Tag
Introduces CCPL, a method to address delayed and stochastic consequences in constrained reinforcement learning using a delay-corrected Bellman operator and an Interventional Consequence Net for causal attribution, with a contraction proof under unknown delays.
OpenAI proposes standardizing constrained RL as the formalism for safe exploration and introduces Safety Gym, a benchmark suite for evaluating safe deep RL algorithms in high-dimensional continuous control tasks with safety constraints.