AI Paper Atlas

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

Causal Curiosity: RL Agents Discovering Self-supervised Experiments for Causal Representation Learning

Sumedh A Sontakke, Arash Mehrjou, Laurent Itti, Bernhard Schölkopf

ICML 2021 · Multi-agent systems and games › Human-AI and goal-conditioned · approach: Language-grounded reinforcement agents · 10 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
Pre-train the planner with causal curiosity before fine-tuning on a downstream taskCausal curiosity planner · Liftingtraining convergenceImproved
Pre-train the planner with causal curiosity before fine-tuning on a downstream taskCausal curiosity planner · Traveltraining convergenceImproved
Add the causal curiosity reward while optimizing the external task reward (ablation)Causal curiosity planner · Lifting and Traveldownstream task performanceHurt
Provide learned causal representations as policy inputsPPO policy · TransferMass and TransferSizeMasszero-shot generalizationImproved
Use self-supervised causal-curiosity experiments instead of supervised plannersExperiment planner · Downstream behavior pre-trainingdata efficiency: approximately 2.5 times less dataImproved
Provide only the z coordinate to the curiosity module (ablation)Causal curiosity experiment planner · Causal World block experimentsbehavior relevance to supplied coordinateImproved
Provide only the x coordinate to the curiosity module (ablation)Causal curiosity experiment planner · Causal World block experimentsbehavior relevance to supplied coordinateImproved

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