Description
Generative modeling is certainly one of the most up to date subjects in AI. It’s now conceivable to show a gadget to excel at human endeavors equivalent to portray, writing, and composing track. With this sensible e-book, gadget-finding out engineers and information scientists will uncover find out how to re-create probably the most such a lot spectacular examples of generative deep finding out fashions, equivalent to variational autoencoders,generative hostile networks (GANs), encoder-decoder fashions, and global fashions.
Author David Foster demonstrates the internal workings of each and every method, beginning with the fundamentals of deep finding out prior to advancing to probably the most such a lot state of the art algorithms within the box. Thru guidelines and tips, you’ll consider find out how to make your fashions be told extra successfully and turn into extra ingenious.
- Discover how variational autoencoders can modification facial expressions in photos
- Build sensible GAN examples from scratch, together with CycleGAN for taste switch and MuseGAN for track generation
- Create recurrent generative fashions for textual content era and learn to strengthen the fashions the usage of attention
- Understand how generative fashions can assist marketers to perform duties inside a reinforcement finding out setting
- Explore the structure of the Transformer (BERT, GPT-2) and symbol era fashions equivalent to ProGAN and StyleGAN
