From the course: Programming Generative AI: From Variational Autoencoders to Stable Diffusion with PyTorch and Hugging Face

Unlock this course with a free trial

Join today to access over 24,800 courses taught by industry experts.

Look ma, no features!

Look ma, no features!

From the course: Programming Generative AI: From Variational Autoencoders to Stable Diffusion with PyTorch and Hugging Face

Start my 1-month free trial Buy for my team

Look ma, no features!

“

- One of the really powerful features of neural networks and deep learning in general is their ability to perform automatic feature learning. Gone are the days where you have to handcraft features about what are the important aspects of text, images, or whatever input data. This applies both to special neural network architectures like convolutional neural networks, transformers, et cetera, but also for any standard multi-layer perceptron or feedforward network. When it comes to images, we can think of this natural image manifold that acts almost as the idealized generative process of images, the natural or universal thing that makes images. In this case, it might be a very complex, very high dimensional space, and at one point, we might have artworks. We have things like the Mona Lisa, we have abstract art, we have stained glass, and it's impossible to capture the complexity of all of the space of natural images. So you'll have to use your imagination to think of something much more…

Contents