February Fourier Talks 2012

Collaboration and structure provide a natural way to incorporate signal and task priors into the problems at hand. In this talk I will describe how collaboration in signal modeling and processing is critical and can render very unstable and difficult problems into stable and much simpler ones.  I will describe a number of such models, the principles underlying them, theoretical results, and numerous applications, including activity recognition in video, image and audio restoration, new sensing paradigms, HIV/viruses analysis, multi-modality collaboration in classification tasks, and prediction in recommender systems with social network collaboration.