Title: Time Series Learning: From Forecasting to Reasoning and Discovery

Abstract: Time series learning has become a central problem in modern machine learning, with applications spanning science, engineering, healthcare, and finance. While much of the existing research has focused on improving forecasting accuracy, accurate prediction is only the starting point. This talk explores a broader view of time series learning, moving from forecasting toward reasoning and scientific discovery. I will first discuss our work on efficient time series forecasting, introducing cross-architecture knowledge distillation to develop lightweight models that achieve strong predictive performance with reduced computational costs. I will then explore spatio-temporal reasoning, presenting methods that integrate time series, spatial structures, and large language models to understand how events evolve and propagate across connected systems. Finally, I will discuss whether AI can move beyond prediction to discover the underlying mathematical and causal relationships governing temporal dynamics. These directions highlight the potential of time series learning to advance from accurate prediction toward deeper understanding and scientific discovery.

Bio: Wei Jin is an Assistant Professor in the Department of Computer Science and the Department of Biostatistics and Bioinformatics at Emory University. His research focuses on time series analysis and AI for health. His honors and achievements include two NIH R01 awards as principal investigator, IJCAI-ECAI Early Career Spotlights, AAAI New Faculty Highlights, KAUST Rising Stars, the INNS Doctoral Dissertation Runner-up Award, recognition for Most Influential Papers in KDD and WWW by Paper Digest, and top finishes in three NeurIPS competitions. His research has been published in Nature Portfolio journals, PNAS, and leading conferences including ICLR, KDD, ICML, and NeurIPS. He has also organized multiple tutorials and workshops at major AI conferences.