DifferentialEvolutionOptimizer.WeightDithering Property

public DifferentialWeightDithering WeightDithering { get; set; }

Public Property WeightDithering As DifferentialWeightDithering
	Get
	Set

public:
property DifferentialWeightDithering WeightDithering {
	DifferentialWeightDithering get ();
	void set (DifferentialWeightDithering value);
}

member WeightDithering : DifferentialWeightDithering with get, set

Property Value

Dithering randomly varies the differential weight during optimization to potentially improve convergence. The default is None (no dithering).

Different dithering strategies can have varying effects on the optimization process. Uniform dithering adds a random value within a specified range to the differential weight, which can help in escaping local optima by introducing more variability. Jitter dithering applies a small percentage variation around the base value, providing a balance between stability and exploration. Range dithering varies the weight within a range around the base value, offering a more aggressive exploration strategy.

The optimal dithering strategy depends on the nature of the optimization problem. For problems with many local optima, more aggressive strategies like Range dithering may be beneficial. For smoother landscapes, Jitter or Uniform dithering might suffice. It is often useful to experiment with different strategies to find the most effective one for a given problem.

The DitheringRange property allows you to further configure the dithering strategy.

The default value is None.

Reference