Preprint
Jul 2026
dtControl2+$\varepsilon$: Trading Optimality for Explainability in MDPs via Decision Trees
This work constructs a smaller decision tree, distilling the essence of the controller, while still guaranteeing its $\varepsilon$-optimality, which enables the tool to provide tunably simpler explanations, omitting a controllable amount of detail.
Tereza Kinská, Jan Křetinský, Tobias Meggendorfer et al.
· 0 citations