Notes

What to say:

- PRACTICE ROHIT PRACTICE 6 times

- reinforcement learning - concerned with how intelligent agents ought to take actions in an environment in order to maximize the notion of cumulative reward.
- CVSS - The Common Vulnerability Scoring System (CVSS) is an open framework for communicating the characteristics and severity of software vulnerabilities
Derivation nodes (visualized as circles) correspond to inter-action rules and represent the reason for a fact to becometrue, they can be viewed as attack steps (Nr={2,4}).Primitive facts (visualized as rectangles) correspond to theinformation given as input (Np={5,6,7,8}), while derivedfacts (visualized as diamonds) are the results of applyinginteraction rules on the primitive facts (can be viewed as attackstep consequences; i.e.,Nd={1,3}).


find possible attack routes that an attacker coulduse to compromise the security of the network.
Then we simplifythe attack graph and generate a transition graph [5]. The final step is to use the transition graph tomodel the environment and the reward system for the agent (attacker) and apply Q-learning which returns the possible routesfor the attackerto reach the goal.

Thediamond vertices (e.g. nodes 9, 25, and 37 in Fig. 3) represent potential privileges or access that an attacker can obtain in the system, e.g., code execution privilege on the web server. The elliptical vertices (e.g. nodes 10, 16, and 24 in Fig. 3) linkpreconditions to postconditions. As an example, it is necessary for an attacker to have access to a machine that has a vulnerability, to be able to exploit the vulnerability and obtain privileges.

- Reinforcement learning
- Reward metric used
- DQN
- Diagrams
EACH TIME TALK ABOUT PURPOSE
- demo
-- small to big
-- example of pauls 3host
-- assisting with attack steps (pathways,pivots) and attack step consequences - leading to the target server