AI Paper Atlas

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

Perceiving the Knowledge Boundary: Uncertainty-Guided Exploration and Imagination for World Models

Zhenxian Liu, Peixi Peng, Yangru Huang, Yonghong Tian

AAAI 2026 · Reinforcement learning › Policy gradient and actor-critic · approach: Model-based RL with guarantees · 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 uncertainty-guided exploration and imagination filteringDreamerV3 · CARLA highwayepisode reward: 375 -> 476Improved
add uncertainty-guided exploration and imagination filteringDreamerV3 · DMC (6 tasks avg, 1M steps)return: 456.5 -> 607.9Improved
add both modulesDreamerV3 · Atari100k Gopherscore: 2977 -> 3122Improved
model-based method vs model-freeOurs · DMC Finger Spinreturn: 756.9 vs DrQ-v2 846.7Hurt
add exploration only or filtering only (ablation)DreamerV3 · CARLA highwayreward curveImproved
combine exploration and filtering (ablation)DreamerV3 · CARLA highwayreward curveImproved
use bootstrap sampling for ensemble (ablation)ensemble disagreement · CARLA highwayreward curveNo clear effect
increase ensemble size from K=3 to 5 (ablation)ensemble disagreement · CARLA highwayreward curveImproved
increase ensemble size beyond K=5 (ablation)ensemble disagreement · CARLA highwayreward curveNo clear effect
stepwise instead of aggregated uncertainty statistics (ablation)imagination filtering · CARLA highwayreward curveImproved

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