Class RBFNet

JSAT/src/jsat/classifiers/neuralnetwork/RBFNet.java:77–888

This provides a highly configurable implementation of a Radial Basis Function Neural Network. A RBF network is a type of neural network that contains one hidden layer, and is related to the LVQ algorithm. In a classical RBF Network, the distance between two data points is generally the {@lin

 * @author Edward Raff
 */
public class RBFNet implements Classifier, Regressor, DataTransform, Parameterized
{

    private static final long serialVersionUID = 5418896646203518062L;
    private int numCentroids;
    private Phase1Learner p1l;
    private Phase2Learner p2l;
    private double alpha;
    private int p;
    private DistanceMetric dm;
    private boolean normalize = true;
    
    private Classifier baseClassifier;
    private Regressor baseRegressor;
    
    private List<Double> centroidDistCache;
    private List<Vec> centroids;
    private double[] bandwidths;

    /**
     * Creates a new RBF Network suitable for binary classification or
     * regression and uses 100 hidden nodes. One of the other constructors
     * should be used if you need classification for multi-class or if you need
     * probability outputs. <br>
     * <br>
     * This will use {@link Phase1Learner#K_MEANS} for neuron selection and
     * {@link Phase2Learner#NEAREST_OTHER_CENTROID_AVERAGE} for activation
     * tuning. The {@link EuclideanDistance} will be use as the metric.
     *
     */
    public RBFNet()
    {
        this(100);
    }
    
    /**
     * Creates a new RBF Network suitable for binary classification or 
     * regression. One of the other constructors should be used if you need 
     * classification for multi-class or if you need probability outputs. <br>
     * <br>
     * This will use {@link Phase1Learner#K_MEANS} for neuron selection and 
     * {@link Phase2Learner#NEAREST_OTHER_CENTROID_AVERAGE} for activation tuning. 
     * The {@link EuclideanDistance} will be use as the metric. 
     * 
     * @param numCentroids the number of centroids or neurons to use in the 
     * network's hidden layer
     */
    public RBFNet(int numCentroids)
    {
        this(numCentroids, Phase1Learner.K_MEANS, Phase2Learner.NEAREST_OTHER_CENTROID_AVERAGE, 3, 3, new EuclideanDistance(), (Classifier) new DCDs());
    }

    /**
     * Creates a new RBF Network for classification tasks. If the classifier can
     * also perform regression, then the network will be able to perform both.
     * 
     * @param numCentroids the number of centroids or neurons to use in the 
     * network's hidden layer