FastIca.Whiten Property

public FastIcaWhiteningMode Whiten { get; set; }

Public Property Whiten As FastIcaWhiteningMode
	Get
	Set

public:
property FastIcaWhiteningMode Whiten {
	FastIcaWhiteningMode get ();
	void set (FastIcaWhiteningMode value);
}

member Whiten : FastIcaWhiteningMode with get, set

Property Value

Whitening is a pre-processing step that decorrelates the data and equalizes variance. Three modes are supported, matching scikit-learn's behavior:

• UnitVariance: After whitening, each component has unit variance. This is the default and matches scikit-learn's 'unit-variance' mode. The extracted sources will also have unit variance.
• ArbitraryVariance: Whitening is applied but no variance normalization is performed afterward. This matches scikit-learn's 'arbitrary-variance' mode (whiten=True in older versions).
• Skip: No whitening is applied. Use this only if the data is already whitened or if you want to supply your own whitening via InitialUnmixingMatrix. This matches scikit-learn's whiten=False mode.

Whitening is generally recommended as it improves convergence and numerical stability.

The default value is UnitVariance.

Reference