DETAILED ACTION
This action is in response to the claims filed 05/14/2024 for Application number 18/663,200. Claims 1-20 are currently pending.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 02/06/2025 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 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding claim 1,
Step 1 Analysis: Claim 1 is directed to a process, which falls within one of the four statutory categories.
Step 2A Prong 1 Analysis: Claim 1 recites, in part, The limitations of:
serializing the tabular data into a natural language string in a natural language format to be an evaluation in the human mind,
combining the natural language string and a prompt [as an input to a pretrained large language model (LLM)] to generate a predicted result, [wherein a set of learned vectors are added into the pretrained LLM for fine-tuning the pre-trained LLM] can be considered to be an evaluation in the human mind
generating, in response to the request for the prediction task, a prediction result for the test tabular data [using the fine-tuned LLM.]
These limitations as drafted, are processes that, under broadest reasonable interpretation, covers performance of the limitation in the mind or with the aid of pen and paper which falls within the “Mental Processes” grouping of abstract ideas.
fine-tuning the pretrained LLM using a differential privacy stochastic gradient descent (SGD) process, wherein fine-tuning the pretrained LLM comprises: determining values of the learned vectors that minimize a difference between the predicted result and the ground truth can be considered to be a mathematical calculation.
These limitations as drafted, are processes that, under broadest reasonable interpretation, covers mathematical calculations which falls within the “Mathematical concepts” grouping of abstract ideas.
Accordingly, the claim recites an abstract idea.
Step 2A Prong 2 Analysis: This judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements – “by one or more computing devices”, “an input to a pretrained large language model (LLM), wherein a set of learned vectors are added into the pretrained LLM for fine-tuning the pre-trained LLM” and “using the fine-tuned LLM”. Thus, these elements in the claim are recited at a high level of generality such that they amount to no more than mere instructions to apply the exception using a generic computer component. Please see MPEP 2106.05(f). Accordingly, these additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea.
The claim further recites: receiving, by one or more computing devices, tabular data and receiving a request including test tabular data for a predication task; This limitation is a mere data gathering step and thus is an insignificant extra-solution activity. Accordingly, these additional elements do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. The claim as a whole is directed to an abstract idea.
Step 2B Analysis: The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to integration of the abstract idea into a practical application, the additional elements of utilizing one or more computing devices, a pretrained large language model and a fine tuned LLM to perform the steps of the claimed process amount to no more than mere instructions to apply the exception using a generic computer component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept. Furthermore, the limitations of receiving, by one or more computing devices, tabular data and receiving a request including test tabular data for a predication task are well-understood, routine, and conventional, as evidenced by MPEP §2106.05(d)(II)(I), “receiving or transmitting data over a network”. These limitations therefore remain insignificant extra-solution activity even upon reconsideration, and does not amount to significantly more. Even when considered in combination, these additional elements amount to mere instructions to apply the exception using generic computer components and insignificant extra-solution activity, which cannot provide an inventive concept. The claim is not patent eligible.
Regarding claim 2, the rejection of claim 1 is further incorporated, and further, the claim recites: converting a function of each learned vector into a linear function; determining a loss function with respect to the learned vector using the linear function of the learned vector; and determining the values of the learned vector that minimize the loss function in an iterative stochastic gradient descent process with differential privacy. This claim recites additional mathematical steps in addition to the judicial exception identified in the rejection of claim 1, thus recites a judicial exception.
The claim does not include any additional elements that amount to an integration of the judicial exceptions into a practical application, nor to significantly more than the judicial exceptions. The claim is not patent eligible.
Regarding claim 3, the rejection of claim 2 is further incorporated, and further, the claim recites: determining a gradient for the loss function; obtaining a masked gradient by adding noise into the gradient; and determining the values of the learned vectors based on the masked gradient, wherein the iterative stochastic gradient descent process is terminated if the values of the learned vectors minimize the loss function. This claim recites additional mathematical steps in addition to the judicial exception identified in the rejection of claim 1, thus recites a judicial exception.
The claim does not include any additional elements that amount to an integration of the judicial exceptions into a practical application, nor to significantly more than the judicial exceptions. The claim is not patent eligible.
Regarding claim 4, the rejection of claim 3 is further incorporated, and further, the claim recites: determining an amount of the noise to add to the gradient based on noise budget parameters, wherein the noise budget parameters comprise a privacy loss parameter and a leakage probability parameter. This claim recites additional mental steps in addition to the judicial exception identified in the rejection of claim 1, thus recites a judicial exception.
