DeepXplain: XAI-Guided Autonomous Defense Against Multi-Stage APT Campaigns

ArXi:2603.21296v1 Announce Type: cross Advanced Persistent Threats (APTs) are stealthy, multi-stage attacks that require adaptive and timely defense. While deep reinforcement learning (DRL) enables autonomous cyber defense, its decisions are often opaque and difficult to trust in operational environments. This paper presents DeepXplain, an explainable DRL framework for stage-aware APT defense.