GMMOptions<T>

Options & Configuration · AiDotNet.Clustering.Options

Configuration options for Gaussian Mixture Model clustering.

Inheritance: ModelOptions → ClusteringOptions<T> → GMMOptions<T>

For Beginners

GMM is like finding overlapping groups in data.

Imagine drops of different colored paint on paper:

  • K-Means: Draws hard boundaries between colors
  • GMM: Allows colors to blend, giving probability of belonging to each

Key features:

  • Soft clustering: Each point has probabilities for all clusters
  • Captures different cluster shapes and sizes
  • Based on statistical modeling

When to use GMM:

  • Clusters have different sizes or shapes
  • You need probability of cluster membership
  • Data might have overlapping clusters

How It Works

GMM assumes data is generated from a mixture of several Gaussian distributions. It finds the parameters of these distributions using the Expectation-Maximization (EM) algorithm.

Constructors

Constructor Summary
GMMOptions() Initializes a new instance of GMMOptions with GMM-appropriate defaults.

Properties

Property Summary
CovarianceType CovarianceType { get; set; } Gets or sets the type of covariance parameters to use.
GMMInitMethod InitMethod { get; set; } Gets or sets the initialization method.
bool AllowLowWeights { get; set; } Gets or sets whether to allow components with very low weights.
bool ComputeLowerBound { get; set; } Gets or sets whether to compute the lower bound during training.
double MinWeight { get; set; } Gets or sets the minimum weight threshold for components.
double RegularizationCovariance { get; set; } Gets or sets the regularization added to the diagonal of covariance.
double? WeightConcentrationPrior { get; set; } Gets or sets the weight concentration prior (for Dirichlet process).
int NumComponents { get; set; } Gets or sets the number of mixture components.