Part B Overview & Diagram

The Principles of Diffusion Models

By Chieh-Hsin Lai, Yang Song, Dongjun Kim, Yuki Mitsufuji, Stefano Ermon

Front Matter

Acknowledgements
Preface and Roadmap
Notations

Part A: Introduction to Deep Generative Modeling

Chapter 1: Deep Generative Modeling

Part B: Foundations and Perspectives of Diffusion Models

Part B Overview & Diagram
A First Operational View of Diffusion Model Training and Sampling
Chapter 2: Variational Perspective: From VAEs to DDPMs
Chapter 3: Score-Based Perspective: From EBMs to NCSN
Chapter 4: Diffusion Models Today: Score SDE Framework
Chapter 5: Flow-Based Perspective: From NFs to Flow Matching
Chapter 6: A Unified and Systematic Lens on Diffusion Models
Chapter 7: (Optional) Diffusion Models and Optimal Transport

Part C: Sampling of Diffusion Models

Part C & D Overview & Diagram
Chapter 8: Guidance and Controllable Generation
Chapter 9: Numerical Solvers for Fast Diffusion Sampling

Part D: Toward Learning Fast Diffusion-Based Generators

Chapter 10: Distillation-Based Methods for Fast Sampling
Chapter 11: Learning Fast Generators from Scratch

Epilogue & Outlook

Chapter 12: A Unifying Principle and the Road Ahead

Appendices

Appendix A: Crash Course on Differential Equations
Appendix B: Density Evolution: From Change of Variable to Fokker–Planck
Appendix C: Behind the Scenes of Diffusion Models: Itô’s Calculus and Girsanov’s Theorem
Appendix D: Supplementary Materials and Proofs

Part B Overview & Diagram

Part B Overview & Diagram

← Previous Chapter Chapter 1: Deep Generative Modeling Next Chapter → A First Operational View of Diffusion Model Training and Sampling