Abstract:Independent component analysis(ICA) is a typical and important blind source separation(BSS) method.It is to recover source signals which are considered statistically independent given only the outputs of a number of sensor mixed signals.How to choose a model of the score functions of the unknown source signals is a basic and important problem in most ICA algorithm.This is usually achieved based on the parametric and nonparametric density estimation approaches.Classical ICA algorithms fit parametric models for the hidden sources.Moreover,most ICA algorithms always assume that the distribution of source signals has a priori knowledge,but this is difficult to obtain in many situations.It has been realized that the unknown distributions of hidden sources can be estimated by nonparametric methods,some authors have discussed the nonparametric ICA method,but the kernel they used was fixed which made the algorithm does not robust in some conditions.How to develop unified nonparametric ICA algorithms,which enable to separate hybrid source signals including symmetric and asymmetric distributions with self-adaptive score functions,is very important.In this paper,efficient ICA algorithm based on generalized Gaussian density(GGD) kernel is proposed using a linear ICA neural network which can separate hybrid mixture of source signals which include both symmetric and asymmetric distribution sources.The proposed ICA algorithm is able to separate the hybrid mixtures of source signals using only a flexible model.Moreover,the GGD kernel can adaptively adjust its shaping and scaling parameter according to the high order statistics of the source signals.It paves the way to wider applications of ICA methods to real world signal processing.Simulations confirm the effectiveness of the proposed algorithm.

Supervised-unsupervised Combined Neural Learning for Independent Component Analysis

ICA Based Multi-Component Study of Remote Sensing Image

A novel compound neural network for fault diagnosis

Connecting geometric independent component analysis to unsupervised learning algorithms

Combination Of Independent Component Analysis And Multi-Way Principal Component Analysis For Batch Process Monitoring

Two-stage series-based neural network approach to nonlinear independent component analysis

A Hybrid Method for Noise Cancellation Based on Independent Component Analysis and Neural Networks

Hyperspectral Data Classification by Independent Component Analysis and Neural Network

Efficient independent component analysis algorithm using linear neural network

Nonlinear independent component analysis for principled disentanglement in unsupervised deep learning

Hebbian And Anti-Hebbian Learning For Independent Component Analysis

Method of Fault Diagnosis Based on ICA and Probabilistic Neural Network

Performance-Driven Ensemble Learning Ica Model For Improved Non-Gaussian Process Monitoring

Adaptive Independent Component Analysis under Multisensing

Classification of Hyperspectral Data Based on Improved Independent Component Analysis

Mutual Information Based Approach for Nonnegative Independent Component Analysis

Method for Unsupervised Remote Imagery Classification Based on Independent Component Analysis

A Neural Learning Algorithm of Blind Separation of Noisy Mixed Images Based on Independent Component Analysis

A supervised independent component analysis algorithm for motion imagery-based brain computer interface

Biologically plausible single-layer networks for nonnegative independent component analysis

Ica-Based Neural Network Approach To Classification Of Hyperspectral Image