AI Paper Atlas

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

HodgeFlow Policy Search by Topologically Dissecting Temporal-Difference Signals in Non-Markovian Environments

Zuyuan Zhang, Sizhe Tang, Tian Lan

ICML 2026 · Reinforcement learning › Policy gradient and actor-critic · approach: Regularized scalable deep value learning

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
project TD errors onto their integrable component (HFPS) instead of standard TD learningHFPS · Nonmarkov LunarLanderFinal@T return: 145.6±12.9 vs DQN 10.1±34.3; DRQN 169.1±3.1Mixed
use HFPS instead of recurrent baselinesHFPS · Nonmarkov PendulumFinal@T return: -106.2±121.9 vs DRQN -235.8±47.8Improved
use HFPS instead of recurrent baselinesHFPS · Nonmarkov AcrobotFinal@T return: -82.3±9.7 vs DRQN -500.0±0.0Improved
use HFPS instead of DQNHFPS · Nonmarkov PointMassFinal@T return: -207.1±49.8 vs DQN -673.6±366.5Improved
use linear HFPS instead of linear TD in random-feature MDPHFPS · Random-feature MDP (50 states)MSVEImproved
make the ring MDP non-integrable by perturbing one edge rewardHFPS · Tabular ring MDPresidual normHurt

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