Seth loves teaching and learning cutting edge machine learning concepts, applying them to solve companies' problems, and teaching others to do the same. Seth discovered Data Science and machine learning while working in consulting in early 2014. After taking virtually every course Udacity and Coursera had to offer on Data Science, he joined Trunk Club as their first Data Scientist in December 2015. There, he worked on lead scoring, recommenders, and other projects, before joining Metis in April 2017 as a Senior Data Scientist, teaching the Chicago full-time course. Over the past six months, he has developed a passion for neural nets and deep learning, working on writing a neural net library from scratch and sharing what he has learned with others via blog posts (on sethweidman.com), as well as speaking at Meetups and conferences.
Generative adversarial networks (GANs) are widely considered one of the most interesting developments in machine learning and AI in the last decade. In this wide-ranging talk, we'll start by covering the fundamentals of how and why they work, reviewing basic neural network and deep learning terminology in the process; we’ll then cover the latest applications of GANs, from generating art from drawings to advancing research areas such as Semi-Supervised Learning, and even generating audio. We’ll also examine the progress on improving GANs themselves, showing the tricks researchers have used to increase the realism of the images GANs generate.
Throughout, we’ll touch on many related topics, such as different ways of scoring GANs, and many of the Deep Learning-related tricks that have been found to improve training. Finally, we’ll close with some speculation from the leading minds in the field on where we are most likely to see GANs applied next.
Attendees will leave with a better understanding of the latest developments in this exciting area and the technical innovations that made those developments possible. Emphasis will be placed throughout on illuminating why the latest achievements have worked, not just what they are. Furthermore, a link to a clean, documented GitHub repo with a working GAN will be provided for attendees to see how to code one up. Attendees will thus leave feeling more confident and empowered to apply these same tricks to solve problems they face in personal projects or at work.