Prosecution Insights
Last updated: November 04, 2026
Application No. 19/328,168

Methods and Systems for Suspected Threat Prioritization and Remediation

Non-Final OA §101§103
Filed
Sep 14, 2025
Priority
Mar 09, 2025 — provisional 63/769,088
Examiner
MOLES, JAMES P
Art Unit
2494
Tech Center
2400 — Computer Networks
Assignee
Veriti Security Ltd.
OA Round
1 (Non-Final)
67%
Grant Probability
Favorable
1-2
OA Rounds
1y 8m
Est. Remaining
95%
With Interview

Examiner Intelligence

Grants 67% — above average
67%
Career Allowance Rate
32 granted / 48 resolved
+8.7% vs TC avg
Strong +29% interview lift
Without
With
+28.6%
Interview Lift
resolved cases with interview
Typical timeline
2y 9m
Avg Prosecution
12 currently pending
Career history
66
Total Applications
across all art units

Statute-Specific Performance

§101
8.2%
-31.8% vs TC avg
§103
62.6%
+22.6% vs TC avg
§102
8.2%
-31.8% vs TC avg
§112
17.6%
-22.4% vs TC avg
Black line = Tech Center average estimate • Based on career data from 48 resolved cases

Office Action

§101 §103
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 . This office action is in response to the applicant’s filing on 09/14/2025. Claims 1-20 are pending. Claims 1, 19, and 20 are independent. Priority Acknowledgement is made of applicant’s claiming of priority to U.S. Provisional application No. 63/769,088 filed on 03/09/2025. 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 an abstract idea without significantly more. Claim 19 recites “obtain security data representing a suspected threat to a client, extract a plurality of technical features for the suspected threat from the security data, classify, by a first trained machine learning model, an authenticity of the suspected threat based on the plurality of technical features, estimate, by a second trained machine learning model, an impact on the client of addressing the suspected threat based on the plurality of extracted technical features, and determine by a third trained machine learning model, a remediation for the suspected threat based on the authenticity classification and the estimated impact”. But for the use of generic machine learning models, the limitations of classifying an authenticity of a suspected threat based on a plurality of technical features by a first trained machine learning model; estimating an impact on a client of addressing the suspected threat based on the plurality of extracted technical features by a second trained machine learning model; and determining a remediation for the suspected threat based on the outputs of the classifying and estimating by a third trained machine learning model, under their broadest reasonable interpretation, are mental processes. Furthermore, obtaining security data representing a suspected threat to a client, and extracting a plurality of technical features for the suspected threat from the security data, under their broadest reasonable interpretation, are mental processes. That is, other than reciting the use of a processor, memory, and multiple generic trained machine learning models, nothing in the claim elements precludes the steps from practically being performed in the mind. If a claim limitation, covers the performance of the limitation in the mind but for the recitation of generic computer technology, then it falls within the “Mental Processes” grouping of abstract ideas. Accordingly, the claim recites a judicial exception. The additional elements of the claim do not integrate the judicial exception into a practical application. The claim recites “a computerized processor; and a non-transitory storage medium for storing instructions that, when executed by the computerized processor, cause the system to: …by a first trained machine learning model … by a second trained machine learning model … by a third trained machine learning model.” The additional elements of a processor, memory, and multiple generic trained machine learning models amount to no more than mere instructions to apply the exception using a generic computer and/or computer components or trained machine learning model. Hence, the claim is directed to an abstract idea. The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As explained above with respect to integration of the abstract ideas into a practical application, the use of a processor, memory with instructions, and multiple generic machine learning models amount to no more than mere instructions to apply the exception using a generic computer and/or computer component. Therefore, the claim is not patent eligible. Claim 1 is a method that recites the steps corresponding to the functions recited in claim 19. Hence, claim 1 is patent ineligible for substantially the same reasons as claim 19. Claim 20 is a computer usable non-transitory storage medium having a computer program embodied thereon for causing a computer to execute the functionality recited in claim 19. The use of a generic computer component as a tool to implement the abstract idea does not integrate the exception into a practical application, nor does it amount to significantly more than the abstract idea. Dependent claims 2-6 merely recite what the features extracted from the security data include, along with how some of the features are calculated, and is part of the mental process. Dependent claims 7-8 recite where the security data is obtained from, how it is aggregated, and how it is processed to identify the suspected threat, and are part of the mental process. Official notice is taken that obtaining security data from these sources is well-understood, routine, and conventional activity in the art. Dependent claim 9 recites processing the obtained security data, and extracting the technical features from the processed data, which falls into the mental process. Dependent claim 10 recites how weights are applied to the authenticity classification and the estimated impact processed by the third trained machine learning model, which is part of the mental process. Dependent claims 11 and 12 specify the type of machine learning model that the first and second trained machine learning model embody, specifically, a classifier and regression model respectively, and the claims further discuss the form of their output. Official notice is taken that these types of ML models are well-understood, routine and conventional types of ML models; therefore, the elements do not integrate the judicial exception into a practical application. Dependent claims 13-15 recite how the first and second models are trained, what data they are trained on, and receiving user feedback for training. Official notice is taken that these are all conventional AI activities; hence, these limitations do not integrate the judicial exception into a practical application. Dependent claim 16 recites what the remediation comprises, however, it is part of the mental process as it merely recites what the remediation may be. Dependent claim 17 recites generating an alert or report and that the remediation is included in the alert or report, and this is insignificant post-solution activity. Dependent claim 18 recites how the third machine learning model determines a score based on the authenticity classification and the estimated impact, which falls into the mental process. Hence, these dependent claims are also not eligible. Claim Rejections - 35 USC § 103 In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status. The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action: A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made. 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. This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention. Claims 1-4, 7-9, 11, 13-16, and 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over Ringlein et al. (US PGPub No. 2020/0401696; hereinafter “Ringlein”) in view of Bazalgette et al. (US PGPub No. 2024/0223592; hereinafter “Bazalgette”). As per claim 1: Ringlein discloses a computer-implemented method for determining remediations for suspected threats, the method comprising: obtaining security data representing a suspected threat to a client (receive, from a source computing device of a monitored computing environment, a security incident comprising a security incident data structure comprising metadata describing properties of the security incident, and a corresponding security knowledge graph [Ringlein ¶ 0007, ¶ 0059]; The security event data stored in the security log 112 may specify various events associated with the particular managed computing resources 101 that represent events of interest to security evaluations, e.g., failed login attempts, password changes, network traffic patterns, system configuration changes [Ringlein ¶ 0062]; FIG. 1B, the new security incidents/alerts and their corresponding security knowledge graphs 150 may be input directly into the SID computing system 140. That is, similar to the operation of the SIEM system 110 during the training operation, the SIEM system 110 generates new security incidents/alerts, which are the basis for the generation of security knowledge graphs 150 which, during runtime operation, are sent directly to the SID computing system 140 for feature extraction and metric generation 143 by the FEE 144 [Ringlein ¶ 0082]); extracting a plurality of technical features for the suspected threat from the security data (The security incident disposition system further operates to process the security incident data structure and security knowledge graph to extract, from the metadata of the security incident data structure and data corresponding to the nodes and edges of the security knowledge graph, a set of security incident features corresponding to the security incident [Ringlein ¶ 0007, ¶ 0009, ¶ 0033-0034, ¶ 0045]; calculating one or more metrics based on the extracted set of security incident features, and adding the one or more metrics as additional security incident features to the set of security incident features prior to inputting the extracted set of security incident features into the trained security incident machine learning model. The calculated metrics provide more insight into the nature of the security incident that permits a more accurate prediction of the security incident disposition [Ringlein ¶ 0009]; While traversing the topology of the security knowledge graph, various counts may be maintained for some of the extracted features to assist in generating metric features. For example, counts of each type of node, counts of each type of edge, counts of toxic nodes, counts of numbers of nodes having particular reputations, e.g., malware, etc [Ringlein ¶ 0072]; These features are added to features extracted from the metadata associated with the security incident or alert itself. For the security incident or alert itself, the metadata associated with the security incident or alert may be analyzed to extract features such as "magnitude", "severity", and "credibility" from the metadata. In one illustrative embodiment, the features extracted from the security incident/alert and the security knowledge graph are features that may be encoded in a numeric, binary, or normal fashion [Ringlein ¶ 0073, ¶ 0076-0077]); classifying, by a first trained machine learning model, an authenticity of the suspected threat based on the plurality of technical features (Moreover, the security incident disposition system operates to input the extracted set of security incident features into a trained security incident machine learning model of the security incident disposition system and generate, by the trained security incident machine learning model of the security incident disposition system, a disposition classification output based on results of processing the extracted set of security incident features by the trained security incident machine learning model. The security incident disposition system outputs the disposition classification output to the source computing device [Ringlein ¶ 0007]; the