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 .
Response to Arguments
Applicant's arguments filed 3/4/2026 have been fully considered but they are not persuasive.
Regarding applicants arguments on 35 U.S.C. 101 rejection, the applicant argues in page 13 of remarks “Claims 1-20 are rejected under 35 USC§ 101 as being directed to an abstract idea.
Applicant respectfully traverses these rejections in view of the claim amendments made herein.
Claims 19 and 20 Applicant notes that these claims were not present in the application as filed.
Claim 1 has been amended to clarify that each MC module is hardware-based and configured to process a pre-trained machine learning main classifier having at least one expert class and a plurality of non-expert classes, and that each EC module is hardware-based and configured to process a pre-trained machine learning expert classifier having two classes including an associated expert class and a residual class. This amendment is supported, at least, by paragraph [0114] of the specification. Applicant submits that this amendment clarifies that the MC and EC modules are not software modules for execution on a general-purpose computer. In the telephone interview conducted on February 18, 2026, the Examiner indicated that this distinction would help to overcome the rejection of the claim under 35 USC 101.
Claims 2-9 depend from claim 1, and are patentable for at least the same reasons.” However calling the modules hardware-based does not make them overcome the 101 rejection. During the interview on 2/18/2026 that the applicant would have to make the distinction that they are hardware as the specification explain the modules to be software and hardware. Further applicant added claims 19 and 20 that have not been examined. The arguments presented by the applicant are not persuasive and are moot regarding claim 19 and 20.
Claim Objections
Claim 19 is objected to because of the following informalities:
The limitation “an MC predicted class for the least one expert class” should be at least one. Appropriate correction is required.
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 rejected under 35 U.S.C. 101 because the claimed invention is directed to abstract idea without significantly more. The claim(s) recite(s) significantly more. The subject matter eligibility test for products and process is describe below for claim 1 in view of dependent claims.
Regarding claim 1:
Step 1: Is the claim to a process machine manufacture or composition of matter?
Yes – Claim 1 recites a hardware accelerator, which is a system that is under the statutory categories.
Step 2A Prong 1: Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes – The claim recites the following:
“each MC module configured to predict an MC predicted class based on input data, determine an MC certainty, and output the MC predicted class and the MC certainty;” - The limitations recites a mental process of predicting a class and determine the certainty (see MPEP 2106.04(a)(2)III).
“each EC module is configured to predict an EC predicted class based on the input data, and output the EC predicted class;”- The limitations recites a mental process of determining an EC predicted class (see MPEP 2106.04(a)(2)III).
“determine a final predicted class and a final certainty based on each MC predicted class, each MC certainty and each EC predicted class, and output the final predicted class and the final certainty,” and ““where, when each MC certainty indicates that the MC predicted class is certain and each MC predicted class is the same, the final predicted class is the MC predicted class, and the final certainty indicates that the final predicted class is certain, and otherwise, when at least one MC predicted class is an expert class and at least one EC predicted class is the expert class, the final predicted class is the EC predicted class, and the final certainty indicates that the final predicted class is certain.”
The limitations recites a mental process of determining the final predicted class and certainty based on the MC certainty and each EC predicted class and how it can be determined based on the method given above (see MPEP 2106.04(a)(2)III).
Step 2 Prong 2: Does the claim recite additional elements that integrate the judicial exception into a particular application? No –
The claim includes the additional element(s):
“A hardware accelerator, comprising: a plurality of main classifier (MC) modules, where: each MC module is hardware-based and configured to process a pre-trained[[,]] machine learning main classifier having at least one expert class and a plurality of non-expert classes,”
The additional elements fall under “apply it” as using a generic computer to process a pre-trained, machine learning model. See Mere Instructions to Apply an Exemption (see MPEP 2106.05(f)).
“an expert classifier (EC) module associated with each expert class, where: each EC module is hardware-based and configured to process a pre-trained[[,]] machine learning expert classifier having two classes including an associated expert class and a residual class that includes any non-associated expert classes and the plurality of non-expert classes,”
The additional elements fall under “apply it” as using a generic computer to process a pre-trained machine learning expert classifier. See Mere Instructions to Apply an Exemption (see MPEP 2106.05(f)).
