Temporal Gradient Inversion for Private Trajectory Reconstruction in Embodied Reinforcement Learning

2026-09-24

AI Summary

This study addresses privacy vulnerabilities in distributed training for embodied reinforcement learning, demonstrating that transmitting only policy gradients still leaks private trajectories due to temporal structure. This work proposes TRACE, an attack framework that formally characterizes cross-timestep gradient correlations and proves exact action recovery under low-entropy regularization. By leveraging amortized temporal gradient inversion with conditional mutual information bound analysis and closed-form action recovery solutions, the method autoregressively reconstructs private observation-action sequences. Experimental results show that TRACE achieves 18.8 dB PSNR with near-perfect action recovery, improves computational efficiency by several orders of magnitude over baselines, and generalizes across diverse network architectures, effectively overcoming the limitations of conventional single-frame attacks.

Abstract

Distributed learning in embodied reinforcement-learning agents offers a degree of privacy by retaining raw sensor data on-device and transmitting only policy gradients to the server. Yet temporal structure can amplify this leakage beyond single-frame attacks. We introduce Temporal Reconstruction Attack on Consecutive Encodings (TRACE), an amortized temporal gradient-inversion attack that autoregressively reconstructs the sequence of private observation-action trajectories from per-step policy-learning gradients. The attack exploits two structural signals ignored by prior single-frame methods: (i) cross-time correlation between successive embodied gradients, which we formalize via a conditional mutual-information bound, and (ii) closed-form action recovery from policy-head gradient structure, which we prove exact when standard entropy regularization is sufficiently small. On held-out embodied scenes, TRACE reaches $18.8$ dB PSNR with near-perfect action recovery at $3$-$4.5$ ms per reconstructed frame, dominating the learning-based baseline across all reconstruction metrics and exceeding optimization attacks while running orders of magnitude faster. Further evaluation demonstrates TRACE's broader applicability across recurrent, residual, and compact transformer victim architectures, multi-modal inputs, and larger discrete action spaces. Defense experiments suggest that protecting temporal gradient streams may require sequence-aware privacy mechanisms.

Problem

• Embodied Reinforcement Learning
• Gradient Inversion
• Privacy Leakage
• Trajectory Reconstruction
• Distributed Learning

Innovation

• Temporal Gradient Inversion
• Embodied Reinforcement Learning
• Trajectory Reconstruction
• Amortized Attack
• Privacy Leakage