Adversarial observations for probabilistic state space models for robust reinforcement learning
Decision-making under partial or adversarial observability requires accurate inference of the environment’s latent state and its associated uncertainty. This work analyzes adversarial attacks on linear probabilistic state-space models, commonly integrated within reinforcement learning architectures, where the attacker alters observations under likelihood constraints that ensure the perturbations remain consistent with the observation model of the underlying decision-making system. We analyze how such adversarial yet realistic observation shifts propagate through the latent state inference process and influence policy decisions. This perspective provides a principled pathway toward building more robust reinforcement learning systems, with direct relevance to safety-critical domains such as robotics, where reliable operation under sensor noise, partial failures, and adversarial conditions is essential.
Palabras clave: Adversarial Machine Learning Reinforcement Learning State Space Models Kalman Filter