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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 2/18/2026 has been entered.
Claims 1, 8, and 15 have been amended. Claims 1-6, 8-13, and 15-22 are pending and have been examined.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-5, 8-12, 15-19, and 22 are rejected under 35 U.S.C. 103 as being unpatentable over Fidel et al., “When Explainability Meets Adversarial Learning: Detecting Adversarial Examples using SHAP Signatures”, hereinafter “Fidel” in view of Chawla et al., “SMOTE: Synthetic Minority Over-sampling Technique”, hereinafter “Chawla”, further in view of Bergstra et al., “Random Search for Hyper-Parameter Optimization”, hereinafter “Bergstra”.
Regarding Claim 1, Fidel teaches:
A computer system, comprising:
one or more processors; and one or more machine-readable medium coupled to the one or more processors and storing computer program code comprising sets of instructions for executable by the one or more processors (Fidel trains and tests models demonstrating that Fidel performs their method on a computer, in which processor, memory, and storage devices are inherent, p. 6, col. 2, paragraph 3, “Table II presents the number or normal and adversarial samples used for training and testing the detection model”, p. 7, Figure 5 showing performance metrics) to:
obtain a historical feature contribution score dataset comprising a number of sets of scores (SHAP values are sets of scores, p. 5, col. 1, paragraph 2, “we utilize SHAP to generate an XAI signature for each sample in the dataset (both normal and adversarial). Specifically, we apply the SHAP DeepExplainer [25] to interpret the neurons of the penultimate layer f [l−1](·). The outcome of this application is n SHAP values for each output in f [l−1](·), where n represent the number of target classes (i.e., SHAP produces a single value for each output and class)”) generated by one or more previously trained machine learning models (SHAP values are generated with a previously trained classification model, p. 3, col. 1, paragraph 4, “compute the importance scores of the neurons of the penultimate layer of the classification model”, p. 6, col. 2, paragraph 1, “evaluated our proposed detection method using… CIFAR-10 dataset [26] with the ResNet-56 classification model [34]. The model achieves a 93.39% accuracy on the CIFAR-10 test set”)
produce a training dataset including feature contribution scores and corresponding category classification labels extracted from the historical feature contribution score dataset and… (training dataset is SHAP values and adversarial or normal categories used to train detector model, p. 6, col. 1, paragraph 3, “we train a supervised binary detector to discriminate between normal and adversarial samples… We use the SHAP values from our generated dataset as the samples’ features to train the classifier. Any standard supervised model can be used to train the detector based on these features… compute the sample’s SHAP values (in the XAI signature phase) and feed the output into our binary classifier to classify the sample as adversarial or normal”), the category classification labels categorizing the corresponding feature contribution scores based on a degree of influence on a prediction of a target feature (classification labels of adversarial or normal categorize the SHAP values which are based on a degree of influence on a prediction of a target feature, p. 6, col. 1, paragraph 3, “classify the sample as adversarial or normal”, p. 3, col. 1, paragraph 1, “SHAP [25] method, which is a unified approach that aims to explain the model output using shapely values”, p. 3, col. 1, paragraph 4, “we utilize SHAP [25] to compute the importance scores of the neurons of the penultimate layer of the classification model”);
train a classification machine learning model to predict the category classification labels using the training dataset (p. 6, col. 1, paragraph 3, “we train a supervised binary detector to discriminate between normal and adversarial samples… We use the SHAP values from our generated dataset as the samples’ features to train the classifier”); and
apply an input feature contribution score set to the classification machine learning model to obtain predicted category classification labels for one or more feature contribution scores in the input feature contribution score set (p. 6, col. 1, paragraph 3, “At inference time, given a sample to classify as normal or adversarial, we compute the sample’s SHAP values (in the XAI signature phase) and feed the output into our binary classifier to classify the sample as adversarial or normal”).
