| Class | Description |
|---|---|
| Gaussian |
This function may be better than LogCosh when the independent
components are highly super-Gaussian, or when robustness is very important.
|
| Kurtosis |
This function is justified on statistical grounds only for
estimating sub-Gaussian independent components when there are no outliers.
|
| LogCosh |
A good general-purpose function for ICA.
|