DETAILED ACTION
Response to Arguments
Applicant’s arguments, see pages 10-12 filed on 2/27/26, with respect to the §103 rejections of the amended claims have been fully considered and are persuasive. Therefore, the rejections have been withdrawn. However, upon further consideration, a new grounds of rejection is made in view of Ionescu et al. US 20190286242 and Ivanov US 20070127825.
Claim Rejections - 35 USC § 103
Claims 1-3, 8, 16-18 and 21-22 are rejected under 35 U.S.C. 103 as being unpatentable over Ionescu et al. US 20190286242 (hereinafter Ionescu ‘242) in view of Kar et al. US 20210182603 (hereinafter Kar ‘603) and Ivanov US 20070127825 (hereinafter Ivanov ‘825).
As per claim 1, Ionescu ‘242 discloses an identity authentication method, comprising:
obtaining first operation behavior data of a to-be-authenticated user; obtaining second operation behavior data of the to-be-authenticated user (para 0135, “In FIG. 12A (as discussed above for FIG. 11), at step S110, the processor of the mobile device is configured by executing one or more software modules, including the feature extraction module, to extract a set of features from the collected motion signals 1105 using one or more of: statistical feature extraction technique 1110, correlation features extraction technique 1115, Mel Frequency Cepstral Coefficients (MFCC) 1120, Shifted Delta Cepstral (SDC) 1125, Histogram of Oriented Gradients (HOG) 1130, Markov Transition Matrix 1135 and deep embeddings extracted with Convolutional Neural Networks (CNN) 1140. The set of features extracted from the collected motion signals can be in the form of concatenate feature vectors 1145.”);
obtaining, by inputting the first operation behavior data into a first authentication model, a first recognition result output by the first authentication model; obtaining, by inputting the second operation behavior data into a second authentication model, a second recognition result output by the second authentication model (para 0137, “Turning now to FIG. 12B, once the subset of discriminative features has been selected, at step S120, the processor of the mobile device is configured by executing one or more software modules, including the classification module, to classify the user as a genuine user or an imposter user based on a classification score generated by the classification algorithm(s) from an analysis of the subset of discriminative features. One or more of the following classifiers are used as classification algorithms for step S120: Naïve Bayes classifier 1305, Support Vector Machine (SVM) classifier 1310, Multi-layer Perception classifier 1315, Random Forest classifier 1320, and Kernel Ridge Regression (KRR) 1325.”),
wherein the first authentication model is a one-class support vector machine (SVM) (para 0037, “Employing a meta-classifier which uses the class labels and the score returned by both one-class and two-class classifiers is an original approach that improves the user identity verification accuracy.”; para 0072, “Furthermore, the system can then perform classification of the so processed data (step S120). For instance, classification can be performed using a meta-classifier that uses as features the classification scores and labels of several binary (two-class) classifiers (Support Vector Machines, Naive Bayes, Random Forrest, Feed-forward Neural Networks, Kernel Ridge Regression) and a one-class classifier (one-class Support Vector Machines).”; para 0126, “In at least one embodiment disclosed herein, as classification techniques, the present systems and methods use binary classifiers that distinguish between two classes, a positive (+1) class corresponding to the Genuine User and a negative (−1) class corresponding to Impostor Users. The Genuine User class represents the user to be verified, while the Impostor User class represents the attackers who try to impersonate the actual user during the verification process.”); and
inputting the first recognition result and the second recognition result into a decision fusion model to output an identity authentication result, wherein the decision fusion model is used to determine the identity authentication result based on (fig. 12B, reference no. 1330; para 0125, “In at least one embodiment described herein, the meta-learning approach at step S120 is organized in two layers. The first layer provides multiple classifications of the user interaction using the features selected by the PCA algorithm, while the second layer classifies the user interaction using the information (output) given by the first layer. It should be note that, different from the standard approach, the features used in the second layer are composed of both the predicted labels (−1 or +1) and the classification scores (continuous real values) produced by the classifiers from the first layer.”).
Although Ionescu ‘242 does not expressly disclose the one-class SVM is an anomaly detection model, one-class SMVs are conventionally recognized as outlier classifiers. Moreover, as known to one of ordinary skilled in the art, one-class SVMs are trained by establishing a boundary around normal data points. For example, Kar ‘603 discloses a one-class SVM as an anomaly detection model (para 0062, “Two, one class machine learning models namely one class support vector machine (OC-SVM) and support vector data description (SVDD) are used for imposter detection. The one class classifiers (OC-SVM and SVDD) generate binary output i.e. +1 if the test sample lies within the decision boundary or −1 if the sample is an outlier. In imposter detection scenario data of each registered user are used to train individual one class models. If there are r number of registered users, there will be r number of trained one class models. The test sample is predicted as imposter (anomaly) if all the trained one class classifiers return −1 i.e. the test sample is anomalous to all the one class classifiers.”). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention for the one-class SVM to be an anomaly detection model. One-class SVM anomaly detection models for user authentication has the benefit of a well-defined training set; i.e., an authorized user’s normal sample data.
