Diffusion Models: A Comprehensive Survey of Methods and Applications

By Ling Yang, Zhilong Zhang, Shenda Hong, Runsheng Xu, Yue Zhao, Yingxia Shao, Wentao Zhang, Ming-Hsuan Yang, Bin Cui. ACM Computing Surveys, 2022

2,329 citations recorded by the source corpus.

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 processing, 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: https://github.com/YangLing0818/Diffusion-Models-Papers-Survey-Taxonomy

Subfields: Computational Theory and Mathematics, Computer Vision and Pattern Recognition, Modeling and Simulation

View Paper: https://www.semanticscholar.org/paper/e342165a614588878ad0f4bc9bacf3905df34d08

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

1. Sheng Wang
2. Jianzhu Ma
3. Jian Peng
4. Jimmy J. Lin
5. Fang Wen
6. Thomas Bo Schön
7. Xiang Wang
8. Jeff Zhang

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Xiaogang Wang
3. Dong Chen
4. Xiaoou Tang
5. Fang Wen
6. Dacheng Tao
7. Ming-Hsuan Yang
8. Jianmin Bao

Citations in the corpus

• I2VGen-XL: High-Quality Image-to-Video Synthesis via Cascaded Diffusion Models (2023, methodology): “There are also methods like LDM [35], LSGM [43] and RDM [19] that utilize distribution learning in the latent space to significantly reduce computational overhead and enable the generation of high-resolution image [10,51].”
• Parameter-Efficient Fine-Tuning for Large Models: A Comprehensive Survey (2024, background): “Diffusion models has various applications [62], [199], [32], [38], [155], while the most notable is stable diffusion [153], which bridges the gap between text and image with its robust capability to generate coherent and contextually relevant images directly from textual descriptions.”
• Explainable Artificial Intelligence (XAI) 2.0: A Manifesto of Open Challenges and Interdisciplinary Research Directions (2023, methodology): “Generative AI models, such as those employed for diffusion denoising [124,125] or the family of GPT models for large-scale language generation [126], are disrupting many sectors.”
• Image Restoration with Mean-Reverting Stochastic Differential Equations (2023, methodology): “Among the commonly used formulations (Yang et al., 2022), we adopt that of using the diffusion models defined via stochastic differential equations (SDEs, Song (1)Department of Information Technology, Uppsala University, Sweden.”
• Text-to-image Diffusion Models in Generative AI: A Survey (2023, methodology): “Therefore, the optimization objective of model is as follows(quoted from [33]):”
• Diffusion models in medical imaging: A comprehensive survey (2022, background): “There have also been review articles on diffusion models surfacing recently for computer vision tasks [20, 21, 22].”
• RODIN: A Generative Model for Sculpting 3D Digital Avatars Using Diffusion (2022, background): “Diffusion models [14, 24, 61, 73] have recently shown unprecedented generative ability and compositional power.”
• Foundation Models Defining a New Era in Vision: A Survey and Outlook (2025, neutral): “Note that we do not cover image generative models aimed at model data distribution such as GANs, VAEs, and Diffusion models owing to dedicated surveys already existing in this area [38], [39], [40], [41] and because the former model class can cover a broader range of downstream applications.”
• Antigen-Specific Antibody Design and Optimization with Diffusion-Based Generative Models for Protein Structures (2022, methodology): “To address these challenges, we propose a diffusion-based generative model [Sohl-Dickstein et al., 2015, Song and Ermon, 2019, Ho et al., 2020, Yang et al., 2022] capable of jointly sampling antibody CDR sequences and structures.”
• Post-Training Quantization on Diffusion Models (2022, background): “A diffusion process transforms real data gradually into Gaussian noise, and then the process is reversed to generate real data from Gaussian noise (denoising process) [10, 39].”
• Retrieval-Augmented Generation for AI-Generated Content: A Survey (2024, neutral): “The entire process can be defined as follows: 3) Diffusion Model : Diffusion models [64] are a family of deep generative models that can create realistic and diverse samples of data [65]–[72], such as images [73]–[79], texts [80]–[83], videos [84]–[88], and molecules [89]–[93].”
• CoMoSpeech: One-Step Speech and Singing Voice Synthesis via Consistency Model (2023, background): “, 2020a] [Yang et al., 2022] of the diffusion model is that it requires many iterations for the generation.”
• What the DAAM: Interpreting Stable Diffusion Using Cross Attention (2022, background): “Diffusion neural networks trained on billions of image– caption pairs represent the state of the art in text-to-image generation [1], some achieving realism comparable to actual photographs in human evaluation.”