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 .
Specification
The disclosure is objected to because of the following informalities:
Specification recites the following: “which allows application vendors and rapidly deploy and take advantage of the latest…” when “to” should be before “rapidly”, so the sentence should be phrased as “which allows application vendors to rapidly deploy and take advantage of the latest…” (Page 18 Line 9)
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 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.
Step 1: Claims 1-20 are within the four statutory categories of patent eligible subject matter (a process, machine, manufacture or composition of matter).
Claim 1:
Step 2A Prong 1:
computing, by a device, a plurality of predictions based on session telemetry data for an application, each of the plurality (This is an abstract idea of a “Mental Process.” The “computing” of “predictions based on session telemetry data” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of the telemetry data within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.).
computing, by the device and for each respective feedback of a plurality of collected feedbacks correlated with the session telemetry data, a residual error vector that quantifies a level of mis-prediction of each respective machine learning model as compared to each respective feedback; (This is an abstract idea of a “Mental Process.” The “computing” of “for each respective feedback” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of the collected feedback within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.).
clustering, by the device, each residual error vector into one or more clusters based on similarity to thereby identify one or more outlier residual error vectors that are not within any of the one or more clusters; and (This is an abstract idea of a “Mental Process.” The “clustering” of “each residual error vector” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of the clusters for residual error vectors within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.).
performing, by the device, one or more mitigation actions responsive to the one or more outlier residual error vectors. (This is an abstract idea of a “Mental Process.” The “performing” of “mitigation action” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of mitigation actions within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.).
Step 2A Prong 2 and 2B: The claim recites the following additional elements:
predictions computed by inputting a particular input of the session telemetry data into a respective machine learning model of a plurality of machine learning models; (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). The limitation recites “predictions computed by inputting a particular input” represents generic computer component and/or computer instructions. Performing remediation is well-understood, routine, conventional activity -see MPEP 2106.05(d)(II)(vi)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 2
Step 2A Prong 1: Claim 2, which incorporates the rejection of claim 1, recites further limitation such as:
wherein the one or more mitigation actions are selected from a group consisting of: marking one or more specific feedbacks of the plurality of collected feedbacks associated with the one or more outlier residual error vectors as anomalous; (This is an abstract idea of a “Mental Process.” The “selected” and “marking” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of collected feedbacks within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.).
marking one or more specific sources of the plurality of collected feedbacks associated with the one or more outlier residual error vectors as anomalous; (This is an abstract idea of a “Mental Process.” The “marking” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of collected feedbacks within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.).
marking one or more specific sources of the plurality of collected feedbacks associated with the one or more outlier residual error vectors as unreliable; (This is an abstract idea of a “Mental Process.” The “marking” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of collected feedbacks within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.).
investigating circumstances that led to the one or more outlier residual error vectors; and (This is an abstract idea of a “Mental Process.” The “investigating” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of collected feedbacks within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.).
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
performing remediation against an adversarial feedback attack based on the one or more outlier residual error vectors. (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). The limitation recites “performing remediation against an adversarial feedback attack” represents generic computer component and/or computer instructions. Performing remediation is well-understood, routine, conventional activity -see MPEP 2106.05(d)(II)(vi)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 3:
Step 2A Prong 1: Claim 3, which incorporates the rejection of claim 1, doesn’t recite an abstract idea.
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
The method of claim 1, wherein the plurality of machine learning models are based on a single model trained using a single method and architecture but with different initializations, subsampling, and/or feature masking. (This limitation recites using a single trained model for a plurality of machine learning models which is mere data gathering recited at a high level of generality, and thus is an insignificant extra-solution activity (MPEP 2106.05(g)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 4:
Step 2A Prong 1: Claim 4, which incorporates the rejection of claim 1, doesn’t recite an abstract idea.
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
The method of claim 1, wherein the plurality of machine learning models are based on a plurality of different models trained on different methods and/or architectures. (This limitation recites using a plurality trained model for a plurality of machine learning models which is mere data gathering recited at a high level of generality and thus is an insignificant extra-solution activity (MPEP 2106.05(g)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 5
Step 2A Prong 1: Claim 5, which incorporates the rejection of claim 1, doesn’t recite an abstract idea.
