AIC Seminar: Jaromír Janisch

Accept our invitation to the AIC seminar with Jaromír Janisch on April 6 at 14:30. It will take place in the room KN:E-301 and we will discuss autonomous penetration testing. 

Abstract

Autonomous Penetration Testing using Deep Reinforcement Learning with Invariant Neural Architectures and Zero-Shot Generalization

In this seminar, Jaromír will take you to his latest research about autonomous penetration testing. In this talk, you will discover NASimEmu, a Network Attack Simulator and Emulator, designed to train and test offensive agents using deep reinforcement learning with focus on 0-shot transfer to unseen scenarios. You will learn how the simulator and emulator work and why both parts are important. Also, you will learn what size-and-permutation-invariant architectures are and why they are important for this problem. The presentation will end with a summary of the current research and what still needs to be done.