The claim does not include any additional elements that amount to an integration of the judicial exceptions into a practical application, nor to significantly more than the judicial exceptions. The claim is not patent eligible.
Regarding claim 5, the rejection of claim 3 is further incorporated, and further, the claim recites: clipping the gradient based on a clipping threshold; and adding the noise into the clipped gradient to obtain the masked gradient. This claim recites additional mathematical steps in addition to the judicial exception identified in the rejection of claim 1, thus recites a judicial exception.
The claim does not include any additional elements that amount to an integration of the judicial exceptions into a practical application, nor to significantly more than the judicial exceptions. The claim is not patent eligible.
Regarding claim 6, the rejection of claim 2 is further incorporated, and further, the claim recites: wherein converting a function of each learned vector into a linear function comprises: converting an element-wise multiplication into a linear function. This claim recites additional mathematical steps in addition to the judicial exception identified in the rejection of claim 1, thus recites a judicial exception.
The claim does not include any additional elements that amount to an integration of the judicial exceptions into a practical application, nor to significantly more than the judicial exceptions. The claim is not patent eligible.
Regarding claim 7, the rejection of claim 6 is further incorporated, and further, the claim recites: wherein converting the element-wise multiplication comprises: converting the learned vector into a diagonal matrix. This claim recites additional mathematical steps in addition to the judicial exception identified in the rejection of claim 1, thus recites a judicial exception.
The claim does not include any additional elements that amount to an integration of the judicial exceptions into a practical application, nor to significantly more than the judicial exceptions. The claim is not patent eligible.
Regarding claim 8, the rejection of claim 1 is further incorporated, and further, the claim recites: the tabular data corresponds to user profile data for a social media platform and wherein the prediction results include a recommendation of content to provide to the user. This limitation amounts to generally linking the judicial exception to a field of use or technological environment. Please see MPEP 2106.05(h).
The claim does not include any additional elements that amount to an integration of the judicial exception into a practical application, nor to significantly more than the judicial exception. The claim is not patent eligible.
Regarding Claims 9-16, they recite features similar to claims 1-8 and are rejected for at least the same reasons therein.
Regarding Claims 17-20, they recite features similar to claims 1-4 and are rejected for at least the same reasons therein.
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)(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.
Claims 1, 9 and 17 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Edemacu et al. ("Privacy Preserving Prompt Engineering: A Survey", a newer version was cited in the IDS filed 02/06/2025 however examiner relies on an earlier version as the provided version was published on the same date as the EFD, hereinafter "Edemacu").
Regarding claim 1, Edemacu teaches A computer-implemented method comprising:
receiving, by one or more computing devices, tabular data (“The capabilities of LLMs have been extended to include tabular data analysis, leveraging the principles of ICL and prompt tuning” [pg. 9, § B.4]);
serializing the tabular data into a natural language string in a natural language format (“In performing this task, the common practice involves first serializing the tabular data into text before using them to prompt the LLM.” [pg. 9, § B.4]);
combining the natural language string and a prompt as an input to a pretrained large language model (LLM) to generate a predicted result (“Similarly, concatenating the serialized texts with a query to facilitate ICL. The top left part of Figure 4 depicts the private demonstration example generation in LDP-TabICL” [pg. 9, § B.4]), wherein a set of learned vectors are added into the pretrained LLM for fine-tuning the pre-trained LLM (“A continuous prompt consists of a continuous set of task specific embeddings. (corresponds to a set of “learned vectors”) Note that all discrete input tokens to LLMs are internally transformed into continuous input embed dings that the LLM then processes.” [pg. 3, left col]);
fine-tuning the pretrained LLM using a differential privacy stochastic gradient descent (SGD) process (“The procedure of deep learning model training is to minimize the output of a loss function through numerous stochastic gradient descent (SGD) steps. DP-SGD uses a clipping bound on l2 norm of the gradient from an individual input, aggregates the clipped updates, and then adds Gaussian noise to the aggregate.” [pg. 5, bottom left col – top right col]), wherein fine-tuning the pretrained LLM comprises:
determining values of the learned vectors that minimize a difference between the predicted result and the ground truth (“The procedure of deep learning model training is to minimize the output of a loss function through numerous stochastic gradient descent (SGD) steps.” [pg. 5, bottom left col; minimizing the loss would correspond to “minimizing a difference between the predicted result and ground truth”]);
receiving a request including test tabular data for a predication task (“…then serialized into text-based demonstration examples along with a test query to facilitate ICL.” [pg. 12, bottom left col])); and
generating, in response to the request for the prediction task, a prediction result for the test tabular data using the fine-tuned LLM (“In this manner, the soft prompts are optimized to capture task-specific details, enabling them to be used to prompt the pre-trained model to generate suitable responses tailored to the task at hand.” [pg. 13, top right col, see Fig. 4]).
