AI Paper Atlas

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

Mean Flow Policy Optimization

Xiaoyi Dong, Xi Sheryl Zhang, Jian Cheng

ICML 2026 · Reinforcement learning › Policy gradient and actor-critic · approach: Regularized scalable deep value learning · 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
use MeanFlow policy (average velocity) instead of standard flow-matching policy (ablation)MFPO · HalfCheetah-v3episode returnImproved
use MeanFlow model for the divergence integral instead of instantaneous divergence network (ablation)MFPO · HalfCheetah-v3episode returnImproved
add average divergence network for action likelihood instead of zeroing the divergence integral (ablation)MFPO · HalfCheetah-v3episode returnImproved
combine policy proposal and Gaussian proposal SNIS estimators instead of Gaussian only (ablation)MFPO · HalfCheetah-v3episode returnImproved
use only the policy proposal for velocity estimation (ablation)MFPO · HalfCheetah-v3episode returnHurt
sample ratio K1:K2 = 1:2 favoring Gaussian proposal (ablation)MFPO · HalfCheetah-v3episode returnImproved
combine two SNIS estimators with ESS weights (ablation)velocity estimator · HalfCheetah-v3 after 120k iterationsestimation varianceImproved
auto-tune temperature with target entropy rho=0.5 instead of fixed temperature (ablation)MFPO · HalfCheetah-v3episode returnImproved
vary target entropy coefficient rho (0.4, 0.5, 0.6) (ablation)MFPO · HalfCheetah-v3episode returnMixed
add average divergence network to likelihood estimationMFPO · training timetraining time: 5% increase in training timeHurt
replace multi-step diffusion policies with 2-step MeanFlow policydiffusion-based online RL · MuJoCo benchmarkstraining time: approximately 50% reductionImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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