A Jacobi-Like Algorithm for the General Joint Diagonalization Problem with Its Application to Blind Source Separation

The general problem of the approximate joint diagonalization of non-Hermitian matrices is considered. This problem mainly arises in the data model of the joint blind separation for two datasets. Based on a special parameterization of the two diagonalizing matrices and adapted approximations of the classical cost function, we establish a Jacobi-like algorithm. It may serve for the canonical polyadic decomposition (CPD) of a third-order tensor, and in some scenarios they can outperform traditional CPD methods. Simulation results show the competitive performance of the proposed algorithm.

[1]  Tülay Adali,et al.  Complex-valued independent vector analysis: Application to multivariate Gaussian model , 2012, Signal Process..

[2]  Andreas Ziehe,et al.  A Fast Algorithm for Joint Diagonalization with Non-orthogonal Transformations and its Application to Blind Source Separation , 2004, J. Mach. Learn. Res..

[3]  Eric Moreau,et al.  A generalization of joint-diagonalization criteria for source separation , 2001, IEEE Trans. Signal Process..

[4]  Eric Moreau,et al.  New Jacobi-like algorithms for non-orthogonal joint diagonalization of Hermitian matrices , 2016, Signal Process..

[5]  Bijan Afsari,et al.  Some Gradient Based Joint Diagonalization Methods for ICA , 2004, ICA.

[6]  Tülay Adali,et al.  Joint blind source separation by generalized joint diagonalization of cumulant matrices , 2011, Signal Process..

[7]  Eric Moreau,et al.  A Coordinate Descent Algorithm for Complex Joint Diagonalization Under Hermitian and Transpose Congruences , 2014, IEEE Transactions on Signal Processing.

[8]  Andrzej Cichocki,et al.  Non-orthogonal tensor diagonalization , 2014, Signal Process..

[9]  Xun Chen,et al.  Underdetermined Joint Blind Source Separation for Two Datasets Based on Tensor Decomposition , 2016, IEEE Signal Processing Letters.

[10]  Florian Roemer,et al.  A semi-algebraic framework for approximate CP decompositions via simultaneous matrix diagonalizations (SECSI) , 2013, Signal Process..

[11]  Vince D. Calhoun,et al.  Joint Blind Source Separation by Multiset Canonical Correlation Analysis , 2009, IEEE Transactions on Signal Processing.

[12]  De-Shuang Huang,et al.  Graphical Representation for DNA Sequences via Joint Diagonalization of Matrix Pencil , 2013, IEEE Journal of Biomedical and Health Informatics.