trained security incident machine learning model is a neural network trained on a set of training data comprising security incidents, corresponding security knowledge graphs for the security incidents in the set of training data, and corresponding correct disposition classifications for the security incidents [Ringlein ¶ 0011]; The features are input to the trained security incident ML model which then generates a prediction of a disposition, e.g., a responsive action to perform, based on a cognitive evaluation of the patterns of features present in the input. For example, the disposition may be to escalate or not to escalate the security incident, where an escalation of the security incident indicates that the security incident is considered to be a likely threat or attack on the computing system [Ringlein ¶ 0035-0036]; The predicted disposition may be a single disposition such as from a binary classifier type neural network, e.g., a binary output value indicating whether to escalate/not escalate, or a vector of output values indicating corresponding probabilities that a disposition associated with the output value is a correct disposition for the security incident. For example, the security incident ML model may output a vector having a plurality of vector slots, where each vector slot corresponds to a particular disposition, and a value in the corresponding vector slot represents a probability (or confidence) that the corresponding disposition is the correct disposition for the security incident [Ringlein ¶ 0038]; whether the security alert should be escalated or not escalated (it can be appreciated that a SIEM system will not generate an alert for events that do not represent potential threats and thus, security alerts only represent either true threats (escalate) or false positives (do not escalate)) [Ringlein ¶ 0058]; This machine learning may be a supervised or unsupervised machine learning operation, such as by executing the CNN model on the extracted features/metrics for security alerts to thereby generate a disposition classification of the security incident/alert as to whether it represents a true threat requiring escalation or a false positive that should not be escalated, and then comparing the result generated by the CNN model to the correct disposition classification for the security alert as specified by the security analyst 131 [Ringlein ¶ 0080, ¶ 0082]; The trained predictive model 148 operates on the features input 160 comprising the extracted features from the security incident/alert and the security knowledge graph, as well as generated metrics based on these extracted features, to apply the trained feature analysis operations, e.g., the activation functions, of the nodes/neurons of the predictive model, e.g., the CNN, to the features input 160 and generate a disposition classification for the security incident/alert, e.g., whether to escalate or not escalate the security incident/alert. The output disposition classification generated by the predictive model 148 may be provided to a security analyst 131, such as via the threat monitoring interface 130 in FIG. 1A, for further evaluation, logged in the security alert database 138, or otherwise made available for further use in determining a responsive action to take if deemed necessary [Ringlein ¶ 0083, ¶ 0088]); [estimating, by a second trained machine learning model, an impact on the client of addressing the suspected threat based on the plurality of extracted technical features]; and determining, [by a third trained machine learning model], a remediation for the suspected threat based on the authenticity classification (In some illustrative embodiments, responsive actions may be automatically initiated in response to the disposition output generated by the security incident ML model. For example, in cases where the confidence or probability associated with the disposition output is greater than a predetermined threshold, i.e. the security incident ML model has significant confidence that the disposition is correct for the security incident, automated processes may be initiated to perform the responsive action to protect the computing resources of the monitored environment. For example, the disposition recommendation made by the security incident ML model may be provided back to the SIEM system with an instruction to initiate SIEM system functionality for responding to the detected security threat, e.g., blocking access to computing resources by a device associated with a particular address, or the like [Ringlein ¶ 0044]; The output disposition classification generated by the predictive model 148 may be provided to a security analyst 131, such as via the threat monitoring interface 130 in FIG. 1A, for further evaluation, logged in the security alert database 138, or otherwise made available for further use in determining a responsive action to take if deemed necessary [Ringlein ¶ 0083]; The responsive actions may take many different forms, such as generating alert notifications, inhibiting operation of particular computer components, or the like [Ringlein ¶ 0025]) [and the estimated impact]. Ringlein discloses the claimed subject matter as discussed above but does not explicitly disclose estimating, by a second trained machine learning model, an impact on the client of addressing the suspected threat based on the plurality of extracted technical features; determining, by a third trained machine learning model, a remediation for the suspected threat based on the authenticity classification and the estimated impact. However, Bazalgette teaches estimating, by a second trained machine learning model, an impact on the client of addressing the suspected threat based on the plurality of extracted technical features (The cyber-attack simulator 105 with its Artificial Intelligence trained on how to conduct and perform cyberattack in a simulation in either a simulator or in a clone creator spinning up virtual instances on virtual machines will take a sequence of actions and then evaluate the actual impact after each action in the sequence, in order to yield a best possible result to contain/mitigate the detected threat while minimizing the impact on other network devices and users that are i) currently active and ii) not in breach, from different possible actions to take. Again, multiple simulations can be run in parallel so that the different sequences of mitigation actions and restoration actions can be evaluated essentially simultaneously. The cyber-attack simulator 105 with Artificial Intelligence-based simulations in the cyber-attack simulator 105 is configured to use one or more mathematical functions to generate a score and/or likelihood for each of the possible actions and/or sequence of multiple possible actions that can be taken in order to determine which set of actions to choose among many possible actions to initiate. The one or more possible actions to take and their calculated scores can be stacked against each other to factor 1) a likelihood of containing the detected threat acting abnormal with each possible set of actions, 2) a severity level of the detected threat to the network, and 3) the impact of taking each possible set of actions i) on users and ii) on devices currently active in the network not acting abnormal to the normal behavior of the network, and then communicate with the cyber threat detection engine, the autonomous response engine, and the cyber-security restoration engine, respectively, to initiate the chosen set of actions to cause a best targeted change of the behavior of the detected threat acting abnormal to the normal pattern of life on the network while minimizing the impact on other network devices and users that are i) currently active and ii) not in breach of being outside the normal behavior benchmark. The cyber-attack simulator cooperates with the AI models modelling a normal pattern of life for entities/nodes in the system being protected [Bazalgette ¶ 0195, Examiner’s Note: scores to factor the impact of taking each possible set of actions i) on users and ii) on devices currently active in the network not acting abnormal to the normal behavior of the network]; This risk assessment and ranking is conducted to avoid frequent “false positive” alerts that diminish the degree of reliance/confidence on the cyber security appliance 100. The cyber-attack simulator 105 may include and cooperate with one or more AI models trained with machine learning on the contextual knowledge of the organization. These trained AI models may be configured to identify data points from the contextual knowledge of the organization and its entities, which may include, but is not limited to, language-based data, email/network connectivity and behavior pattern data, and/or historic knowledgebase data. The cyber-attack simulator 105 may use the trained AI models to cooperate with one or more AI classifier(s) by producing a list of specific organization-based classifiers for the AI classifier [Bazalgette ¶ 0110]; the cyber-attack simulator 105 using Artificial Intelligence-based simulations is communicatively coupled to a cyber security appliance 100, an open source (OS) database server 790, an email system 796, one or more endpoint computing devices 791A-B, and an IT network system 792 with one or more entities, over one or more networks 791/792 in the system being protected [Bazalgette ¶ 0185]; The cyber-attack simulator 105 with Artificial Intelligence-based simulations is configured to integrate with the cyber security appliance 100 and cooperate with components within the cyber security appliance 100 installed and protecting the network from cyber threats by making use of outputs, data collected, and functionality from two or more of a data store, other modules, and one or more AI models already existing in the cyber security appliance 100 [Bazalgette ¶ 0186]); determining, by a third trained machine learning model, a remediation for the suspected threat based on the authenticity classification and the estimated impact (an autonomous response engine using Artificial Intelligence algorithms configured and trained to perform a second machine-learned task of taking one or more mitigation actions to mitigate the cyber threat, iii) a cyber-security restoration engine using Artificial Intelligence algorithms configured and trained to perform a third machine-learned task of remediating the system being protected back to a trusted operational state [Bazalgette ¶ 0013]; The cyber-attack simulator 105 with Artificial Intelligence-based simulations in the cyber-attack simulator 105 is configured to use one or more mathematical functions to generate a score and/or likelihood for each of the possible actions and/or sequence of multiple possible actions that can be taken in order to determine which set of actions to choose among many possible actions to initiate. The one or more possible actions to take and their calculated scores can be stacked against each other to factor 1) a likelihood of containing the detected threat acting abnormal with each possible set of actions, 2) a severity level of the detected threat to the network, and 3) the impact of taking each possible set of actions i) on users and ii) on devices currently active in the network not acting abnormal to the normal behavior of the network, and then communicate with the cyber threat detection engine, the autonomous response engine, and the cyber-security restoration engine, respectively, to initiate the chosen set of actions to cause a best targeted change of the behavior of the detected threat acting abnormal to the normal pattern of life on the network while minimizing the impact on other network devices and users that are i) currently active and ii) not in breach of being outside the normal behavior benchmark [Bazalgette ¶ 0195]; This risk assessment and ranking is conducted to avoid frequent “false positive” alerts that diminish the degree of