Behavior-based host modeling
Methods and systems for modeling host behavior in a network include determining a first probability function for observing each of a set of process-level events at a first host based on embedding vectors for the first event and the first host. A second probability function is determined for the first host issuing each of a set of network-level events connecting to a second host based on embedding vectors for the first host and the second host. The first and second probability functions are maximized to determine a set of likely process-level and network-level events for the first host. A security action is performed based on the modeled host behavior.
1. A method for modeling host behavior in a network, comprising:
determining a first probability function for observing each of a set of process-level events at a first host based on embedding vectors for the first event and the first host;
determining a second probability function for the first host issuing each of a set of network-level events connecting to a second host based on embedding vectors for the first host and the second host;
maximizing the first and second probability functions to determine a set of process-level and network-level events for the first host by performing a negative sampling of host-event pairs and a mini-batch gradient descent using the negative sampling of the host-event pairs; and
performing a security action based on the modeled host behavior.
2. The method of claim 1 , wherein the set of process-level events and the set of network level events are historical events detected at the first host.
3. The method of claim 1 , wherein the negative sampling approximates a maximized conditional probability that an event will occur at a target host.
4. The method of claim 1 , wherein performing the security action further comprises automatically performing at least one security action selected from the group consisting of shutting down devices, stopping or restricting certain types of network communication, raising alerts to system administrators, and changing a security policy level.
5. The method of claim 1 , wherein all pairs of hosts and events are embedded into a common latent space.
6. The method of claim 1 , wherein each host is modeled as a context of process-level events and as a context of all hosts reached by network events of said host.
7. The method of claim 1 , wherein each process-level event on a host is modeled as a conditional probability of said host given the event.
8. The method of claim 1 , wherein each network-level event is modeled as a conditional probability that one host issues a network event that connects to another host.
9. A system for modeling host behavior in a network, comprising:
a host behavior module comprising a processor configured to determine a first probability function for observing each of a set of process-level events at a first host based on embedding vectors for the first event and the first host, to determine a second probability function for the first host issuing each of a set of network-level events connecting to a second host based on embedding vectors for the first host and the second host, and to maximize the first and second probability functions to determine a set of process-level and network-level events for the first host by performing a negative sampling of host-event pairs and a mini-batch gradient descent using the negative sampling of the host-event pairs; and
a security module configured to perform a security action based on the modeled host behavior.
10. The system of claim 9 , wherein the set of process-level events and the set of network level events are historical events detected at the first host.
11. The system of claim 9 , wherein the negative sampling approximates a maximized conditional probability that an event will occur at a target host.
12. The system of claim 9 , wherein the security module is further configured to perform at least one security action selected from the group consisting of shutting down devices, stopping or restricting certain types of network communication, raising alerts to system administrators, and changing a security policy level.
13. The system of claim 9 , wherein all pairs of hosts and events are embedded into a common latent space.
14. The system of claim 9 , wherein each host is modeled as a context of process-level events and as a context of all hosts reached by network events of said host.
15. The system of claim 9 , wherein each process-level event on a host is modeled as a conditional probability of said host given the event.
16. The system of claim 9 , wherein each network-level event is modeled as a conditional probability that one host issues a network event that connects to another host.