Shapley value vectors are clustered and linearly regressed to extract decision boundaries from deep RL policies, but the proof and evaluation are not convincing.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
From Explainability to Interpretability: Interpretable Policies in Reinforcement Learning Via Model Explanation
Shapley value vectors are clustered and linearly regressed to extract decision boundaries from deep RL policies, but the proof and evaluation are not convincing.