Tag
The paper introduces a protocol to decompose auditability in reinforcement learning policies into six testable predicates, demonstrating that symbolic rule overlap does not guarantee behavioral agreement across independently trained agents.
The paper analyzes the safety of per-agent policy composition in multi-agent reinforcement learning, proving that independent composition can fail, and proposes MA-USFA, a hierarchical method that enables safe and flexible successor-feature transfer in cooperative MARL.
This paper proposes AutoSafe, a safety-aware policy architecture for safe online reinforcement learning that integrates structured safety monitoring and intervention directly into action generation, enabling smooth, risk-dependent transitions between performance and safety behaviors, demonstrated on benchmarks and a physical cart-pole system.