A Framework for Adversarial Analysis of Decision Support Systems Prior to Deployment

TL;DR

This paper presents a comprehensive framework for analyzing and securing Deep Reinforcement Learning-based decision support systems before deployment, focusing on understanding vulnerabilities and assessing adversarial attack impacts through simulation.

Contribution

It introduces a novel framework for pre-deployment adversarial analysis of DRL decision systems, including methods for visualizing behavior and systematically evaluating attack impacts.

Findings

Framework effectively visualizes agent behavior and vulnerabilities.

Systematic attack impact ranking reveals critical observation indices.

Adversarial transferability varies across architectures and training algorithms.

Abstract

This paper introduces a comprehensive framework designed to analyze and secure decision-support systems trained with Deep Reinforcement Learning (DRL), prior to deployment, by providing insights into learned behavior patterns and vulnerabilities discovered through simulation. The introduced framework aids in the development of precisely timed and targeted observation perturbations, enabling researchers to assess adversarial attack outcomes within a strategic decision-making context. We validate our framework, visualize agent behavior, and evaluate adversarial outcomes within the context of a custom-built strategic game, CyberStrike. Utilizing the proposed framework, we introduce a method for systematically discovering and ranking the impact of attacks on various observation indices and time-steps, and we conduct experiments to evaluate the transferability of adversarial attacks across agent architectures and DRL training algorithms. The findings underscore the critical need for robust adversarial defense mechanisms to protect decision-making policies in high-stakes environments.