SR-Reward replaces the environment reward with the L2 norm of a successor representation learned from demonstrations, enabling offline RL without reward labels.
Successor feature sets: Generalizing successor representations across policies
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
SR-Reward: Taking The Path More Traveled
SR-Reward replaces the environment reward with the L2 norm of a successor representation learned from demonstrations, enabling offline RL without reward labels.