Fidel does not expressly teach:
materialize additional feature contribution score sets…
…such that the size of each additional feature contribution score set is a randomly selected value within a set-size range
…additional feature contribution score sets…
However, Chawla teaches
materialize additional feature contribution score sets… and …additional feature contribution score sets… (Additional contribution score set is the data generated, Chawla, p. 328, ¶2, “generate synthetic examples”);
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Chawla generating synthetic examples with the labeled dataset of SHAP values and adversarial or normal labels taught by Fidel. The modification would have been motivated to help with data imbalance (Chawla, p.322, ¶2, “performance of machine learning algorithms is typically evaluated using predictive accuracy. However, this is not appropriate when the data is imbalanced”, p. 328, ¶2, “We propose an over-sampling approach in which the minority class is over-sampled by creating “synthetic” examples rather than by over-sampling with replacement”, Fidel, p. 1, col. 2, paragraph 4, “different adversarial example generation algorithms”).
Fidel in view of Chawla does not expressly teach:
…such that the size of each additional feature contribution score set is a randomly selected value within a set-size range
However, Fidel in view of Chawla further in view of Bergstra teaches:
…such that the size of each additional feature contribution score set is a randomly selected value within a set-size range (Chawla, p. 328, paragraph 4, “We have approximately 9831 examples in the majority class and 233 examples in the minority class for the training set used in 10-fold cross-validation. The minority class was over-sampled at 100%, 200%, 300%, 400% and 500% of its original size”, Bergstra, p. 302, paragraph 4, “To investigate the effect of one hyper-parameter of interest X, we recommend random search (instead of grid search) for optimizing over other hyper-parameters”, p. 281, Abstract, “we find that random search over the same domain is able to find models that are as good or better within a small fraction of the computation time”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Chawla oversampling across multiple percentages and randomly selecting from a set of candidate values of Bergstra. The motivation to do so would be to select an oversampling percentage without needing to deterministically identify the best percentage, which will save on computation time (p. 281, Abstract, “we find that random search over the same domain is able to find models that are as good or better within a small fraction of the computation time”).
Regarding Claim 2, Fidel in view of Chawla and Bergstra teaches the computer system of Claim 1 as referenced above. In the combination as set forth above in Claim 1, Chawla teaches:
wherein materializing additional feature contribution score sets includes randomly generating scores based on a number of sample score- ranges (The values of the data which are the scores are based on a random number between 0 and 1, Chawla, p. 328, ¶2, “Take the difference between the feature vector (sample) under consideration and its nearest neighbor. Multiply this difference by a random number between 0 and 1, and add it to the feature vector under consideration”.
Regarding Claim 3, Fidel in view of Chawla and Bergstra teaches the computer system of Claim 1 as referenced above. In the combination as set forth above in Claim 1, Chawla teaches:
wherein the materializing of the additional feature contribution score sets includes normalizing score values of the additional feature contribution score (The score value of a feature is normalized with its nearest neighbor, Chawla, p. 328, ¶2, “Take the difference between the feature vector (sample) under consideration and its nearest neighbor”).
Regarding Claim 4, Fidel in view of Chawla and Bergstra teaches the computer system of Claim 1 as referenced above. In the combination as set forth above in Claim 1, Fidel teaches:
wherein the category classification labels are assigned to the scores of additional feature contribution score sets after they are materialized (Additional feature contribution score sets are part of detector dataset, Fidel, p. 7, col. 1, paragraph 2, “Each sample in the train/test sets was represented using its SHAP values with the class label set to be “1” for adversarial example and “0” otherwise”).
Regarding Claim 5, Fidel in view of Chawla and Bergstra teaches the computer system of Claim 1 as referenced above. In the combination as set forth above in Claim 1, Chawla teaches:
determine a deficit number based on the number of the sets of scores in the feature contribution score dataset and a predefined number of feature contribution score sets, the materializing of the additional feature contribution score sets based on the deficit number. (A deficit number to materialize additional feature contribution score sets is based on class size and class imbalance, Chawla, p.321, ¶1, “dataset is imbalanced if the classes are not approximately equally represented”, p.328, ¶2, “The minority class is over-sampled… introducing synthetic examples… if the amount of over-sampling needed is 200%”).
