A modified underdetermined blind source separation algorithm using competitive learning

The problern of underdetermined blind source sepamtion is addressed. A n adnanced classification method based upon competitive leainin,g is proposed for automatically determining the number of active sources over the obseruatior~. Its ihtroduction in underdetermined blind source separation successfully overcomes th,e drawbock of an existing method, in which the goal of sepamtiny more sources than the number of available mixtures is achiezled by exploiting the sparsity of the non-stationaq sources in the time-frequency domain. Sim,ulation studies are presented to support the proposed approach.