The Little Book of Generative AI Foundations: An Intuitive Mathematical Primer
Points and comments are a snapshot, not live.
Tianhua Chen's book offers a derivation-focused primer on generative AI mathematical foundations.
The 195-page preprint introduces the mathematics behind major generative model families, including PCA, variational autoencoders, diffusion models, normalising flows, GANs, Wasserstein GANs, and energy-based models. It focuses on derivations and relationships between models rather than surveying all recent architectures. The revised version adds visual explanations. Aimed at researchers, practitioners, and students seeking an accessible yet substantive foundation.
What commenters are saying
Commenters compared it to François Fleuret's 'The Little Book of Deep Learning,' noting that Fleuret's book covers basic deep learning while this one addresses generative AI foundations. Several readers appreciated the phone-formatted PDF option for Fleuret's book. One commenter shared a reformatted version. Another noted the book's relevance to diffusion models and its potential intersection with transformer architectures like DiT. A reader shared a large collection of related whitepapers organized for RAG workflows.