Diffusion Models: A Comprehensive Survey of Methods and Applications

Ling Yang,Zhilong Zhang,Yang Song,Shenda Hong,Runsheng Xu,Yue Zhao,Wentao Zhang,Bin Cui,Ming-Hsuan Yang

2024-06-24

Abstract:Diffusion models have emerged as a powerful new family of deep generative models with record-breaking performance in many applications, including image synthesis, video generation, and molecule design. In this survey, we provide an overview of the rapidly expanding body of work on diffusion models, categorizing the research into three key areas: efficient sampling, improved likelihood estimation, and handling data with special structures. We also discuss the potential for combining diffusion models with other generative models for enhanced results. We further review the wide-ranging applications of diffusion models in fields spanning from computer vision, natural language generation, temporal data modeling, to interdisciplinary applications in other scientific disciplines. This survey aims to provide a contextualized, in-depth look at the state of diffusion models, identifying the key areas of focus and pointing to potential areas for further exploration. Github: <a class="link-external link-https" href="https://github.com/YangLing0818/Diffusion-Models-Papers-Survey-Taxonomy" rel="external noopener nofollow">this https URL</a>.

Machine Learning,Artificial Intelligence,Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The paper attempts to address the issue of providing a comprehensive review of Diffusion Models, an emerging and powerful family of deep generative models. Specifically, the paper aims to:

1. **Overview the research progress of diffusion models**: The paper categorizes the research in the field of diffusion models into three key areas: efficient sampling, improved likelihood estimation, and handling data with special structures (such as discrete data, data with invariant structures, and data on manifolds).

2. **Explore the combination with other generative models**: The paper also discusses how to combine diffusion models with other generative models (such as Variational Autoencoders, Generative Adversarial Networks, Normalizing Flows, Autoregressive Models, and Energy-Based Models) to enhance performance.

3. **Summarize application areas**: The paper reviews the wide applications of diffusion models in various fields, including computer vision, natural language processing, time series modeling, multimodal learning, robust learning, and interdisciplinary applications (such as drug design, material design, and medical image reconstruction).

4. **Identify future research directions**: The paper finally looks forward to the future research directions in the field of diffusion models, including revisiting assumptions, theoretical understanding, latent representations, AI-generated content (AIGC), and foundational diffusion models.

Through these contents, the paper aims to provide a comprehensive introductory guide for new researchers entering the field, while also offering a broader perspective for experienced researchers.