AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Inverse Reinforcement Learning with Natural Language Goals

Li Zhou, Kevin Small

AAAI 2021 · Reinforcement learning › Visual and representation RL · approach: Graph search and compositional goals · 22 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 variational goal generation with EGR, HGR, and HGSLangGoalIRL · R2R seen validationsuccess rate: 0.449 ± 0.0130 -> 0.530 ± 0.0138Improved
Add self-supervised fine-tuning in the unseen environmentLangGoalIRL · R2R unseen validationsuccess rate: 0.357 ± 0.0089 -> 0.491 ± 0.0065Improved
Add attention over natural-language goalsLangGoalIRL_BASE · R2R seen validationsuccess rate: 0.359 ± 0.0117 -> 0.449 ± 0.0130Improved
Add EGR alone (ablation)LangGoalIRL · R2R seen validationsuccess rate: 0.449 ± 0.0130 -> 0.450 ± 0.0122No clear effect
Add HGR alone (ablation)LangGoalIRL · R2R seen validationsuccess rate: 0.449 ± 0.0130 -> 0.427 ± 0.0048Hurt
Add HGS alone (ablation)LangGoalIRL · R2R seen validationsuccess rate: 0.449 ± 0.0130 -> 0.399 ± 0.0124Hurt
Add HGR to EGR (ablation)LangGoalIRL · R2R seen validationsuccess rate: 0.450 ± 0.0122 -> 0.504 ± 0.0185Improved
Add HGS to EGR and HGR (ablation)LangGoalIRL · R2R seen validationsuccess rate: 0.504 ± 0.0185 -> 0.530 ± 0.0138Improved
Add EGR to HGR and HGS (ablation)LangGoalIRL · R2R seen validationsuccess rate: 0.503 ± 0.0117 -> 0.530 ± 0.0138Improved
Add HER to HGR without EGR or HGS (ablation)LangGoalIRL · R2R seen validationsuccess rate: 0.427 ± 0.0048 -> 0.101 ± 0.0790Hurt
Add HER to the full EGR, HGR, and HGS method (ablation)LangGoalIRL · R2R seen validationsuccess rate: 0.530 ± 0.0138 -> 0.521 ± 0.0093Hurt
Collect new-environment demonstrations with the goal-agnostic policy instead of random exploration (ablation)LangGoalIRL_SS · R2R unseen validation, 15,000 demonstrationssuccess rate: 0.347 ± 0.0098 -> 0.491 ± 0.0065Improved

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