Comparison between principal component analysis (PCA) and independent component analysis (ICA). The coordinates of each dot are their values at a single time point. PCA is a popular unsupervised learning technique that picks out a direction that bisects the two signals, maximally mixing them, and the PCA axes are always perpendicular to each other. ICA finds the axes that fall along the directions of the dots, representing the separated signals, which may not be perpendicular.
No comments:
Post a Comment