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 .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 11/03/2025 has been entered.
Information Disclosure Statement
The examiner has considered the information disclosure statements (IDS) submitted on 11/12/2025 and 11/17/2025.
Response to Amendment
The amendment filed 10/03/2025 has been entered. Claims 1-20 remain pending in the application.
Response to Arguments
Applicant’s arguments, filed 10/03/2025, with respect to the rejections of claims 1 and 13-14 under 103 have been fully considered and are persuasive because of the amendments. Therefore, the rejections have been withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of Rai et al. (US Pub. 2021/0216683) in view of Zoldi et al. (US. Pub. 2019/0354853).
Applicant’s arguments, filed 10/03/2025, with respect to the rejections of the claims under 101 have been fully considered and are not persuasive.
Applicant argues (pages 7-10)
Technical Improvements
Applicant notes that the claimed embodiments provide numerous technical benefits, including, for example (see paragraphs [0017]-[0022] of the original application):
representations for modeling real-valued combinatorial functions over categorical variables;
techniques for learning a surrogate model for the generic Black Box Optimization (BBO) problem;
techniques for optimizing black box functions;
methods for using a version of simulated annealing that utilizes a surrogate model for internal cost-free evaluations before producing the next black box query;
methods for using a version of Monte Carlo tree search (MCTS) in conjunction with a surrogate model as a reward function of the terminal states during intermediate tree traversals/backups in order to improve the sample efficiency of the search algorithm; and
biological (RNA) sequence optimization with competitive or superior performance for the disclosed methods over state-of-the-art counterparts, while substantially reducing the computation time and sample efficiency, respectively.
The claimed embodiments provide the technical improvements that solve these technical issues, as described below.
Step 2A Prong 2: Even if Claim 1 is not 101-compliant at Step 2A Prong 1, it is respectfully asserted that it is clearly 101-compliant at Step 2A, Prong 2. As understood, the Examiner alleges that the alleged judicial exception is not integrated into a practical application.2 Applicant respectfully disagrees.
The limitations of the independent claims are indicative of integration of the respective improvements into a practical application at least because the limitations are directed to improvements in the relevant technical field (computerized machine learning) and are patentable subject matter under the 2-part rubric of MPEP 2106.04(d)(l). The improvements are set forth in the specification (e.g., 0017-0022).
In response
The claim limitations of the current Application recite mathematical concepts, and can be performed mentally or with the aid of pencil and paper, as was discussed above in the 101 section below. The Applicant argues that “the limitations are directed to improvements in the relevant technical field (computerized machine learning)”. The examiner respectfully disagrees.
The claim recites a concept of using a generated surrogate machine learning model to predict the optimization value for the black box evaluator/function, and analyzing the black box evaluator/function with newly selected sample data to optimize the black box evaluator/function. There is no improvement on the surrogate machine learning model itself. The surrogate machine learning model does not improve, the claim does not recite how the surrogate model is operated to get improvement, the surrogate machine learning model is only used as a tool to perform the executing process using certain data such as “new candidate point for the categorical values” to maybe improve the optimization process. However, an improvement to the abstract idea is not consider an improvement on the functioning of a computer or to any other technology or technical field.
Thus, at Step 2A, prong two, even when considering the elements in combination, the claim as a whole does not integrate the recited exception into a practical application.
Applicant further argues
Furthermore, the independent claims include the components or steps of the invention that provide, for example, the improvements to techniques for learning a surrogate model for a generic Black Box Optimization (BBO) problem; and techniques for optimizing black box functions, as achieved by the steps of:
accessing, by a computing device, a black box evaluator;
generating, by the computing device, a surrogate machine learning model that provides estimates for the optimization of categorical values for the black box evaluator, the surrogate machine learning model being based upon observations from previous executions of the black box evaluator;
optimizing the black box evaluator by selecting, by an acquisition function executing on the computing device, a new candidate point for the categorical values; and
executing, by the computing device, the black box evaluator with the new candidate point for the categorical values, wherein data values are represented using a Fourier expansion, where characters of each representation are considered as experts and respective coefficients of the characters are updated via an exponential weight update rule each time the black box evaluator is executed.
In the Response the Arguments, the Examiner asserted that "an improvement in an abstract idea itself is not an improvement in technology; The claims must recite additional elements which provide the improvement."3
Respectfully, improvements to computerized machine learning are patent eligible under Section 101, as recently recognized by the PTAB in Ex parte Desjardins, Appeal 2024-000567, Decision on Request for Rehearing, BP AI September 26, 2025. Applicant notes that the independent claims recite "the surrogate machine learning model being based upon observations from previous executions of the black box evaluator" thereby providing the cited improvement of "the surrogate machine learning model being based upon observations from previous executions of the black box evaluator" and recite "executing, by the computing device, the black box evaluator with the new candidate point for the categorical values, wherein data values are represented using a Fourier expansion, where characters of each representation are considered as experts and respective coefficients of the characters are updated via an exponential weight update rule each time the black box evaluator is executed' thereby providing the cited improvement of "techniques for optimizing black box functions."
Thus, under the 2-part rubric of MPEP 2106.04(d)(l), even if reciting a judicial exception, the present claims are not directed to a judicial exception, since the claims as a whole integrate the recited judicial exception into a practical application of that exception.
In response
The claim does not provide any steps that the improvements to techniques for learning a surrogate model for a generic Black Box Optimization (BBO) problem. As mentioned above, the surrogate machine learning model is only used as a tool to perform the executing process using certain data such as “new candidate point for the categorical values” to maybe improve the abstract idea of “optimization process”. Further, a process of selecting new data to analyze a model/function based on the model/function output to improve the output does not provide an improvement to the surrogate machine learning model. Wherein, the limitation of “data values are represented using a Fourier expansion, where characters of each representation are considered as experts” amounts to linking the use of a judicial exception to a particular technological environment or field of use which does not integrate the exception into a practical application (see MPEP 2106.05(h)).
The Applicant also argues that the independent claims include the components or steps of the invention that provide, for example, the improvements to techniques for optimizing black box functions. Since the process of “optimizing black box functions” recites a mental process as stated in the 101 rejections section below, it is clearly that the improvement is on the abstract idea of “optimizing black box functions”.
Thus, at Step 2A, prong two, even when considering the elements in combination, the claim as a whole does not integrate the recited exception into a practical application.
Applicant then argues
Claim 5, for example, recites, in part, "utilizing the black box evaluator to generate one or more candidate biological molecule sequences that have desirable properties, which are constructed using a vocabulary of fixed size, and synthesizing the optimal biological molecule sequence" and is evidence that the alleged judicial exception is integrated the recited judicial exception into a practical application.
Thus, even if reciting a judicial exception, the present claims are not directed to a judicial exception, since the claims as a whole integrate the recited judicial exception into a practical application of that exception. In particular, according to MPEP 2106.04(d)(l), "the 'improvements' analysis in Step 2A determines whether the claim pertains to an improvement to the functioning of a computer or to another technology without reference to what is well-understood, routine, conventional activity. That is, the claimed invention may integrate the judicial exception into a practical application by demonstrating that it improves the relevant existing technology although it may not be an improvement over well-understood, routine, conventional activity." In the present case, the operations outlined above improve the relevant existing technology of computerized machine learning using novel techniques. Note, respectfully, applicant does not admit that there is anything routine or conventional about the limitations of Claim 1; applicant merely points out that even if that were the case, Section 101 is still satisfied at Step 2A prong 2.
For the above reasons, it is respectfully requested that the rejection of claims 1-20 under 35 U.S.C. § 101 be withdrawn.
