SPS, Thu 13/10/2011, TVNUM, 10h45:
Speaker: Prasad Sudhakar
Title: "The use of sparsity hypothesis for source separation"
Abstract: Sparsity has been traditionally exploited for data compression, and a popular hypothesis to solve under-determined linear inverse systems (sparse recovery problem). Considerable amount of work has been done, both theoretical and algorithmic, in this regard. Of late, sparsity is being used to perform more complicated tasks such as source separation, learning, etc. The focus of this talk will be the usage of sparsity hypothesis for source separation.
In Sparse Component Analysis (SCA), the separation problem is often formulated as a sparse recovery problem, and the objective is to seek sparse components from their mixtures, when the mixing parameters are known. The sparse component principle can be used even to estimate the mixing parameters blindly in certain simple settings. In a generic setting of convolutive mixtures, SCA throws up some interesting problems. Although several algorithms and improvements have been proposed for SCA, there is a lack of strong theoretical foundations. In this talk, an overview of the SCA principles will be given followed by the questions they raise in generic settings.
Last updated September 25, 2013, at 09:12 AM