AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Offline Meta Reinforcement Learning -- Identifiability Challenges and Effective Data Collection Strategies

Ron Dorfman, Idan Shenfeld, Aviv Tamar

NeurIPS 2021 · Reinforcement learning › Visual and representation RL · approach: Task-inference meta-RL · 7 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
reduce data diversity: uniform to initial states excluding semi-circle (ablation)BOReL · Ant-Semi-circleaverage return: 171.8 -> 102.8Hurt
reduce data diversity: uniform to fixed initial position (ablation)BOReL · Ant-Semi-circleaverage return: 171.8 -> 99.2Hurt
add CQL to critic training, uniform data (ablation)BOReL · Ant-Semi-circleaverage return: 171.8 -> 176.0Improved
add CQL to critic training, excluding s.c. data (ablation)BOReL · Ant-Semi-circleaverage return: 102.8 -> 116.6Improved
add CQL to critic training, fixed-position data (ablation)BOReL · Ant-Semi-circleaverage return: 99.2 -> 112.4Improved
remove reward relabelling (ablation)BOReL · domains with varying rewardsaverage rewardHurt
remove policy replaying (ablation)BOReL · domains with non-overlapping identifying statesaverage rewardHurt
remove policy replaying (ablation)BOReL · domains with overlapping identifying statesaverage rewardNo clear effect

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