GaussianMixtureModel<T>

Models & Types · AiDotNet.Clustering.Probabilistic

Gaussian Mixture Model clustering using Expectation-Maximization.

Inheritance: ClusteringBase<T> → GaussianMixtureModel<T>

Implements: IGeneratedParameterRegistrar<T>

For Beginners

GMM is like soft K-Means.

Instead of saying “this point belongs to cluster 2”, GMM says “this point has 70% chance of cluster 2, 25% chance of cluster 1, 5% chance of cluster 3”.

This is useful when:

  • Clusters overlap (points could belong to multiple groups)
  • Clusters have different shapes (some wide, some narrow)
  • You need uncertainty estimates for cluster assignments

The EM algorithm works by:

  • E-step: For each point, estimate probability of belonging to each cluster
  • M-step: Update cluster parameters based on these probabilities
  • Repeat until stable

How It Works

GMM represents data as a mixture of Gaussian distributions. Each cluster is characterized by its mean, covariance, and mixing weight. The EM algorithm iteratively refines these parameters.

Algorithm steps (EM):

  1. Initialize means, covariances, and weights
  2. E-step: Compute responsibility of each component for each point
  3. M-step: Update parameters to maximize expected log-likelihood
  4. Repeat until convergence

Example

using AiDotNet;
using AiDotNet.Data.Loaders;
using AiDotNet.Clustering.Probabilistic;
using AiDotNet.Tensors.LinearAlgebra;

var data = new Matrix<double>(6, 2);
double[][] rows = { new[] { 1.0, 1.0 }, new[] { 1.2, 0.9 }, new[] { 1.1, 1.1 },
                    new[] { 8.0, 8.0 }, new[] { 8.2, 7.9 }, new[] { 7.9, 8.1 } };
for (int i = 0; i < 6; i++) { data[i, 0] = rows[i][0]; data[i, 1] = rows[i][1]; }

var result = await new AiModelBuilder<double, Matrix<double>, Vector<double>>()
    .ConfigureModel(new GaussianMixtureModel<double>())
    .ConfigureDataLoader(DataLoaders.FromMatrix(data))
    .BuildAsync();

var labels = result.Predict(data);
Console.WriteLine($"GaussianMixtureModel: clustered {labels.Length} points.");

Constructors

Constructor Summary
GaussianMixtureModel(GMMOptions<T> options) Initializes a new GaussianMixtureModel instance.

Properties

Property Summary
Matrix<T> Means { get; } Gets the component means.
T LowerBound { get; } Gets the lower bound (ELBO) from the last training.
Tensor<T> Covariances { get; } Gets the component covariances.
Vector<T> Weights { get; } Gets the mixture weights.

Methods

Method Summary
IFullModel<T, Matrix<T>, Vector<T>> WithParameters(Vector<T> parameters)
Matrix<T> PredictProba(Matrix<T> x) Predicts probability of each component for each sample.
Matrix<T> Sample(int numSamples) Samples from the fitted mixture model.
ModelOptions GetOptions()
T Score(Matrix<T> x) Computes log-likelihood of data under the model.
Vector<T> FitPredict(Matrix<T> x)
Vector<T> Predict(Matrix<T> x)
void Train(Matrix<T> x, Vector<T> y)