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
The method of claim 1, wherein the plurality of predictions comprise quality-of- experience predictions, and wherein the plurality of collected feedbacks comprise quality-of-experience feedbacks. (This limitation recites using a plurality quality-of-experience predictions which is mere data gathering recited at a high level of generality, and thus is an insignificant extra-solution activity (MPEP 2106.05(g)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 6
Step 2A Prong 1: Claim 6, which incorporates the rejection of claim 1, doesn’t recite an abstract idea.
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
The method of claim 1, further comprising: attaching contextual troubleshooting information to the one or more outlier residual error vectors. (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)). The limitation recites “attaching contextual troubleshooting information” represents generic computer component and/or computer instructions. Attaching contextual troubleshooting is well-understood, routine, conventional activity -see MPEP 2106.05(d)(II)(vi)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 7
Step 2A Prong 1: Claim 7, which incorporates the rejection of claim 6, recites further limitation such as:
The method of claim 6, wherein the contextual troubleshooting information is selected from a group consisting of: additionally solicited feedback from sources of the plurality of collected feedbacks associated with the one or more outlier residual error vectors; (This is an abstract idea of a “Mental Process.” The “grouping” of “outlier residual error vectors” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of contextual troubleshooting information within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.).
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
additionally solicited feedback from new sources; indications of predominant signals used in computing the plurality of predictions; (This limitation is directed to transmitting data of sources of feedback data which is well-understood, routine, and conventional activity, see MPEP 2106.05(d)(II)(i)).
and information on how the plurality of collected feedbacks was provided. (This limitation is directed to transmitting data of collected feedback which is well-understood, routine, and conventional activity, see MPEP 2106.05(d)(II)(i)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 8
Step 2A Prong 1: Claim 8, which incorporates the rejection of claim 1, doesn’t recite an abstract idea.
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
The method of claim 1, wherein clustering is based on density-based spatial clustering of applications with noise. (This limitation recites using clustering based on density-based spatial clustering which is mere data gathering recited at a high level of generality, and thus is an insignificant extra-solution activity (MPEP 2106.05(g)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 9
Step 2A Prong 1: Claim 9, which incorporates the rejection of claim 1, doesn’t recite an abstract idea.
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
The method of claim 1, wherein clustering is limited based on a temporal sliding window. (This limitation recites using clustering based on temporal sliding window which is mere data gathering recited at a high level of generality, and thus is an insignificant extra-solution activity (MPEP 2106.05(g)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 10
Step 2A Prong 1: Claim 10, which incorporates the rejection of claim 1, recites further limitation such as:
to thereby further identify one or more additional outlier residual error vectors. (This is an abstract idea of a “Mental Process.” The “identify” of “outlier residual error vectors” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of outlier residual error vectors within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.).
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
The method of claim 1, further comprising: removing the one or more outlier residual error vectors from consideration in clustering; (This limitation recites “removing” which is mere instruction to apply the abstract idea on a generic computer to process the data elements, see MPEP 2106.05(f)).
and re-clustering each residual error vector other than the one or more outlier residual error vectors (This limitation recites “re-clustering” which is mere data gathering recited at a high level of generality and thus is an insignificant extra-solution activity (MPEP 2106.05(g)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 11
Step 2A Prong 1: Claim 11, which incorporates the rejection of claim 1, doesn’t recite an abstract idea.
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
The method of claim 1, wherein the one or more mitigation actions comprise discarding one or more specific feedbacks of the plurality of collected feedbacks associated with the one or more outlier residual error vectors from use in training the plurality of machine learning models. (This limitation recites using discarding feedbacks with outlier residual error vectors which is mere data gathering recited at a high level of generality, and thus is an insignificant extra-solution activity (MPEP 2106.05(g)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 12
Step 2A Prong 1: Claim 12, which incorporates the rejection of claim 1, doesn’t recite an abstract idea.