Regarding Claim 8, Fidel teaches:
one or more non-transitory computer-readable medium storing computer program code comprising sets of instructions (Fidel trains and tests models demonstrating that Fidel performs their method on a computer, in which processor, memory, and storage devices are inherent, p. 6, col. 2, paragraph 3, “Table II presents the number or normal and adversarial samples used for training and testing the detection model”, p. 7, Figure 5 showing performance metrics) to:
obtain a historical feature contribution score dataset comprising a number of sets of scores (SHAP values are sets of scores, p. 5, col. 1, paragraph 2, “we utilize SHAP to generate an XAI signature for each sample in the dataset (both normal and adversarial). Specifically, we apply the SHAP DeepExplainer [25] to interpret the neurons of the penultimate layer f [l−1](·). The outcome of this application is n SHAP values for each output in f [l−1](·), where n represent the number of target classes (i.e., SHAP produces a single value for each output and class)”) generated by one or more previously trained machine learning models (SHAP values are generated with a previously trained classification model, p. 3, col. 1, paragraph 4, “compute the importance scores of the neurons of the penultimate layer of the classification model”, p. 6, col. 2, paragraph 1, “evaluated our proposed detection method using… CIFAR-10 dataset [26] with the ResNet-56 classification model [34]. The model achieves a 93.39% accuracy on the CIFAR-10 test set”)
produce a training dataset including feature contribution scores and corresponding category classification labels extracted from the historical feature contribution score dataset and… (training dataset is SHAP values and adversarial or normal categories used to train detector model, p. 6, col. 1, paragraph 3, “we train a supervised binary detector to discriminate between normal and adversarial samples… We use the SHAP values from our generated dataset as the samples’ features to train the classifier. Any standard supervised model can be used to train the detector based on these features… compute the sample’s SHAP values (in the XAI signature phase) and feed the output into our binary classifier to classify the sample as adversarial or normal”), the category classification labels categorizing the corresponding feature contribution scores based on a degree of influence on a prediction of a target feature (classification labels of adversarial or normal categorize the SHAP values which are based on a degree of influence on a prediction of a target feature, p. 6, col. 1, paragraph 3, “classify the sample as adversarial or normal”, p. 3, col. 1, paragraph 1, “SHAP [25] method, which is a unified approach that aims to explain the model output using shapely values”, p. 3, col. 1, paragraph 4, “we utilize SHAP [25] to compute the importance scores of the neurons of the penultimate layer of the classification model”);
train a classification machine learning model to predict the category classification labels using the training dataset (p. 6, col. 1, paragraph 3, “we train a supervised binary detector to discriminate between normal and adversarial samples… We use the SHAP values from our generated dataset as the samples’ features to train the classifier”); and
apply an input feature contribution score set to the classification machine learning model to obtain predicted category classification labels for one or more feature contribution scores in the input feature contribution score set (p. 6, col. 1, paragraph 3, “At inference time, given a sample to classify as normal or adversarial, we compute the sample’s SHAP values (in the XAI signature phase) and feed the output into our binary classifier to classify the sample as adversarial or normal”).
Fidel does not expressly teach:
materialize additional feature contribution score sets…
…such that the size of each additional feature contribution score set is a randomly selected value within a set-size range
…additional feature contribution score sets…
However, Chawla teaches
materialize additional feature contribution score sets… and …additional feature contribution score sets… (Additional contribution score set is the data generated, Chawla, p. 328, ¶2, “generate synthetic examples”);
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Chawla generating synthetic examples with the labeled dataset of SHAP values and adversarial or normal labels taught by Fidel. The modification would have been motivated to help with data imbalance (Chawla, p.322, ¶2, “performance of machine learning algorithms is typically evaluated using predictive accuracy. However, this is not appropriate when the data is imbalanced”, p. 328, ¶2, “We propose an over-sampling approach in which the minority class is over-sampled by creating “synthetic” examples rather than by over-sampling with replacement”, Fidel, p. 1, col. 2, paragraph 4, “different adversarial example generation algorithms”).
