AI Paper Atlas

Research map / Probabilistic and scientific modeling / Causal discovery

TopoDistill: Distilling Global System Topology for Causal Discovery in Multivariate Time Series

Zehao Liu, Pengfei Jiao, Yuhan Wu, Jianqi Yang, Yuyu Yin

ICML 2026 · Probabilistic and scientific modeling › Causal discovery · approach: Causal discovery from time series

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 topology-guided distillation from the global embedder (ablation)TopoDistill · fMRIAUC-PR: 0.596 -> 0.744Improved
add topology-guided distillation from the global embedder (ablation)TopoDistill · Causal RiversAUC-PR: 0.415 -> 0.508Improved
add temporal contrastive learning (ablation)TopoDistill · fMRIAUC-PR: 0.642 -> 0.744Improved
add temporal contrastive learning (ablation)TopoDistill · Causal RiversAUC-PR: 0.452 -> 0.508Improved
use topology and contrastive losses instead of window reconstruction objective (ablation)TopoDistill · Causal RiversAUC-ROC: 0.672 -> 0.813Improved
use topology and contrastive losses instead of window reconstruction objective (ablation)TopoDistill · fMRISHD: 0.100 -> 0.067Improved
add topology distillation (ablation)TopoDistill · Causal RiversSHD: 0.200 -> 0.150Improved
replace TCN delay embedder with MLPTopoDistill · Double PendulumAUC-ROC: 0.888 -> 0.912No clear effect
replace TCN delay embedder with MLPTopoDistill · Causal RiversAUC-PR: 0.508 -> 0.525No clear effect
replace time-delay embedding with learned TopoDistill embedderCCM pipeline · Lorenz-96 with noiseAUC-ROC/AUC-PRImproved
replace CCM with TopoDistill-TCNCCM · Double PendulumSHD: 0.167 -> 0.067Improved
replace CCM with TopoDistill-TCNCCM · Causal RiversAUC-ROC: 0.594 -> 0.794Improved

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