Systems and methods for dynamic vulnerability scoring
The present disclosure provides systems and methods for substantially continuous and dynamic vulnerability scoring. According to the present disclosure, the method includes detecting one or more vulnerabilities. The method includes determining a contextual prioritization score (CPS) for each of the one or more vulnerabilities based on historical data, the historical data including a series of contextual features corresponding to each one of the one or more vulnerabilities. The method may include, in response to detection of an event, determining a partial CPS score by an agent. The method may include, if a new partial CPS is determined, generating an updated CPS based on the CPS and the new partial CPS and transmitting the updated CPS to each of one or more computing devices.
1 . A system for dynamically assessing and ranking security vulnerabilities of a selected entity including a network having a plurality of computing devices linked thereto, the system comprising:
a hardware processor; and
a memory coupled to the hardware processor with instructions stored thereon, that cause the hardware processor to execute a contextual vulnerability prioritization engine configured to:
dynamically detect a vulnerability for the selected entity;
determine a contextual prioritization score (CPS) for the vulnerability in relation to the selected entity based on an aggregation of a plurality of partial CPSs generated for each of (a) a plurality of contextual features associated with the selected entity and related to the vulnerability and (b) one or more non-contextual features related to the vulnerability;
scan for occurrences of events related to the vulnerability on the plurality of computing devices of the selected entity;
in response to detection of an occurrence of an event related to the vulnerability on the plurality of computing devices, direct data corresponding to a detected event related to the vulnerability to corresponding agents of a plurality of agents, wherein the directing is based on a data type of the data corresponding to the detected event;
wherein each of the plurality of partial CPSs of each of the plurality of contextual features are dynamically computed by a corresponding agent of the plurality of agents based on the plurality of contextual features related to the vulnerability and data associated with the corresponding agent collected from the detected event;
determine, via each corresponding agent receiving the data corresponding to the detected event, a new partial CPS for each contextual feature applicable to data corresponding to the occurrence of the detected event, wherein the plurality of agents each comprise a machine learning model that generates the new partial CPS based on the data corresponding to the detected event and the selected entity, wherein the new partial CPS for each contextual feature is weighted based on the data type of the data corresponding to the detected event during the aggregation of the plurality of partial CPSs,
wherein the new partial CPS for each contextual feature is further weighted based on an importance of the corresponding agent;
determine an updated CPS representative of a present state of the vulnerability in relation to the selected entity based on the new partial CPSs; and
transmit the updated CPS to each of the plurality of computing devices.
2 . The system of claim 1 , wherein the plurality of agents each comprise one or more of trained classifiers, statistical models, or probabilistic models.
3 . The system of claim 2 , wherein the trained classifiers are generated via another plurality of additional contextual features related to the vulnerability and a plurality of additional non-contextual features related to the vulnerability and one or more of a supervised machine learning algorithm or an unsupervised machine learning algorithm.
4 . The system of claim 1 , wherein the plurality of computing devices include one or more of a switch, an access point, servers, storage devices, or user devices.
5 . The system of claim 1 , wherein the one or more non-contextual features include a common vulnerability scoring system (CVSS) score related to the vulnerability.
6 . The system of claim 1 , wherein the plurality of contextual features include one or more of a payload analysis, exploitation ease, detection reliability, threat intelligence, or vulnerability trends.
7 . The system of claim 6 , wherein the plurality of contextual features further include one or more of exposure data, software and services affected by the vulnerability, behavioral analytics related to the vulnerability, website content, scan frequency, a network attack surface, a pivotal potential, an outstanding asset detection, or a remediation time.
8 . The system of claim 1 , wherein the data corresponding to the detected event comprises one or more of exposure data, software and services affected by the vulnerability, behavioral analytics related to the vulnerability, website content, scan frequency, a network attack surface, a pivotal potential, an outstanding asset detection, or a remediation time.
9 . The system of claim 1 , wherein the contextual vulnerability prioritization engine is further configured to dynamically prioritize a plurality of vulnerabilities based on one or more dynamically updated CPSs for one or more vulnerabilities of the plurality of vulnerabilities.