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
The method of claim 1, wherein the one or more mitigation actions comprise preventing feedback participation from one or more specific sources of the plurality of collected feedbacks associated with the one or more outlier residual error vectors that are deemed to be adversarial. (This limitation recites using discarding sources of collected feedbacks with outlier residual error vectors which is mere data gathering recited at a high level of generality, and thus is an insignificant extra-solution activity (MPEP 2106.05(g)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 13
Step 2A Prong 1: Claim 13, which incorporates the rejection of claim 1, recites further limitation such as:
confirming that the new plurality of predictions is an improvement on the plurality of predictions. (This is an abstract idea of a “Mental Process.” The “confirming” of “new plurality of predictions” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of confirming a new plurality of predictions within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.)
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
The method of claim 1, further comprising: recomputing a new plurality of predictions based on the session telemetry data with one or more specific feedbacks of the plurality of collected feedbacks associated with the one or more outlier residual error vectors being removed; and (This limitation recites “recomputing a new plurality of predictions” which is merely a post-solution step of storing the data—a nominal addition to the claim that does not meaningfully limit the claim. The method storing is recited at a high level of generality. Simply implementing the abstract idea in a generic method is not a practical application of the abstract idea. Therefore, storing step is an insignificant extra-solution activity. See MPEP 2106.05(g)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 14
Step 2A Prong 1: Claim 14, which incorporates the rejection of claim 1, recites further limitation such as:
The method of claim 1, further comprising: collecting the plurality of collected feedbacks based on polling users about their subjective experience during an application session; and (This is an abstract idea of a “Mental Process.” The “collecting” of “collected feedbacks” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of collecting users subjective experience within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.)
correlating each of the plurality of collected feedbacks with a respective input of the session telemetry data (This is an abstract idea of a “Mental Process.” The “correlating” of “collected feedbacks” step, as drafted, and under its broadest reasonable interpretation, covers concepts that can be practically performed in the human mind, see MPEP 2106.04(a)(2) § III. This limitation encompasses a mental process of evaluation or determination of correlating collected feedbacks with telemetry data within a threshold to cluster/group generically recited machine learning models into clusters/groups based on an observed model performance.)
Step 2A Prong 2 and 2B: this claim recites the following additional elements:
correlating each of the plurality of collected feedbacks with a respective input of the session telemetry data (This limitation recites correlating feedbacks to input telemetry data which is mere instruction to apply the abstract idea on a generic computer to process the data elements, see MPEP 2106.05(f)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 15:
Step 2A Prong 1: The claim recites similar limitations as corresponding claim 1. Therefore, the same analysis (the abstract idea of “a mental process”) that was utilized under step 2A prong 1 for claim 1, as described above is equally applicable to claim 15.
Step 2A Prong 2 and 2B: This claim recites the following additional elements:
An apparatus, comprising :one or more network interfaces to communicate with a network; a processor coupled to the one or more network interfaces and configured to execute one or more processes; and a memory configured to store a process that is executable by the processor, the process comprising: (The claimed additional elements represent generic computer components and computer instructions to apply the abstract idea. Therefore, these elements are sufficient to integrate the judicial exception into a practical application, see MPEP 2106.05(f)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
Claim 16
The claim recites similar limitations as corresponding claim 2. Therefore, the same subject matter analysis that was utilized for claim 2, as described above, is equally applicable to claim 16. Therefore, the claim is ineligible.
Claim 17
The claim recites similar limitations as corresponding claim 5. Therefore, the same subject matter analysis that was utilized for claim 5, as described above, is equally applicable to claim 17. Therefore, the claim is ineligible.
Claim 18
The claim recites similar limitations as corresponding claim 6. Therefore, the same subject matter analysis that was utilized for claim 6, as described above, is equally applicable to claim 18. Therefore, the claim is ineligible.
Claim 19
The claim recites similar limitations as corresponding claim 11. Therefore, the same subject matter analysis that was utilized for claim 11, as described above, is equally applicable to claim 19. Therefore, the claim is ineligible.
Claim 20
Step 2A Prong 1: The claim recites similar limitations as corresponding claim 1. Therefore, the same analysis (the abstract idea of “a mental process”) that was utilized under step 2A prong 1 for claim 1, as described above is equally applicable to claim 20.
