- 2017-07-13 15:00-17:00
- Dr. Zhao Yunbin
Dr Yunbin Zhao received the PhD degree from the Chinese Academy of Sciences (CAS) in 1998. He joined to the University of Birmingham in 2007. His research interests include the operations research, computational optimization, numerical analysis, numerical linear algebra, and their applications to sparse data reconstruction.
Seeking sparse or the sparsest point of a polyhedral set is a long-lasting challenging problem in the fields of sparse data representation, signal recovery, statistical learning, and mathematical optimization. The weighted/reweighted L1-algorithm is one of the efficient computational methods for locating a sparse point in polyhedrons. In this talk, a traditional framework of reweighted l1-methods is reviewed. Following that, a new design of reweighted l1-method from dual perspective will be proposed. This design will be based on the observation that seeking for a sparse point in a polyhedron can be reformulated as the equivalent problem of finding the corresponding densest slack variable of the dual problem of certain weighted L1-problems. This algorithmic development is remarkably different from existing sparsity-seeking methods. The efficiency of the algorithm has been demonstrated by empirical simulations.