Diffusion Models

Back to Generative Models

Generate data by learning to reverse a gradual noising process. Start from pure noise and iteratively denoise to produce realistic outputs. State-of-the-art for image generation (Stable Diffusion, DALL-E, Midjourney). More stable training than GANs.

  • GANs (earlier approach to image generation)
  • VAEs (also learn latent representations)

deep-learning generative diffusion image-generation