Step 2A Prong 2 and 2B:
A tangible, non-transitory, computer-readable medium storing program instructions that cause a device to execute a process comprising: (The claimed additional elements represent generic computer components and computer instructions to apply the abstract idea. Therefore, these elements are sufficient to integrate the judicial exception into a practical application, see MPEP 2106.05(f)).
The claim as a whole, looking at the additional elements individually and in combination, does not integrate the judicial exception into a practical application and does not amount to significantly more than the identified judicial exception.
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.
Claim(s) 1-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Kursun et al, (Pub. No.: US 20200387602 A1) in view of Yue et al., (CTRL: Clustering Training Losses for Label Error Detection).
Regarding Claim 1, Kursun teaches the following:
A method, comprising: computing, by a device, a plurality of predictions based on session telemetry data for an application, each of the plurality of predictions computed by inputting a particular input of the session telemetry data into a respective machine learning model of a plurality of machine learning models; (Kursun teaches “the system is also configured to generate an ensemble of machine learning models in order to provide a combined group of models that is determined to be accurate and efficient in identifying relevant patterns for determining irregular activity or suspected malfeasance. In this way, the system may identify and counteract the effects of schemes employed by users who attempt to inject irregular data, or “poison” any single machine learning model provided by the system.” [0019]).
performing, by the device, one or more mitigation actions responsive to the one or more outlier residual error vectors. (Kursun teaches “the reported data 401 may be received from users 102, entity systems 120, or third party systems 140. Suspected malfeasance or irregular/abnormal activity may be detected during a transaction or interaction with the entity, and may be reported to the entity so that it may later be labeled as fraudulent or abnormal input data…”[0051] and “As discussed previously, the ensemble 540 of machine learning models may be reconfigured based on determining that a particular model has been trained with abnormally injected data…” [0065]).
Kursun does not explicitly disclose:
computing, by the device and for each respective feedback of a plurality of collected feedbacks correlated with the session telemetry data, a residual error vector that quantifies a level of mis-prediction of each respective machine learning model as compared to each respective feedback;
clustering, by the device, each residual error vector into one or more clusters based on similarity to thereby identify one or more outlier residual error vectors that are not within any of the one or more clusters; and
However Yue teaches the limitation:
computing, by the device and for each respective feedback of a plurality of collected feedbacks correlated with the session telemetry data, a residual error vector that quantifies a level of mis-prediction of each respective machine learning model as compared to each respective feedback; (Yue teaches “detecting their label errors can significantly increase their efficacy. We propose a novel framework, called CTRL1 (Clustering TRaining Losses for label error detection), to detect label errors in multiclass datasets.” Page 1 Abstract Paragraph 1).
clustering, by the device, each residual error vector into one or more clusters based on similarity to thereby identify one or more outlier residual error vectors that are not within any of the one or more clusters; and (Yue teaches “We introduce a label error detection method that finds noisy labels of samples by clustering their training loss trajectories.” Page 1 Section 1 Introduction Paragraph 4 and “Among the six image datasets, Animal-10N and Food-101N contain real-world label errors to which we can directly apply our method without introducing artificial label noise.” Page 7 Section IV. Experimental Results Paragraph 1. Examiner notes that “noisy labels” refers to outliers and “training loss” refers to residual error vectors.).
Therefore, it would have been obvious to one of ordinary skill in the art, before the effective filling date of the claimed invention, to have modified a plurality of predictions computed by inputting a particular input of the session telemetry data into a respective machine learning model of a plurality of machine learning models of Kursun with clustering each residual error vector into one or more clusters based on similarity to thereby identify one or more outlier residual error vectors of Yue with a reasonable expectation of success. One of ordinary skill in the art would have been motivated to combine to identify and clean any classification dataset that may potentially have label errors (Yue Page 1 Abstract Paragraph 2).