Claim 9 recites features similar to claim 1 and is rejected for at least the same reasons therein. Claim 9 additionally requires A non-transitory computer-readable medium encoded with instructions that, when executed by one or more computers, cause the one or more computers to perform operations comprising:… (Edemacu, “Leveraging parallel computing technologies such as GPU and FPGA can speed up their execution. Furthermore, computing architectures such as in-memory and in-storage can be explored to boost the speed” [pg. 18, top left col, implies use of computers/processors])
Claim 17 recites features similar to claim 1 and is rejected for at least the same reasons therein. Claim 17 additionally requires A system comprising one or more computers and one or more storage devices on which are stored instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations:… (Edemacu, “Leveraging parallel computing technologies such as GPU and FPGA can speed up their execution. Furthermore, computing architectures such as in-memory and in-storage can be explored to boost the speed” [pg. 18, top left col, implies use of computers/processors])
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 2-7, 10-15 and 18-20 are rejected under 35 U.S.C. 103 as being unpatentable over Edemacu in view of Liu et al. ("Few-Shot Parameter-Efficient Fine-Tuning is Better and Cheaper than In-Context Learning", hereinafter "Liu").
Regarding claim 2, Edemacu teaches The computer-implemented method of claim 1, wherein fine-tuning the pretrained LLM comprises:
determining a loss function with respect to the learned vector using the linear function of the learned vector (“A soft prompt is prepended to the private data embeddings. This combination is forwarded to an LLM. The LLM makes predictions ˆy and computes loss accordingly. Next, LLM uses the loss to compute the gradients associated with the soft prompt” [pg. 12, Fig. 7 caption]); and
determining the values of the learned vector that minimize the loss function in an iterative stochastic gradient descent process with differential privacy. (“PromptDPSGP that leverages the DP-SGD algorithm [59] to learn soft prompts that are prepended to an LLM’s input with a differential privacy guarantee. An illustration of the process is shown in Figure 7.” [pg. 12, top right col])
However, fails to explicitly teach converting a function of each learned vector into a linear function;
Liu teaches converting a function of each learned vector into a linear function; (“Left: (IA)3 introduces the learned vectors lk,lv, and lff which respectively rescale (via element-wise multiplication, visualized as ) the keys and values in attention mechanisms and the inner activations in position-wise feed-forward networks.” [Figure 1 caption])
It would have been obvious to one of ordinary skill in the art before the effective filing date to modify Edemacu’s teachings by converting a function of each learned vector into a linear function as taught by Liu. One would have been motivated to make this modification in order to scale activations by learned vectors to attain stronger performance. [Abstract, Liu]
Regarding claim 3, Edemacu/Liu teaches The computer-implemented method of claim 2, Edemacu teaches wherein the iterative stochastic gradient descent process with differential privacy comprises:
determining a gradient for the loss function (“DP-SGD uses a clipping bound on l2 norm of the gradient from an individual input” [pg. 5, bottom left col – top right col]);
obtaining a masked gradient by adding noise into the gradient (“For private training, DP noise is added to the gradients associated with these embeddings.” [pg. 12, right col]); and
determining the values of the learned vectors based on the masked gradient, wherein the iterative stochastic gradient descent process is terminated if the values of the learned vectors minimize the loss function. (“The procedure of deep learning model training is to minimize the output of a loss function through numerous stochastic gradient descent (SGD) steps. DP-SGD uses a clipping bound on l2 norm of the gradient from an individual input, aggregates the clipped updates, and then adds Gaussian noise to the aggregate. The clipping truncation controls the sensitivity of the sum of gradients as the sensitivity of gradients and the scale of the noise would otherwise be unbounded.” [pg. 5, bottom left col – top right col])
Regarding claim 4, Edemacu/Liu teaches The computer-implemented method of claim 3, Edemacu teaches further comprising:
determining an amount of the noise to add to the gradient based on noise budget parameters, wherein the noise budget parameters comprise a privacy loss parameter and a leakage probability parameter (“To reduce the effect of noise, the algorithm limits the vocabulary to the tokens present in top-K indices of the next-token probability coming from only the instruction without any private data.” [pg. 10, top right col; See further pg. 12, D. Other Scenarios, ¶1; “An alternative point of leakage is through model outputs. An adversary can infer information about the input data from model outputs. A solution can be achieved through a noisy consensus among an ensemble of an LLM’s responses.”])).