“a final predicted class decision module, coupled to each MC module and each EC module, and configured to receive each MC predicted class, each MC certainty and each EC predicted class,”
The additional elements fall under Insignificant Extra-Solution Activity as mere data gathering by receiving each MC predicted class the certainty. See MPEP 2106.5(g).
“where each MC certainty is a binary value that indicates whether the MC predicted class is certain or uncertain, and the final certainty is a binary value that indicates whether the final predicted class is certain or uncertain.”
The additional elements fall under Insignificant Extra-Solution Activity as it only explain that a binary value is used to represent what MC certainty and the final certainty . See MPEP 2106.5(g).
Step 2B: Does the claim recite additional elements that amount to significantly more than the judicial exception?
No - The claim does not include additional elements that are sufficient to amount to a significantly more than the judicial exemption. As an order whole, the claim is directed an ensemble method for predicting and determining classification results based on certainty. As discussed above with respect to integration of the abstract idea into a practical application, the additional elements of processing, classifying, determining, receiving results fall under using generic computer to apply an exemption and mere data gathering. The method does not improve on the function of a computer, transforms an article into another article, nor is it applied by a particular machine, making the claim not patent eligible.
Regarding claim 2:
Step 2A Prong 1:
“and each MC certainty is calculated based on an entropy of the output node probabilities of the associated classes.” – The limitation recites a mathematical calculation by using the entropy of the probabilities to calculate the MC certainty (see MPEP 2106.04(a)(2).
Step 2A Prong 2, Step 2B: The additional element(s):
“The hardware accelerator according to claim 1, where: each main classifier (MC) module is configured to implement an artificial neural network that includes an input layer, one or more hidden layers and an output layer having a plurality of output nodes, each output node generating a probability for an associated class;”
The additional elements fall under “apply it” as using a generic computer to implement an artificial neural network to output nodes that generate probability. See Mere Instructions to Apply an Exemption (see MPEP 2106.05(f)).
Regarding claim 3:
Step 2A Prong 1:
“The hardware accelerator according to claim 2, where the entropy is calculated based on a sum of each output node probability times .” – The limitation recites a mathematical calculation by a sum of each output node probability times a value approximately equal to a binary logarithm of the output node probability (see MPEP 2106.04(a)(2).
Step 2A Prong 2, Step 2B: The additional element(s):
No additional elements. The judicial exemptions do not integrate into a practical application nor provide an improvement. The process does not provide an inventive concept nor provides a practical application
Regarding claim 4:
Step 2A Prong 1:
“The hardware accelerator according to claim [[3]]2, where each MC certainty is certain when the entropy is less than a predetermined threshold, and uncertain when the entropy is equal to or greater than the predetermined threshold.” – The limitations recites a mental process of to determine if a MC certainty is certain or uncertain (see MPEP 2106.04(a)(2)III).
Step 2A Prong 2, Step 2B: The additional element(s):
No additional elements. The judicial exemptions do not integrate into a practical application nor provide an improvement. The process does not provide an inventive concept nor provides a practical application
Regarding claim 5:
Step 2A Prong 2, Step 2B: The additional element(s):
“The hardware accelerator according to claim 4, where the output node probabilities are between 0 and 1, and the predetermined threshold is determined during training.”
The additional elements fall under “apply it” as using a generic computer to implement neural networks to output probabilities between 0 and 1 . See Mere Instructions to Apply an Exemption (see MPEP 2106.05(f)).
Regarding claim 6:
Step 2A Prong 2, Step 2B: The additional element(s):
“The hardware accelerator according to claim 1, where[[,]] the machine learning main classifier of each MC module is pre-trained using a first learning rate and a dataset labeled for the at least one expert class and the plurality of non-expert classes; and the machine learning expert classifier of the EC module is pre-trained using a second learning rate and the dataset labeled for the associated expert class and residual class, the second learning rate is higher than the first learning rate.”
