BehaviorGuard: Online Backdoor Defense for Deep Reinforcement Learning

2026-05-07

AI Summary

This work addresses the challenge of defending against backdoor attacks in deep reinforcement learning, which are difficult to mitigate online due to complex triggers and high fine-tuning costs. To this end, the authors propose BehaviorGuard, a runtime behavior-aware framework that enables online backdoor defense in both single-agent and multi-agent settings without requiring trigger reconstruction or model fine-tuning. BehaviorGuard detects behavioral anomalies by analyzing high-quantile action distribution statistics and tail deviation, and employs a real-time action suppression mechanism to achieve efficient, trigger-agnostic mitigation. Experimental results across multiple benchmark environments and attack variants demonstrate that BehaviorGuard significantly outperforms existing defenses, offering superior protection efficacy and computational efficiency.

Abstract

Backdoor attacks pose a serious threat to deep reinforcement learning (DRL). Current defenses typically rely on reward anomalies to reverse-engineer triggers and model finetuning to remove backdoors. However, complex trigger patterns undermine their robustness, and fine-tuning entails high costs, limiting practical utility. Therefore, we shift defense concerns to trigger-agnostic backdoor output behaviors and propose BehaviorGuard, an online behavior-based backdoor detection and mitigation framework for DRL. Specifically, we find that regardless of attacks, backdoored policies induce consistent shifts in action distributions to ensure reliable activation, leaving detectable traces in high-quantile regions and distribution tails, even in the absence of triggers. Based on this, we design a novel metric that captures behavioral drift in action distributions to identify and suppress backdoor actions at runtime. To our knowledge, this is the first online backdoor defense that counters attacks both in single- and multi-agent DRL. Evaluated across diverse benchmarks with different backdoor attacks, BehaviorGuard consistently surpasses prior methods in both efficacy and efficiency.

Problem

• backdoor attacks
• deep reinforcement learning
• online defense
• trigger-agnostic
• action distributions

Innovation

• backdoor defense
• deep reinforcement learning
• action distribution
• online detection
• trigger-agnostic