AI Paper Atlas

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

Iterative Gumbel Planning for Continuous Control

Shaohuai Liu, Weirui Ye, Yilun Du, Le Xie

NeurIPS 2026 · Reinforcement learning › Policy gradient and actor-critic · approach: Tree search and policy optimization · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes Iterative Gumbel Planning, a search-based policy improvement method for continuous control that addresses the mismatch between finite-candidate search and parametric continuous policies.

Problem

Search-based planning (AlphaZero/MuZero-style) works well for discrete actions, but in continuous control search over a finite candidate set returns a discrete distribution, which mismatches the parametric continuous policy class.

Approach

Extends Gumbel-based policy improvement to continuous action spaces by iteratively planning over candidate actions so the search output can be distilled into a parametric continuous policy; the abstract is truncated, so the precise mechanism is not stated.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace Gaussian policy with iteratively refined factorized categorical policyEfficientZero-Multitask · DMC-Hard dog-runmax return: 405.874 -> 613.257Improved
replace Gaussian policy with iteratively refined factorized categorical policyEfficientZero-Multitask · DMC-Hard humanoid-walkmax return: 779.404 -> 911.542Improved
replace Gaussian policy with factorized policyEfficientZero-Multitask · HumanoidBench h1hand-slidemax return: 666.507 -> 870.521Improved
replace Gaussian policy with factorized policyEfficientZero-Multitask · HumanoidBench h1hand-reach, crawl, stair, sit_hardmax return: reach 5820.811 -> 5114.852Hurt
use factorized categorical policy instead of refinement-matched Gaussian proposal (ablation)EZ-IGP · DMC humanoid-standreturn: 763.8 -> 903.7Improved
distill search-score-weighted mixture of candidates instead of winner action into Gaussian (ablation)Gaussian policy · DMC humanoidpolicy sigma and return: final sigma 2.67No clear effect
increase bins V from 11 to 31 (ablation)EZ-IGP · HB-Hard normalized scorenormalized score: 0.622 -> 0.743Improved
increase bins V from 31 to 41 (ablation)EZ-IGP · HB-Hard normalized scorenormalized score: 0.743 -> 0.690Hurt
use two refinement rounds instead of one (ablation)EZ-IGP · HB-Hard normalized scorenormalized score: 0.667 -> 0.743Improved
increase refinement rounds from 2 to 4 (ablation)EZ-IGP · HB-Hard normalized scorenormalized score: 0.743 -> 0.689Hurt
reduce candidates/simulations from 8/16 to 2/4 (ablation)EZ-IGP · DMC humanoid averageaverage return: 728.243 -> 296.748Hurt
use factorized policy instead of GaussianEZ-IGP · evaluation action smoothnesstemporal action varianceImproved

Experimental setup

Builds on: AlphaZero MuZero Gumbel MuZero

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