Regarding Claim 2, Kursun in view of Yue, teaches the elements of claim 1 as outlined above, and further teaches:
The method of claim 1, wherein the one or more mitigation actions are selected from a group consisting of: marking one or more specific feedbacks of the plurality of collected feedbacks associated with the one or more outlier residual error vectors as anomalous; (Kursun teaches “injection interactions may be identified and removed from the data training set, along with other data that may be labeled as high risk or outlier data, as shown by block 705.” [0053]).
performing remediation against an adversarial feedback attack based on the one or more outlier residual error vectors. (Kursun teaches “the reported data 401 may be received from users 102, entity systems 120, or third party systems 140. Suspected malfeasance or irregular/abnormal activity may be detected during a transaction or interaction with the entity, and may be reported to the entity so that it may later be labeled as fraudulent or abnormal input data…”[0051] and “As discussed previously, the ensemble 540 of machine learning models may be reconfigured based on determining that a particular model has been trained with abnormally injected data…” [0065]).
Regarding Claim 3, Kursun in view of Yue, teaches the elements of claim 1 as outlined above, and further teaches:
The method of claim 1, wherein the plurality of machine learning models are based on a single model trained using a single method and architecture but with different initializations, subsampling, and/or feature masking. (Kursun teaches “Reconfiguration of the population of machine learning models may further include architectural changes through the addition and/or removal of particular models or model clusters as well as reweighting of models. As such, the present invention provides a technical solution to a technical problem of irregular data injection by implementing artificial intelligence and machine learning technologies in real time in order to shield from and counteract against poisoning of machine learning models designed to detect irregular activity.” [0019]. Examiner notes that “particular models or model clusters” refers to a plurality of models or a single model).
Regarding Claim 4, Kursun in view of Yue, teaches the elements of claim 1 as outlined above, and further teaches:
The method of claim 1, wherein the plurality of machine learning models are based on a plurality of different models trained on different methods and/or architectures. (Kursun teaches “Reconfiguration of the population of machine learning models may further include architectural changes through the addition and/or removal of particular models or model clusters as well as reweighting of models. As such, the present invention provides a technical solution to a technical problem of irregular data injection by implementing artificial intelligence and machine learning technologies in real time in order to shield from and counteract against poisoning of machine learning models designed to detect irregular activity.” [0019] Examiner notes that “particular models or model clusters” refers to a plurality of models or a single model).
Regarding Claim 5, Kursun in view of Yue, teaches the elements of claim 1 as outlined above, and further teaches:
The method of claim 1, wherein the plurality of predictions comprise quality-of- experience predictions, and wherein the plurality of collected feedbacks comprise quality-of-experience feedbacks. (Kursun teaches “In this way, users, who may be customers, employees, analysts, account-holders or the like, may manually self-report certain interaction or transaction data as irregular… As shown, user input 402 and analyst input 403 are subsets of reported data that the system may receive and process via machine learning model 406.”[0051]).
Regarding Claim 6, Kursun in view of Yue, teaches the elements of claim 1 as outlined above, and further teaches:
The method of claim 1, further comprising: attaching contextual troubleshooting information to the one or more outlier residual error vectors. (Yue teaches “However, when the model is robust against noisy labels, it demonstrates different convergences on different label types. In this case, we should assign its masked score a higher weight. We can get model robustness information by checking its loss histogram.” Page 5 Section III Methodology subsection D. Determining the Best Mask Paragraph 1. Examiner notes that using different procedures for different label types that detect noisy labels due to different contexts refers to contextual troubleshooting information).
Regarding Claim 7, Kursun in view of Yue, teaches the elements of claim 6 as outlined above, and further teaches:
The method of claim 6, wherein the contextual troubleshooting information is selected from a group consisting of: …additionally solicited feedback from new sources;…and (Kursun teaches “Different population samples can be generated using the different input groupings and learned label 407 gradients. In addition, ss each of the input streams may change over time, for example, associated losses and outlier patterns that may change over time, the learned label(s) 407 may be altered as well increase model accuracy and to minimize potential losses from irregular or malfeasant activity.” [0053]).
information on how the plurality of collected feedbacks was provided. (Kursun teaches “identifying a specific user or specific user account associated with the abnormally injected data and blocking interaction data associated with the specific user or specific user account. In other embodiments, responding to the abnormally injected data involves identifying a pattern of abnormally injected data based on metadata associated with the interaction data, and blocking he interaction data that fits the identified pattern.” [0065]).
