AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Instructing Goal-Conditioned Reinforcement Learning Agents with Temporal Logic Objectives

Wenjie Qiu, Wensen Mao, He Zhu

NeurIPS 2023 · Reinforcement learning › Visual and representation RL · approach: Graph search and compositional goals · 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
Use GCRL-LTL instead of LTL2Action on in-distribution partially-ordered tasksgoal-conditioned agent · LetterWorldsuccess rate: 0.661 -> 0.996Improved
Increase LTL formula depth beyond training distribution (5 to 15)GCRL-LTL vs LTL2Action · LetterWorldsuccess rate: 0.477 -> 0.986Improved
Increase number of conjuncts (4 to 12) out of distributionGCRL-LTL vs LTL2Action · LetterWorldsuccess rate: 0.567 -> 0.992Improved
Use GCRL-LTL instead of LTL2Action on avoidance tasksgoal-conditioned agent · ZoneEnvsuccess rate: 0.934 -> 0.958No clear effect
Avoid one extra zone in second stage (OOD task)GCRL-LTL vs LTL2Action · ZoneEnvsuccess rate: 0.342 -> 0.928Improved
Long sequence of four consecutive zones (OOD task)GCRL-LTL vs LTL2Action · ZoneEnvsuccess rate: 0.060 -> 0.926Improved
Same OOD sequence task, discounted returnGCRL-LTL vs LTL2Action · ZoneEnvdiscounted return: 0.020 -> 0.220Improved
Use a single goal-conditioned agent instead of per-specification DiRL trainingGCRL-LTL vs DiRL · Ant-16roomstraining steps: 3e6 total vs 3e6 per specificationImproved
Train GCRL-LTL without any LTL task sampler (only random goals)GCRL-LTL vs LTL2Action · LetterWorld/ZoneEnv multi-tasksuccess rateImproved
Apply to omega-regular loop and stability tasks never seen in trainingGCRL-LTL · ZoneEnvomega occurrencesImproved

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