AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Composing Task-Agnostic Policies with Deep Reinforcement Learning

Ahmed H. Qureshi, Jacob J. Johnson, Yuzhe Qin, Taylor Henderson, Byron Boots, Michael C. Yip

ICLR 2020 · Reinforcement learning › Visual and representation RL · approach: Skill learning and diffusion policies · 10 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 the composed policy instead of the strongest standard-RL baselineAnt Random Goal · Ant Random Goaldistance: 0.21 \pm 0.08 -> 0.11 \pm 0.05Improved
Use the composed policy instead of the strongest standard-RL baselineAnt Cross Maze · Ant Cross Mazedistance: 0.78 \pm 0.06 -> 0.11 \pm 0.02Improved
Use the composed policy instead of the strongest standard-RL baselinePusher · Pusherdistance: 0.17 \pm 0.02 -> 0.14 \pm 0.02Improved
Use the composed policy instead of the strongest standard-RL baselineHalfCheetah Hurdle · HalfCheetah Hurdledistance: 0.79 \pm 0.01 -> 0.27 \pm 0.22Improved
Add the attention network (ablation)Composition-SAC · Ant Random Goal, Ant Cross Maze, and PusherdistanceImproved
Add the BRNN skill encoder (ablation)Composition-SAC · Ant Random Goal, Ant Cross Maze, and PusherdistanceImproved
Add both the attention network and BRNN skill encoder (ablation)Composition-SAC · Ant Random Goal, Ant Cross Maze, and PusherdistanceImproved

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