This episode of Fresh from the arXiv is going to be a little different. Normally I skim through all of the AI, computer vision and NLP preprints that came out during the week and pick a few that I consider particularly interesting. Often there is a common theme uniting a few of my choices, but the idea is not really to zoom in on any particular subject. Last week, however, I could not help but fall down a rabbit hole called semi-supervised learning with GANs. I ended up putting together a little introduction into the topic that is not too technical (meaning, it should be understandable to anyone with a vague idea of how a vanilla unsupervised GAN operates), but also provides a few directions to explore in more detail should you be interested.