Fidel in view of Chawla does not expressly teach:
…such that the size of each additional feature contribution score set is a randomly selected value within a set-size range
However, Fidel in view of Chawla further in view of Bergstra teaches:
…such that the size of each additional feature contribution score set is a randomly selected value within a set-size range (Chawla, p. 328, paragraph 4, “We have approximately 9831 examples in the majority class and 233 examples in the minority class for the training set used in 10-fold cross-validation. The minority class was over-sampled at 100%, 200%, 300%, 400% and 500% of its original size”, Bergstra, p. 302, paragraph 4, “To investigate the effect of one hyper-parameter of interest X, we recommend random search (instead of grid search) for optimizing over other hyper-parameters”, p. 281, Abstract, “we find that random search over the same domain is able to find models that are as good or better within a small fraction of the computation time”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Chawla oversampling across multiple percentages and randomly selecting from a set of candidate values of Bergstra. The motivation to do so would be to select an oversampling percentage without needing to deterministically identify the best percentage, which will save on computation time (p. 281, Abstract, “we find that random search over the same domain is able to find models that are as good or better within a small fraction of the computation time”).
Regarding Claim 9, the rejection of Claim 8 is incorporated and further, the claim is rejected for the same reasons as set forth in Claim 2.
Regarding Claim 10, the rejection of Claim 8 is incorporated and further, the claim is rejected for the same reasons as set forth in Claim 3.
Regarding Claim 11, the rejection of Claim 8 is incorporated and further, the claim is rejected for the same reasons as set forth in Claim 4.
Regarding Claim 12, the rejection of Claim 8 is incorporated and further, the claim is rejected for the same reasons as set forth in Claim 5.
Regarding Claim 15, Fidel teaches:
a computer-implemented method, comprising:
obtaining a historical feature contribution score dataset comprising a number of sets of scores (SHAP values are sets of scores, p. 5, col. 1, paragraph 2, “we utilize SHAP to generate an XAI signature for each sample in the dataset (both normal and adversarial). Specifically, we apply the SHAP DeepExplainer [25] to interpret the neurons of the penultimate layer f [l−1](·). The outcome of this application is n SHAP values for each output in f [l−1](·), where n represent the number of target classes (i.e., SHAP produces a single value for each output and class)”) generated by one or more previously trained machine learning models (SHAP values are generated with a previously trained classification model, p. 3, col. 1, paragraph 4, “compute the importance scores of the neurons of the penultimate layer of the classification model”, p. 6, col. 2, paragraph 1, “evaluated our proposed detection method using… CIFAR-10 dataset [26] with the ResNet-56 classification model [34]. The model achieves a 93.39% accuracy on the CIFAR-10 test set”);
producing a training dataset including feature contribution scores and corresponding category classification labels extracted from the historical feature contribution score dataset and… (training dataset is SHAP values and adversarial or normal categories used to train detector model, p. 6, col. 1, paragraph 3, “we train a supervised binary detector to discriminate between normal and adversarial samples… We use the SHAP values from our generated dataset as the samples’ features to train the classifier. Any standard supervised model can be used to train the detector based on these features… compute the sample’s SHAP values (in the XAI signature phase) and feed the output into our binary classifier to classify the sample as adversarial or normal”), the category classification labels categorizing the corresponding feature contribution scores based on a degree of influence on a prediction of a target feature (classification labels of adversarial or normal categorize the SHAP values which are based on a degree of influence on a prediction of a target feature, p. 6, col. 1, paragraph 3, “classify the sample as adversarial or normal”, p. 3, col. 1, paragraph 1, “SHAP [25] method, which is a unified approach that aims to explain the model output using shapely values”, p. 3, col. 1, paragraph 4, “we utilize SHAP [25] to compute the importance scores of the neurons of the penultimate layer of the classification model”);
training a classification machine learning model to predict the category classification labels using the training dataset (p. 6, col. 1, paragraph 3, “we train a supervised binary detector to discriminate between normal and adversarial samples… We use the SHAP values from our generated dataset as the samples’ features to train the classifier”); and
applying an input feature contribution score set to the classification machine learning model to obtain predicted category classification labels for one or more feature contribution scores in the input feature contribution score set (p. 6, col. 1, paragraph 3, “At inference time, given a sample to classify as normal or adversarial, we compute the sample’s SHAP values (in the XAI signature phase) and feed the output into our binary classifier to classify the sample as adversarial or normal”).
