Learning, reasoning, and discovery from complex data.
Building AI that understands complex relational, multimodal, and scientific data — and turns representations into reasoning, personalization, and discovery.
We are DSAIL at KAIST, led by Prof. Chanyoung Park. Our research spans graph and relational learning, personalized recommendation, multimodal intelligence, agentic AI, and AI for science. Across these areas, we study how representations learned from complex data can support generalization, reasoning, decision-making, and discovery.
Looking to join the lab? See Prospective students.
Research at a glance
Described in full on the Research page.
News
- August 2026Four papers got accepted at CIKM 2026 (2 x Research Track, 1 x Applied Research Track, 1 x Short Paper Track).
- August 2026Our paper "Physics-Embedded Graph Neural Operator for Interaction-Controlled Colloidal Aggregation" received the Best Paper Award at KDD 2026 Workshop on Reliable Scientific Foundation Models.
- August 2026Our paper "Personalize-then-Store: Benchmarking and Learning Personalized Memory for Long-Horizon Agents" received the Best Paper Award at KDD 2026 Workshop on Personal Intelligence in the Agentic AI Era (PILA).
- August 2026Ilho Shin and Hyunchul Kim ranked 1st in Task 2: Contextualized Early Detection of Depression at CLEF eRisk 2026.
- June 2026Eleven undergraduate interns joined our lab.
- June 2026Two papers got accepted at ECCV 2026.
- May 2026Six papers got accepted at ICML 2026 Workshop (3 x AI for Science, 1 x CATS, 1 x FAGEN, 1 x GenBIO).
- May 2026Two papers got accepted at KDD 2026 (1 x Research Track and 1 x AI for Science Track).
- May 2026Five papers got accepted at ICML 2026.
- April 2026A paper got accepted by ACM Transactions on Intelligent Systems and Technology.
Contact
Address
DSAIL, Room 4124, E2-2, KAIST
291 Daehak-ro, Yuseong-gu, Daejeon, Republic of Korea
View on map →
DSAIL, Room 4124, E2-2, KAIST
291 Daehak-ro, Yuseong-gu, Daejeon, Republic of Korea
View on map →