reliance/confidence on the cyber security appliance 100. The cyber-attack simulator 105 may include and cooperate with one or more AI models trained with machine learning on the contextual knowledge of the organization. These trained AI models may be configured to identify data points from the contextual knowledge of the organization and its entities, which may include, but is not limited to, language-based data, email/network connectivity and behavior pattern data, and/or historic knowledgebase data. The cyber-attack simulator 105 may use the trained AI models to cooperate with one or more AI classifier(s) by producing a list of specific organization-based classifiers for the AI classifier. The cyber-attack simulator 105 is further configured to calculate, based at least in part on the results of the one or more hypothetical simulations of a possible cyberattack and/or of an actual ongoing cyberattack from a cyber threat determine a risk score for each node (e.g. each device, user account, etc.), the threat risk score being indicative of a possible severity of the compromise prior to an autonomous response action is taken in response to the actual cyberattack of the cyber incident [Bazalgette ¶ 0110]; FIG. 5, the cyber security restoration engine is configured to take one or more remediation actions based on configured and/or Artificial Intelligence assistance to remediate the one or more nodes in the graph of the system being protected back to a trusted operational state in a recovery from the cyber threat. These actions might be fully automatic, or require a specific human confirmation decision before they begin. The cyber security restoration engine is configured to cooperate with the other AI-based engines of the cyber security system, via the interfaces and/or direct integrations, to track and understand the cyber threat identified by the other components as well as track the one or more mitigation actions taken to mitigate the cyber threat during the cyberattack by the other components in order to assist in intelligently restoring the protected system while still mitigating the cyber threat attack back to a trusted operational state; and thus, as a situation develops with an ongoing cyberattack, the cyber security restoration engine is configured to take one or more remediation actions to remediate (e.g. restore) at least one of the nodes in the graph of the protected system back to a trusted operational state while the cyberattack is still ongoing [Bazalgette ¶ 0097]). Ringlein and Bazalgette are analogous art because they are from the same field of endeavor of machine learning security. Therefore, based on Ringlein in view of Bazalgette, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of Bazalgette to the system of Ringlein in order to effectively improve detection and remediation capabilities and limiting an impact of the cyber threat from consuming unauthorized CPU cycles, memory space, and power consumption. Hence, it would have been obvious to combine the references above to obtain the invention as specified in the instant claim. As per claim 2: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Ringlein discloses wherein the plurality of technical features (process the security incident data structure and security knowledge graph to extract, from the metadata of the security incident data structure and data corresponding to the nodes and edges of the security knowledge graph, a set of security incident features corresponding to the security incident [Ringlein ¶ 0007, ¶ 0009, ¶ 0033-0034, ¶ 0045]; These features are added to features extracted from the metadata associated with the security incident or alert itself. For the security incident or alert itself, the metadata associated with the security incident or alert may be analyzed to extract features such as "magnitude", "severity", and "credibility" from the metadata. In one illustrative embodiment, the features extracted from the security incident/alert and the security knowledge graph are features that may be encoded in a numeric, binary, or normal fashion [Ringlein ¶ 0073, ¶ 0076-0077]) includes a Direction of Connection feature (Metric features that may be generated based on the features extracted from the edges, in accordance with one illustrative embodiment, may be a count of each edge type. The metric features may also include a count of a number of QRadar® Advisor with Watson™ (QRA) relevant triples, i.e. node->edge->node combinations as defined above that indicate potential malicious connections. So, domain->resolves-> IP, or ['File', 'CONNECT', 'Uri'], ['EmailContent', 'ATTACHMENT', 'File'], ['File', 'CONNECT', 'IpAddress'], ['File', 'CONNECT', 'DomainName'], for example. For example, if a File connects to a URL, it could be an example of a malware communicating with a command and control server. This is denoted by the triple ['File', 'CONNECT', 'Uri'] where node 1 is of type File, node 2 is of type 'URL' and they are connected by an edge of type 'CONNECT' [Ringlein ¶ 0077]). As per claim 3: Ringlein in view of Bazalgette teach all the limitations of claim 2. Furthermore, Ringlein discloses wherein the Direction of Connection feature is determined by generating a topology of digital assets of the client (In extracting features from the security knowledge graph, selected features may be extracted from the nodes, edges, and overall topology of the graph in accordance with a specific configuration of the feature extraction mechanisms [Ringlein ¶ 0033]; Metric features that may be generated based on the features extracted from the edges, in accordance with one illustrative embodiment, may be a count of each edge type. The metric features may also include a count of a number of QRadar® Advisor with Watson™ (QRA) relevant triples, i.e. node->edge->node combinations as defined above that indicate potential malicious connections. So, domain->resolves->IP, or [‘File’, ‘CONNECT’, ‘Url’], [‘EmailContent’, ‘ATTACHMENT’, ‘File’], [‘File’, ‘CONNECT’, ‘IpAddress’], [‘File’, ‘CONNECT’, ‘DomainName’], for example. For example, if a File connects to a URL, it could be an example of a malware communicating with a command and control server. This is denoted by the triple [‘File’, ‘CONNECT’, ‘Url’] where node 1 is of type File, node 2 is of type ‘URL’ and they are connected by an edge of type ‘CONNECT’ [Ringlein ¶ 0077]; the security incident disposition system further operates to process the security knowledge graph to extract topology features representing an overall topology of the security knowledge graph, and inputting the extracted topology features into the trained security incident machine learning model [Ringlein ¶ 0008]; When extracting features from the security knowledge graph, the security knowledge graph is traversed from the root node and selected features of the encountered nodes, edges, and the overall topology are extracted and stored as part of a feature set representative of the security incident [Ringlein ¶ 0071]; For edge features, an example of an edge feature that may be extracted include the edge type, e.g., “contains”, “connects”, and “resolves”. With regard to topology features, examples of topology features that may be extracted from the security knowledge graph include number of nodes which have at least X incoming edges, number of nodes that have at least X outgoing edges, tuples (such as a triplet tuple (node-edge-node)) indicating combinations that are toxic that indicate potential malicious connections, etc. where X is any suitable value. While traversing the topology of the security knowledge graph, various counts may be maintained for some of the extracted features to assist in generating metric features. For example, counts of each type of node, counts of each type of edge, counts of toxic nodes, counts of numbers of nodes having particular reputations, e.g., malware, etc [Ringlein ¶ 0072]). As per claim 4: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Ringlein discloses wherein the plurality of technical features includes at least one of: a Protection Confidence parameter, a Protection Severity parameter (These features are added to features extracted from the metadata associated with the security incident or alert itself. For the security incident or alert itself, the metadata associated with the security incident or alert may be analyzed to extract features such as "magnitude", "severity", and "credibility" from the metadata. In one illustrative embodiment, the features extracted from the security incident/alert and the security knowledge graph are features that may be encoded in a numeric, binary, or normal fashion [Ringlein ¶ 0073, Examiner’s Note: see specification page 13, lines 12-14, “… this feature may represent the severity of a potential threat such as a security event reflected by the log”]), a Number of Logs with a same direction-protection combination, a Number of Unique Destination IP Addresses for a same direction-protection combination, a Mean Time Deltas between consecutive logs, a Standard Deviation of Time Deltas between consecutive logs, or a Percentage of Outliers in time deltas. As per claim 7: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Ringlein discloses wherein the security data is obtained from one or more security systems of the client, the one or more security systems including at least one of an Intrusion Prevention System (IPS), a firewall, an AntiBot system, an AntiVirus system, or a Security Information and Event Management (SIEM) system (Security Incident and Event Management (SIEM) is an approach to security management that combines security information management with security event monitoring functions into a single security management system. A SIEM system aggregates data from various data sources in order to identify deviations in the operation of the computing devices associated with these data sources from a normal operational state, and then take appropriate responsive actions to the identified deviations. SIEM systems may utilize multiple collection agents that gather security related events from computing devices, network equipment, firewalls, intrusion prevention systems, antivirus systems, and the like. The collection agents may then send this information, or a subset of this information that has been pre-processed to identify only certain events for forwarding, to a centralized management console where security analysts examine the collected event data and prioritize events as to their security threats for appropriate responsive actions [Ringlein ¶ 0025, Fig. 1A, Fig. 1B]; As shown in FIG. 1A, a SIEM computing system 110, which may be positioned in an end user computing environment 105, obtains security log information from managed computing resources 101 in the end user computing environment 105, e.g., servers, client computing devices, computing network devices, firewalls, database systems, software applications executing on computing devices, and the like [Ringlein ¶ 0059, Examiner’s Note: client computing devices]). As per claim 8: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Ringlein discloses wherein obtaining the security data includes aggregating security logs of the client into events (A SIEM system aggregates data from various data sources in order to identify deviations in the operation of the computing devices associated with these data sources from a normal operational state, and then take appropriate responsive actions to the identified deviations. SIEM systems may utilize multiple collection agents that gather security related events from computing devices, network equipment, firewalls, intrusion prevention systems, antivirus systems, and the like. The collection agents may then send this information, or a subset of this information that has been preprocessed to identify only certain events for forwarding, to a centralized management console where security analysts examine the collected event data and