From the course: The AI Ecosystem for Developers: Models, Datasets, and APIs

Unlock this course with a free trial

Join today to access over 24,600 courses taught by industry experts.

Generative architectures: Diffusion and GANs

Generative architectures: Diffusion and GANs

From the course: The AI Ecosystem for Developers: Models, Datasets, and APIs

Start my 1-month free trial Buy for my team

Generative architectures: Diffusion and GANs

“

- [Instructor] Generative AI architecture focus on creating new data instances that resemble the training data. They can generate images, text, and even audio data resembling real-world examples. They are useful for tasks such as image sentences, text generation, and audio creation. Two permanent generative architectures are diffusion models and generative adversarial networks, GANs. Diffusion models are a class of generative models that operate on the principle of gradually adding noise to an image and then learning to reverse this process to generate new data. It is inspired by how diffusion works in physics, where particles spread out from high to low concentration. In diffusion models, the goal is to learn how to reverse this diffusion process, transforming random noise into structured data. The key components of a diffusion model include forward process, where you gradually add noise to the input data until it becomes pure noise. Reverse process, the model learns to reverse the…

Contents