In response
Claim 5 limitation does not integrate the judicial exception into a practical application. The limitation of “utilizing the black box evaluator to generate one or more candidate biological molecule sequences that have desirable properties, which are constructed using a vocabulary of fixed size, and synthesizing the optimal biological molecule sequence" amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception do not integrate into a practical application (see MPEP 2106.05(h)).
Also, limitation of “utilizing the black box evaluator to generate …” is recited at a high-level of generality (i.e., as a generic device performing the generic computer function) such that it amounts no more than mere instructions to apply the exception using the generic computer components (MPEP 2106.05(f)). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
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 a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more.
Claim 1 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites a method which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 1):
The limitation of “generating, by the computing device, a surrogate machine learning model that provides estimates for the optimization of categorical values for the black box evaluator, the surrogate machine learning model being based upon observations from previous executions of the black box evaluator”, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind. That is, other than reciting “by the computing device” nothing in the claim element precludes the step from practically being performed in the mind. For example, “generating” in the context of this claim encompasses the user determining the values for a function base on the previous output of the function. This limitation also recites a mathematical concept.
The limitation of “optimizing the black box evaluator by selecting, by an acquisition function executing on the computing device, a new candidate point for the categorical values”, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind. That is, other than reciting “by an acquisition function executing on the computing device” nothing in the claim element precludes the step from practically being performed in the mind. For example, “optimizing” in the context of this claim encompasses the user selecting a new sample data for the evaluator/function. This limitation also recites a mathematical concept.
The limitation of “executing, by the computing device, the black box evaluator with the new candidate point for the categorical values”, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind. That is, other than reciting “by the computing device” nothing in the claim element precludes the step from practically being performed in the mind. For example, “executing” in the context of this claim encompasses the user analyzing the evaluator/function using the newly selected sample data. This limitation also recites a mathematical concept.
The limitation of “respective coefficients of the characters are updated via an exponential weight update rule each time the black box evaluator is executed” recites a mathematical concept.
Step 2A (prong 2):
This judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of “a computing device” and “a surrogate machine learning model”. The additional elements are recited at a high-level of generality (i.e., as a generic device performing the generic computer functions of generating and executing) such that they amount no more than mere instructions to apply the exception using the generic computer components (MPEP 2106.05(f)). Accordingly, these additional elements do not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
The claim recites the additional element of “accessing, by a computing device, a black box evaluator”. This additional element is recited at a high-level of generality (i.e., as a generic device performing the generic computer functions) such that it amounts no more than mere instructions to apply the exception using the generic computer components (MPEP 2106.05(f)). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
The additional elements of “wherein data values are represented using a Fourier expansion, where characters of each representation are considered as experts” amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not integrate into a practical application (see MPEP 2106.05(h)).
Step 2B:
The claim does 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 “a computing device” and “a surrogate machine learning model” to perform the “generating, optimizing and executing” steps 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.
The additional element of “accessing, by a computing device, a black box evaluator” amounts 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.
The additional elements of “wherein data values are represented using a Fourier expansion, where characters of each representation are considered as experts” amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the judicial exception (see MPEP 2106.05(h)).
Claim 2 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites the method which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 2):
This judicial exception is not integrated into a practical application. The claim recites an additional element of “wherein the data values are represented using a group-theoretic Fourier expansion”. This limitation amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception does not integrate into a practical application (see MPEP 2106.05(h)).
Step 2B:
The claim does 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 element of “wherein the data values are represented using a group-theoretic Fourier expansion” amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the judicial exception (see MPEP 2106.05(h)).
Claim 3 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites the method which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 2):
This judicial exception is not integrated into a practical application. The claim recites an additional element of “wherein the data values are represented using an abridged one-hot encoded Boolean Fourier expansion”. This limitation amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception does not integrate into a practical application (see MPEP 2106.05(h)).
Step 2B:
The claim does 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 element of “wherein the data values are represented using an abridged one-hot encoded Boolean Fourier expansion” amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the judicial exception (see MPEP 2106.05(h)).
Claim 4 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites the method which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 2):
This judicial exception is not integrated into a practical application. The claim recites an additional element of “wherein a one-hot encoding of each variable
PNG
media_image1.png
20
74
media_image1.png
Greyscale
is expressed as a (k - 1)-tuple
PNG
media_image2.png
24
142
media_image2.png
Greyscale
where
PNG
media_image3.png
22
92
media_image3.png
Greyscale
are Boolean variables with a constraint that at most one such variable is equal to -1 for any given
PNG
media_image4.png
20
58
media_image4.png
Greyscale
”. This limitation amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception does not integrate into a practical application (see MPEP 2106.05(h)).
Step 2B:
The claim does 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 element of “wherein a one-hot encoding of each variable
PNG
media_image1.png
20
74
media_image1.png
Greyscale
is expressed as a (k - 1)-tuple
PNG
media_image2.png
24
142
media_image2.png
Greyscale
where
PNG
media_image3.png
22
92
media_image3.png
Greyscale
are Boolean variables with a constraint that at most one such variable is equal to -1 for any given
PNG
media_image4.png
20
58
media_image4.png
Greyscale
” amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the judicial exception (see MPEP 2106.05(h)).
Claim 5 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites the method which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 2):
This judicial exception is not integrated into a practical application.
The claim recites the additional element of “utilizing the black box evaluator to generate one or more candidate biological molecule sequences that have desirable properties, which are constructed using a vocabulary of fixed size, and synthesizing the optimal biological molecule sequence”. This additional element is recited at a high-level of generality (i.e., as a generic device performing the generic computer function) such that it amounts no more than mere instructions to apply the exception using the generic computer components (MPEP 2106.05(f)). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
Step 2B:
The claim does 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 element of “utilizing the black box evaluator to generate one or more candidate biological molecule sequences that have desirable properties, which are constructed using a vocabulary of fixed size, and synthesizing the optimal biological molecule sequence” amounts 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.
Claim 6 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites the method which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 2):
This judicial exception is not integrated into a practical application.
The claim recites the additional element of “utilizing the black box evaluator to design optimal sequences over a combinatorially large search space”. This additional element is recited at a high-level of generality (i.e., as a generic device performing the generic computer function) such that it amounts no more than mere instructions to apply the exception using the generic computer components (MPEP 2106.05(f)). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
Step 2B:
The claim does 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 element of “utilizing the black box evaluator to design optimal sequences over a combinatorially large search space” amounts 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.
Claim 7 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites the method which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 2):
This judicial exception is not integrated into a practical application.
The claim recites the additional element of “utilizing the black box evaluator to find a sequence given a specific structure”. This additional element is recited at a high-level of generality (i.e., as a generic device performing the generic computer function) such that it amounts no more than mere instructions to apply the exception using the generic computer components (MPEP 2106.05(f)). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
Step 2B:
The claim does 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 element of “utilizing the black box evaluator to find a sequence given a specific structure” amounts 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.
Claim 8 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites the method which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 2):
This judicial exception is not integrated into a practical application. The claim recites an additional element of “the generation of the surrogate machine learning model is performed via a hedge algorithm where basis functions act as experts”. This limitation amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception does not integrate into a practical application (see MPEP 2106.05(h)).
Step 2B:
The claim does 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 element of “the generation of the surrogate machine learning model is performed via a hedge algorithm where basis functions act as experts” amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the judicial exception (see MPEP 2106.05(h)).
Claim 9 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites the method which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 2):
This judicial exception is not integrated into a practical application. The claim recites an additional element of “Monte Carlo tree search (MCTS) and the surrogate machine learning model are used as a reward function of terminal states during intermediate tree traversals and backups”. This limitation amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception does not integrate into a practical application (see MPEP 2106.05(h)).
