AI Paper Atlas

Research map / Reinforcement learning / Policy gradient and actor-critic

What Makes Value Learning Efficient in Residual Reinforcement Learning?

Guozheng Ma, Lu Li, Haoyu Wang, Zixuan Liu, Pierre-Luc Bacon, Dacheng Tao

ICML 2026 · Reinforcement learning › Policy gradient and actor-critic · approach: Temporal-difference and discounting theory · 0 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 base-policy warmup transitions to the replay buffer (ablation)minimal residual SAC · ManiSkill (3 tasks)sample efficiency / success: fails entirely without warmup on PegInsertionSide; improves consistently with more warmupImproved
add explicit critic pre-training (Soft Q auto alpha, fixed alpha, Hard Q) on top of implicit warmup (ablation)DAWN-style residual SAC · ManiSkill (3 tasks)sample efficiencyHurt
add exploration noise during warmup data collection instead of base policy only (ablation)residual SAC warmup · ManiSkillsample efficiencyNo clear effect
add Layer Normalization to the critic (ablation)residual SAC · ManiSkill (3 tasks)sample efficiencyImproved
add critic normalization: critic sensitivity to residual action (ablation)residual SAC · PegInsertionSidegradient norm of Q wrt residual: approximately twice the level of the unnormalized baselineImproved
use Layer Norm instead of Hyperspherical Norm in critic (ablation)residual SAC · ManiSkill, varying lambdasuccess rateImproved
replace MSE critic with distributional objectives (C51, QR, TQC) (ablation)residual SAC with warmup and LN · ManiSkill (2 challenging tasks)early-stage successNo clear effect
add normalization to the actor (ablation)DAWN · ManiSkill/Adroitsuccess rateNo clear effect
remove either data-anchored warmup or critic normalization from DAWN (ablation)DAWN · ManiSkill/Adroitlearning efficiencyImproved
use DAWN instead of Policy DecoratorDiffusion Policy residual RL · PegInsertionSidetime to 90% success: approximately 6x fasterImproved

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