prioritize events as to their security threats for appropriate responsive actions [Ringlein ¶ 0025, Fig. 1A]; As shown in FIG. 1A, a SIEM computing system 110, which may be positioned in an end user computing environment 105, obtains security log information from managed computing resources 101 in the end user computing environment 105, e.g., servers, client computing devices, computing network devices, firewalls, database systems, software applications executing on computing devices, and the like. Security monitoring engines 103 may be provided in association with these managed computing resources, such as agents deployed and executing on endpoint computing devices, which collect security events and provide the security event data to the SIEM computing system 110 where it is logged in a security log data structure 112. In some illustrative embodiments, the security monitoring engine(s) 103 themselves may be considered an extension of the SIEM computing system 110 and may apply SIEM rules to perform analysis of the security events to identify event data indicative of suspicious activity that may be indicative of a security attack or vulnerability, triggering a security alert 120 or security log entry to be generated. Moreover, the security monitoring engine(s) 103 may provide such security event log information to the SIEM system 110 for further evaluation. In other illustrative embodiments, the security event data gathered by the security monitoring engines 103 may be provided to the SIEM system 110 for logging in a security log 112 and for generation of security alerts 120 [Ringlein ¶ 0059]). As per claim 9: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Ringlein discloses further comprising: processing the obtained security data to identify the suspected threat, and wherein extracting the plurality of technical features for the suspected threat is performed on the processed security data (the processing of the security incident data structure and securing knowledge graph to extract security incident features further comprises calculating one or more metrics based on the extracted set of security incident features, and adding the one or more metrics as additional security incident features to the set of security incident features prior to inputting the extracted set of security incident features into the trained security incident machine learning model. The calculated metrics provide more insight into the nature of the security incident that permits a more accurate prediction of the security incident disposition [Ringlein ¶ 0009]; given a security incident from an end user's computing environment, e.g., from a Security Incident and Event Management (SIEM) system, or one or more collection agents associated with a SIEM system, extracts features from the security incident, obtains a security knowledge graph for the security incident, extracts features from the nodes, edges, and/or topology of the security knowledge graph, and generates metrics that themselves are regarded as additional features, based on a traversal of the security knowledge graph [Ringlein ¶ 0034]; As shown in FIG. 1A, a SIEM computing system 110, which may be positioned in an end user computing environment 105, obtains security log information from managed computing resources 101 in the end user computing environment 105, e.g., servers, client computing devices, computing network devices, firewalls, database systems, software applications executing on computing devices, and the like. Security monitoring engines 103 may be provided in association with these managed computing resources, such as agents deployed and executing on endpoint computing devices, which collect security events and provide the security event data to the SIEM computing system 110 where it is logged in a security log data structure 112. In some illustrative embodiments, the security monitoring engine(s) 103 themselves may be considered an extension of the SIEM computing system 110 and may apply SIEM rules to perform analysis of the security events to identify event data indicative of suspicious activity that may be indicative of a security attack or vulnerability, triggering a security alert 120 or security log entry to be generated. Moreover, the security monitoring engine(s) 103 may provide such security event log information to the SIEM system 110 for further evaluation. In other illustrative embodiments, the security event data gathered by the security monitoring engines 103 may be provided to the SIEM system 110 for logging in a security log 112 and for generation of security alerts 120 [Ringlein ¶ 0059]). As per claim 11: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Ringlein discloses wherein the first trained machine learning model is a classifier that provides an authenticity classification as a probability vector (The predicted disposition may be a single disposition such as from a binary classifier type neural network, e.g., a binary output value indicating whether to escalate/not escalate, or a vector of output values indicating corresponding probabilities that a disposition associated with the output value is a correct disposition for the security incident. For example, the security incident ML model may output a vector having a plurality of vector slots, where each vector slot corresponds to a particular disposition, and a value in the corresponding vector slot represents a probability (or confidence) that the corresponding disposition is the correct disposition for the security incident [Ringlein ¶ 0038]). As per claim 13: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Ringlein discloses wherein the first trained machine learning model is trained using a labeled dataset of events labeled as legitimate attacks (the trained security incident machine learning model is a neural network trained on a set of training data comprising security incidents, corresponding security knowledge graphs for the security incidents in the set of training data, and corresponding correct disposition classifications for the security incidents [Ringlein ¶ 0011, ¶ 0038-0039]; The security incident ML model, in some illustrative embodiments, is a neural network based model that is trained through a supervised or unsupervised machine learning operation. The neural network model may be a convolutional neural network (CNN), a recurrent neural network (RNN), Long Short-Term Memory (LSTM) neural network, or any other type of neural network model architecture that is currently known or later developed [Ringlein ¶ 0036]; and then label the security alert 120 as to the proper disposition for the security alert 132, e.g., an actual threat which should be escalated 134 or a false positive alert that should not be escalated 136. The labeled security alert data including the security alert, the security knowledge graph, and the security analyst's provided disposition label are then stored in an entry in the security alert database 138 to thereby generate a training dataset comprising security alert entries and corresponding disposition labels as to the correct disposition classification of the security alert [Ringlein ¶ 0063]) or as benign events (During training, the security analyst 131 may view the security incidents/alerts and security knowledge graphs 120, 150 using a user interface that is accessible within the end user computing environment 105 and may provide the security analyst provided disposition labels as to whether the security incidents/alerts represent true threats (escalate) or false positives ( do not escalate). The outputting of such security incidents/alerts, security knowledge graphs, and corresponding correct labels may be performed as a training dataset that is exported directly to the SID system 140 for processing by the predictive model 148 [Ringlein ¶ 0085]; the trained security incident machine learning model is a neural network trained on a set of training data comprising security incidents, corresponding security knowledge graphs for the security incidents in the set of training data, and corresponding correct disposition classifications for the security incidents [Ringlein ¶ 0011]; The security incident ML model, in some illustrative embodiments, is a neural network based model that is trained through a supervised or unsupervised machine learning operation. The neural network model may be a convolutional neural network (CNN), a recurrent neural network (RNN), Long Short-Term Memory (LSTM) neural network, or any other type of neural network model architecture that is currently known or later developed [Ringlein ¶ 0036]; and then label the security alert 120 as to the proper disposition for the security alert 132, e.g., an actual threat which should be escalated 134 or a false positive alert that should not be escalated 136. The labeled security alert data including the security alert, the security knowledge graph, and the security analyst's provided disposition label are then stored in an entry in the security alert database 138 to thereby generate a training dataset comprising security alert entries and corresponding disposition labels as to the correct disposition classification of the security alert [Ringlein ¶ 0063]). As per claim 14: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Bazalgette discloses wherein the second trained machine learning model is trained using one or more of: a labeled dataset of events labeled as legitimate attacks or as benign events, extracted technical features from suspected threats (The high-level understanding of each complex cyber incident in a graph representation can be retrieved from anonymized cyber threat analyst module (e.g., AI analyst) incidents and then be used to train the GNN models [Bazalgette ¶ 0024]; In step 1, an initial training of the Artificial Intelligence model trained on cyber threats can occur using unsupervised learning and/or supervised learning on characteristics and attributes of known potential cyber threats including malware, insider threats, and other kinds of cyber threats that can occur within that domain [Bazalgette ¶ 0113]), configuration files of clients, values of digital assets, remediation costs, downtime consequences, or a number of affected devices (The system trains machine learning algorithms on graphs of cyber incidents to try and classify those graphs into associated categories of graphs, so when an incident in a network with the cyber threat analyst module deployed occurs, it can analyze the series of nodes in the graph, such as boundary network devices and/or endpoint computing devices connected together by linking factors [Bazalgette ¶ 0037]; The self-learning algorithms in the AI can, for example, understand each node's (user account, device, etc.) in an organization's normal patterns of life in about a week, and grows more bespoke with every passing minute. Conventional AI typically relies solely on identifying threats based on historical attack data and reported techniques, requiring data to be cleansed, labelled, and moved to a centralized repository. The detection engine self-learning AI can learn "on the job" from real-world data occurring in the system and constantly evolves its understanding as the system's environment changes. The Artificial Intelligence can use machine learning algorithms to analyze patterns and 'learn' what is the 'normal behavior' of the network by analyzing data on the activity on the network at the device and employee level [Bazalgette ¶ 0204]). As per claim 15: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Ringlein discloses further comprising receiving feedback from the client and tailoring at least one of the trained machine learning models based on the feedback (In one illustrative embodiment, outputting the graphical user interface further comprises outputting a feedback user interface for receiving user input specifying a correctness or incorrectness of the recommended disposition. The feedback