CMAEvolutionStrategyOptimizer.UseActiveCMA Property

public bool UseActiveCMA { get; set; }

Public Property UseActiveCMA As Boolean
	Get
	Set

public:
property bool UseActiveCMA {
	bool get ();
	void set (bool value);
}

member UseActiveCMA : bool with get, set

Property Value

Active Covariance Matrix Adaptation (CMA) is a variant of the CMA-ES algorithm that includes negative updates to the covariance matrix. This can help to accelerate convergence by actively reducing the influence of less promising regions of the search space. When active CMA is enabled, the algorithm performs additional updates using the worst solutions in the population, which can lead to faster adaptation of the covariance matrix.

Use active CMA when you need faster convergence on problems where the search space has many local optima or when the standard CMA-ES algorithm is not performing well. However, be aware that active CMA can sometimes lead to premature convergence, especially in noisy optimization problems or when the population size is too small.

Reference