Step 2B:
The claim does 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 element of “Monte Carlo tree search (MCTS) and the surrogate machine learning model are used as a reward function of terminal states during intermediate tree traversals and backups” amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the judicial exception (see MPEP 2106.05(h)).
Claim 10 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites the method which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 2):
This judicial exception is not integrated into a practical application.
The claim recites the additional element of “performing simulated annealing utilizing the surrogate machine learning model”. This additional element is recited at a high-level of generality (i.e., as a generic device performing the generic computer function) such that it amounts no more than mere instructions to apply the exception using the generic computer components (MPEP 2106.05(f)). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
Step 2B:
The claim does 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 element of “performing simulated annealing utilizing the surrogate machine learning model” amounts 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.
Claim 11 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites the method which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 2):
This judicial exception is not integrated into a practical application.
The claim recites the additional element of “the simulated annealing utilizing the surrogate machine learning model is performed for internal cost-free evaluations before producing a next black box query”. This additional element is recited at a high-level of generality (i.e., as a generic device performing the generic computer function) such that it amounts no more than mere instructions to apply the exception using the generic computer components (MPEP 2106.05(f)). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
Step 2B:
The claim does 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 element of “the simulated annealing utilizing the surrogate machine learning model is performed for internal cost-free evaluations before producing a next black box query” amounts 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.
Claim 12 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites the method which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 2):
This judicial exception is not integrated into a practical application. The claim recites an additional element of “the black box evaluator utilizes a black box machine learning model”. This limitation amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception does not integrate into a practical application (see MPEP 2106.05(h)).
Step 2B:
The claim does 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 element of “the black box evaluator utilizes a black box machine learning model” amounts to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitation that amounts to merely indicating a field of use or technological environment in which to apply a judicial exception does not amount to significantly more than the judicial exception (see MPEP 2106.05(h)).
Claim 13 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites a computer program product which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 1):
The limitation of “generate, by the computing device, a surrogate machine learning model that provides estimates for the optimization of categorical values for the black box evaluator, the surrogate machine learning model being based upon observations from previous executions of the black box evaluator”, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind. That is, other than reciting “by the computing device” nothing in the claim element precludes the step from practically being performed in the mind. For example, “generate” in the context of this claim encompasses the user determining the values for a function base on the previous output of the function. This limitation also recites a mathematical concept.
The limitation of “optimize the black box evaluator by selecting, by an acquisition function executing on the computing device, a new candidate point for the categorical values”, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind. That is, other than reciting “by an acquisition function executing on the computing device” nothing in the claim element precludes the step from practically being performed in the mind. For example, “optimize” in the context of this claim encompasses the user selecting a new sample data for the evaluator/function. This limitation also recites a mathematical concept.
The limitation of “execute, by the computing device, the black box evaluator with the new candidate point for the categorical values”, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind. That is, other than reciting “by the computing device” nothing in the claim element precludes the step from practically being performed in the mind. For example, “execute” in the context of this claim encompasses the user analyzing the evaluator/function using the newly selected sample data. This limitation also recites a mathematical concept.
The limitation of “respective coefficients of the characters are updated via an exponential weight update rule each time the black box evaluator is executed” recites a mathematical concept.
Step 2A (prong 2):
This judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of “A computer program product”, “a computer readable storage medium”, “a computer”, “a computing device” and “a surrogate machine learning model”. The additional elements are recited at a high-level of generality (i.e., as a generic device performing the generic computer functions of generating and executing) such that they amount no more than mere instructions to apply the exception using the generic computer components (MPEP 2106.05(f)). Accordingly, these additional elements do not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
The claim recites the additional element of “access, by a computing device, a black box evaluator”. This additional element is recited at a high-level of generality (i.e., as a generic device performing the generic computer functions) such that it amounts no more than mere instructions to apply the exception using the generic computer components (MPEP 2106.05(f)). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
The additional elements of “wherein data values are represented using a Fourier expansion, where characters of each representation are considered as experts” amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not integrate into a practical application (see MPEP 2106.05(h)).
Step 2B:
The claim does 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 “A computer program product”, “a computer readable storage medium”, “a computer”, “a computing device” and “a surrogate machine learning model” to perform the “generating, optimizing and executing” steps 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.
The additional element of “access, by a computing device, a black box evaluator” amounts 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.
The additional elements of “wherein data values are represented using a Fourier expansion, where characters of each representation are considered as experts” amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the judicial exception (see MPEP 2106.05(h)).
Claim 14 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1:
The claim recites an apparatus which falls within at least one of the four statutory categories of patent eligible subject matter.
Step 2:
Step 2A (prong 1):
The limitation of “generating, by the computing device, a surrogate machine learning model that provides estimates for the optimization of categorical values for the black box evaluator, the surrogate machine learning model being based upon observations from previous executions of the black box evaluator”, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind. That is, other than reciting “by the computing device” nothing in the claim element precludes the step from practically being performed in the mind. For example, “generating” in the context of this claim encompasses the user determining the values for a function base on the previous output of the function. This limitation also recites a mathematical concept.
The limitation of “optimizing the black box evaluator by selecting, by an acquisition function executing on the computing device, a new candidate point for the categorical values”, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind. That is, other than reciting “by an acquisition function executing on the computing device” nothing in the claim element precludes the step from practically being performed in the mind. For example, “optimizing” in the context of this claim encompasses the user selecting a new sample data for the evaluator/function. This limitation also recites a mathematical concept.
The limitation of “executing, by the computing device, the black box evaluator with the new candidate point for the categorical values”, as drafted, is a process that, under its broadest reasonable interpretation, covers performance of the limitation in the mind. That is, other than reciting “by the computing device” nothing in the claim element precludes the step from practically being performed in the mind. For example, “executing” in the context of this claim encompasses the user analyzing the evaluator/function using the newly selected sample data. This limitation also recites a mathematical concept.
The limitation of “respective coefficients of the characters are updated via an exponential weight update rule each time the black box evaluator is executed” recites a mathematical concept.
Step 2A (prong 2):
This judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements of “An apparatus”, “a memory”, “processor”, “a computing device” and “a surrogate machine learning model”. The additional elements are recited at a high-level of generality (i.e., as a generic device performing the generic computer functions of generating and executing) such that they amount no more than mere instructions to apply the exception using the generic computer components (MPEP 2106.05(f)). Accordingly, these additional elements do not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
The claim recites the additional element of “accessing, by a computing device, a black box evaluator”. This additional element is recited at a high-level of generality (i.e., as a generic device performing the generic computer functions) such that it amounts no more than mere instructions to apply the exception using the generic computer components (MPEP 2106.05(f)). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
The additional elements of “wherein data values are represented using a Fourier expansion, where characters of each representation are considered as experts” amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not integrate into a practical application (see MPEP 2106.05(h)).
Step 2B:
The claim does 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 “An apparatus”, “a memory”, “processor”, “a computing device” and “a surrogate machine learning model” to perform the “generating, optimizing and executing” steps 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.
The additional element of “accessing, by a computing device, a black box evaluator” amounts 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.
The additional elements of “wherein data values are represented using a Fourier expansion, where characters of each representation are considered as experts” amount to no more than generally linking the use of a judicial exception to a particular technological environment or field of use. As explained by the Supreme Court, a claim directed to a judicial exception cannot be made eligible "simply by having the applicant acquiesce to limiting the reach of the patent for the formula to a particular technological use." Diamond v. Diehr, 450 U.S. 175, 192 n.14, 209 USPQ 1, 10 n. 14 (1981). Thus, limitations that amount to merely indicating a field of use or technological environment in which to apply a judicial exception do not amount to significantly more than the judicial exception (see MPEP 2106.05(h)).
