Machine Learning Seminar: Dr. Yining Hong, "Bridging Mind and World through Embodied Foundation Models."

Abstract: This talk explores how embodied foundation models can connect the mind and the physical world. Yining will begin with models for perceiving and understanding the 3D world, and then move to embodied reasoning and action, where an agent reasons through a physical body situated in 3D space. She will then discuss closing the perception–action loop, enabling agents to actively explore and interact with the world, observe the consequences of their actions, and update their understanding accordingly. Finally, she will consider how repeated perception–action loops give rise to the life, experience, and long-term memory of embodied agents, and how interactions among multiple embodied agents may support social intelligence and theory of mind.

Bio: Yining Hong is a postdoctoral researcher in Computer Science at Stanford University, advised by Professors Yejin Choi and Leonidas Guibas. She received her Ph.D. from UCLA, advised by Professors Ying Nian Wu, Kai-Wei Chang and Song-Chun Zhu. She got her bachelor degree from Shanghai Jiao Tong University. Her research focuses on building general-purpose embodied agents that can perceive, reason about, and interact with the 3D physical world, spanning 3D foundation models, world models, and embodied reasoning. She is a recipient of the Rising Star Award for Spatial Intelligence, Rising Star in Computational and Data Sciences, and Baidu Scholarship.