user interface further comprises one or more use interface elements for specifying a correct disposition for the security incident. In one illustrative embodiment, the security incident disposition system further operates to receive user input, via the feedback user interface, specifying the recommended disposition to be incorrect and providing a correct disposition for the security incident, generates a dynamic training data set based on the received user input, and retrains the trained security incident disposition system based on the dynamic training data set. Thus, the user, e.g., human analyst, is provided with a mechanism for overriding the predicted disposition of a security incident and in some embodiments generating additional training data to improve the training of the security incident machine learning model by providing dynamic user feedback that is collected into a dynamically generated training dataset [Ringlein ¶ 0013, ¶ 0042-0043, ¶ 0082, ¶ 0084]). As per claim 16: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Ringlein discloses wherein the remediation comprises one of blocking the suspected threat or releasing a blocking of the suspected threat (The output disposition classification generated by the predictive model 148 may be provided to a security analyst 131, such as via the threat monitoring interface 130 in FIG. 1A, for further evaluation, logged in the security alert database 138, or otherwise made available for further use in determining a responsive action to take if deemed necessary [Ringlein ¶ 0083]; The responsive actions may take many different forms, such as generating alert notifications, inhibiting operation of particular computer components, or the like [Ringlein ¶ 0025, Examiner’s Note: inhibiting operation is blocking]). As per claim 19: Ringlein in view of Bazalgette disclose all the limitations of claim 1. Furthermore, Ringlein discloses a computer system for determining remediations for suspected threats, the computer system comprising: a computerized processor (in a data processing system comprising at least one processor and at least one memory, the at least one memory comprising instructions executed by the at least one processor to cause the at least one processor to implement a security incident disposition system [¶ 0007, ¶ 0048, ¶ 0055, ¶ 0110, Fig. 5]); and a non-transitory storage medium for storing instructions that, when executed by the computerized processor, cause the system to (The computer readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer readable storage medium may be, for example, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. A non-exhaustive list of more specific examples of the computer readable storage medium includes the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as punch-cards or raised structures in a groove having instructions recorded thereon, and any suitable combination of the foregoing. A computer readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire [¶ 0051, ¶ 0053, ¶ 0007, Fig. 5]): The limitations of claim 19 are substantially similar to claim 1 above, and therefore the claim is likewise rejected. As per claim 20: Ringlein in view of Bazalgette disclose all the limitations of claim 1. Furthermore, Ringlein discloses a computer usable non-transitory storage medium having a computer program embodied thereon for causing a suitably programmed system to determine remediations for suspected threats by performing the following steps when such program is executed on the system, the steps comprising (The computer readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer readable storage medium may be, for example, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. A non-exhaustive list of more specific examples of the computer readable storage medium includes the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as punch-cards or raised structures in a groove having instructions recorded thereon, and any suitable combination of the foregoing. A computer readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire [¶ 0051, ¶ 0053, ¶ 0007, Fig. 5]): The limitations of claim 20 are substantially similar to claim 1 above, and therefore the claim is likewise rejected. Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Ringlein in view of Bazalgette in view of Rieger et al (US PGPub No. 2020/0175171; hereinafter “Rieger”). As per claim 5: Ringlein in view of Bazalgette teach all the limitations of claim 4. Furthermore, Ringlein discloses wherein [the Mean Time Deltas between consecutive] logs (The security event data stored in the security log 112 may specify various events associated with the particular managed computing resources 101 that represent events of interest to security evaluations, e.g., failed login attempts, password changes, network traffic patterns, system configuration changes [Ringlein ¶ 0062]; FIG. 1B, the new security incidents/alerts and their corresponding security knowledge graphs 150 may be input directly into the SID computing system 140. That is, similar to the operation of the SIEM system 110 during the training operation, the SIEM system 110 generates new security incidents/alerts, which are the basis for the generation of security knowledge graphs 150 which, during runtime operation, are sent directly to the SID computing system 140 for feature extraction and metric generation 143 by the FEE 144 [Ringlein ¶ 0082]) is calculated by dividing a sum of time differences between consecutive logs by a number of logs (The data store 135 cooperates with the process identifier classifier to collect and maintain historical data of processes and their connections, which is updated over time as the network is in operation [Bazalgette ¶ 0083, Examiner’s Note: the process of calculating the mean/average of a set of values is a known mathematical calculation]). Ringlein in view of Bazalgette discloses the claimed subject matter as discussed above but does not explicitly disclose the Mean Time Deltas between consecutive. However, Rieger teaches the Mean Time Deltas between consecutive (In some embodiments, the cyber state metadata 220 may comprise, define, and/or characterize a cyber state at one or more cyber nodes (cyber nodes 124) of the control system 101. The cyber state metadata 220 may comprise, define, and/or characterize any suitable aspect of cyber communication at a cyber node 124. In some embodiments, the cyber state metadata 220 may comprise statistical characteristics of cyber communication at particular cyber nodes 124, which characteristics may include, but are not limited to: communication speed, mean time delta between messages, mean message latency, number of messages per destination, number of message sources, mean message size, number of zero size messages, mean data length, maximum data length, data speed, and/or the like. Alternatively, or in addition, the cyber state metadata 220 may comprise parameters 222 corresponding to cyber communication between particular cyber nodes 124 (e.g., communication between a controller 132 and one or more sensors 144, actuators 146, automation controllers 134, and/or the like), which may include, but are not limited to: communication speed to/from the nodes 124, mean time delta between messages to/from the nodes 124, latency of messages communicated between the nodes 124, mean size of messages communicated between the nodes 124, and/or the like. Although particular examples of cyber state metadata 220 and/or cyber state parameters 222 are described herein, the disclosure is not limited in this regard, and could be adapted to utilize any suitable information pertaining to a cyber state of a cyber-physical system 100, including acquiring, estimating, determining, and/or monitoring any suitable type of cyber state parameter 222 pertaining to any suitable characteristic and/or aspect of the cyber state of the control system 101 [¶ 0064]). Rieger and the instant application are analogous art because they are from the same field of endeavor of device security. Therefore, based on Ringlein in view of Bazalgette in view of Rieger, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of Rieger to the system of Ringlein in view of Bazalgette in order to calculate a metric of time between data points such as messages being logged for improved characteristic data collection. Hence, it would have been obvious to combine the references above to obtain the invention as specified in the instant claim. Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Ringlein in view of Bazalgette in view of Finkelshtein et al. (US PGPub No. 2021/0314350; hereinafter “Finkelshtein”) in view of MAMMADLI et al. (US PGPub No. 2021/0360015; hereinafter “MAMMADLI”). As per claim 6: Ringlein in view of Bazalgette teach all the limitations of claim 4. Furthermore, Ringlein discloses wherein the plurality of technical features includes (The security incident disposition system further operates to process the security incident data structure and security knowledge graph to extract, from the metadata of the security incident data structure and data corresponding to the nodes and edges of the security knowledge graph, a set of security incident features corresponding to the security incident [Ringlein ¶ 0007, ¶ 0009, ¶ 0033-0034, ¶ 0045]; calculating one or more metrics based on the extracted set of security incident features, and adding the one or more metrics as additional security incident features to the set of security incident features prior to inputting the extracted set of security incident features into the trained security incident machine learning model. The calculated metrics provide more insight into the nature of the security incident that permits a more accurate prediction of the security incident disposition [Ringlein ¶ 0009]; While traversing the topology of the security knowledge graph, various counts may be maintained for some of the extracted features to assist in generating metric features. For example, counts of each type of node, counts of each type of edge, counts of toxic nodes, counts of numbers of nodes having particular reputations, e.g., malware, etc [Ringlein ¶ 0072]; These features are added to features extracted from the metadata associated with the security incident or alert itself. For the security incident or alert itself, the metadata associated with the security incident or alert may be analyzed to extract features such as "magnitude", "severity", and "credibility" from the metadata. In one illustrative embodiment, the features extracted from the security incident/alert and the security knowledge graph are features that may be encoded in a numeric, binary, or normal fashion [Ringlein ¶ 0073, ¶ 0076-0077]) [a Standard Deviation of Time Deltas between consecutive] logs (The security event data stored in the security log 112 may specify various events associated with the particular managed computing resources 101 that represent events of interest to security evaluations, e.g., failed login attempts, password changes, network traffic patterns, system configuration changes [Ringlein ¶ 0062]; FIG. 1B, the new security incidents/alerts and their corresponding security knowledge graphs 150 may be input directly into the SID computing system 140. That is, similar to the operation of the SIEM system 110 during the training operation, the SIEM system 110 generates new security incidents/alerts, which are the basis for the generation of security knowledge graphs 150 which, during runtime operation, are sent directly to the SID computing system 140 for feature extraction and metric generation 143 by the FEE 144 [Ringlein ¶ 0082]) and [a Percentage of Outliers in time deltas]. Ringlein in view of Bazalgette discloses the claimed subject matter as discussed above but does not explicitly disclose a Standard Deviation of Time Deltas between consecutive. However, Finkelshtein teaches a Standard Deviation of Time Deltas between