Claim 15: is substantially similar to claim 2 and thus rejected for similar reasons as claim 2.
Claim 16: is substantially similar to claim 3 and thus rejected for similar reasons as claim 3.
Claim 17: is substantially similar to claim 5 and thus rejected for similar reasons as claim 5.
Claim 18: is substantially similar to claim 8 and thus rejected for similar reasons as claim 8.
Claim 19: is substantially similar to claim 9 and thus rejected for similar reasons as claim 9.
Claim 20: is substantially similar to claim 11 and thus rejected for similar reasons as claim 11.
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 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.
Claims 1-2 and 13-15 are rejected under 35 U.S.C. 103 as being unpatentable over Rai et al. (US Pub. 2021/0216683) in view of Zoldi et al. (US. Pub. 2019/0354853).
As per claim 1, Rai teaches a method comprising:
accessing, by a computing device, a black box evaluator [Figs. 14A – 14E, paragraph 0114, “a surrogate model-based optimization approach was adopted, wherein the cost function is considered as a black box problem characterized by only its input and output. How surrogate model-based optimization works is illustrated in FIGS. 14A-14E. The x-axis represents the design variable and the cost function is plotted on the y-axis”];
generating, by the computing device, a surrogate machine learning model that provides estimates for the optimization of categorical values for the black box evaluator, the surrogate machine learning model being based upon observations from previous executions of the black box evaluator [paragraph 0076, “FIG. 16 shows a graph of the performance of surrogate model-based optimization for problem configuration in FIG. 13B. With surrogate model-based optimization, the best objective value can be obtained with 194 evaluations”; paragraphs 0114-0115, “a surrogate model-based optimization approach was adopted, wherein the cost function is considered as a black box problem characterized by only its input and output. How surrogate model-based optimization works is illustrated in FIGS. 14A-14E. The x-axis represents the design variable and the cost function is plotted on the y-axis”; paragraphs 0118-0120, “Surrogate model: Next, a surrogate model is mapped on the initial design space data to create an analytical function mapping the design variables to the cost function values … few candidate points are generated by small perturbations of the current minima of the surrogate model … compares the cost function values of the new sample points and assess the stopping criteria. The method iterates through the process and progressively improves the surrogate model until the stopping criteria are satisfied. Either one of them has to be satisfied to terminate the program. The first stopping criterion is satisfied if the improvement in the optimal cost function is less than 104 for 10 consecutive iterations”; It can be seen that the surrogate model keeps updating/improving based on the cost function values at each iteration];
optimizing the black box evaluator by selecting, by an acquisition function executing on the computing device, a new candidate point for the categorical values [paragraph 0115, “FIG. 14C shows new sample points (■) generated in the next iteration. FIG. 14D shows a surrogate model updated based on information from new sample points. FIG. 14E shows that, after iteratively updating and achieving convergence, an optimum solution C•) obtained”; paragraph 0119, “Selection of new samples: In this step, a new set of sample orientations are generated, and the target values are evaluated at the newly generated orientations. The surrogate model is progressively improved using the new data samples. The desired number of sample points to be selected in each iteration is selected by a user”]; and
executing, by the computing device, the black box evaluator with the new candidate point for the categorical values [paragraphs 0119-0120, “Selection of new samples: In this step, a new set of sample orientations are generated, and the target values are evaluated at the newly generated orientations. The surrogate model is progressively improved using the new data samples. The desired number of sample points to be selected in each iteration … compares the cost function values of the new sample points and assess the stopping criteria. The method iterates through the process and progressively improves the surrogate model until the stopping criteria are satisfied”], wherein data values are represented using a Fourier expansion, where characters of each representation are considered as experts and respective coefficients of the characters [paragraph 0089, “Fourier series is powerful yet simple mathematical technique. It helps establish a scheme to systematically study CPCSs and allows representation of an arbitrary periodic function as a weighted sum of cosine and sine functions. By using this technique, each periodic function has an associated distinctive set of weighted coefficients (or Fourier coefficients) that univocally represent the periodic function. In particular, three-dimensional periodic functions defined within cubic unit cells are considered with side length a. The Fourier series expansion for these periodic functions is given as follows:
PNG
media_image5.png
46
488
media_image5.png
Greyscale
PNG
media_image6.png
44
180
media_image6.png
Greyscale
where n, m, and pare integer numbers, a is the length of the edge of the cubic unit cell, and a000, anmp and bnmp are the Fourier coefficients corresponding to three-dimensional periodic functions. The distance (or period) A at which the cosine or sine functions in Equation 1 repeat in space is given by
PNG
media_image7.png
34
142
media_image7.png
Greyscale
. A parameter d=n2+m2+p2 is defined to group cosine and sine functions with the same spatial period A”].
Rai does not explicitly teach
coefficients of the characters are updated via an exponential weight update rule each time the black box evaluator is executed.
Zoldi teaches
coefficients of the characters are updated via an exponential weight update rule each time the black box evaluator is executed [paragraph 0002, “Many of these machine learning technologies are “black box” in nature”; paragraph 0026, “Each node of the first hidden layer is configured to learn complex relationships among the input variables. These latent features are “black box” in nature due to a non-linear transformation applied to a weighted combination of the input variables”; paragraphs “FIG. 3 shows a schematic representation of an arbitrary objective function C, as a function of two weights, w.sub.1 and w.sub.2. Any of a number of optimization algorithms can be used for minimizing the cost functions to identify the most optimal weights … The weights are then updated in the direction of the gradient using the following update rule”; paragraph 00003, “Often, the optimal value of λ is learned through an iterative selection process called parameter tuning. This yields the following update rule to update the weights”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include coefficients of the characters are updated via an exponential weight update rule of Zoldi. Doing so would help identifying the most optimal weights for minimizing the cost function (Zoldi, 0035).
As per claim 2, Rai and Zoldi teach the method of claim 1.
Rai further teaches
the data values are represented using a group-theoretic Fourier expansion [paragraph 0089, “Fourier series is powerful yet simple mathematical technique. It helps establish a scheme to systematically study CPCSs and allows representation of an arbitrary periodic function as a weighted sum of cosine and sine functions. By using this technique, each periodic function has an associated distinctive set of weighted coefficients (or Fourier coefficients) that univocally represent the periodic function. In particular, three-dimensional periodic functions defined within cubic unit cells are considered with side length a. The Fourier series expansion for these periodic functions is given as follows:
PNG
media_image5.png
46
488
media_image5.png
Greyscale
PNG
media_image6.png
44
180
media_image6.png
Greyscale
where n, m, and pare integer numbers, a is the length of the edge of the cubic unit cell, and a000, anmp and bnmp are the Fourier coefficients corresponding to three-dimensional periodic functions. The distance (or period) A at which the cosine or sine functions in Equation 1 repeat in space is given by
PNG
media_image7.png
34
142
media_image7.png
Greyscale
. A parameter d=n2+m2+p2 is defined to group cosine and sine functions with the same spatial period A”; Since the claim does not define what the group-theoretic Fourier expansion is, and the specification is also silent of that term, the examiner interprets the group-theoretic Fourier expansion as the Fourier analysis on groups (for example, group of periodic functions, and group cosine and sine functions)].
