AI Paper Atlas

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

Max-Min Off-Policy Actor-Critic Method Focusing on Worst-Case Robustness to Model Misspecification

Takumi Tanabe, Rei Sato, Kazuto Fukuchi, Jun Sakuma, Youhei Akimoto

NeurIPS 2022 · Reinforcement learning › Policy gradient and actor-critic · approach: Off-policy actor-critic with expressive policies · 2 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 N=1 uncertainty parameter instead of N=5 (ablation)M2TD3 · InvertedPendulum 1worst-case return (x10^2): 8.22 -> 5.77Hurt
increase N from 5 to 10 (ablation)M2TD3 · InvertedPendulum 1worst-case return (x10^2): 8.22 -> 9.07Improved
increase N from 5 to 10 (ablation)M2TD3 · InvertedPendulum 1average return (x10^2): 9.66 -> 9.89Improved
remove distance-based refreshing strategy (w/o DRS) (ablation)M2TD3 · InvertedPendulum 1worst-case return (x10^2): 8.22 -> 9.07No clear effect
remove distance-based refreshing strategy (w/o DRS) (ablation)M2TD3 · InvertedPendulum 1average return (x10^2): 9.66 -> 9.08Hurt
remove probability-based refreshing strategy (w/o PRS) (ablation)M2TD3 · InvertedPendulum 1worst-case return (x10^2): 8.22 -> 5.41Hurt
remove probability-based refreshing strategy (w/o PRS) (ablation)M2TD3 · InvertedPendulum 1average return (x10^2): 9.66 -> 6.90Hurt
use M2TD3 instead of DR (TD3) for worst-caseM2TD3 · InvertedPendulum 1worst-case return (x10^2): 3.18 -> 8.22Improved
use M2TD3 instead of DR (TD3)M2TD3 · Ant 2worst-case return (x10^3): 1.64 -> 4.13Improved
use M2TD3 instead of DR (TD3)M2TD3 · Small HalfCheetah 1worst-case return (x10^3): 6.76 -> 5.27Hurt
use M2TD3 instead of DR (TD3)M2TD3 · HumanoidStandup 1worst-case return (x10^4): 9.68 -> 9.33Hurt
use M2TD3 instead of DR (TD3)M2TD3 · Small Hopper 1worst-case return (x10^3): 3.42 -> 2.88Hurt

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