Fidel does not expressly teach:
materializing additional feature contribution score sets…
…such that the size of each additional feature contribution score set is a randomly selected value within a set-size range
…additional feature contribution score sets…
However, Chawla teaches
materializing additional feature contribution score sets… and …additional feature contribution score sets… (Additional contribution score set is the data generated, Chawla, p. 328, ¶2, “generate synthetic examples”);
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Chawla generating synthetic examples with the labeled dataset of SHAP values and adversarial or normal labels taught by Fidel. The modification would have been motivated to help with data imbalance (Chawla, p.322, ¶2, “performance of machine learning algorithms is typically evaluated using predictive accuracy. However, this is not appropriate when the data is imbalanced”, p. 328, ¶2, “We propose an over-sampling approach in which the minority class is over-sampled by creating “synthetic” examples rather than by over-sampling with replacement”, Fidel, p. 1, col. 2, paragraph 4, “different adversarial example generation algorithms”).
Fidel in view of Chawla does not expressly teach:
…such that the size of each additional feature contribution score set is a randomly selected value within a set-size range
However, Fidel in view of Chawla further in view of Bergstra teaches:
…such that the size of each additional feature contribution score set is a randomly selected value within a set-size range (Chawla, p. 328, paragraph 4, “We have approximately 9831 examples in the majority class and 233 examples in the minority class for the training set used in 10-fold cross-validation. The minority class was over-sampled at 100%, 200%, 300%, 400% and 500% of its original size”, Bergstra, p. 302, paragraph 4, “To investigate the effect of one hyper-parameter of interest X, we recommend random search (instead of grid search) for optimizing over other hyper-parameters”, p. 281, Abstract, “we find that random search over the same domain is able to find models that are as good or better within a small fraction of the computation time”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Chawla oversampling across multiple percentages and randomly selecting from a set of candidate values of Bergstra. The motivation to do so would be to select an oversampling percentage without needing to deterministically identify the best percentage, which will save on computation time (p. 281, Abstract, “we find that random search over the same domain is able to find models that are as good or better within a small fraction of the computation time”).
Regarding Claim 16, the rejection of Claim 15 is incorporated and further, the claim is rejected
for the same reasons as set forth in Claim 2.
Regarding Claim 17, the rejection of Claim 15 is incorporated and further, the claim is rejected
for the same reasons as set forth in Claim 3.
Regarding Claim 18, the rejection of Claim 15 is incorporated and further, the claim is rejected
for the same reasons as set forth in Claim 4.
Regarding Claim 19, the rejection of Claim 15 is incorporated and further, the claim is rejected
for the same reasons as set forth in Claim 5.
Regarding Claim 22, Fidel in view of Chawla and Bergstra teaches the method of Claim 15 as referenced above. Fidel further teaches:
wherein training the classification machine learning model comprises:
identifying the category classification label as a target feature (Fidel, p. 6, col. 1, paragraph 3, “train a supervised binary detector to discriminate between normal and adversarial samples”);
identifying one or more features that are not the target feature as input features for the supervised learning algorithm (Fidel, p. 6, col. 1, paragraph 3, “We use the SHAP values from our generated dataset as the samples’ features to train the classifier”); and
performing a supervised learning algorithm using the target feature and the input features (Fidel, p. 6, col. 1, paragraph 3, “we train a supervised binary detector to discriminate between normal and adversarial samples… We use the SHAP values from our generated dataset as the samples’ features to train the classifier).
Claims 6, 13, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Fidel, in view of Chawla, further in view of Anwaar, “Machine Learning Tutorial — Feature Engineering Tabular Data”, hereinafter “Anwaar”.
