SOURCE DISTRIBUTION ADAPTIVE MAXIMUM LIKELIHOOD ESTIMATION OF ICA MODEL
Jan Eriksson , Juha Karvanen , and Visa Koivunen
fjan.eriksson,juha.karvanen,visa.koivuneng@hut.
In this paper a new approach for performing Indepen-
dent Component Analysis (ICA) is introduced. The
Extended Generalized Lambda Distribution (EGLD) is
employed for modeling source distributions. The major
bene t of the EGLD is that it also takes into account
the skewness of the distributions. We brie
y review
maximum likelihood approach in ICA and study how
the parameters of EGLD may be estimated. The score
function of EGLD based ICA is presented and algo-
rithms for its maximization are proposed. The sim-
ulation examples illustrate that the proposed method
reliably separates the sources in situations where some
widely used contrast functions may perform poorly.