Independent component analysis (ICA) is similar to PCA in terms of dimensionality reduction. However, it originated from the signal processing world wherein they had this problem that multiple signals were being transmitted from a number of sources, and there were a number of devices set up to capture it. However, the problem was that the captured signal by the device was not very clear as it happened to be a mix of a number of sources. They needed to have clear and independent reception for the different devices that gave birth to ICA. Heralt and Jutten came up with this in.
The difference between PCA and ICA is that PCA focuses upon finding uncorrelated factors, whereas ICA is all about deriving independent factors. Confused? Let me help you. Uncorrelated factors imply that there is no linear relationship between them, whereas independence means that two factors have got no bearing on each other. For example, scoring good marks in mathematics is independent...