How do Diffusion Models relate to unsupervised learning?

Diffusion models learn data distributions by reversing a noise-adding process without labeled data.

Diffusion models define a forward process that gradually adds Gaussian noise to data until it becomes pure noise. A neural network is then trained to reverse this process step-by-step, learning to denoise and reconstruct the original data distribution. This is fundamentally unsupervised because it does not require labels—only raw data distributions. The model learns score functions or noise predictions that approximate the gradient of the data distribution, enabling high-quality generative sampling.