Generative adversarial networks and TF-GAN (ML Tech Talks)

In this session of Machine Learning Tech Talks, Research Engineer Joel Shor will discuss a very cool development and technique in machine learning called Generative Adversarial Networks (GANs) and a library that offers open source to help make training and evaluating GANs easier. Chapters: 0:00 - Introduction 1:29 - Demos from Google 11:42 - What is a GAN? 27:14 - What are GANs good for? 41:59 - Deep dive: Metrics 54:05 - Deep dive: Self-attention GAN 57:47 - How to get started Resources: Boundless video → https://goo.gle/3AfOjf0 GANSynth project page → https://goo.gle/3xcm8vr Batch equalization paper → https://goo.gle/2TgkudL Superresolution colab → https://goo.gle/3y5xbqi Image-to-image translation colab → https://goo.gle/3hmwYrZ CycleGAN colab → https://goo.gle/3dwKyrG Inception Score implementation → https://goo.gle/3ydL322 Frechet Inception Distance implementation → https://goo.gle/2UMTfYo Self-Attention GAN implementation → https://goo.gle/3627o6q TF-GAN examples → https://goo.gle/3w3xvEE Catch more ML Tech Talks → http://goo.gle/ml-tech-talks Subscribe to TensorFlow → https://goo.gle/TensorFlow