AI Paper Atlas

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

State2Explanation: Concept-Based Explanations to Benefit Agent Learning and User Understanding

Devleena Das, Sonia Chernova, Been Kim

NeurIPS 2023 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Theory-of-mind social reasoning

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 S2E concept-based reward shaping to sparse-reward MuZeroMuZero · Connect 4training samples to similar Win%: ~200k less training samplesImproved
replace expert-defined reward shaping with S2E-informed shapingMuZero · Lunar LanderReward and Win%No clear effect
13.9% incorrect explanation retrievals vs expert shaping upper bound (ablation)MuZero+S2E-RS · Connect 4learning rate: 13.9\%No clear effect
show S2E concept explanation instead of no explanationConnect 4 users · ATS improvementt-test: t(70) = 3.08Improved
show S2E concept explanation instead of action explanationConnect 4 users · ATS improvementt-test: t(70)=-3.25Improved
show S2E concept explanation instead of value explanationConnect 4 users · ATS improvementt-test: t(70)=4.03Improved
S2E-retrieved explanation vs ground-truth concept explanation (ablation)Connect 4 users · ATS improvementATSNo clear effect
add InF and TeG abstraction to S2E concept explanations vs noneLunar Lander users · ATS improvementt-test: t(98) = 3.15Improved
add InF and TeG abstraction vs unabstracted concept explanation (ablation)Lunar Lander users · ATS improvementt-test: t(98)=-3.19Improved
add InF and TeG vs value-based explanationLunar Lander users · ATS improvementt-test: t(98)=3.59Improved
use unabstracted S2E concept explanation in high-frequency Lunar Lander (ablation)Lunar Lander users · ATS improvementATSNo clear effect
InF thresholds x-position 0.15, tilt 0.01 and -0.05 sensitivity (ablation)InF · Lunar Landerfraction of concepts filteredNo clear effect

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