Speaker |
Zijing Ou |
|---|---|
Affiliation |
Imperial College London |
Date |
Friday, 04 April 2025 |
Time |
12:00-13:00 |
Location |
UCL Centre for Artificial Intelligence, 1st Floor, 90 High Holborn, London WC1V 6BH |
Link |
https://ucl.zoom.us/j/99748820264 |
Event series |
Jump Trading/ELLIS CSML Seminar Series |
Abstract |
Traditional diffusion models are typically trained to predict only the mean of the denoised distribution given a noisy sample. But what if we go beyond the mean? This talk explores how incorporating additional information—such as predicting the covariance of the denoised distribution—can significantly accelerate sampling and improve density estimation. We’ll dive into different techniques for covariance prediction, their theoretical connection, and practical benefits for more efficient and expressive generative modelling. |
Biography |
Zijing is a PhD student in the Department of Computing at Imperial College London, learning to train energy-based models under the supervision of Yingzhen Li. He completed his undergraduate studies in the School of Computer Science and Engineering at Sun Yat-sen University and previously worked as a research intern at Apple MLR, Shell AI, Tencent AI Lab, and Tencent Jarvis Lab. |