Safe Reinforcement Learning
A branch of reinforcement learning that aims to learn policies while simultaneously satisfying safety constraints and avoiding unacceptable violations. It is typically formulated on the constrained Markov decision process framework.
Timeline (last 90 days)
Safe reinforcement learning constrains the probability or cumulative cost of violating safety constraints, rather than solely pursuing robustness to perturbations.
FAITH is a model-free safe reinforcement learning framework that approximates the optimal state-action safety value and uses a feedforward network to amortize the minimal-intervention filtering process.
Safe Reinforcement Learning (Safe RL) explicitly incorporates safety constraints into the optimization objective, ensuring that agents do not cross dangerous boundaries during exploration.
All Facts (3)
Safe reinforcement learning constrains the probability or cumulative cost of violating safety constraints, rather than solely pursuing robustness to perturbations.
50%UnverifiedFAITH is a model-free safe reinforcement learning framework that approximates the optimal state-action safety value and uses a feedforward network to amortize the minimal-intervention filtering process.
50%UnverifiedSafe Reinforcement Learning (Safe RL) explicitly incorporates safety constraints into the optimization objective, ensuring that agents do not cross dangerous boundaries during exploration.
50%