As per claim 13, Rai teaches
access, by a computing device, a black box evaluator [Figs. 14A – 14E, paragraph 0114, “a surrogate model-based optimization approach was adopted, wherein the cost function is considered as a black box problem characterized by only its input and output. How surrogate model-based optimization works is illustrated in FIGS. 14A-14E. The x-axis represents the design variable and the cost function is plotted on the y-axis”];
generate, by the computing device, a surrogate machine learning model that provides estimates for the optimization of categorical values for the black box evaluator, the surrogate machine learning model being based upon observations from previous executions of the black box evaluator [paragraph 0076, “FIG. 16 shows a graph of the performance of surrogate model-based optimization for problem configuration in FIG. 13B. With surrogate model-based optimization, the best objective value can be obtained with 194 evaluations”; paragraphs 0114-0115, “a surrogate model-based optimization approach was adopted, wherein the cost function is considered as a black box problem characterized by only its input and output. How surrogate model-based optimization works is illustrated in FIGS. 14A-14E. The x-axis represents the design variable and the cost function is plotted on the y-axis”; paragraphs 0118-0120, “Surrogate model: Next, a surrogate model is mapped on the initial design space data to create an analytical function mapping the design variables to the cost function values … few candidate points are generated by small perturbations of the current minima of the surrogate model … compares the cost function values of the new sample points and assess the stopping criteria. The method iterates through the process and progressively improves the surrogate model until the stopping criteria are satisfied. Either one of them has to be satisfied to terminate the program. The first stopping criterion is satisfied if the improvement in the optimal cost function is less than 104 for 10 consecutive iterations”; It can be seen that the surrogate model keeps updating/improving based on the cost function values at each iteration];
optimize the black box evaluator by selecting, by an acquisition function executing on the computing device, a new candidate point for the categorical values [paragraph 0115, “FIG. 14C shows new sample points (■) generated in the next iteration. FIG. 14D shows a surrogate model updated based on information from new sample points. FIG. 14E shows that, after iteratively updating and achieving convergence, an optimum solution C•) obtained”; paragraph 0119, “Selection of new samples: In this step, a new set of sample orientations are generated, and the target values are evaluated at the newly generated orientations. The surrogate model is progressively improved using the new data samples. The desired number of sample points to be selected in each iteration is selected by a user”]; and
execute, by the computing device, the black box evaluator with the new candidate point for the categorical values [paragraphs 0119-0120, “Selection of new samples: In this step, a new set of sample orientations are generated, and the target values are evaluated at the newly generated orientations. The surrogate model is progressively improved using the new data samples. The desired number of sample points to be selected in each iteration … compares the cost function values of the new sample points and assess the stopping criteria. The method iterates through the process and progressively improves the surrogate model until the stopping criteria are satisfied”], wherein data values are represented using a Fourier expansion, where characters of each representation are considered as experts and respective coefficients of the characters [paragraph 0089, “Fourier series is powerful yet simple mathematical technique. It helps establish a scheme to systematically study CPCSs and allows representation of an arbitrary periodic function as a weighted sum of cosine and sine functions. By using this technique, each periodic function has an associated distinctive set of weighted coefficients (or Fourier coefficients) that univocally represent the periodic function. In particular, three-dimensional periodic functions defined within cubic unit cells are considered with side length a. The Fourier series expansion for these periodic functions is given as follows:
PNG
media_image5.png
46
488
media_image5.png
Greyscale
PNG
media_image6.png
44
180
media_image6.png
Greyscale
where n, m, and pare integer numbers, a is the length of the edge of the cubic unit cell, and a000, anmp and bnmp are the Fourier coefficients corresponding to three-dimensional periodic functions. The distance (or period) A at which the cosine or sine functions in Equation 1 repeat in space is given by
PNG
media_image7.png
34
142
media_image7.png
Greyscale
. A parameter d=n2+m2+p2 is defined to group cosine and sine functions with the same spatial period A”].
Rai does not explicitly teach
coefficients of the characters are updated via an exponential weight update rule each time the black box evaluator is executed.
Zoldi teaches
a computer program product for federated learning, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a computer to cause the computer to [paragraph 0064, “the term "machine-readable medium" refers to any computer program product, apparatus and/or device, such as for example magnetic discs, optical disks, memory, and Programmable Logic Devices (PLDs), used to provide machine instructions and/or data to a programmable processor, including a machine-readable mediun1 that receives machine instructions as a machine readable signal. The term "machine-readable signal" refers to any signal used to provide machine instructions and/or data to a programmable processor. The machine-readable medium can store such machine instructions non-transitorily”];
coefficients of the characters are updated via an exponential weight update rule each time the black box evaluator is executed [paragraph 0002, “Many of these machine learning technologies are “black box” in nature”; paragraph 0026, “Each node of the first hidden layer is configured to learn complex relationships among the input variables. These latent features are “black box” in nature due to a non-linear transformation applied to a weighted combination of the input variables”; paragraphs “FIG. 3 shows a schematic representation of an arbitrary objective function C, as a function of two weights, w.sub.1 and w.sub.2. Any of a number of optimization algorithms can be used for minimizing the cost functions to identify the most optimal weights … The weights are then updated in the direction of the gradient using the following update rule”; paragraph 00003, “Often, the optimal value of λ is learned through an iterative selection process called parameter tuning. This yields the following update rule to update the weights”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include coefficients of the characters are updated via an exponential weight update rule of Zoldi. Doing so would help identifying the most optimal weights for minimizing the cost function (Zoldi, 0035).
As per claim 14, Rai teaches
accessing, by a computing device, a black box evaluator [Figs. 14A – 14E, paragraph 0114, “a surrogate model-based optimization approach was adopted, wherein the cost function is considered as a black box problem characterized by only its input and output. How surrogate model-based optimization works is illustrated in FIGS. 14A-14E. The x-axis represents the design variable and the cost function is plotted on the y-axis”];
generating, by the computing device, a surrogate machine learning model that provides estimates for the optimization of categorical values for the black box evaluator, the surrogate machine learning model being based upon observations from previous executions of the black box evaluator [paragraph 0076, “FIG. 16 shows a graph of the performance of surrogate model-based optimization for problem configuration in FIG. 13B. With surrogate model-based optimization, the best objective value can be obtained with 194 evaluations”; paragraphs 0114-0115, “a surrogate model-based optimization approach was adopted, wherein the cost function is considered as a black box problem characterized by only its input and output. How surrogate model-based optimization works is illustrated in FIGS. 14A-14E. The x-axis represents the design variable and the cost function is plotted on the y-axis”; paragraphs 0118-0120, “Surrogate model: Next, a surrogate model is mapped on the initial design space data to create an analytical function mapping the design variables to the cost function values … few candidate points are generated by small perturbations of the current minima of the surrogate model … compares the cost function values of the new sample points and assess the stopping criteria. The method iterates through the process and progressively improves the surrogate model until the stopping criteria are satisfied. Either one of them has to be satisfied to terminate the program. The first stopping criterion is satisfied if the improvement in the optimal cost function is less than 104 for 10 consecutive iterations”; It can be seen that the surrogate model keeps updating/improving based on the cost function values at each iteration];
optimizing the black box evaluator by selecting, by an acquisition function executing on the computing device, a new candidate point for the categorical values optimizing the black box evaluator by selecting, by an acquisition function executing on the computing device, a new candidate point for the categorical values [paragraph 0115, “FIG. 14C shows new sample points (■) generated in the next iteration. FIG. 14D shows a surrogate model updated based on information from new sample points. FIG. 14E shows that, after iteratively updating and achieving convergence, an optimum solution C•) obtained”; paragraph 0119, “Selection of new samples: In this step, a new set of sample orientations are generated, and the target values are evaluated at the newly generated orientations. The surrogate model is progressively improved using the new data samples. The desired number of sample points to be selected in each iteration is selected by a user”]; and
executing, by the computing device, the black box evaluator with the new candidate point for the categorical values [paragraphs 0119-0120, “Selection of new samples: In this step, a new set of sample orientations are generated, and the target values are evaluated at the newly generated orientations. The surrogate model is progressively improved using the new data samples. The desired number of sample points to be selected in each iteration … compares the cost function values of the new sample points and assess the stopping criteria. The method iterates through the process and progressively improves the surrogate model until the stopping criteria are satisfied”], wherein data values are represented using a Fourier expansion, where characters of each representation are considered as experts and respective coefficients of the characters [paragraph 0089, “Fourier series is powerful yet simple mathematical technique. It helps establish a scheme to systematically study CPCSs and allows representation of an arbitrary periodic function as a weighted sum of cosine and sine functions. By using this technique, each periodic function has an associated distinctive set of weighted coefficients (or Fourier coefficients) that univocally represent the periodic function. In particular, three-dimensional periodic functions defined within cubic unit cells are considered with side length a. The Fourier series expansion for these periodic functions is given as follows:
PNG
media_image5.png
46
488
media_image5.png
Greyscale
PNG
media_image6.png
44
180
media_image6.png
Greyscale
where n, m, and pare integer numbers, a is the length of the edge of the cubic unit cell, and a000, anmp and bnmp are the Fourier coefficients corresponding to three-dimensional periodic functions. The distance (or period) A at which the cosine or sine functions in Equation 1 repeat in space is given by
PNG
media_image7.png
34
142
media_image7.png
Greyscale
. A parameter d=n2+m2+p2 is defined to group cosine and sine functions with the same spatial period A”].