Regarding claim 5, Edemacu/Liu teaches The computer-implemented method of claim 3, Edemacu teaches wherein adding noise to the gradient comprises:
clipping the gradient based on a clipping threshold; and adding the noise into the clipped gradient to obtain the masked gradient. (“Next, LLM uses the loss to compute the gradients associated with the soft prompt. These gradients are then clipped, and noise is added to them before being used to update the soft prompt” [Fig. 7 caption; See further: “The clipping truncation controls the sensitivity of the sum of gradients as the sensitivity of gradients and the scale of the noise would otherwise be unbounded.” [pg. 5, top right col]])
Regarding claim 6, Edemacu/Liu teaches The computer-implemented method of claim 2, Liu teaches wherein converting a function of each learned vector into a linear function comprises: converting an element-wise multiplication into a linear function (“Left: (IA)3 introduces the learned vectors lk,lv, and lff which respectively rescale (via element-wise multiplication, visualized as ) the keys and values in attention mechanisms and the inner activations in position-wise feed-forward networks.” [Figure 1 caption]).
Same motivation to combine the teachings of Edemacu/Liu as claim 2.
Regarding claim 7, Edemacu/Liu teaches The computer-implemented method of claim 6, Liu teaches wherein converting the element-wise multiplication comprises: converting the learned vector into a diagonal matrix. (“We also note that, in the event that a model will only be used on a single task, the modifications introduced by (IA)3 can also be applied to weight matrices permanently so that no elementwise multiplication is required and the model’s architecture remains unchanged. This possible because element-wise multiplications performed in (IA)3 always co-occur with a matrix multiplication, and l Wx = (l W)x.” [pg. 6, ¶3])
Same motivation to combine the teachings of Edemacu/Liu as claim 2.
Regarding claims 10-15, they are substantially similar to claims 2-7 respectively, and are rejected in the same manner, the same art, and reasoning applying.
Regarding claims 18-20, they are substantially similar to claims 2-4 respectively, and are rejected in the same manner, the same art, and reasoning applying.
Claims 8 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Edemacu in view of Tan et al. ("User Modeling in the Era of Large Language Models: Current Research and Future Direction", hereinafter "Tan").
Regarding claim 8, Edemacu teaches The computer-implemented method of claim 1, however fails to explicitly teach the tabular data corresponds to user profile data for a social media platform and wherein the prediction results include a recommendation of content to provide to the user.
Tan teaches the tabular data corresponds to user profile data for a social media platform (“User modeling (UM) aims to discover patterns or learn representations from user data about the characteristics of a specific user, such as profile, preference, and personality.” [Abstract; See further Introduction “social networks”]) and wherein the prediction results include a recommendation of content to provide to the user. (“The user models enable personalization and suspiciousness detection in many online applications such as recommendation, education, and healthcare.” [Abstract])
It would have been obvious to one of ordinary skill in the art before the effective filing date to modify Edemacu’s teachings in order to use user profile data for a social media platform as taught by Tan. One would have been motivated to make this modification as “LLMs have shown superior performance on generating, understanding, and even reasoning over text data. The approaches of user modeling have been equipped with LLMs and soon become outstanding. This article summarizes existing research about how and why LLMs are great tools of modeling and understanding UGC.” [Abstract, Tan]
Regarding claim 16, it is substantially similar to claim 8 respectively, and is rejected in the same manner, the same art, and reasoning applying.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Fang et al. (“Large Language Models(LLMs) on Tabular Data: Prediction, Generation, and Understanding- A Survey”) discloses various techniques involving the use of tabular data with LLMs.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MICHAEL H HOANG whose telephone number is (571)272-8491. The examiner can normally be reached Mon-Fri 8:30AM-4:30PM.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Kakali Chaki can be reached at (571) 272-3719. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/MICHAEL H HOANG/PRIMARY EXAMINER, Art Unit 2122
Prosecution Insights