Signals & structures
Time series, anomaly detection, and causal inference reveal patterns and relationships in complex systems.
Explore time-series researchSAIL · Tsinghua University
Superintelligence And Industrial Lab
Superintelligence
for complex systems.
We advance research in superintelligence (SI), combining statistics and machine learning to understand complex data and guide better decisions.
Department of Industrial Engineering, Tsinghua University
Research directions
Understand. Learn. Decide.
Models that generalize
Foundation models, multimodal learning, and statistical methods connect knowledge across data and domains.
Explore foundation modelsIntelligence in action
Adaptive sampling, optimization, and control turn insights into decisions for manufacturing, transportation, and healthcare.
Explore decision methods
From the lab
Latest news
Paper acceptances, research milestones, and lab updates.
2026
Paper Accepted by Nature Communications
Our paper “A Computer-Aided System for Enhanced M-protein Diagnosis and Monitoring via Multimodal Learning” has been accepted by Nature Communications. Paper link forthcoming.
2026
Paper Accepted by NeurIPS 2026
Our paper “VETime: Vision Enhanced Zero-Shot Time Series Anomaly Detection” has been accepted by NeurIPS 2026. [arXiv]
2026
Paper Accepted by NeurIPS 2026
Our paper “Supervision Recovery for Time Series Anomaly Detection via Counterfactual Pairing” has been accepted by NeurIPS 2026. [OpenReview]
2026
Paper Accepted by NeurIPS 2026
Our paper “Learning Collision-Free Dispatch Policies for Route-Wise Decision-Dependent Anomaly Detection” has been accepted by NeurIPS 2026. [OpenReview]
2026
Paper Accepted by IEEE TPAMI
Our paper “Quickest Causal Change Point Detection by Adaptive Intervention” has been accepted by IEEE TPAMI. [arXiv]
Research in focus
Recent publications
-
01
↗
2026 · Nature Communications — accepted
A Computer-Aided System for Enhanced M-protein Diagnosis and Monitoring via Multimodal Learning
-
02
↗
2026 · NeurIPS 2026 — accepted
VETime: Vision Enhanced Zero-Shot Time Series Anomaly Detection
-
03
↗
2026 · NeurIPS 2026 — accepted
Supervision Recovery for Time Series Anomaly Detection via Counterfactual Pairing
-
04
↗
2026 · NeurIPS 2026 — accepted
Learning Collision-Free Dispatch Policies for Route-Wise Decision-Dependent Anomaly Detection
-
05
↗
2026 · IEEE Transactions on Pattern Analysis and Machine Intelligence — accepted
Quickest Causal Change Point Detection by Adaptive Intervention
-
06
↗
2026 · Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2, 3436-3447
SCALE: Style-Causal Disentanglement with Adaptive Lifelong Expert for Online Latent-domain Anomaly Detection
Work with us
Ideas grow through collaboration.
Explore research opportunities and collaborations in superintelligence and industrial systems.