| Paper: | MLSP-P1.10 | ||
| Session: | Blind Source Separation II | ||
| Time: | Tuesday, May 16, 14:00 - 16:00 | ||
| Presentation: | Poster | ||
| Topic: | Machine Learning for Signal Processing: Blind Signal Separation and Independent Component Analysis | ||
| Title: | Frequency Domain Blind Source Separation Exploiting Higher-Order Dependencies | ||
| Authors: | Taesu Kim, Korea Advanced Institute of Science and Technology, Republic of Korea; Hagai Attias, Golden Metallic, Inc., United States; Soo-Young Lee, Korea Advanced Institute of Science and Technology, Republic of Korea; Te-Won Lee, University of California, San Diego, United States | ||
| Abstract: | We propose a novel approach to the blind source separation (BSS) that exploits frequency dependencies within a source. In contrast to conventional algorithms that separate the sources independently in each frequency bin, we assume that dependencies exist between frequency bins in a source signal. In this manner, we can reduce or eliminate the well-known frequency permutation problem. We derive the learning algorithm by defining a cost function as an extension of mutual information between multivariate random variables and by introducing a source prior that models the inherent frequency dependencies. This results in a simple form of a multivariate score function. In simulations and real recording experiments, we evaluate the performance of the proposed method and compare it against other well-known algorithms under various conditions. Our results indicate that modeling dependencies yields improved performance and robust scaling to higher number of sources and mixtures. | ||
©2018 Conference Management Services, Inc. -||- email: webmaster@icassp2006.org -||- Last updated Friday, August 17, 2012