■R2017a and later versions:
Although it is not the Independent Component Analysis (ICA) itself, Statistics and Machine Learning Toolbox has the function rica , which is the derivation of the ICA called Reconstruction ICA (RICA). This algorithm is based on ICA, but with a reconstruction cost. Here, the cost means a penalty or regularization term.
By replacing the normal orthonormal constraints of the ICA with the smoother cost term, this algorithm can improve the convergence speed of computation and enables feature extraction with not local but global orthogonality. The details of the algorithms of ICA with the reconstruction cost, please refer to the following documentation:
- Reconstruction ICA algorithm
https://www.mathworks.com/help/stats/feature-extraction.html■R2016b or earlier versions:
The ability to perform Independent Component Analysis is not available in MATLAB.
Please refer the following URL of MATLAB Central:
Note that The MathWorks does not guarantee or warrant the use or content of these submissions. Any questions, issues, or complaints should be directed to the contributing author.
・The FastICA package for MATLAB
http://research.ics.aalto.fi/ica/fastica/・NLPCAMATLAB
http://www.nlpca.org/ica_independent_component_analysis.html・ICALAB Toolboxes
https://web.archive.org/web/20191229160146/https://bsp.teithe.gr/members/downloads/ICALABSignal.htmlhttps://web.archive.org/web/20191229160146/https://bsp.teithe.gr/members/downloads/ICALABImage.html・PCA and ICA Package
https://www.mathworks.com/matlabcentral/fileexchange/38300-pca-and-ica-package
