AI Paper Atlas

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

A Long $N$-step Surrogate Stage Reward for Deep Reinforcement Learning

Junmin Zhong, Ruofan Wu, Jennie Si

NeurIPS 2023 · 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
add LNSS surrogate reward (N=100)DDPG · DMC Humanoid Walk, no noiseaverage reward: 1.3 -> 250.9Improved
add LNSS surrogate reward (N=100)TD3 · DMC Acrobot SwingUp, no noiseaverage reward: 5.2 -> 114.8Improved
add LNSS surrogate reward (N=100)D4PG · DMC Acrobot SwingUp, no noiseaverage reward: 133.5 -> 406.3Improved
add LNSS surrogate reward (N=100)DDPG · DMC Finger TurnHard, no noiseaverage reward: 301.8 -> 943.3Improved
add LNSS surrogate reward (N=100)TD3 · DMC Fish Swim, no noiseaverage reward: 483.1 -> 676.8Improved
replace 5-step n-step return with LNSSTD3 · DMC Fish Swim, no noiseaverage reward: 731.7 -> 676.8Hurt
use 5-step n-step return (n5) instead of single stepTD3 · DMC Humanoid Walk, no noiseaverage reward: 184.2 -> 1.01Hurt
use 5-step n-step return (n5) instead of single stepDDPG · DMC Fish Swim, no noiseaverage reward: 272.9 -> 150.5Hurt
add LNSS surrogate reward under 10% reward noiseDDPG · DMC Humanoid Walk, 10% noiseaverage reward: 0.76 -> 235.4Improved
add LNSS surrogate reward under 10% reward noiseTD3 · DMC Humanoid Walk, 10% noiseaverage reward: 1.5 -> 200.2Improved
increase LNSS horizon N from 5, 10 to 50, 100 (ablation)TD3-LNSS · DMC tasksfinal reward and learning speedImproved
increase LNSS N to 100 (vs N = 5, 50) (ablation)TD3-LNSS · DMC taskscv(Q)Improved

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