10 . A computer-implemented method of dynamically assessing and ranking security vulnerabilities of a selected entity including a network having a plurality of computing devices linked thereto, performed using a processing device, the method comprising:
dynamically detecting a vulnerability for the selected entity;
determining a contextual prioritization score (CPS) for the vulnerability in relation to the selected entity based on an aggregation of a plurality of partial CPSs generated for each of (a) a plurality of contextual features associated with the selected entity and related to the vulnerability and (b) one or more non-contextual features related to the vulnerability;
scanning for occurrences of events related to the vulnerability on the plurality of computing devices of the selected entity;
in response to detection of an occurrence of an event related to the vulnerability on the plurality of computing devices, directing data corresponding to a detected event related to the vulnerability to corresponding agents of a plurality of agents, wherein the directing is based on a data type of the data corresponding to the detected event;
wherein each of the plurality of partial CPSs of each of the plurality of contextual features are dynamically computed by a corresponding agent of the plurality of agents based on the plurality of contextual features related to the vulnerability and data associated with the corresponding agent collected from the detected event;
determining, via each corresponding agent receiving the data corresponding to the detected event, a new partial CPS for each contextual feature applicable to data corresponding to the occurrence of the detected event, wherein the plurality of agents each comprise a machine learning model that generates the new partial CPS based on the data corresponding to the detected event and the selected entity, wherein the new partial CPS for each contextual feature is weighted based on the data type of the data corresponding to the detected event during the aggregation of the plurality of partial CPSs,
wherein the new partial CPS for each contextual feature is further weighted based on an importance of the corresponding agent;
determining an updated CPS representative of a present state of the vulnerability in relation to the selected entity based on the new partial CPSs; and
transmitting the updated CPS to each of the plurality of computing devices.
11 . The method of claim 10 , wherein the plurality of agents each comprise one or more of trained classifiers, statistical models, or probabilistic models.
12 . The method of claim 11 , wherein the trained classifiers are generated via another plurality of additional contextual features related to the vulnerability and a plurality of additional non-contextual features related to the vulnerability and one or more of a supervised machine learning algorithm or an unsupervised machine learning algorithm.
13 . The method of claim 10 , wherein the plurality of computing devices include one or more of a switch, an access point, servers, storage devices, or user devices.
14 . The method of claim 10 , wherein the one or more non-contextual features include a common vulnerability scoring system (CVSS) score related to the vulnerability.
15 . The method of claim 10 , wherein the plurality of contextual features include one or more of a payload analysis, exploitation ease, detection reliability, threat intelligence, or vulnerability trends.
16 . A non-transitory machine-readable storage medium that has instructions stored thereon that, responsive to execution by a processing device, cause the processing device to perform operations of dynamically assessing and ranking security vulnerabilities of a selected entity including a network having a plurality of computing devices linked thereto, the operations comprising:
dynamically detecting a vulnerability for the selected entity;
determining a contextual prioritization score (CPS) for the vulnerability in relation to the selected entity based on an aggregation of a plurality of partial CPSs generated for each of (a) a plurality of contextual features associated with the selected entity and related to the vulnerability and (b) one or more non-contextual features related to the vulnerability;
scanning for occurrences of events related to the vulnerability on the plurality of computing devices of the selected entity;
in response to detection of an occurrence of an event related to the vulnerability on the plurality of computing devices, directing data corresponding to a detected event related to the vulnerability to corresponding agents of a plurality of agents, wherein the directing is based on a data type of the data corresponding to the detected event;
wherein each of the plurality of partial CPSs of each of the plurality of contextual features are dynamically computed by a corresponding agent of the plurality of agents based on the plurality of contextual features related to the vulnerability and data associated with the corresponding agent collected from the detected event;
determining, via each corresponding agent receiving the data corresponding to the detected event, a new partial CPS for each contextual feature applicable to data corresponding to the occurrence of the detected event, wherein the plurality of agents each comprise a machine learning model that generates the new partial CPS based on the data corresponding to the detected event and the selected entity, wherein the new partial CPS for each contextual feature is weighted based on the data type of the data corresponding to the detected event during the aggregation of the plurality of partial CPSs,
wherein the new partial CPS for each contextual feature is further weighted based on an importance of the corresponding agent;
determining an updated CPS representative of a present state of the vulnerability in relation to the selected entity based on the new partial CPSs; and
transmitting the updated CPS to each of the plurality of computing devices.
17 . The non-transitory machine-readable storage medium of claim 16 , wherein the plurality of agents each comprise one or more of trained classifiers, statistical models, or probabilistic models.
18 . The non-transitory machine-readable storage medium of claim 17 , wherein the trained classifiers are generated via another plurality of additional contextual features related to the vulnerability and a plurality of additional non-contextual features related to the vulnerability and one or more of a supervised machine learning algorithm or an unsupervised machine learning algorithm.
19 . The non-transitory machine-readable storage medium of claim 16 , wherein the plurality of computing devices include one or more of a switch, an access point, servers, storage devices, or user devices.
20 . The non-transitory machine-readable storage medium of claim 16 , wherein the one or more non-contextual features include a common vulnerability scoring system (CVSS) score related to the vulnerability.