Regarding Claim 8, Kursun in view of Yue, teaches the elements of claim 1 as outlined above, and further teaches:
The method of claim 1, wherein clustering is based on density-based spatial clustering of applications with noise. (Yue teaches “Based on speed and robustness considerations, we use Kmeans as the core clustering algorithm... Among popular clustering algorithms, …clusters as noisy-labeled” Page 4 Section III Methodology subsection B. Clustering Algorithm Paragraph 1-2).
Regarding Claim 9, Kursun in view of Yue, teaches the elements of claim 1 as outlined above, and further teaches:
The method of claim 1, wherein clustering is limited based on a temporal sliding window. (Yue teaches “We then split L into equal-length intervals by epoch number…We group the samples into k clusters and assign s clusters with larger areas under the loss curve as noisy. k and s are input parameters. In every window, we have a mask vaYue for each sample. Finally, we sum up the mask scores across windows and classify a sample as noisy if it is marked as noisy in enough windows.” Page 3 Section III Methodology Paragraph 2. Examiner notes that splitting labels in “equal-length intervals by epoch number” much like temporal sliding windows, allows the cluster to be divided accordingly in a timely manner for processing labels and in this case, splitting by windows of epochs, allows the clustering algorithm to do so.).
Regarding Claim 10, Kursun in view of Yue, teaches the elements of claim 1 as outlined above, and further teaches:
The method of claim 1, further comprising: removing the one or more outlier residual error vectors from consideration in clustering; and (Yue teaches “obtain the loss curve for each sample. Then, we apply clustering algorithms to the training losses to group samples. After label error detection, we remove samples with noisy labels and retrain the model…but a common problem is that many datasets have inaccuracies in their labeling…This algorithm can identify and clean any classification dataset that may potentially have label errors.” Page 1 Abstract Paragraph 1-2).
re-clustering each residual error vector other than the one or more outlier residual error vectors to thereby further identify one or more additional outlier residual error vectors. (Yue teaches “obtain the loss curve for each sample. Then, we apply clustering algorithms to the training losses to group samples. After label error detection, we remove samples with noisy labels and retrain the model…but a common problem is that many datasets have inaccuracies in their labeling…This algorithm can identify and clean any classification dataset that may potentially have label errors.” Page 1 Abstract Paragraph 1-2. Examiner notes that retraining the model entails re-clustering training loss or residual error as that is a part of the algorithms steps.).
Regarding Claim 11, Kursun in view of Yue, teaches the elements of claim 1 as outlined above, and further teaches:
The method of claim 1, wherein the one or more mitigation actions comprise discarding one or more specific feedbacks of the plurality of collected feedbacks associated with the one or more outlier residual error vectors from use in training the plurality of machine learning models. (Kursun teaches “responding to the abnormally injected data further comprises identifying a specific user or specific user account associated with the abnormally injected data; and blocking interaction data associated with the specific user or specific user account.” [0004]).
Regarding Claim 12, Kursun in view of Yue, teaches the elements of claim 1 as outlined above, and further teaches:
The method of claim 1, wherein the one or more mitigation actions comprise preventing feedback participation from one or more specific sources of the plurality of collected feedbacks associated with the one or more outlier residual error vectors that are deemed to be adversarial. (Kursun teaches “responding to the abnormally injected data further comprises receiving metadata associated with the interaction data; identifying a pattern of abnormally injected data, wherein the pattern of abnormally injected data is based on the metadata associated with the interaction data; and blocking interaction data associated with the identified pattern of abnormally injected data.” [0005]).
