Fast Ica.Contrast Function Property
Gets or sets the contrast (nonlinearity) function used in the FastICA fixed-point iteration.
Definition
Namespace: Numerics.NET.Statistics.Multivariate
Assembly: Numerics.NET (in Numerics.NET.dll) Version: 9.1.5
C#
Assembly: Numerics.NET (in Numerics.NET.dll) Version: 9.1.5
public FastIcaContrastFunction ContrastFunction { get; set; }
Property Value
FastIcaContrastFunctionRemarks
The contrast function determines how non-Gaussianity is measured during component extraction:
- LogCosh: Robust, general-purpose function. Good for most data types including sub-Gaussian and super-Gaussian sources. This is the default and recommended choice for most applications.
- Exponential: Optimized for super-Gaussian (heavily-tailed) distributions. Use when sources are known to have high kurtosis or extreme outliers.
- Cubic: Simple polynomial function. Works well for sub-Gaussian (light-tailed) distributions. Fastest but least robust to outliers.
The LogCoshAlpha parameter controls the behavior of the LogCosh function.
The default value is LogCosh.