First Principles with Christian Keil cover image

#12: Deep Prasad - How to Build a Physics-Fluent AI System

First Principles with Christian Keil

NOTE

Learning Coherent Images through Diffusion Models

Diffusion models incrementally add noise to a model and train a loss function to generate coherent images from noisy ones. By teaching AI to work backwards from noisy images to original ones, it learns to create any coherent image. Injecting noise into the system forces it to come up with realistic constructions. Combining crystal diffusion and variational autoencoder results in a model that improves with more data.

00:00
Transcript
Play full episode

Remember Everything You Learn from Podcasts

Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
App store bannerPlay store banner