National AI Research Lab Invited Talk Series #2
Invited Talk : Fluid Dynamic Models in Machine Learning
Seminar Overview
- The fluid dynamics model incorporates feature learning through inverse diffusion models and stochastic gradient descent, playing an important role in understanding the dynamics of recent machine learning models.
- By explaining the connection between fluid flow models and probabilistic models, it emphasizes the role of conservation laws in both contexts.
- The Fokker–Planck equation provides a framework for describing how velocity flow fields and diffusion processes influence the spatial and temporal evolution of probability distributions.
- These concepts are also discussed in relation to how they can be applied to modeling neural weights and kernel integral operators in neural networks during feature learning.
Speaker: Professor Daniel Lee
- Tisch University Professor of Electrical and Computer Engineering at Cornell Tech.
- Fellow of IEEE and AAAI, and a recipient of the NSF CAREER Award and the Lindback Award for Distinguished Teaching.
- He has organized the U.S.–Japan Frontiers of Engineering Symposium and the Neural Information Processing Systems (NeurIPS) Conference.
- His main research areas include robotics and autonomy, information, network and decision systems, and artificial intelligence.
- Speaker
- Prof. Daniel Dongyuel Lee
- Date
- November 12, 2024
- Time
- 2:30 PM - 3:30 PM
- Location
- Kim Jae-chul, AI Graduate School, KAIST Seoul Campus, Building 1, Kim Dong-myeong Lecture Room
