Summary
In this chapter, we discussed GANs. A GAN typically consists of two networks; one is trained to forge synthetic data that looks authentic, and the second is trained to discriminate authentic data against forged data. The two networks continuously compete, and in doing so, they keep improving each other. We reviewed an open source code, learning to forge MNIST and CIFAR-10 images that look authentic. In addition, we discussed WaveNet, a deep generative network proposed by Google DeepMind for teaching computers how to reproduce human voices and musical instruments with impressive quality. WaveNet directly generates raw audio with a parametric text-to-speech approach based on dilated convolutional networks. Dilated convolutional networks ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access