报告人： Shiyu Wang（ByteDance Inc.）
时间：2025-09-25 15:30-17:00
地点：Room 217, Guanghua Building 2
Abstract:
In recent years, with the rapid development of deep learning, time series modeling has entered a new era. A series of deep time series models, represented by Autoformer, have emerged and significantly advanced the field. More recently, the success of large models has further inspired creativity in the time series community. New paradigms are quickly taking shape, exemplified by models such as TimeMoE, TimeMixer, and U-Cast. This talk will provide an overview of these latest developments, highlighting their design principles, key innovations, and potential impact on real-world applications.
About the Speaker:
Shiyu Wang is a Senior Staff Researcher at ByteDance Inc., where he serves as the Director of the Supply Chain Algorithm Team. Previously, he was a Staff Researcher at Ant Group. His research primarily focuses on time series analysis with deep learning and the development of multimodal large models for time series. He is dedicated to building powerful learning machines from large-scale data to address real-world application challenges. At ByteDance, he leads the development of a large-scale time series forecasting platform and drives innovative research in time series algorithms, enhancing the integration of forecasting and decision-making capabilities. He has published extensively in premier conferences such as NeurIPS, ICLR, ICML, KDD, AAAI, IJCAI, WSDM, ICDM, and DASFAA, and actively contributes to the research community as a Senior Area Chair for KDD and as a reviewer and program committee member for multiple venues. He has also been recognized with Outstanding Reviewer Awards for his contributions.
Your participation is warmly welcomed!
欢迎扫码关注北大统计科学中心公众号，了解更多讲座信息!
