AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Dual Goal Representations

Seohong Park, Deepinder Mann, Sergey Levine

ICLR 2026 · Reinforcement learning › Visual and representation RL · approach: Learned state and action representations

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
replace original goal with dual goal representationGCIVL · OGBench state-based (13 tasks avg)success rate: 34 -> 41Improved
replace original goal with dual goal representationCRL · OGBench state-based (13 tasks avg)success rate: 33 -> 39Improved
replace original goal with dual goal representationGCFBC · OGBench state-based (13 tasks avg)success rate: 20 -> 23Improved
replace original goal with dual goal representationGCIVL · cube-double-playsuccess rate: 35 -> 60Improved
replace original goal with dual goal representationGCIVL · antsoccer-arena-navigatesuccess rate: 47 -> 31Hurt
replace original goal with dual goal representationGCIVL · pixel-based OGBenchsuccess rate: 66 -> 78 (visual-antmaze-medium)Improved
use dual goal representation on visual puzzle tasksGCIVL · visual-puzzle-4x4success rate: 65 -> 0Hurt
use early fusion of state and goal images instead of late fusion (ablation)vanilla GCIVL · visual-puzzle-4x4success rate: 0 -> 65Improved
use inner product aggregation instead of symmetric metric (ablation)Dual representation · OGBench state-based avgsuccess rate: 38 -> 42Improved
add separate downstream GCRL value function instead of extracting policy directly from f (ablation)Dual representation · scene-playsuccess rate: 50 -> 71Improved
direct policy extraction from temporal distance function (ablation)Dual representation · antmaze-largesuccess rate: 28 -> 58Hurt
use dual representation for noisy evaluation goalsGCIVL · noisy OGBench environmentssuccess rateImproved

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