Rai does not explicitly teach
An apparatus comprising: a memory; and at least one processor, coupled to said memory, and operative to perform operations comprising:
coefficients of the characters are updated via an exponential weight update rule each time the black box evaluator is executed.
Zoldi teaches
An apparatus comprising: a memory; and at least one processor, coupled to said memory, and operative to perform operations comprising [paragraph 0009, “computer systems are also described that may include one or more processors and one or more memories coupled to the one or more processors. A memory, which can include a non-transitory computer-readable or machine-readable storage medium, may include, encode, store, or the like one or more programs that cause one or more processors to perform one or more of the operations described herein”]:
coefficients of the characters are updated via an exponential weight update rule each time the black box evaluator is executed [paragraph 0002, “Many of these machine learning technologies are “black box” in nature”; paragraph 0026, “Each node of the first hidden layer is configured to learn complex relationships among the input variables. These latent features are “black box” in nature due to a non-linear transformation applied to a weighted combination of the input variables”; paragraphs “FIG. 3 shows a schematic representation of an arbitrary objective function C, as a function of two weights, w.sub.1 and w.sub.2. Any of a number of optimization algorithms can be used for minimizing the cost functions to identify the most optimal weights … The weights are then updated in the direction of the gradient using the following update rule”; paragraph 00003, “Often, the optimal value of λ is learned through an iterative selection process called parameter tuning. This yields the following update rule to update the weights”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include coefficients of the characters are updated via an exponential weight update rule of Zoldi. Doing so would help identifying the most optimal weights for minimizing the cost function (Zoldi, 0035).
As per claim 15, Rai and Zoldi teach the apparatus of claim 14.
Rai further teaches
the data values are represented using a group-theoretic Fourier expansion [paragraph 0089, “Fourier series is powerful yet simple mathematical technique. It helps establish a scheme to systematically study CPCSs and allows representation of an arbitrary periodic function as a weighted sum of cosine and sine functions. By using this technique, each periodic function has an associated distinctive set of weighted coefficients (or Fourier coefficients) that univocally represent the periodic function. In particular, three-dimensional periodic functions defined within cubic unit cells are considered with side length a. The Fourier series expansion for these periodic functions is given as follows:
PNG
media_image5.png
46
488
media_image5.png
Greyscale
PNG
media_image6.png
44
180
media_image6.png
Greyscale
where n, m, and pare integer numbers, a is the length of the edge of the cubic unit cell, and a000, anmp and bnmp are the Fourier coefficients corresponding to three-dimensional periodic functions. The distance (or period) A at which the cosine or sine functions in Equation 1 repeat in space is given by
PNG
media_image7.png
34
142
media_image7.png
Greyscale
. A parameter d=n2+m2+p2 is defined to group cosine and sine functions with the same spatial period A”; Since the claim does not define what the group-theoretic Fourier expansion is, and the specification is also silent of that term, the examiner interprets the group-theoretic Fourier expansion as the Fourier analysis on groups (for example, group of periodic functions, and group cosine and sine functions)].
Claims 3 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Rai et al. in view of Zoldi et al. and further in view of Cao et al. (FTCLNet: Convolutional LSTM with Fourier Transform for Vulnerability Detection).
As per claim 3, Rai and Zoldi teach the method of claim 1.
Rai and Zoldi do not teach
the data values are represented using an abridged one-hot encoded Boolean Fourier expansion.
Cao teaches
the data values are represented using an abridged one-hot encoded Boolean Fourier expansion [abstract, “This article proposes a deep convolutional LSTM neural network with Fourier transform for vulnerability detection. The discrete Fourier transform method convert code space into frequency domain”; Fig. 2, page 539, Col. 2, 2nd paragraph to page 540, Col. 1, 1st paragraph, “In this article, FTCLNet, a deep convolutional LSTM neural network with Fourier transform is proposed for more efficient vulnerability detection. In order to learn more significant foreground information with diverse patterns of the code, the discrete Fourier transform (OFT) method is introduced to convert source code into frequency domain … Comparing with the one-hot encoding, distributed representation can help the model capture more context information of vulnerabilities. It also uses code anonymization to solve the out-of-vocabulary problem. And after the inverse discrete Fourier transform (IDFT), the attention mechanism is used to balance the weights of each element in code space. The main contributions of this paper are summarized as follows: • The Fourier transform is introduced to convert codes into frequency domain. And the source code is rewritten by anonymization and transformed to code embedding without domain knowledge before the Fourier transform”; page 540, Col. 2, last paragraph, “The overall structure of the FTCLNet is shown in the Fig. 2. In order to feed the source code to the neural network, a method is required to convert the source code into a vector representation which can save more information about the source code”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include the data values are represented using an abridged one-hot encoded Boolean Fourier expansion of Cao. Doing so would help converting code space into frequency domain to helps deep model learn remarkable patterns (Cao, abstract).
As per claim 16, Rai and Zoldi teach the apparatus of claim 14.
Rai and Zoldi do not teach
the data values are represented using an abridged one-hot encoded Boolean Fourier expansion.
Cao teaches
the data values are represented using an abridged one-hot encoded Boolean Fourier expansion [abstract, “This article proposes a deep convolutional LSTM neural network with Fourier transform for vulnerability detection. The discrete Fourier transform method convert code space into frequency domain”; Fig. 2, page 539, Col. 2, 2nd paragraph to page 540, Col. 1, 1st paragraph, “In this article, FTCLNet, a deep convolutional LSTM neural network with Fourier transform is proposed for more efficient vulnerability detection. In order to learn more significant foreground information with diverse patterns of the code, the discrete Fourier transform (OFT) method is introduced to convert source code into frequency domain … Comparing with the one-hot encoding, distributed representation can help the model capture more context information of vulnerabilities. It also uses code anonymization to solve the out-of-vocabulary problem. And after the inverse discrete Fourier transform (IDFT), the attention mechanism is used to balance the weights of each element in code space. The main contributions of this paper are summarized as follows: • The Fourier transform is introduced to convert codes into frequency domain. And the source code is rewritten by anonymization and transformed to code embedding without domain knowledge before the Fourier transform”; page 540, Col. 2, last paragraph, “The overall structure of the FTCLNet is shown in the Fig. 2. In order to feed the source code to the neural network, a method is required to convert the source code into a vector representation which can save more information about the source code”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include the data values are represented using an abridged one-hot encoded Boolean Fourier expansion of Cao. Doing so would help converting code space into frequency domain to helps deep model learn remarkable patterns (Cao, abstract).