The additional elements fall under “apply it” as using a generic computer to train the MC module using a first learning rate and use a second learning rate for the EC module. See Mere Instructions to Apply an Exemption (see MPEP 2106.05(f)).
Regarding claim 7:
Step 2A Prong 1:
“The hardware accelerator according to claim [[6]] 1, where, when each MC certainty indicates that the MC predicted class is certain, at least one MC predicted class is different, at least one MC predicted class is an expert class and at least one EC predicted class is the expert class, the final predicted class is the EC predicted class, and the final certainty indicates that the final predicted class is certain.” The limitation recites a mental process determining the final certainty and the final predicted class is certain given the recited conditions mentioned above (see MPEP 2106.04(a)(2)III).
Step 2A Prong 2, Step 2B: The additional element(s):
No additional elements. The judicial exemptions do not integrate into a practical application nor provide an improvement. The process does not provide an inventive concept nor provides a practical application
Regarding claim 8:
Step 2A Prong 1:
“The hardware accelerator according to claim 7, where, when at least one MC certainty indicates that the MC predicted class is uncertain, at least one MC predicted class is an expert class and at least one EC predicted class is the expert class, the final predicted class is the EC predicted class, and the final certainty indicates that the final predicted class is certain.” - The limitations of recites a mental process determining the final certainty and the final predicted class is certain given the recited conditions mentioned above (see MPEP 2106.04(a)(2)III).
Step 2A Prong 2, Step 2B: The additional element(s):
No additional elements. The judicial exemptions do not integrate into a practical application nor provide an improvement. The process does not provide an inventive concept nor provides a practical application
Regarding claim 9:
Step 2A Prong 1:
“The hardware accelerator according to claim 1, when each MC certainty indicates that the MC predicted class is certain and each MC predicted class is the same, each EC module does not predict and output the EC predicted class.” - The limitations recites a mental process of determining when the EC module will be used given the conditions mentioned above (see MPEP 2106.04(a)(2)III).
Step 2A Prong 2, Step 2B: The additional element(s):
No additional elements. The judicial exemptions do not integrate into a practical application nor provide an improvement. The process does not provide an inventive concept nor provides a practical application
Claims 10-18 recite a method and are analogous to the system claims 1-9. Therefore, the rejections of claim 1-9 above applies to claims 10-18.
Regarding claim 19:
Step 1: Is the claim to a process machine manufacture or composition of matter?
Yes – Claim 1 recites a hardware accelerator, which is a system that falls under the statutory categories.
Step 2A Prong 1: Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes – The claim recites the following:
“determining, based on the input data to be classified, a probability and an MC predicted class for the least one expert class and the plurality of non-expert classes of the MC module;” - The limitations recites a mental process of determining a probability and an MC predicted class (see MPEP 2106.04(a)(2)III).
“determining an MC certainty based on the probabilities associated with the plurality of MC classes, wherein the MC certainty is a binary value that indicates whether the MC predicted class is certain or uncertain;”- The limitations recites a mental process of determining an MC certainty (see MPEP 2106.04(a)(2)III).
“determining an EC predicted class based on the input data;”
The limitations recites a mental process of determining an EC predicted class (see MPEP 2106.04(a)(2)III).
“determining a final predicted class and a final certainty based on each MC predicted class, each MC certainty and each EC predicted class for the input data, wherein the final certainty is a binary value that indicates whether the final predicted class is certain or uncertain, wherein the final predicted class and final certainty have greater prediction accuracy than any individual MC predicted class and MC certainty.”
The limitations recites a mental process of determining an the final predicted class and certainty based on the MC certainty and each EC predicted class (see MPEP 2106.04(a)(2)III).
Step 2 Prong 2: Does the claim recite additional elements that integrate the judicial exception into a particular application? No –
The claim includes the additional element(s):
“A computer-implemented method of classifying input data in a machine learning classifier, wherein the machine learning classifier includes: a plurality of main classifier (MC) modules of a machine learning classifier, each MC module having a plurality of classes including at least one expert class and a plurality of non-expert classes, said an expert classifier (EC) module associated with each expert class, each EC module having two classes including an associated expert class and a residual class;”
The additional elements fall under “apply it” as using a generic computer implement a machine learning classifier. See Mere Instructions to Apply an Exemption (see MPEP 2106.05(f)).
