Lecture
Score-Based Generative Models
Description
This lecture explores the concept of score-based generative models, focusing on the learning of natural distributions without a defined score. Topics include Langevin Monte Carlo algorithms, Hyvärinen's trick for learning the score, and the theoretical guarantees of sampling. The lecture also delves into the impact of neural network architecture on robustness, initialization strategies, and the trade-offs between width and depth in deep learning.
Official source
About this result
This page is automatically generated and may contain information that is not correct, complete, up-to-date, or relevant to your search query. The same applies to every other page on this website. Please make sure to verify the information with EPFL's official sources.