AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

GALOIS: Boosting Deep Reinforcement Learning via Generalizable Logic Synthesis

Yushi Cao, Zhiming Li, Tianpei Yang, Hao Zhang, YAN ZHENG, Yi Li, Jianye HAO, Yang Liu

NeurIPS 2022 · Reinforcement learning › Visual and representation RL · approach: Scalable RL systems and environments

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 a hierarchical white-box program with explicit cause-effect logic instead of neural DRL or fixed-sequence program synthesisGALOIS · Complex MiniGrid tasks, including BoxKey, UnlockPickup, and MultiRoomaverage returnImproved
Use explicit cause-effect program logic rather than a fixed sequence of subprogramsGALOIS · BoxKey semantic modificationaverage return: 0.119±0.020 -> 0.976±0.001Improved
Use explicit cause-effect program logic rather than a fixed sequence of subprogramsGALOIS · UnlockPickup semantic modificationaverage return: 0.000±0.000 -> 0.983±0.003Improved
Represent observations as valuation vectors rather than raw environment observationsneural DRL baselines · MiniGrid test environments with different sizesaverage returnImproved
Add hierarchical program structure to policy learningprogram synthesis and hierarchical methods · UnlockPickup and BoxKeylearning speedImproved
Reuse a learned GALOIS knowledge base from a task with overlapping logic as initializationGALOIS · DoorKey to BoxKey and other related environmentslearning efficiencyImproved
Reuse only the WHERE subprogram from a related taskGALOIS · BoxKey transferlearning efficiencyImproved
Reuse the BoxKey toggle rule in UnlockPickup, where the object needs to be picked up insteadGALOIS · BoxKey to UnlockPickup transfertask completionHurt
Use valuation-vector inputs instead of raw observationsDQN, SAC, PPO, and hDQN · DoorKey 8*8 trainingaverage return: 0.473±0.130 -> 0.919±0.071; 0.966±0.019 -> 0.938±0.052; 0.919±0.017 -> 0.958±0.008; 0.979±0.002 -> 0.928±0.114Mixed

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