AI Paper Atlas

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

Improve Agents without Retraining: Parallel Tree Search with Off-Policy Correction

Gal Dalal, Assaf Hallak, Steven Dalton, iuri frosio, Shie Mannor, Gal Chechik

NeurIPS 2021 · Reinforcement learning › Policy gradient and actor-critic · approach: Tree search and policy optimization · 1 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
Apply vanilla tree search at inference to pretrained RainbowRainbow agent · Atari (Frostbite, Breakout, BeamRider, SpaceInvaders, Asteroids)rewardHurt
Add off-policy correction (BCTS) to tree search over vanilla TSRainbow agent with TS · Atari Frostbite, Breakout, BeamRider, SpaceInvadersreward: d = 4: score more than doubles; BeamRider about x5Improved
Increase tree depth with BCTSRainbow with BCTS · Atari inferencerewardImproved
BCTS vs vanilla TS on AsteroidsRainbow with TS · Atari AsteroidsrewardNo clear effect
Increase tree depth in DQN training with TS (d=1 to d=4)DQN with TS · Atari Asteroidsaverage score: 2, 093 -> 17, 929Improved
Increase tree depth in DQN training with TS (d=1 to d=4)DQN with TS · Atari Breakoutaverage score: 385 -> 620Mixed
Increase tree depth in DQN training with TS (d=1 to d=4)DQN with TS · Atari MsPacmanaverage score: 1, 644 -> 4, 021Improved
Increase tree depth in DQN training with TS (d=1 to d=4)DQN with TS · Atari SpaceInvadersaverage score: 675 -> 2, 550Improved
Increase tree depth in DQN training with TS (d=1 to d=4)DQN with TS · Atari VideoPinballaverage score: 229, 129 -> 345, 742Mixed
DQN with TS (d=4) vs plain DQN under equal 2-day computeDQN with TS · Atari Asteroidsaverage score: 1, 664 -> 17, 929Improved
DQN with TS (d=4) vs Rainbow under equal computeDQN with TS · Atari VideoPinballaverage score: 641, 235 -> 345, 742Hurt
DQN with TS (d=4) vs Rainbow under equal computeDQN with TS · Atari MsPacmanaverage score: 3, 600 -> 4, 021Mixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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