consecutive (In various examples, a possible representation of the window can include a superset or a subset of any the following values: a total movement Euclidean distance in pixels, a window angle between starting point and final point, a mean x coordinate location, a standard deviation of x coordinate location, a mean y coordinate location, a standard deviation of y coordinate location, a mean time delta between events in milliseconds, a standard deviation (STD) of time delta between events, a percentage of events of each type that includes a value for each event type, an average time delta between event types that includes a value for each possible pair of events, and a standard deviation (STD) of time delta between event types that includes a value for each possible pair of events [¶ 0031]); a Percentage (In various examples, a possible representation of the window can include a superset or a subset of any the following values: a total movement Euclidean distance in pixels, a window angle between starting point and final point, a mean x coordinate location, a standard deviation of x coordinate location, a mean y coordinate location, a standard deviation of y coordinate location, a mean time delta between events in milliseconds, a standard deviation (STD) of time delta between events, a percentage of events of each type that includes a value for each event type, an average time delta between event types that includes a value for each possible pair of events, and a standard deviation (STD) of time delta between event types that includes a value for each possible pair of events [¶ 0031]). Finkelshtein and the instant application are analogous art because they are from the same field of endeavor of attack detection. Therefore, based on Ringlein in view of Bazalgette in view of Finkelshtein, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of Finkelshtein to the system of Ringlein in view of Bazalgette in order to effectively use mathematical features of log data for improved analysis. Hence, it would have been obvious to combine the references above to obtain the invention as specified in the instant claim. Ringlein in view of Bazalgette in view of Finkelshtein discloses the claimed subject matter as discussed above but does not explicitly disclose Outliers in time deltas. However, MAMMADLI teaches Outliers in time deltas (At operation 315, the threat detection server 110 may conduct operations for removing outliers from time-deltas (e.g., data representing time differences between successive pairs of communication events) based on factors such as absolute mean deviation. Other operations of outlier identification may be used [¶ 0083]). MAMMADLI and the instant application are analogous art because they are from the same field of endeavor of anomaly identification. Therefore, based on Ringlein in view of Bazalgette in view of Finkelshtein in view of MAMMADLI, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of MAMMADLI to the system of Ringlein in view of Bazalgette in view of Finkelshtein in order to effectively detect suspicious activity through outliers in the data. Hence, it would have been obvious to combine the references above to obtain the invention as specified in the instant claim. Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over Ringlein in view of Bazalgette in view of MARBOUTI et al. (US PGPub No. 2024/0015168; hereinafter “MARBOUTI”). As per claim 10: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Ringlein and Bazalgette discloses wherein the third trained machine learning model is configured to (an autonomous response engine using Artificial Intelligence algorithms configured and trained to perform a second machine-learned task of taking one or more mitigation actions to mitigate the cyber threat, iii) a cyber-security restoration engine using Artificial Intelligence algorithms configured and trained to perform a third machine-learned task of remediating the system being protected back to a trusted operational state [Bazalgette ¶ 0013]) [apply one or more weights] on the authenticity classification and the estimated impact (a disposition classification output based on results of processing the extracted set of security incident features by the trained security incident machine learning model. The security incident disposition system outputs the disposition classification output to the source computing device [Ringlein ¶ 0007]; The trained predictive model 148 operates on the features input 160 comprising the extracted features from the security incident/alert and the security knowledge graph, as well as generated metrics based on these extracted features, to apply the trained feature analysis operations, e.g., the activation functions, of the nodes/neurons of the predictive model, e.g., the CNN, to the features input 160 and generate a disposition classification for the security incident/alert, e.g., whether to escalate or not escalate the security incident/alert. The output disposition classification generated by the predictive model 148 may be provided to a security analyst 131, such as via the threat monitoring interface 130 in FIG. 1A, for further evaluation, logged in the security alert database 138, or otherwise made available for further use in determining a responsive action to take if deemed necessary [Ringlein ¶ 0083, ¶ 0088]; then evaluate the actual impact after each action in the sequence, in order to yield a best possible result to contain/mitigate the detected threat while minimizing the impact on other network devices and users that are i) currently active and ii) not in breach, from different possible actions to take. Again, multiple simulations can be run in parallel so that the different sequences of mitigation actions and restoration actions can be evaluated essentially simultaneously. The cyber-attack simulator 105 with Artificial Intelligence-based simulations in the cyber-attack simulator 105 is configured to use one or more mathematical functions to generate a score and/or likelihood for each of the possible actions and/or sequence of multiple possible actions that can be taken in order to determine which set of actions to choose among many possible actions to initiate. The one or more possible actions to take and their calculated scores can be stacked against each other to factor 1) a likelihood of containing the detected threat acting abnormal with each possible set of actions, 2) a severity level of the detected threat to the network, and 3) the impact of taking each possible set of actions i) on users and ii) on devices currently active in the network not acting abnormal to the normal behavior of the network, and then communicate with the cyber threat detection engine, the autonomous response engine, and the cyber-security restoration engine, respectively, to initiate the chosen set of actions to cause a best targeted change of the behavior of the detected threat acting abnormal to the normal pattern of life on the network while minimizing the impact on other network devices and users that are i) currently active and ii) not in breach of being outside the normal behavior benchmark. The cyber-attack simulator cooperates with the AI models modelling a normal pattern of life for entities/nodes in the system being protected [Bazalgette ¶ 0195]). Ringlein in view of Bazalgette discloses the claimed subject matter as discussed above but does not explicitly disclose apply one or more weights. However, MARBOUTI teaches apply one or more weights (First model component 304 and second model component 306 may receive inputs and generate outputs. This output may be processed by third model component 308. System 300 may further include an aggregation function that may average outputs from first model component 304 and second model component 306. For example, that system may determine a weight for an output from first model component 304 and/or second model component 306 is based on a weighted average, in which the weight is based on an amount of available data (e.g., the amount of content and/or the amount of length of a user interaction history). In some embodiments, the system may determine a weight based on a number of sub-components included within each of first model component 304 and second model component 306. For example, first model component 304 and second model component 306 may include sub-components that each generate an output. The system may determine the weight based on the number of these sub-components [¶ 0048]). MARBOUTI and the instant application are analogous art because they are from the same field of endeavor of threat detection. Therefore, based on Ringlein in view of Bazalgette in view of MARBOUTI, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of MARBOUTI to the system of Ringlein in view of Bazalgette in order to effectively weigh the importance of data based on the amount of data available. Hence, it would have been obvious to combine the references above to obtain the invention as specified in the instant claim. Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Ringlein in view of Bazalgette in view of Baek (US PGPub No. 2021/0089963; hereinafter “Baek”). As per claim 12: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Bazalgette discloses wherein the second trained machine learning model is a regression model (In step 1, an initial training of the Artificial Intelligence model trained on cyber threats can occur using unsupervised learning and/or supervised learning on characteristics and attributes of known potential cyber threats including malware, insider threats, and other kinds of cyber threats that can occur within that domain … In this example, one of the AI model(s) 160 trained on identifying a specific cyber threat can be trained with machine learning such as Linear Regression, Regression Trees, Non-Linear Regression, Bayesian Linear Regression, Deep learning, etc. to learn and understand the characteristics and attributes in that category of cyber threats [Bazalgette ¶ 0113]) that provides the estimated impact (The cyber-attack simulator 105 with Artificial Intelligence-based simulations in the cyber-attack simulator 105 is configured to use one or more mathematical functions to generate a score and/or likelihood for each of the possible actions and/or sequence of multiple possible actions that can be taken in order to determine which set of actions to choose among many possible actions to initiate. The one or more possible actions to take and their calculated scores can be stacked against each other to factor 1) a likelihood of containing the detected threat acting abnormal with each possible set of actions, 2) a severity level of the detected threat to the network, and 3) the impact of taking each possible set of actions i) on users and ii) on devices currently active in the network not acting abnormal to the normal behavior of the network, and then communicate with the cyber threat detection engine, the autonomous response engine, and the cyber-security restoration engine, respectively, to initiate the chosen set of actions to cause a best targeted change of the behavior of the detected threat acting abnormal to the normal pattern of life on the network while minimizing the impact on other network devices and users that are i) currently active and ii) not in breach of being outside the normal behavior benchmark. The cyber-attack simulator cooperates with the AI models modelling a normal pattern of life for entities/nodes in the system being protected [Bazalgette ¶ 0195]) [as a normalized score]. Ringlein in view of Bazalgette discloses the claimed subject matter as discussed above but does not explicitly disclose as a normalized score. However, Baek teaches as a normalized score (A technical solution is provided herein to normalize relevancy scores for content items from different sub-models such that the normalized relevancy scores are comparable for ranking purposes. The technical