Shielding the policy improvement process in offline RL yields policies that are safe with high probability while outperforming unshielded baselines in both average and worst-case performance, especially under limited data.
During policy iteration, DUIPI updates the baseline policy by iteratively increasing the probability of the action with the highest penalized action-value ๐
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Robust Probabilistic Shielding for Safe Offline Reinforcement Learning
Shielding the policy improvement process in offline RL yields policies that are safe with high probability while outperforming unshielded baselines in both average and worst-case performance, especially under limited data.