Source separation using higher order moments

By Jean-François Cardoso. 1989

Subfields: Statistics and Probability, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistics and Probability

In Signal Processing

Citations in the corpus

Beyond CCA: Moment Matching for Multi-View Models (2016, positive): “As in Podosinnikova et al. (2015), we use as evaluation metric the normalized (cid:96) 1 - error between a recovered matrix (cid:98) D and the true matrix D with the best permutation of columns err 1 ( (cid:98) D, D ) := min π ∈ PERM The minimization is over the possible permutations π ∈ PERM of the”
Rethinking LDA: Moment Matching for Discrete ICA (2015, neutral): “This procedure was referred to as the spectral algorithm for LDA [3] and the fourth-order3 blind identification algorithm for ICA [17, 18].”
Eigen-structure of the fourth-order cumulant tensor with application to the blind source separation problem (1990, methodology): “Simulations of high-resolution separation in the noisy case can be found in [ 9 ].”
Source separation in post-nonlinear mixtures (1999, background): “The source separation problem has been intensively studied over the last 12 years for linear instantaneous (memoryless) mixtures [2], [5], [12], [16], [18], [20], [22], [25], [29] and, more recently, since 1990, for linear convolutive mixtures [13], [15], [19], [24], [36], [37], [42].”
Invariant co‐ordinate selection (2009, methodology): “Theorem 6 represents a generalization of an early ICA algorithm that was proposed by Cardoso (1989) based on the spectral value decomposition of a kurtosis matrix.”
AMUSE: a new blind identification algorithm (1990, methodology): “Herault-Jutten [5] and Cardoso [ 6 ] have dcvclopcd algorithms that can be applied to general structure of the paramctcr matrix.”
Natural image statistics and neural representation. (2001, background): “Over the past decade, a number of researchers have developed techniques for estimating this final rotation matrix (e.g. Cardoso 1989, Jutten & Herauult 1991, Comon 1994).”
Probabilistic independent component analysis for functional magnetic resonance imaging (2004, background): “Theoretical work on highorder moments provided one of the first solutions to a BSS problem [9].”
Sign and rank covariance matrices (2000, methodology): “The separation is achieved by finding statistically independent components from these linear mixtures, see Cardoso (1989) , Oja, E. (1997).”
Mathematics in signal processing V (2003, methodology): “Some less efficientearly methods were based on contracted versions of the array [ 6 ] (Cardoso 1989) or on noiseless observations[14] (Comon 1989).”
Independent Component Analysis via Distance Covariance (2013, background): “Information theory (see Hyv¨arinen and Oja 1997), the maximum likelihood principal (see Hastie and Tibshirani 2003), generalized decorrelation (see Cardoso 1989; Bach and Jordan 2003), and characteristic functions (see Eriksson and Koivunen 2003; Chen and Bickel 2005) are four broad methods for ICA…”
Non-orthogonal joint diagonalization in the least-squares sense with application in blind source separation (2002, methodology): “" Hermitian " problems usually occur when the target matrices are empirical correlation matrices , such as in [7], or cumulants matrices such as in [1], [3].”