solution has a benefit that the different sub-models do not need to be re-learned and the learning of the different sub-models do not need to be specially constrained. This benefit improves the operation of content management systems that employ the technical solution because computing resources (e.g., processor, networking, and storage resources) are conserved by not having to re-learn the sub-models or otherwise specially constrain the learning of the sub-models. Instead, the sub-models that were used to score content items before an implementation of the technical solution can continue to be used after the implementation without modification. Thus, the technical solution may be viewed as a non-destructive augmentation to a collection of sub-models used to compute relevancy scores for different types of content items for purposes of making personalized content suggestions [¶ 0074]). Ringlein in view of Bazalgette and Baek are analogous art because they are from the same field of endeavor of machine learning. Therefore, based on Ringlein in view of Bazalgette in view of Baek, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of Baek to the system of Ringlein in view of Bazalgette in order to allow for direct comparison of the score with other models scores as a result of normalization. Hence, it would have been obvious to combine the references above to obtain the invention as specified in the instant claim. Claim 17 is rejected under 35 U.S.C. 103 as being unpatentable over Ringlein in view of Bazalgette in view of IVENSO et al. (US PGPub No. 2023/0075065; hereinafter “IVENSO”). As per claim 17: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Ringlein discloses further comprising generating an alert or a report, wherein [the remediation is incorporated into the alert or report] (The responsive actions may take many different forms, such as generating alert notifications, inhibiting operation of particular computer components, or the like [Ringlein ¶ 0025, Examiner’s Note: inhibiting operation is blocking]). Ringlein in view of Bazalgette discloses the claimed subject matter as discussed above but does not explicitly disclose the remediation is incorporated into the alert or report. However, IVENSO teaches the remediation is incorporated into the alert or report (In one embodiment, the recommended mitigation technique is included in the generated alert message. In this way, the alert may include the ratio (iFM) and a recommendation of a technique to mitigate the signal following in the machine learning model selected from one or more of increasing a number of training vectors in the ML model, filtering signal inputs to the ML model to reduce noise, and increasing a number of signals [¶ 0086]). IVENSO and the instant application are analogous art because they are from the same field of endeavor of anomaly detection. Therefore, based on Ringlein in view of Bazalgette in view of IVENSO, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of IVENSO to the system of Ringlein in view of Bazalgette in order to communicate the remediation action to a system or user for a timely response. Hence, it would have been obvious to combine the references above to obtain the invention as specified in the instant claim. Claim 18 is rejected under 35 U.S.C. 103 as being unpatentable over Ringlein in view of Bazalgette in view of Jenks et al. (US PGPub No. 2023/0300161; hereinafter “Jenks”). As per claim 18: Ringlein in view of Bazalgette teach all the limitations of claim 1. Furthermore, Ringlein and Bazalgette disclose wherein the third trained machine learning model determines [a priority score for the suspected threat based] on the authenticity classification and the estimated impact (an autonomous response engine using Artificial Intelligence algorithms configured and trained to perform a second machine-learned task of taking one or more mitigation actions to mitigate the cyber threat, iii) a cyber-security restoration engine using Artificial Intelligence algorithms configured and trained to perform a third machine-learned task of remediating the system being protected back to a trusted operational state [Bazalgette ¶ 0013]; The cyber-attack simulator 105 with Artificial Intelligence-based simulations in the cyber-attack simulator 105 is configured to use one or more mathematical functions to generate a score and/or likelihood for each of the possible actions and/or sequence of multiple possible actions that can be taken in order to determine which set of actions to choose among many possible actions to initiate. The one or more possible actions to take and their calculated scores can be stacked against each other to factor 1) a likelihood of containing the detected threat acting abnormal with each possible set of actions, 2) a severity level of the detected threat to the network, and 3) the impact of taking each possible set of actions i) on users and ii) on devices currently active in the network not acting abnormal to the normal behavior of the network, and then communicate with the cyber threat detection engine, the autonomous response engine, and the cyber-security restoration engine, respectively, to initiate the chosen set of actions to cause a best targeted change of the behavior of the detected threat acting abnormal to the normal pattern of life on the network while minimizing the impact on other network devices and users that are i) currently active and ii) not in breach of being outside the normal behavior benchmark [Bazalgette ¶ 0195]; This risk assessment and ranking is conducted to avoid frequent “false positive” alerts that diminish the degree of reliance/confidence on the cyber security appliance 100. The cyber-attack simulator 105 may include and cooperate with one or more AI models trained with machine learning on the contextual knowledge of the organization. These trained AI models may be configured to identify data points from the contextual knowledge of the organization and its entities, which may include, but is not limited to, language-based data, email/network connectivity and behavior pattern data, and/or historic knowledgebase data. The cyber-attack simulator 105 may use the trained AI models to cooperate with one or more AI classifier(s) by producing a list of specific organization-based classifiers for the AI classifier. The cyber-attack simulator 105 is further configured to calculate, based at least in part on the results of the one or more hypothetical simulations of a possible cyberattack and/or of an actual ongoing cyberattack from a cyber threat determine a risk score for each node (e.g. each device, user account, etc.), the threat risk score being indicative of a possible severity of the compromise prior to an autonomous response action is taken in response to the actual cyberattack of the cyber incident [Bazalgette ¶ 0110]; FIG. 5, the cyber security restoration engine is configured to take one or more remediation actions based on configured and/or Artificial Intelligence assistance to remediate the one or more nodes in the graph of the system being protected back to a trusted operational state in a recovery from the cyber threat. These actions might be fully automatic, or require a specific human confirmation decision before they begin. The cyber security restoration engine is configured to cooperate with the other AI-based engines of the cyber security system, via the interfaces and/or direct integrations, to track and understand the cyber threat identified by the other components as well as track the one or more mitigation actions taken to mitigate the cyber threat during the cyberattack by the other components in order to assist in intelligently restoring the protected system while still mitigating the cyber threat attack back to a trusted operational state; and thus, as a situation develops with an ongoing cyberattack, the cyber security restoration engine is configured to take one or more remediation actions to remediate (e.g. restore) at least one of the nodes in the graph of the protected system back to a trusted operational state while the cyberattack is still ongoing [Bazalgette ¶ 0097]; The trained predictive model 148 operates on the features input 160 comprising the extracted features from the security incident/alert and the security knowledge graph, as well as generated metrics based on these extracted features, to apply the trained feature analysis operations, e.g., the activation functions, of the nodes/neurons of the predictive model, e.g., the CNN, to the features input 160 and generate a disposition classification for the security incident/alert, e.g., whether to escalate or not escalate the security incident/alert. The output disposition classification generated by the predictive model 148 may be provided to a security analyst 131, such as via the threat monitoring interface 130 in FIG. 1A, for further evaluation, logged in the security alert database 138, or otherwise made available for further use in determining a responsive action to take if deemed necessary [Ringlein ¶ 0083, ¶ 0088]). Ringlein in view of Bazalgette discloses the claimed subject matter as discussed above but does not explicitly disclose a priority score for the suspected threat based. However, Jenks teaches a priority score for the suspected threat based (With reference to FIG. 1, based on a set of inputs, the threat exposure engine 106 generates a prioritized exposure score 104. The exposure engine 106 can further include a modeling engine 105, which can be configured to operate as described herein. Inputs to the threat exposure engine 106 can include organizational threat model information 101, fleet exposure information 102, and defense surface posture information 103. As non-limiting examples, defense surface posture information can include configuration, raw logs, detection logic, security control and policies. Based on these inputs, the exposure engine 106 can be configured to generate one or more of the prioritized threat exposure score 104 and a ranked, or prioritized, list of recommended actions and mitigations 107. As used herein, ranking, sorting, and prioritization are considered interchangeably. The ranked list of actions and mitigations 107 can also be provided to a responder 108, which can issue commands to a target organization defense surface, as explained herein. As changes to the target organization defense surface are caused by the commands, these are reflected in the organizational threat model information 101, fleet exposure information 102, and defense surface posture information 103 [¶ 0029]; Modeling engine 105 can be configured to consume, as non-limiting examples, machine learning models 503, statistical analysis 504, algorithms 505, enrichment context 506, and/or ensemble or deep learning models 507. The model outcomes 502 can include trends or comparisons 508, scoring 509, recommendations 510, prioritized lists 511, predictive features 512, and relationships 513 [¶ 0076]). Jenks are analogous art because they are from the same field of endeavor of machine learning. Therefore, based on Ringlein in view of Bazalgette in view of Jenks, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of Jenks to the system of Ringlein in view of Bazalgette in order to improve the efficiency of risk mitigation in the system (¶ 0109). Hence, it would have been obvious to combine the references above to obtain the invention as specified in the instant claim. Conclusion Any inquiry concerning this communication or earlier communications from the examiner should be directed to JAMES P MOLES whose telephone number is (703)756-1043. The examiner can normally be reached M-F 8:00am-5:00pm. 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, Jung Kim can be reached at (571) 272-3804. 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. /JAMES P MOLES/Examiner, Art Unit 2494 /JUNG W KIM/Supervisory Patent Examiner, Art Unit 2494
Read full office action

