Chapter 6. Generative Networks
Generative networks are backed by a famous quotation of Richard Feynman, professor of undergraduate physics at Caltech Institute of Technology and a Nobel Prize winner: "What I cannot create, I cannot understand." Generative networks are one of the most promising approaches to having a system that can understand the world and store knowledge within it. As their name indicates, generative networks learn the pattern of the true data distribution and try to generate new samples that look like the samples from this true data distribution.
Generative models are a subcategory of unsupervised learning since they learn the underlying pattern by trying to generate samples. They do this by pushing the low-dimensional latent vector and parameter vector to learn the important features it requires to generate the image back. The knowledge that the network acquired while generating images is essentially knowledge about the system and the environment...
