environmental perturbation

state correction (1) value-aware correction (1) value correction (1) manipulation task (1) out-of-distribution state (1) out-of-distribution action (1) dynamic programming (1) value function (1) policy learning (1) offline reinforcement learning (1)