Independent component analysis, a new concept?
Signal Processing - Special issue on higher order statistics
Adaptive Blind Signal and Image Processing: Learning Algorithms and Applications
Adaptive Blind Signal and Image Processing: Learning Algorithms and Applications
Learning Overcomplete Representations
Neural Computation
Natural Gradient Learning for Over-and Under-Complete Bases in ICA
Neural Computation
Fast and robust fixed-point algorithms for independent component analysis
IEEE Transactions on Neural Networks
Mining latent sources of causal time series using nonlinear state space modeling
ACIIDS'11 Proceedings of the Third international conference on Intelligent information and database systems - Volume Part I
Hi-index | 0.00 |
Separation of independent sources using independent component analysis (ICA) requires prior knowledge of the number of independent sources. Performing ICA when the number of recordings is greater than the number of sources can give erroneous results. To improve the quality of separation, the most suitable recordings have to be identified before performing ICA. Techniques employed to estimate suitable recordings require estimation of number of independent sources or require repeated iterations. However there is no objective measure of the number of independent sources in a given mixture. Here, a technique has been developed to determine the number of independent sources in a given mixture. This paper demonstrates that normalised determinant of the global matrix is a measure of the number of independent sources, N, in a mixture of M recordings. It has also been shown that performing ICA on N randomly selected recordings out of M recordings gives good quality of separation.