Chapter 1. AIGC and Related Technologies

1.1 Introduction to AIGC 2
1.2 Introduction to Diffusion Modeling 4

Chapter 2.
Fundamentals of Diffusion Modeling

2.1 Noise Removal Diffusion Probability Model 10
2.2 Score-based generative model 22
2.3 Stochastic Differential Equations 25

Chapter 3.
Efficient sampling for diffusion modeling

3.1 Differential Equations 50
3.2 Deterministic Sampling 52
3.2.1 Stochastic Differential Equation Solver 52
3.2.2 Ordinary Differential Equation Solver 60
3.3 Learning-based sampling 66
3.3.1 Individual Mode 66
3.3.2 Cut diffusion 68
3.3.3 Knowledge Distillation 75

Chapter 4.
Likelihood maximization for diffusion models

4.1 Maximizing the Likelihood Function 82
4.2 Optimizing the Noise Addition Strategy 84
4.3 Inverse Distributed Learning 88
4.4 Exact log-likelihood estimation 100

Chapter 5.
Applying diffusion modeling to data with special structures

5.1 Discontinuous Data 106
5.2 Data with Immutable Structures 110
5.3 Data in Streaming Structure 118
5.3.1 Known Flow Patterns 118
5.3.2 Unknown Flow Pattern 119

Chapter 6.
Diffusion model related to other generative models

6.1 Various Autoencoders and Diffusion Models 126
6.2 Creating Adversarial Networks and Diffusion Models 128
6.3 Normalized Flow and Diffusion Models 135
6.4 Autoregressive and Diffusion Models 139
6.5 Energy-based models and diffusion models 140

Chapter 7.
Application of diffusion modeling

7.1 Comparison of Unconditional and Conditional Diffusion Models 144
7.2 Computer Vision 145
7.2.1 Image Super-Resolution, Image Restoration, and Image Translation 145
7.2.2 Semantic Segmentation 151
7.2.3 Video Creation 154
7.2.4 Point Cloud Completion and Point Cloud Generation 157
7.2.5 Anomaly Detection 160
7.3 Natural Language Processing 162
7.4 Temporal Data Modeling 168
7.4.1 Time Series Interpolation 168
7.4.2 Time Series Forecasting 170
7.5 Multimodal Learning 172
7.5.1 Text-to-Image Generation 172
7.5.2 Creating Text-to-Speech 180
7.5.3 From Scene Graph to Image Generation 182
7.5.4 Creating Text-3D Content 184
7.5.5 Creating Text-Body Motion 185
7.5.6 Text-to-Video Generation 185
7.6 Robust Learning 187
7.7 Convergence Applications 188
7.7.1 Artificial Intelligence Drug Development 188
7.7.2 Medical Imaging 195

Chapter 8.
The Future of Diffusion Modeling - GPT and Large Models

8.1 Introduction to Pre-Learning Techniques 201
8.1.1 Generative Pretraining and Contrastive Pretraining 202
8.1.2 Parallel Learning Techniques 206
8.1.3 Fine-tuning techniques 209
8.2 GPT and Large Model 210
8.2.1 GPT-1 211
8.2.2 GPT-2 214
8.2.3 GPT-3 and Large Model 217
8.2.4 InstructGPT and ChatGPT 225
8.2.5 Visual Chat GPT 229
8.3 GPT and Large-Scale Model-Based Diffusion Modeling 232
8.3.1 Algorithm Study 232
8.3.2 Applying the Paradigm 233

Search 235

This book comprehensively introduces how to apply the DIFFUSION MODEL to various fields, and covers AIGC (AI Generated Content) and related technologies, the basics of diffusion models, efficient sampling of diffusion models, maximizing the potential of diffusion models, applying diffusion models to data with special structures, the correlation between diffusion models and other generative models, applications of diffusion models, and the future of diffusion models. It will help you understand the main research fields of diffusion models.

The structure of this book is as follows:


Chapter 1 introduces AIGC and related technologies. Chapter 2 introduces the basic theory and algorithm of the diffusion model from three perspectives, and introduces the neural network architecture and code practice for the diffusion model.


Chapters 3, 4, and 5 systematically introduce the characteristics of the diffusion model and the corresponding improvement work from three perspectives: efficient sampling of the diffusion model, maximizing the potential of the diffusion model, and applying the diffusion model to data with special structures, respectively.


Chapter 6 discusses the relationship between diffusion models and other generative models, including variational autocoders, generative adversarial networks, regularized flow, autoregressive models, and energy-based models.


Chapter 7 introduces application examples of the diffusion model.
Chapter 8 discusses GPT and large-scale models, the future of diffusion models.

This book is written for students majoring in computer science, artificial intelligence, and machine learning, as well as developers of big data and artificial intelligence applications.
It will also be useful for undergraduate students, graduate students, faculty, and researchers at research institutions.