Framework for experimental analysis of adversarial example attacks on policy learning in Deep RL. Attack methodologies are based on our paper “Vulnerability of Deep Reinforcement Learning to Policy Induction Attacks” (Behzadan & Munir, 2017 – https://arxiv.org/abs/1701.04143 ).
I am a (minor) contributor to Nettacker. OWASP Nettacker project aims to automate information gathering, vulnerability scanning, and report generation for networked systems, covering services, bugs, vulnerabilities, misconfigurations, and other security-related information. This software utilizes TCP SYN, ACK, ICMP and many other protocols in order to detect and bypass Firewall/IDS/IPS devices. Leveraging a unique method, Nettacker is able to discover protected services and devices such as SCADA. It provides make a competitive edge in automated pentesting.