AI Paper Atlas

Research map / Multi-agent systems and games / Human-AI and goal-conditioned

Generating High-Quality Explanations for Navigation in Partially-Revealed Environments

Gregory J. Stein

NeurIPS 2021 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Language-grounded reinforcement agents · 2 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 explanation-based training loss Lcomp (vs none) (ablation)learned planner · Guided Mazeaverage distance: 19.17 -> 14.94Improved
add explanation-based training loss Lcomp (vs none) (ablation)learned planner · University Buildingsaverage distance: 44.34 -> 42.46Improved
limit gradient to 4 most important subgoal properties instead of all (ablation)learned planner · University Buildingsaverage distance: 43.52 -> 42.46Improved
limit gradient to 4 subgoal properties instead of all (ablation)learned planner · Guided Mazeaverage distance: 14.94 -> 14.94No clear effect
train via explaining (4 properties) vs non-learned baselinelearned planner · University Buildingsaverage distance: 46.70 -> 42.46Improved
train via explaining (all properties) vs non-learned baselinelearned planner · Guided Mazeaverage distance: 19.45 -> 14.94Improved
learned baseline without Lcomp vs non-learned baselinelearned planner · Guided Mazeaverage distance: 19.45 -> 19.17No clear effect
intervene using explanation (4 properties)learned planner · University Buildings worst 50 trialsaverage distance: 124.24 -> 93.77Improved
intervene using explanation (all properties)learned planner · University Buildings worst 50 trialsaverage distance: 124.24 -> 92.81Improved
explain with all subgoal properties instead of 4 during intervention (ablation)learned planner · University Buildings worst 50 trialssuccessful interventions: 37/50 -> 43/50Improved
intervened agent vs non-learned baselinelearned planner · University Buildings worst 50 trialsaverage distance: 85.05 -> 93.77Hurt

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