AI Paper Atlas

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

ScaleMoE: Mixture-of-Experts for Scalable Continuous Control in Actor-Critic Reinforcement Learning

Yi Ma, Chenjun Xiao, Hongyao Tang, Yaodong Yang, Jinyi Liu, Jing Liang, Jiye Liang

ICML 2026 · Reinforcement learning › Policy gradient and actor-critic · approach: Actor-critic theory

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
scale from 1 to 4 experts (feature-level gating) (ablation)ScaleMoE-Simba · DMC dog-runIQM return: +32%Improved
scale to 8 experts (ablation)ScaleMoE-Simba · DMC humanoid-runIQM return: up to +60%Improved
use 8 small experts (width 512) instead of single BRC critic width 4096BRC · DMC DOGSIQM return: 835 -> 849Improved
use 8 small experts instead of width 4096 monolithic critic (15% parameters)BRC · DMC HUMANOIDSIQM return: 709 -> 730Improved
use 16 experts with full activation instead of width 4096 BRCBRC · HumanoidBenchIQM normalized return: 0.49 -> 0.62Improved
switch from feature-level to output-level gating in single-task (ablation)ScaleMoE · DMC single-taskreturnHurt
switch gating level in multi-task (ablation)ScaleMoE · multi-task DMCreturnNo clear effect
use top-1 expert activation (K=1) (ablation)ScaleMoE-Output · DMC single and multi-tasknormalized returnHurt
increase total experts N from 2 to 4 to 8 (K=2) (ablation)ScaleMoE-Simba · DMC single-taskmean final score: 675 -> 697 -> 698Mixed
replace top-K gating MoE with SoftMoE (ablation)BRC · MetaWorld 50 tasksIQM success rateHurt
replace naive BRC ensemble with ScaleMoE (ablation)BRC · MetaWorld 50 tasksIQM success rateImproved
vary auxiliary regularization weight lambda from 0.005 to 1 (ablation)ScaleMoE · MetaWorld 50 tasksIQM success rateNo 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