Claims 5-8, 17-18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Rai et al. in view of Zoldi et al. and further in view of Angermueller et al. (Population-Based Black-Box Optimization for Biological Sequence Design).
As per claim 5, Rai and Zoldi teach the method of claim 1.
Rai and Zoldi do not teach
utilizing the black box evaluator to generate one or more candidate biological molecule sequences that have desirable properties, which are constructed using a vocabulary of fixed size, and synthesizing the optimal biological molecule sequence.
Angermueller teaches
utilizing the black box evaluator to generate one or more candidate biological molecule sequences that have desirable properties, which are constructed using a vocabulary of fixed size, and synthesizing the optimal biological molecule sequence [page 2, section 2, Col. 2, Lines 1-8, "We define sequences as elements of VL, where V is a finite vocabulary (for DNA, |V| = 4; for proteins, |V| = 20) and L is the sequence length. For variable length sequences, we assume that sequences are padded to length L by an end-of-sequence token. Sequence design aims to maximize a function f: VL ➔ R, which can be evaluated on batches of sequences X ⃀ VL size B = |X|, but only a limited number of times T"].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include utilizing the black box evaluator to generate one or more candidate biological molecule sequences that have desirable properties of Angermueller. Doing so would help maximizing a function f which can be evaluated on batches of sequences (Angermueller, page, section 2).
As per claim 6, Rai and Zoldi teach the method of claim 1.
Rai and Zoldi do not teach
utilizing the black box evaluator to design optimal sequences over a combinatorially large search space.
Angermueller teaches
utilizing the black box evaluator to design optimal sequences over a combinatorially large search space [page 2, section 1, Col. 1, Lines 12-19, "We evaluate P3BO and Adaptive-P3BO empirically on over 100 batched black-box optimization problems, and show that P3BO and Adaptive-P3BO are considerably more robust, generate more diverse batches of sequences, and find distinct optima faster than any single method in their population. Adaptive-P3BO improves upon P3BO results, and furthermore is able to recover from a poor initial population of methods"].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include utilizing the black box evaluator to design optimal sequences over a combinatorially large search space of Angermueller. Doing so would help producing more diverse batches and finding distinct optima faster (Angermueller, page 2, section 1).
As per claim 7, Rai and Zoldi teach the method of claim 1.
Rai and Zoldi do not teach
utilizing the black box evaluator to find a sequence given a specific structure.
Angermueller teaches
utilizing the black box evaluator to find a sequence given a specific structure [page 2, section 2.1, Col. 2, Lines 9-17, "Most discrete black-box optimization methods have associated hyper-parameters. Throughout the paper, an algorithm A refers to an instance of a particular method class (e.g., evolutionary search), which is instantiated by a specific hyper-parameter configuration (e.g., mutation rate for evolutionary search). As P3BO ensembles heterogeneous algorithms, including global model-based optimizers and local search strategies, we make the following assumptions about the interface of algorithms"].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include utilizing the black box evaluator to find a sequence given a specific structure of Angermueller. Doing so would help proposing higher quality sequences as well as more diverse batches (Angermueller, Introduction).
As per claim 8, Rai and Zoldi teach the method of claim 1.
Rai and Zoldi do not teach
the generation of the surrogate machine learning model is performed via a hedge algorithm where basis functions act as experts.
Angermueller teaches
the generation of the surrogate machine learning model is performed via a hedge algorithm where basis functions act as experts [page 2, section 1, Col. 1, Lines 27-30, "a population-based optimization framework for discrete batched black-box function optimization that ensembles over algorithms to hedge against brittleness and improve diverse sequence discovery"].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include the generation of the surrogate machine learning model is performed via a hedge algorithm where basis functions act as experts of Angermueller. Doing so would help optimizing discrete batched black-box function that ensembles over algorithms to hedge against brittleness and improving diverse sequence discovery (Angermueller, page 2, section 1).
As per claim 17, Rai and Zoldi teach the apparatus of claim 14.
Rai and Zoldi do not teach
utilizing the black box evaluator to generate one or more candidate biological molecule sequences that have desirable properties, which are constructed using a vocabulary of fixed size, and synthesizing the optimal biological molecule sequence;
Angermueller teaches
utilizing the black box evaluator to generate one or more candidate biological molecule sequences that have desirable properties, which are constructed using a vocabulary of fixed size, and synthesizing the optimal biological molecule sequence [page 2, section 2, Col. 2, Lines 1-8, " We define sequences as elements of VL, where V is a finite vocabulary (for DNA, |V| = 4; for proteins, |V| = 20) and L is the sequence length. For variable length sequences, we assume that sequences are padded to length L by an end-of-sequence token. Sequence design aims to maximize a function f: VL ➔ R, which can be evaluated on batches of sequences X ⃀ VL size B = |X|, but only a limited number of times T"];
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include utilizing the black box evaluator to generate one or more candidate biological molecule sequences that have desirable properties, which are constructed using a vocabulary of fixed size, and synthesizing the optimal biological molecule sequence of Angermueller. Doing so would help maximizing a function f which can be evaluated on batches of sequences (Angermueller, page, section 2).
As per claim 18, Rai and Zoldi teach the apparatus of claim 14.
Rai and Zoldi do not teach
the generation of the surrogate machine learning model is performed via a hedge algorithm where basis functions act as experts.
Angermueller teaches
the generation of the surrogate machine learning model is performed via a hedge algorithm where basis functions act as experts [page 2, section 1, Col. 1, Lines 27-30, "a population-based optimization framework for discrete batched black-box function optimization that ensembles over algorithms to hedge against brittleness and improve diverse sequence discovery"].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include the generation of the surrogate machine learning model is performed via a hedge algorithm where basis functions act as experts of Angermueller. Doing so would help optimizing discrete batched black-box function that ensembles over algorithms to hedge against brittleness and improving diverse sequence discovery (Angermueller, page 2, section 1).
As per claim 20, Rai and Zoldi teach the apparatus of claim 14.
Rai and Zoldi do not teach
performing simulated annealing utilizing the surrogate machine learning model for internal cost-free evaluations before producing a next black box query.
Angermueller teaches
Performing the simulated annealing utilizing the surrogate machine learning model is performed for internal cost-free evaluations before producing a next black box query [page 3, section 3, Col. 1, Lines 15-20, "We introduce P3BO, a robust black-box optimization method that constructs batches of sequences with which to query f (x) using a population of heterogeneous optimization algorithms. By sharing data between algorithms, algorithms benefit from each other's distinct exploration strategies"].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include the simulated annealing utilizing the surrogate machine learning model is performed for internal cost-free evaluations before producing a next black box query of Angermueller. Doing so would help optimizing batched sequence design (Angermueller, section 3).
Claims 9 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Rai et al. in view of Zoldi et al. and further in view of Guez et al. (Learning to Search with MCTSnets).
As per claim 9, Rai and Zoldi teach the method of claim 1.
Rai and Zoldi do not teach
Monte Carlo tree search (MCTS) and the surrogate machine learning model are used as a reward function of terminal states during intermediate tree traversals and backups.
