Semi-Supervised Learning

In this lesson, Ian Goodfellow will lead you through implementing a semi-supervised GAN model in TensorFlow. Semi-supervised models are used when you only have a few labeled data points. These perform surprisingly well even though the amount of labeled data you have is relatively tiny.

Here we'll use the SVHN dataset again and attempt to classify the images with a large proportion of the labels dropped.

As a heads up, this material is cutting edge and is more difficult to understand and implement than what you've seen before. But, you'll learn a lot and see some really cool results.

Getting the notebooks

Ian will lead you through a set of notebooks, one where you can try implementing the network and another notebook with his solutions. The notebooks are available from our public GitHub repo, in the semi-supervised folder. Either download the notebooks from there, or do a git pull if you already have the repo.