外观
论文arXiv 2610.0946221 天前
TMT：面向未见任务中 Vision-Language-Action 策略的运行时后门检测
原文链接：TMT: Runtime Backdoor Detection for Vision-Language-Action Policies on Unseen Tasks（arXiv 2610.09462 · 发布日期：2026-10-07）
TMT: Runtime Backdoor Detection for Vision-Language-Action Policies on Unseen Tasks
TMT用令牌流形与潜在转移建模检测VLA策略运行时后门，并在未见任务上验证。
核心要点
- 动机是后门VLA可保持正常表现，触发后输出看似合理的恶意动作。
- TMT在良性轨迹上训练，结合输入令牌结构与相邻层潜在动态预测误差。
- 发现可疑轨迹后由潜在偏差确认，并指导后续监测中的转移选择。
- 与十个基线后验比较，TMT在三类VLA后门攻击的未见任务上达到最优表现；项目页已公开。
验证范围与局限：验证范围为未见任务上的三类VLA后门攻击，摘要未说明仿真或真机设置。
英文摘要 (Abstract)
Backdoored vision-language-action (VLA) policies can preserve benign task performance while producing malicious actions when a trigger appears. Detecting such activation is difficult because malicious behavior can comprise individually plausible actions, while unfamiliar tasks introduce legitimate changes in observations and behavior. We introduce TMT, a runtime backdoor detector based on Token Manifold and latent Transition modeling. Trained on benign rollouts, its two branches assess input-token structure and prediction errors in adjacent-layer latent dynamics. A suspicious rollout identified by the token manifold branch, once confirmed through latent deviations, guides transition selection for subsequent monitoring. We further explore policy purification through self-distillation: a frozen copy of the backdoored policy provides benign-input actions to supervise a student on paired benign and triggered observations, without requiring a separate clean reference policy. For evaluation, we adapt traditional backdoor detectors and repurpose anomaly and failure detection methods as VLA backdoor detectors. In a post-hoc comparison with ten baselines, TMT achieves state-of-the-art backdoor detection performance on unseen tasks across three VLA backdoor attacks. Our project page is available at https://zzr42.github.io/tmt/.
作者：Zirun Zhou, Jingfeng Zhang, Haochuan Xu, Xizhe Zhang, Elliott Wen, Jing Sun, Hong Jia
来源
本页事实对应的公开页面，以原文为准。