Acceleration and Personalization of VLMs

The convergence of language, vision, and generative models is a captivating and rapidly advancing domain for AI applications. In this talk, we explore the evolving landscape of generative AI through the lenses of model distillation and personalization. We begin with recent breakthroughs in distillation techniques that dramatically reduce the size and inference cost of large generative models while preserving their capabilities. Building on this, we delve into emerging methods for model personalization that enable fine-grained customization of generative behavior for individual users, tasks, or domains. Together, these threads highlight how efficiency and adaptability are shaping the next generation of practical, scalable, and user-centric generative AI systems.

Speaker

Yu-Chiang Frank Wang received his Ph.D. in ECE from Carnegie Mellon University (2009). He joined National Taiwan University in 2017 and became Professor in 2019; since August 2022 he has been the Research Director for Deep Learning and Computer Vision at NVIDIA, leading NVIDIA Research Taiwan. His research covers vision and language, 3D vision, and explainable AI. He has served as area chair for CVPR, ICCV, ECCV and ACCV.

Co-hosted as KAIST Kim Jaechul Graduate School of AI Colloquium (2nd session).