“said method comprising: for each MC module of a plurality of MC modules: training the MC module, based on training data for the module, to determine a probability and a predicted class for the least one expert class and the plurality of non-expert classes of the MC module;”
The additional elements fall under “apply it” as using a generic computer to train the MC module based on training data. See Mere Instructions to Apply an Exemption (see MPEP 2106.05(f)).
“for each expert class: training the associated EC module to determine an EC predicted class based on training data for the module;”
The additional elements fall under “apply it” as using a generic computer to train the EC module based on training data. See Mere Instructions to Apply an Exemption (see MPEP 2106.05(f)).
Step 2B: Does the claim recite additional elements that amount to significantly more than the judicial exception?
No - The claim does not include additional elements that are sufficient to amount to a significantly more than the judicial exemption. As an order whole, the claim is directed an ensemble method for predicting and determining classification results based on certainty. As discussed above with respect to integration of the abstract idea into a practical application, the additional elements of processing, classifying, determining, receiving results fall under using generic computer to apply an exemption and mere data gathering. The method does not improve on the function of a computer, transforms an article into another article, nor is it applied by a particular machine, making the claim not patent eligible.
Claims 20 recite a computer-implemented method and is analogous to the system claim 6. Therefore, the rejections of claim 6 above applies to claims 19 and 20.
Allowable Subject Matter
Claim 1-20 would be allowable if rewritten or amended to overcome the rejection(s) under 35 U.S.C. 101 set forth in this Office action.
Regarding claim 1, the prior art of record teaches limitations as noted in the previous Office Actions.
The prior art of record Aizpurua et al., "Improved power transformer condition monitoring under uncertainty through soft computing and probabilistic health index," Applied Soft Computing, Volume 85, 2019, 105530, ISSN 1568-4946 (“Aizpurua”) teaches “Aizpurua page 10 3.2.3. Ensemble of Classifiers para 1, 3 and 4,
DGA focuses on a supervised learning problem with five input gases (H2 , CH4, C2~, C2H2, C2H6) and four possible health states (normal degradation, arcing fault, partial discharge fault, and thermal fault).
For instance, assume that a model has been trained to classify certain faults. So long as the test data is comprised of faults which are similar to the trained model, it should return a prediction with high confidence. However, if the model is tested on an unseen fault class, the model should be able to quantify this with uncertainty levels, which can convey information about the confidence of the diagnosis of the model. This information is completely lost with black-box (BB) models, which only represent purely numerical connections, but they tend to have a high classification accuracy. Conversely, white-box (WB) models capture expert knowledge either as a causal model or through first principle models. They generate the uncertainty associated with the decision-making process by quantifying the PDF of the likelihood of different diagnostics states. This function represents the strength of the diagnosis of the model, i.e. the wider the variance, the less the confidence in the diagnostics outcome and vice-versa.
In this context it is possible to combine the accuracy of BB models with the uncertainty information of WB models so as to improve the final classification accuracy. Figure 7 shows the implemented uncertainty-aware ensemble diagnostics framework with two BB source classifiers (artificial neural networks - ANN, support vector machines - SVM) and a WB classifier, i.e. Gaussian Bayesian Networks (GEN). For each group of DGA data samples, ANN and SVM models generate deterministic probability estimates (
m
A
N
N
,
m
s
v
n
), and the GEN model generates a probability) density function for each fault type which is preprocessed to infer maximum likelihood information and uncertainty information (
m
B
N
,
m
u
)
Page 11,
PNG
media_image1.png
193
730
media_image1.png
Greyscale
” Aizpurua teaches using an ensemble method for classification to determine a certainty and probability of a fault.