Ionescu ‘242 does not expressly disclose, but Ivanov ‘825 discloses the identity authentication result based on weight parameters of the first recognition result and the second recognition result (para 0015, “The combining 110 can also use an adaptation of an approximate Bayesian combination. The Baysian combination uses some measure of classifier confidence to weigh the prediction probabilities of each weak classifier with respect to an expected accuracy of the weak classifier for each of the classes”; para 0021, “The AdaBoost process trains each classifier in a combination with increasingly more difficult data, and then uses a weighted score. During the training, the combined classifiers are examined, in turn, with replacement. At every iteration, a greedy selection is made. The combined classifier that yields a minimal error rate on data misclassified during a previous iteration is selected, and the weight is determined as a function of the error rate. The AdaBoost process iterates until one of the following conditions is met: a predetermined number of iterations has been made, a pre-determined number of classifiers have been selected, the error rate decreases to a pre-determined threshold, or no further improvement to the error rate can be made.”). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the method of Ionescu ‘242 such that the identity authentication result is based on weight parameters of the first recognition result and the second recognition result as taught by Ivanov ‘825. One would have been motivated to do so to achieve a combination strong classifier from a set combination of weak classifiers that achieves a lower error rate. See Ivanov ‘825, para 0006.
As per claim 2, Ionescu ‘242 in view of Kar ‘603 and Ivanov ‘825 discloses the method according to claim 1 (supra). Moreover, Ionescu ‘242 and Kar ‘603 discloses wherein a model parameter of the first authentication model is obtained by performing training based on sample data (Ionescu ‘242, para 0047 “In one or more embodiments, the processor of the mobile device can be configured to examine the collected motion signals and measure the quality of those signals. For example, for an explicit gesture or interaction, motion signals of the user corresponding to the explicit gesture can be measured against sample motion signals for that specific explicit gesture.”; para 0128, “Support Vector Machines (SVM)—Support Vector Machines try to find the vector of weights that defines the hyperplane that maximally separates the training examples belonging to the two classes. The training samples that fall inside the maximal margin are called support vectors.”; Kar ‘603, para 0063, “Two, one class machine learning models namely one class support vector machine (OC-SVM) and support vector data description (SVDD) are used for imposter detection. The one class classifiers (OC-SVM and SVDD) generate binary output i.e. +1 if the test sample lies within the decision boundary or −1 if the sample is an outlier. In imposter detection scenario data of each registered user are used to train individual one class models.”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention for the one-class SVM to be an anomaly detection model. One-class SVM anomaly detection models for user authentication has the benefit of a well-defined training set; i.e., an authorized user’s normal sample data.
As per claim 3, Ionescu ‘242 in view of Kar ‘603 and Ivanov ‘825 discloses the method according to claim 1. In addition, Ionescu ‘242 discloses wherein at least one of the first operation behavior data or the second operation behavior data are data collected by a sensor (para 0116, “FIG. 11 presents the computation flow of the feature extraction step (S110) of the present method of verifying a user based on the interaction with a device by measuring it with mobile sensors, e.g. accelerometer and gyroscope, in accordance with one or more embodiments described herein.”).
As per claim 8, Ionescu ‘242 in view of Ivanov ‘825 disclose the method according to claim 1 (supra). In addition, Ionescu ‘242 discloses wherein the first operation behavior data or the second operation behavior data comprises at least one of the following types of data: an x/y-axis coordinate of a touch point, a touch area, touch pressure, a touch speed, a touch acceleration, a touch track slope, a touch length, a touch displacement, a touch angle, a touch direction, acceleration x/y/z-axis data, or gyroscope x/y/z-axis data (para 0044, “To capture this physical phenomenon, the present system is configured to collect multi-axis signals (motion signals) corresponding to the physical movement of the user during a specified time domain from motion sensors (e.g. accelerometer and gyroscope) of the mobile device. ”).
Claims 16-18 and 21-22 are apparatus and computer readable medium claims that correspond to claims 1-3. Hence, claims 16-18 and 21-22 are rejected for the same reasons as claims 1-3.
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 JUNG W KIM whose telephone number is (571)272-3804. The examiner can normally be reached Monday-Friday, 10 a.m. - 6 p.m..
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, Amy Cohen Johnson can be reached at 571-272-2238. 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.
/JUNG W KIM/Supervisory Patent Examiner, Art Unit 2494
Prosecution Insights