26/01/2017
Abstract: The automatic interpretation of video sequences at higher level of abstraction poses some of the biggest challenges in computer science today. It requires processing large amounts of visual data as fast as possible while being able to learn the meaningful information as accurately as possible. It uses the most advanced high performance computers available today and the most advanced machine learning algorithms. The spatiotemporal coherence present in video also gives the possibility to conceive methods that learn in an unsupervised manner – another one of the important, still unsolved problems in artificial intelligence. By putting together many interesting challenges, and bringing together several fields in science and engineering, the problem of automatic video understanding can lead to the creation of novel technologies and also shed more light on our understanding of how the mind works. The brain does, in large part, vision, and that is what we also do, in our computer vision group. I will present some of the tasks we address, challenges that we face and solutions which we have found together with my students at the Institute of Mathematics of the Romanian Academy and University Politehnica of Bucharest. I will talk about several aspects of automatic video understanding, ranging from unsupervised learning to automatic translation of video content into language.
Bio: I am an Associate Professor at the University Politehnica of Bucharest and senior researcher at the Institute of Mathematics of the Romanian Academy. I am interested in the nature of intelligence, life and consciousness. In particular, my research focuses on computer vision, machine learning and robotics. At the university I teach the graduate level computer vision and robotics classes.
I have received a Ph.D. in Robotics from Carnegie Mellon University in 2009 and Bachelor degrees in Mathematics and Computer Science from the City University of New York, in 2003. My research has made contributions to learning and optimization for graph matching and probabilistic graphical models, object recognition and tracking, 3D modeling of urban scenes, boundary detection, optical flow, activity recognition, feature selection, object discovery and classification in video. In 2014 the Romanian Academy awarded me the “Grigore Moisil” Prize in Mathematics for my work on unsupervised learning for graph matching.