Diffusion Papers Reading List

An incomplete reading list of diffusion models papers

Useful things to know:

• Entropy: https://en.wikipedia.org/wiki/Entropy_(information_theory)
• ELBO: https://en.wikipedia.org/wiki/Evidence_lower_bound
• Fourier transform: https://en.wikipedia.org/wiki/Fourier_transform
• Learning in the fourier domain: https://arxiv.org/pdf/2006.10739
• Score functions: https://jmlr.org/papers/volume6/hyvarinen05a/hyvarinen05a.pdf

Fundamental papers

• DDPM: https://arxiv.org/pdf/2006.11239
• Improving DDPM schedules: https://arxiv.org/pdf/2102.09672
• DDIM: Faster + deterministic (ish) sampler for DDPM: https://arxiv.org/pdf/2010.02502
• Classifier guidance: https://arxiv.org/pdf/2105.05233
• Classifier-free guidance: https://arxiv.org/pdf/2207.12598
• Score-based methods: https://arxiv.org/pdf/2011.13456
• Denoising in a latent space: https://arxiv.org/pdf/2112.10752

Building off of the fundamental papers:

• ControlNet (condition on different stuff): https://arxiv.org/pdf/2302.05543
• Sampling visual anagrams: https://arxiv.org/pdf/2311.17919
• Solving linear inverse problems: https://arxiv.org/pdf/2201.11793
• Solving non-linear inverse problems: https://arxiv.org/pdf/2209.14687
• Learning clean distributions from noisy data: https://arxiv.org/pdf/2305.19256
• Generative image dynamics: https://arxiv.org/pdf/2309.07906
• Text to shape images with constraints: https://arxiv.org/pdf/2503.14720

Generalizations:

• Sample with any equilibrium: https://arxiv.org/pdf/2208.09392
• Score-based deblurring: https://arxiv.org/pdf/2206.13397