报告人：Shuo Li 教授 美国凯斯西储大学

报告时间：2025年7月15日（周二）下午15:00-16:30

报告地点：av电影 九龙湖校区计算机楼513报告厅

报告摘要：Kolmogorov-Arnold Networks (KANs) have emerged as a promisingalternative to traditional multilayer perceptrons (MLPs), offering afundamentally different approach to function approximation. In this talk, I wilreview the core principles and motivations behind KANs, highlighting theimathematical foundation, architectural distinctions, and practical implicationscompared to MLPs. We will explore where KANs outperform, where they falshort, and what this means for real-world applications. I will also presenseveral of our recent research efforts applying KANs to tasks in scientificomputing and medical Al, providing insight into their strengths, limitationsand open challenges, This talk aims to stimulate discussion on whether KAN.or a disruptive shift-in the future of neura]are a complementary tool-network design.

报告人简介：Dr. Li is a global leader in conducting multi-disciplinary research to enable artificialintelligence (Al) in healthcare. He is a Leonard Case, Jr. endowed professor at CaseWestern Reserve University, Before that, he was an associate professor at WesteriUniversity (Canada) and a scientist at the Lawson Health Research Institute. He wasa scientist at GE Healthcare (2006-2015). He is a committee member in multiplhighly influential conferences and societies, He is most notable for serving on theprestigious board of directors in the MICCAl society (2015-2024), where he is alscthe general chair for the MICCAI 2022 conference, which is the most influential Alin-imaging conference. He has over 300 publications, acted as the editor for sixSpringer books, and seryes as an associate editor for several prestigious iournals irthe field,Throuchout his career,he has received severalawards from GE.varlousHe is a Fellow of SPIE, AAIA, IET, andinstitutes, and international organizations.AIMBE.