Regarding Claim 13, Kursun in view of Yue, teaches the elements of claim 1 as outlined above, and further teaches:
The method of claim 1, further comprising: recomputing a new plurality of predictions based on the session telemetry data with one or more specific feedbacks of the plurality of collected feedbacks associated with the one or more outlier residual error vectors being removed; and (Kursun teaches “injection interactions may be identified and removed from the data training set, along with other data that may be labeled as high risk or outlier data, as shown by block 705. At this point, the model may be retrained based on interactions cleared by the trained models, as shown at block 706. In this way, the system ensures that abnormal data injection is not harming the training process of the machine learning model(s) 406 by being accounted for as legitimate interaction data.” [0062]).
confirming that the new plurality of predictions is an improvement on the plurality of predictions. (Kursun teaches “injection interactions may be identified and removed from the data training set, along with other data that may be labeled as high risk or outlier data, as shown by block 705. At this point, the model may be retrained based on interactions cleared by the trained models, as shown at block 706. In this way, the system ensures that abnormal data injection is not harming the training process of the machine learning model(s) 406 by being accounted for as legitimate interaction data.” [0062] and “In addition, ss each of the input streams may change over time, for example, associated losses and outlier patterns that may change over time, the learned label(s) 407 may be altered as well increase model accuracy and to minimize potential losses from irregular or malfeasant activity.”[0053]).
Regarding Claim 14, Kursun in view of Yue, teaches the elements of claim 1 as outlined above, and further teaches:
The method of claim 1, further comprising: collecting the plurality of collected feedbacks based on polling users about their subjective experience during an application session; and (Kursun teaches “the user 102 is a user interacting with, maintaining, or employing a population of machine learning models, wherein the system enables the user to reconfigure the population based on user-specified criteria and policies…” [0033] and “…ensemble 540, are deployed and configured to adaptively monitor interactions between one or more users 102 and one or more entities, such as entity systems 120, third party systems 140, and other users 102. As shown at block 812, the system may receive interaction data for interactions between the one or more users and one or more entities…” [0064]).
correlating each of the plurality of collected feedbacks with a respective input of the session telemetry data. (Kursun teaches “These configurations may be adjusted at any time by the user according to further assessment of the effectiveness and accuracy of particular models or effectiveness and accuracy of the ensemble 540. The models 1-N and ensemble 540 are monitored by the system in order to assess accuracy and irregularities in predictive output.” [0057]).
Regarding Claim 15, the claim recites similar limitations as claim 1 and is rejected for similar reasons as claim 1 using similar teachings and rationale. Claim 1 is directed to a method and claim 15 is directed to an apparatus. Kursun in view of Yue teaches the elements of claim 15, and further teaches:
An apparatus, comprising: one or more network interfaces to communicate with a network; a processor coupled to the one or more network interfaces and configured to execute one or more processes; and a memory configured to store a process that is executable by the processor, the process comprising: (Kursun teaches “may be embodied as an apparatus (including, for example, a system, a machine, a device, a computer program product, and/or the like)”[0066]).
Regarding Claim 16, the claim recites similar limitation as claim 2 and is rejected for similar reasons as claim 2 using similar teachings and rationale.
Regarding Claim 17, the claim recites similar limitation as claim 5 and is rejected for similar reasons as claim 5 using similar teachings and rationale.
Regarding Claim 18, the claim recites similar limitation as claim 6 and is rejected for similar reasons as claim 6 using similar teachings and rationale.
Regarding Claim 19, the claim recites similar limitation as claim 11 and is rejected for similar reasons as claim 11 using similar teachings and rationale.
Regarding Claim 20, the claim recites similar limitations as claim 1 and is rejected for similar reasons as claim 1 using similar teachings and rationale. Claim 1 is directed to a method and claim 20 is directed to a non-transitory computer-readable medium. Kursun in view of Yue teaches the elements of claim 20, and further teaches:
A tangible, non-transitory, computer-readable medium storing program instructions that cause a device to execute a process comprising (Kursun teaches “As used herein, a processor may be “configured to” perform a certain function in a variety of ways, including, for example, by having one or more special-purpose circuits perform the functions by executing one or more computer-executable program code portions embodied in a computer-readable medium, and/or having one or more application-specific circuits perform the function.”[0066]).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Nadeem A Abdulmelik whose telephone number is (571)270-3584. The examiner can normally be reached Monday-Friday (8:00 am - 5:00 pm) EST Off every other Friday.
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, Li B Zhen can be reached at (571) 272-3768. 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.
/NADEEM AMIR ABDULMELIK/Examiner, Art Unit 2121
/Li B. Zhen/Supervisory Patent Examiner, Art Unit 2121
