event handlers (the term “document data store” refers to a data structure that describes data associated with controlled text document domain [0059]; In addition, or alternatively, the generative service may include a remote service that is implemented by a remote computing system [0158]; The Examiner notes “document data store” refers to a data structure that can be remotely operated in response to a user interaction).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have modified the method of Modified Jain to incorporate the teachings of Kelly for the benefit of the Large Language Model (LLM) which may be iteratively retrained to continuously and automatically improve without a human in the loop (Kelly [0091]).
Regarding claim 14, claim 14 is similar to claim 4. It is rejected in the same manner and reasoning applying.
Regarding claim 15, claim 15 is similar to claim 5. It is rejected in the same manner and reasoning applying.
Regarding claim 17, claim 17 is similar to claim 7. It is rejected in the same manner and reasoning applying.
Regarding claim 19, claim 19 is similar to claim 9. It is rejected in the same manner and reasoning applying.
Regarding claim 20, claim 20 is similar to claim 10. It is rejected in the same manner and reasoning applying.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MORIAM MOSUNMOLA GODO whose telephone number is (571)272-8670. The examiner can normally be reached Monday-Friday 8am-5pm EST.
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, Michelle T Bechtold can be reached on (571) 431-0762. 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.
/M.G./Examiner, Art Unit 2148
/MICHELLE T BECHTOLD/Supervisory Patent Examiner, Art Unit 2148
Prosecution Insights