Prosecution Timeline

Sep 14, 2025
Application Filed
Oct 02, 2026
Non-Final Rejection mailed — §101, §103 (current)

Precedent Cases

Applications granted by this same examiner with similar technology

Patent 12755051
Aggregating Permissions across Multiple Platforms
1y 10m to grant Granted Oct 06, 2026
Patent 12743521
Systems and Methods for Merging Performance and Security into a Unit Testing Environment
4y 4m to grant Granted Sep 22, 2026
Patent 12732388
NON-FUNGIBLE TOKEN (NFT) BASED INTELLIGENT DOCUMENT PROTOCOLS
3y 4m to grant Granted Sep 08, 2026
Patent 12724884
Adaptive Incident Prioritization Based on User Feedback
2y 3m to grant Granted Sep 01, 2026
Patent 12705362
METHOD, APPARATUS, SYSTEM AND COMPUTER PROGRAM FOR IDENTIFYING AND RESPONDING TO QUANTUM VULNERABILITY USING DYNAMIC ANALYSIS FOR APPLICATION
2y 2m to grant Granted Aug 11, 2026
Study what changed to get past this examiner. Based on 5 most recent grants.

Strategy Recommendation AI-generated — please review before filing

Get a prosecution strategy drawn from examiner precedents, rejection analysis, and claim mapping.
Typically takes 5-10 seconds — AI-generated, attorney review required before filing

Prosecution Projections

1-2
Expected OA Rounds
67%
Grant Probability
95%
With Interview (+28.6%)
2y 9m (~1y 8m remaining)
Median Time to Grant
Low
PTA Risk
Based on 48 resolved cases by this examiner. Grant probability derived from career allowance rate.

Sign in with your work email

Enter your email to receive a magic link. No password needed.

Personal email addresses (Gmail, Yahoo, etc.) are not accepted.

Free tier: 3 strategy analyses per month