Guez teaches
Monte Carlo tree search (MCTS) and the surrogate machine learning model are used as a reward function of terminal states during intermediate tree traversals and backups [Figure 1, page 5, Lines 1-6, "This diagram shows an execution of a search with M = 4. (Top) The evolution of the search tree rooted at s0 after each simulation, with the last simulation path highlighted in red. (Bottom) The computation graph in MCTSnet resulting from these simulations. Black arrows represent the application of the embedding network ϵ (s) to initialize hat tree nodes. Red arrows represent the forward tree traversal during a simulation using the simulation policy (based on last memory state) and the environment model until a leaf node is reached. Blue arrows correspond to the backup network β, which updates the memory statistics h along the traversed simulation path based on the child statistic and the last updated parent memory (in addition to transition information such as reward)"].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include Monte Carlo tree search (MCTS) and the surrogate machine learning model are used as a reward function of terminal states during intermediate tree traversals and backups of Guez. Doing so would help training MCTSnet to predict the action a from state s (Guez, page 4, section 3.5).
As per claim 19, Rai and Zoldi teach the apparatus of claim 14.
Rai and Zoldi do not teach
Monte Carlo tree search (MCTS) and the surrogate machine learning model are used as a reward function of terminal states during intermediate tree traversals and backups.
Guez teaches
Monte Carlo tree search (MCTS) and the surrogate machine learning model are used as a reward function of terminal states during intermediate tree traversals and backups [Figure 1, page 5, Lines 1-6, "This diagram shows an execution of a search with M = 4. (Top) The evolution of the search tree rooted at s0 after each simulation, with the last simulation path highlighted in red. (Bottom) The computation graph in MCTSnet resulting from these simulations. Black arrows represent the application of the embedding network ϵ (s) to initialize hat tree nodes. Red arrows represent the forward tree traversal during a simulation using the simulation policy (based on last memory state) and the environment model until a leaf node is reached. Blue arrows correspond to the backup network β, which updates the memory statistics h along the traversed simulation path based on the child statistic and the last updated parent memory (in addition to transition information such as reward)"].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include Monte Carlo tree search (MCTS) and the surrogate machine learning model are used as a reward function of terminal states during intermediate tree traversals and backups of Guez. Doing so would help training MCTSnet to predict the action a from state s (Guez, page 4, section 3.5).
Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over Rai et al. in view of Zoldi et al. and further in view of Wang et al. (US Pub. 2021/0124854).
As per claim 10, Rai and Zoldi teach the method of claim 1.
Rai and Zoldi do not teach
performing simulated annealing utilizing the surrogate machine learning model.
Wang teaches
performing simulated annealing utilizing the surrogate machine learning model [paragraph 0056, “search for the optimal model parameters using the surrogate model … the surrogate model’s output may include active power, reactive power or both. In some cases, the optimizer may be gradient based method including Newton-like methods. For example, the optimizer may be gradient free method including, simulated annealing”].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include performing simulated annealing utilizing the surrogate machine learning model of Wang. Doing so would help search for the optimal model parameters (Wang, 0056).
Claim 11 is rejected under 35 U.S.C. 103 as being unpatentable over Rai et al. in view of Zoldi et al. in view of Wang et al. and further in view of Angermueller et al. (Population-Based Black-Box Optimization for Biological Sequence Design).
As per claim 11, Rai, Zoldi and Wang teach the method of claim 10.
Rai, Zoldi and Wang do not teach
the simulated annealing utilizing the surrogate machine learning model is performed for internal cost-free evaluations before producing a next black box query;
Angermueller teaches
the simulated annealing utilizing the surrogate machine learning model is performed for internal cost-free evaluations before producing a next black box query [page 3, section 3, Col. 1, Lines 15-20, "We introduce P3BO, a robust black-box optimization method that constructs batches of sequences with which to query f (x) using a population of heterogeneous optimization algorithms. By sharing data between algorithms, algorithms benefit from each other's distinct exploration strategies"].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include the simulated annealing utilizing the surrogate machine learning model is performed for internal cost-free evaluations before producing a next black box query of Angermueller. Doing so would help optimizing batched sequence design (Angermueller, section 3).
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Rai et al. in view of Zoldi et al. and further in view of Guidotti et al. (A Survey of Methods for Explaining Black Box Models).
As per claim 12, Rai and Zoldi teach the method of claim 1.
Rai and Zoldi do not explicitly teach
the black box evaluator utilizes a black box machine learning model.
Guidotti teaches
the black box evaluator utilizes a black box machine learning model [page 1, Introduction, "These black box systems exploit sophisticated machine-learning models to predict individual information that may also be sensitive. We can consider credit score, insurance risk, health status, as examples. Machine learning algorithms build predictive models that are able to map user features into a class (outcome or decision) thanks to a learning phase. This learning process is made possible by the digital traces that people leave behind them while performing everyday activities (e.g., movements, purchases comments in social networks, etc.). This enormous amount of data may contain human biases and prejudices. Thus, decision models learned on them may inherit such biases, possibly leading to unfair and wrong decisions"].
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to have modified the surrogate model-based optimization framework of Rai to include the black box evaluator utilizes a black box machine learning model of Guidotti. Doing so would help exploiting machine-learning models to predict individual information that may also be sensitive (Guidotti, Introduction).
Allowable Subject Matter
Claim 4 is objected to as being dependent upon the rejected base claims, but would be allowable if rewritten in independent form including all of the limitations of the base claims and any intervening claims, and the 101 rejections above were overcome.
The following is a statement of reasons for the indication of allowable subject matter:
Claim 4 is allowable for disclosing
The method of claim 3, wherein a one-hot encoding of each variable
PNG
media_image1.png
20
74
media_image1.png
Greyscale
is expressed as a (k - 1)-tuple
PNG
media_image2.png
24
142
media_image2.png
Greyscale
where
PNG
media_image3.png
22
92
media_image3.png
Greyscale
are Boolean variables with a constraint that at most one such variable is equal to -1 for any given
PNG
media_image4.png
20
58
media_image4.png
Greyscale
.
The closest references found
Handa (US Pub. 2021/0365605) in paragraphs 0124-0125 teaches [“The one-hot constraint is a constraint that “there is just a single variable having a value of 1 in a certain set of variables”, This constraint is applied to various problems such as a quadratic assignment problem … The one-hot constraint permits just one bit to have “1” among the bits belonging to the same group”].
Okada et al. (US Pub. 2019/0392326) in paragraphs 100 teaches [“The one-hot constraint represents a constraint in which only one hot state value “1” appears in the binary variables xi representing the respective multivalued variables Si and a second term on the right side of the evaluation function H0 represents this constraint. The second term on the right side of the evaluation function H0 indicates a constraint term added so that the evaluation function H0 becomes the lowest when only one binary variable xi is set as the hot state value “1” in each multivalued variable Si”].
However, the prior art of record do not teach or suggest, individually or in combination
a one-hot encoding of each variable
PNG
media_image1.png
20
74
media_image1.png
Greyscale
is expressed as a (k - 1)-tuple
PNG
media_image2.png
24
142
media_image2.png
Greyscale
where
PNG
media_image3.png
22
92
media_image3.png
Greyscale
are Boolean variables with a constraint that at most one such variable is equal to -1 for any given
PNG
media_image4.png
20
58
media_image4.png
Greyscale
.
Therefore, the combination of features is considered to be allowable.
Prior Art
The prior art made of record and not relied upon is considered pertinent to applicant’s disclosure.
Lenc et al. (US Pub. 2020/0234142) describes a black box optimization technique that maintains and adapts a distribution over the weights that are of interest.
Dalli et al. (US Pub. 2019/0057308) describes a method of using a predictor function to predict the output of the black-box model.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to TRI T NGUYEN whose telephone number is 571-272-0103. The examiner can normally be reached M-F, 8 AM-5 PM, (CT).
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, OMAR FERNANDEZ can be reached at 571-272-2589. 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.
/TRI T NGUYEN/Examiner, Art Unit 2128
/OMAR F FERNANDEZ RIVAS/Supervisory Patent Examiner, Art Unit 2128