AI Paper Atlas

Research map / Reinforcement learning / MDPs and planning

A Meta-MDP Approach to Exploration for Lifelong Reinforcement Learning

Francisco Garcia, Philip S. Thomas

NeurIPS 2019 · Reinforcement learning › MDPs and planning · approach: Average-reward and regularized MDPs · 11 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
add learned advisor exploration policy to REINFORCEREINFORCE · Pole-balance (d)average return: 20.32 -> 28.52Improved
add learned advisor exploration policy to REINFORCEREINFORCE · Animataverage return: -779.62 -> -387.27Improved
add advisor to PPOPPO · Pole-balance (d)average return: 27.87 -> 46.29Improved
add advisor to PPOPPO · Animataverage return: -751.40 -> -631.97Improved
add advisor to PPOPPO · Pole-balance (c)average return: 29.95 -> 438.13Improved
add advisor to PPOPPO · Hopperaverage return: 13.82 -> 164.43Improved
add advisor to PPOPPO · Antaverage return: -42.75 -> 83.76Improved
PPO+Advisor versus MAMLPPO+Advisor · Antaverage return: 113.33 (MAML) vs 83.76Hurt
use advisor for fine-tuning to new carsSelf-driving agent · Self-driving fine-tuningepisodes to threshold: 42 vs 12 (first half); 5 vs 18 (second half)Mixed
use advisor exploration policy alone as exploitation policy instead of task-specific policy (ablation)Advisor policy · Pole-balancing and animat unseen tasksreturn / steps to goalHurt

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