About the talk
Computer Vision research achieved impressive progress in recent years, mostly thanks to large scale deep-learning techniques. Machines now perform on-par with humans for tasks such as image classification and object detection, which are almost considered as solved. These problems share the common characteristic that the computer is required to predict a single output or a set of uncorrelated values. However, many problems of practical interest necessitate to predict several variables at the same time while enforcing a specific structure. Pierre Baqué will explain why standard learning techniques fail in this context when not used carefully. Using intuitive examples, he will present several classes of methods used to solve it.
About the speaker
Pierre Baqué received an engineering degree in Applied Mathematics and a Masters degree in Operations Research from Ecole Polytechnique, in 2013. After working for Credit-Suisse in London, he joined the Computer Vision Laboratory at EPFL. Since 2014 he is pursuing his PhD under the supervision of Prof. Pascal Fua and François Fleuret. His research focuses on Structured Learning and Variational Inference applied to Computer Vision.
18:00 - Welcome
18:15 - Talk
19:00 - Beers & Snacks - Networking
** All talks are in English
**You want to present or suggest a speaker? Get in touch!
Looking forward to seeing you all! Cheers!