Future of graphical models: more modeling power, parallelization, scalable solvers. (Talk)
We propose a new computational framework for combinatorial problems arising in machine learning and computer vision. This framework is a special case of Lagrangean (dual) decomposition, but allows for efficient dual ascent (message passing) optimization. In a sense, one can understand both the framework and the optimization technique as a generalization of those for standard undirected graphical models (conditional random fields). We will make an overview of our recent results and plans for the nearest future.
Biography: Bogdan Savchynskyy is a senior researcher in TU Dresden. His main research interests are optimization problems in computer vision and machine learning. In particular, he is an author of a number of papers on exact and approximate inference for discrete graphical models. One of his recent works in this field has got an award at CVPR 14 conference.