Isaac Symes Thompson

Research Assistant at The Alan Turing Institute, researching reinforcement learning and autonomous cyber defence.

[Isaac Symes Thompson @IsaacThss]

25/ The goal here to try to persuade that the established 'monolithic' approach used in cyber defence simulators is a dead end. It would be beneficial if all network defence environments supported an entity-based approach natively.

[Isaac Symes Thompson @IsaacThss]

1/ A thread to describe our recent workshop paper and repository on RL in autonomous cyber (network) defence: "Entity-based Reinforcement Learning for Autonomous Cyber Defence", with @mavroudisv , @AlbiCaron and @hkscy. https://t.co/EdzUKlfqnk
https://t.co/grYUXP9EVQ

[Isaac Symes Thompson @IsaacThss]

24/ These experiments aren't exhaustive, and are more intended to back up results that already exist in various forms. Also, Yawning Titan is an abstract and simple environment that is quite a long way from representing a real network defence problem.