AI Paper Atlas

Research map / Reinforcement learning / MDPs and planning

Best-Effort Policies for Robust Markov Decision Processes

Alessandro Abate, Thom Badings, Giuseppe De Giacomo, Francesco Fabiano

AAAI 2026 · Reinforcement learning › MDPs and planning · approach: Robust MDP policy gradient

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add a second value-iteration pass under best-case transition probabilities as tie-breakerPRISM robust value iteration on interval MDPs · slippery gridworld IMDP, nu=0.0best-effort actions [%]: 21.9 -> 100.0Improved
add best-case tie-breaker passPRISM robust value iteration · slippery gridworld IMDP, |S|=10000runtime [s]: 48.9 -> 54.4Mixed
add best-case tie-breaker passPRISM robust value iteration · slippery gridworld IMDP, |S|=100runtime [s]: 2.0 -> 3.9Hurt
add best-case tie-breaker pass to robust value iterationRVI on s-rectangular RMDPs · slippery gridworld RMDP, |S|=100, nu=0.0best-effort actions [%]: 0.0 -> 100.0Improved
add derivative-based tie-breaker to robust value iterationRVI on s-rectangular RMDPs · slippery gridworld RMDP, |S|=100, nu=0.5best-effort actions [%]: 49.2 -> 100.0Improved
add derivative-based tie-breaker to robust value iterationRVI on s-rectangular RMDPs · slippery gridworld RMDP, |S|=900runtime [s]: 163.6 -> 172.0No clear effect
add best-case tie-breaker to robust value iterationRVI on s-rectangular RMDPs · slippery gridworld RMDP, |S|=900runtime [s]: 163.6 -> 274.4Hurt
vary order in which best-effort action is defined (nu) (ablation)PRISM robust value iteration · slippery gridworld IMDP, |S|=100best-effort actions [%]: 21.9 -> 89.9Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy