The theory and practice of Generative Adversarial Networks (GANs) a talk by Jakub Langr
Friday, 14 June, 12:30 in Club
Until recently, generative modelling of any kind has had limited success. But now that Generative Adversarial Networks (GANs) have recently reached few tremendous milestones (and truly exponential growth in the interest in this technology), we are now closer to a general purpose framework for generating new data.
Now GANs can achieve a variety of applications such as synthesising full-HD synthetic faces, to semi-supervised learning as well as defending and mastering adversarial examples, we can discuss them in this talk. In this talk, we will start with the basics of generative models, but eventually we will explore the state of the art in generating full-HD images and dive into adversarial attacks and why this matters to all computer vision algorithms.
GANs are a novel approach to generating new data or on a variety of adjacent problems that leverages the power of deep learning and two competing agents to achieve breath-taking results.
Prerequisites: Barebones basics of deep learning (at least conceptually)
Jakub Langr
I graduated from the University of Oxford where I also taught at OU Computing Services.
I have worked in data science since 2013, most recently as a Data Science Tech Lead at Filtered.com and as an R&D Data Scientist at Mudano.
I am a co-author of GANs in Action by Manning Publications. I also designed and teach Data Science courses at the University of Birmingham, numerous private companies and I am a guest lecturer at the University of Oxford.
