Reproduce “Machine Theory of Mind for Autonomous Cyber-Defence” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Luke Swaby, Matthew Stewart, Daniel Harrold, Chris Willis, Gregory Palmer
Researchers at BAE Systems Applied Intelligence Labs developed GIGO-ToM, a Graph Neural Network (GNN)-based Machine Theory of Mind (MToM) model for autonomous cyber-defence, and introduced a novel, topology-aware evaluation metric called Network Transport Distance (NTD). GIGO-ToM accurately characterizes cyber attackers, predicts their high-value targets, and forecasts their attack trajectories in variable network topologies.
Key insights
- GNN-based ToM architectures can effectively characterize and predict adversarial agent goals and trajectories across complex, variable-sized graph environments.
- Understanding an adversary's intentions and beliefs provides a strategic advantage, enabling a shift from reactive to proactive cyber-defence strategies.
- The Network Transport Distance (NTD) provides a robust and interpretable quantitative measure for evaluating predictions of agent behavior on network topologies, critical for practical application.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.