Regarding Claim 6, Fidel in view of Chawla and Bergstra teaches the computer system of Claim 1 as referenced above. In the combination as set forth above in Claim 1, Fidel in view of Chawla and Bergstra does not teach, but Anwaar teaches:
derive engineered features based on the historical feature contribution score dataset, the additional feature contribution score sets, and one or more of a maximum feature contribution score (Derives engineered features based on existing features which includes historical feature contribution score dataset and additional feature contribution score sets, Anwaar, p. 5, ¶1, “Creating new features by performing simple statistical calculations on the raw features including… max”), a minimum feature contribution score, a mean feature contribution score, a distance to the maximum feature contribution score, a distance to the minimum feature contribution score, a distance to the mean feature contribution score, and a variance of feature contribution scores.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Anwaar feature engineering with historical feature contribution score sets and additional feature contribution score sets taught by Fidel in view of Chawla and Bergstra. The modification would have been motivated to generate better results (Anwaar, p. 1, ¶1, “Feature engineering is the method of transforming variables in meaningful information that support our predictive model and generate good results.”).
Regarding Claim 13, the rejection of Claim 8 is incorporated and further, the claim is rejected
for the same reasons as set forth in Claim 6.
Regarding Claim 20, the rejection of Claim 15 is incorporated and further, the claim is rejected
for the same reasons as set forth in Claim 6.
Claims 21 is rejected under 35 U.S.C. 103 as being unpatentable over Fidel, in view of Chawla, further in view of Zoppi et al., “Meta-Learning to Improve Unsupervised Intrusion Detection in Cyber-Physical Systems”, hereinafter “Zoppi”.
Regarding Claim 21, Fidel in view of Chawla and Bergstra teaches the method of Claim 15 as referenced above. Fidel in view of Chawla and Bergstra does not teach, however Zoppi teaches:
wherein the category classification labels comprise at least one of: weak, moderate, or strong (weak label is shown with not sure and strong label is shown with yes or no, Zoppi, p. 10, paragraph 4, “wrap the binary decision, converting it into a ternary {yes, no, not sure}, and re-modulating the confusion matrix accordingly”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Zoppi turning binary decisions into a ternary {yes, no, not sure} with the binary labels of Fidel. The modification would have been motivated to have a binary classifier answer only if they are confident enough (Zoppi, p. 10, paragraph 4, “classifiers should answer either “yes” or “no” if and only if they are confident enough”).
Response to Arguments
35 U.S.C. 103
Argument 1: Chawla does not teach or suggest “materializing additional feature contribution score sets that the size of each feature contribution score set is a randomly selected value within a set-size range” which shows that the random values determine or influence how many elements the set will contain.
Examiner Response: Examiner agrees. Chawla does not expressly teach the size of the feature contribution score set is a randomly selected value, however Chawla in view of Bergstra teaches this. Chawla teaches using different amounts of oversampling percentages, Chawla, p. 328, paragraph 4, “We have approximately 9831 examples in the majority class and 233 examples in the minority class… minority class was over-sampled at 100%, 200%, 300%, 400% and 500% of its original size”, Bergstra teaches choosing a random option from a plurality of options, Bergstra, p. 302, paragraph 4, “To investigate the effect of one hyper-parameter of interest X, we recommend random search (instead of grid search) for optimizing over other hyper-parameters”. The motivation for this combination is to be able to select an oversampling percentage without needing to deterministically identify the best oversampling percentage which will save on computation time, Bergstra p. 281, Abstract, “we find that random search over the same domain is able to find models that are as good or better within a small fraction of the computation time”.
From applicant remarks, “random values determine or influence how many elements the set will contain”, the combination of Chawla and Bergstra shows how a randomly chosen value from the percentages will determine how many elements the set will contain. This combination clearly teaches the claim language of the size being a randomly selected value and there is nothing stated in the claim language that would cause this combination to differ from the claims.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JESSE CHEN COULSON whose telephone number is (571)272-4716. The examiner can normally be reached Monday-Friday 8:30-5:30.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Kakali Chaki can be reached at (571) 272-3719. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/JESSE C COULSON/
Examiner, Art Unit 2122
/KAKALI CHAKI/Supervisory Patent Examiner, Art Unit 2122
Prosecution Insights