AI Paper Atlas

Research map / Reinforcement learning / MDPs and planning

Improved Sample Complexity for Incremental Autonomous Exploration in MDPs

Jean Tarbouriech, Matteo Pirotta, Michal Valko, Alessandro Lazaric

NeurIPS 2020 · Reinforcement learning › MDPs and planning · approach: Rich-observation exploration · 9 citations (OpenAlex)

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
Use ALGO's model-based incremental exploration instead of UCBExploreIncremental autonomous exploration · DCL sample complexitysample complexity: \widetilde{O}(L^6 S_{L + \varepsilon} A/\varepsilon^3) for UCBExplore; ALGO bound: \widetilde{O}(L^5 \Gamma_{L + \varepsImproved
Target near-optimal shortest-path policies on the final discovered state setIncremental autonomous exploration · DC*objective capabilityImproved
Use SSP policies rather than finite-horizon policies for policy evaluationIncremental autonomous exploration · DCL sample complexitysample complexity: \widetilde{O}(\varepsilon^{-3}) for UCBExplore; \widetilde{O}(\varepsilon^{-2}) for ALGOImproved
Recompute policies using the final expanded set of controllable statesALGO · Incrementally L-controllable statesgoal-reaching policy optimalityImproved
Use ALGO's exploration data for zero-shot planning with arbitrary supported costsALGO · Cost-sensitive SSP on incrementally L-controllable statespolicy suboptimality: \varepsilon/c_{\min}Improved

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