C. M. Intisar and Q. Zhao, "A Selective Modular Neural Network Framework," 2019 IEEE 10th International Conference on Awareness Science and Technology (iCAST), Morioka, Japan, 2019, pp. 1-6, doi: 10.1109/ICAwST.2019.8923334. (“Instisar”) teaches “Instisar page 2, III. Key Idea of this study para 1, In this study, we aim to solve multi-class classification problems. The basic idea or motivation is, instead of running each and every baseline module in the ensemble, we run only a selected (constant) number of baseline modules. To do so, we propose a framework, where selection of the modules are initiated based on a routing module. This routing module is trained to be accurate enough to include the correct answer, say in the top-2 outputs (Here, 2 is called the redundancy rate, which is denoted by Rr in this paper. We can use Rr larger than 2 to improve the accuracy). After we obtain the top-2 most likely classes based on the soft-max confidence values of the routing module, the routing module selects 3 or 4 expert modules. These expert modules are then used to make the final decision. For each predicted class, two modules participate in further evaluation. After evaluating the expert modules, the final prediction is calculated based on the aggregated values of experts and routing module soft-max scores, which is termed as the Corrected Activation (CA).” Instisar teaches a final prediction class based on values from experts.
Costello (US20200065384A1) (“Costello”) teaches a method of determining an ensemble of models that output predicted results of a likely class and a comparing a metric corresponding to each of the ensembles “Costello [0035], In a third aspect, the present disclosure further provides a computer-implemented process or method for identifying one or a combination of models for an intent classification method with favorable accuracy based on a dataset. The process or the method comprises the following steps:
[0036] providing a plurality of models;
[0037] training each of the plurality of models using data from a train set of the dataset to thereby obtain a plurality of trained model sets, wherein each of the plurality of trained model sets is based on a different model and comprises a plurality of trained models, wherein each of the plurality trained models in the each of the plurality of trained model sets is obtained by training a same model with a different initialization condition over the dataset;
[0038] feeding an input text in a test set of the dataset into each of the plurality trained models in the each of the plurality of trained model sets to thereby obtain a plurality of prediction results, each indicating a likely intent class of the input text, wherein the plurality of prediction results correspond respectively to the plurality of trained models in each of the plurality of trained model sets;
[0039] ensembling prediction results corresponding to a same trained model set for each of the plurality of trained model sets to thereby obtain a plurality of first-layer ensembles corresponding respectively to the plurality of trained model sets;
[0040] ensembling the plurality of first-layer ensembles to thereby obtain a plurality of second-layer ensembles, wherein each of the plurality of second-layer ensembles is obtained by ensembling one of all possible combinations among the plurality of first-layer ensembles;
[0041] evaluating each of the plurality of first-layer ensembles and each of the plurality of second-layer ensembles using pre-determined labels from the test set; and
[0042] comparing a metric corresponding to each of the plurality of first-layer ensembles and each of the plurality of second-layer ensembles, and identifying the one or the combination of the models for the intent classification method if the one or the combination of the models implicated in one of the plurality of first-layer ensembles or the plurality of second-layer ensembles has a most favorable metric compared with others.
[0043] Herein optionally, the metric can comprise an unweighted F1 score.”
However Aizpurua, Instisar, and Costello, taken alone or in combination with the prior art record fail to teach or suggest “determine a final predicted class and a final certainty based on each MC predicted class, each MC certainty and each EC predicted class, and output the final predicted class and the final certainty, where each MC certainty is a binary value that indicates whether the MC predicted class is certain or uncertain, and the final certainty is a binary value that indicates whether the final predicted class is certain or uncertain,” as recited in claim 1, in combination with the reaming features and elements of the claimed invention.
Independent claims 10 and 19 would be allowable for the same reasons cited in claim 1. The remaining claims would be allowable because they depend on one of allowable independent claims 1, 10 and 19.
Pertinent Prior Art
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Wang et al. (US20170243124A1) (“Wang”) – teaches an ensemble of binary classifiers train in a method as shown in Figure 3A.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ALFREDO CAMPOS whose telephone number is (571)272-4504. The examiner can normally be reached 7:00 - 4:00 pm M - F.
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, Michael J. Huntley can be reached at (303) 297-4307. 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.
/ALFREDO CAMPOS/Examiner, Art Unit 2129
/MICHAEL J HUNTLEY/Supervisory Patent Examiner, Art Unit 2129
Prosecution Insights