Blind source separation via the second characteristic function

By Arie Yeredor. 2000

112 citations recorded by the source corpus.

Subfields: Analytical Chemistry, 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 Analytical Chemistry

1. A. Hero
2. Koby Todros
3. Simon Lacoste-Julien
4. Anastasia Podosinnikova
5. Francis R. Bach
6. Alon Slapak
7. Shun-ichi Amari
8. Andrzej S. Cichocki

In Statistics and Probability

1. Simon Lacoste-Julien
2. Andrzej S. Cichocki
3. A. Hero
4. Koby Todros
5. Alon Slapak
6. Masashi Sugiyama
7. Francis R. Bach
8. Anastasia Podosinnikova

In Signal Processing

1. Andrzej S. Cichocki
2. Simon Lacoste-Julien
3. Francis R. Bach
4. Anastasia Podosinnikova
5. Shun-ichi Amari
6. A. Hero
7. Koby Todros
8. Anh Huy Phan

Citations in the corpus

• Beyond CCA: Moment Matching for Multi-View Models (2016, positive): “Following Yeredor (2000); Slapak & Yeredor (2012b), we use the most direct way of defining the finite sample estimators of the generalized expectation (20) and covariance matrix (21).”
• Tensor Networks for Dimensionality Reduction and Large-scale Optimization: Part 2 Applications and Future Perspectives (2017, background): “To illustrate the efficiency of higher-order derivatives of the second GCF in blind identification we considered a system of two mixtures, I = 2, linearly composed by R signals of length T = 100ˆ 2R, the entries of which can take the values 1 or ´1, i.e., sr,t = 1 or ´1.”
• Charrelation and Charm: Generic Statistics Incorporating Higher-Order Information (2012, methodology): “Four years later, in 2000, Yeredor independently proposed the use of only the 2-order (analytic) derivative of the Log-GCF at τ 6= 0, termed τ as a processing-point, and the idea was abstracted to Hessians of the Log-GCF at off-origin processing-points [6], which were later on re-termed as charrelat”
• Measure Transformed Independent Component Analysis (2013, methodology): “When we use an exponential MT-function the corresponding measure-transformed covariance matrix of the observation vector is equal to the Hessian of the cumulant-generating-function, resulting in the ICA method proposed in (Yeredor, 2000), which we call here exponential-MTICA.”
• Adaptive blind signal and image processing (2002, background): “In order to improve performance and/or to extend the JAD approach to various kinds of source signals several extensions and generalizations have been proposed [1222, 1334, 1336, 1337].”