Diffusion Model
Definition
A generative model that learns to create data by gradually denoising a random noise signal, reversing a process that progressively adds noise to training data.
Diffusion models work by defining a forward process that gradually corrupts data with Gaussian noise over many steps, and then training a neural network to reverse this process step by step. During generation, the model starts with pure random noise and iteratively denoises it into a coherent output. This approach produces remarkably high-quality images with better diversity and training stability compared to GANs. Diffusion models power leading image generators like DALL-E 3, Midjourney, and Stable Diffusion. They have been extended to video generation (Sora), audio synthesis, and 3D object creation. Techniques like classifier-free guidance and latent diffusion have made them both more controllable and more computationally efficient.
Related Terms
GAN (Generative Adversarial Network)
A generative model architecture consisting of two neural networks — a generator and a discriminator ...
VAE (Variational Autoencoder)
A generative model that learns a compressed latent representation of data and can generate new sampl...
Generative AI
AI systems that can create new content such as text, images, audio, video, and code, rather than sim...
Text-to-Image
AI systems that generate images from natural language descriptions, typically using diffusion models...