Class Prior

  • All Implemented Interfaces:
    java.io.Serializable, RevisionHandler
    Direct Known Subclasses:
    GaussianPriorImpl, LaplacePriorImpl

    public abstract class Prior
    extends java.lang.Object
    implements java.io.Serializable, RevisionHandler
    This is an interface to plug various priors into the Bayesian Logistic Regression Model.
    Version:
    $Revision: 1.2 $
    Author:
    Navendu Garg (gargnav@iit.edu)
    See Also:
    Serialized Form
    • Constructor Summary

      Constructors 
      Constructor Description
      Prior()  
    • Method Summary

      All Methods Instance Methods Concrete Methods 
      Modifier and Type Method Description
      void computelogLikelihood​(double[] betas, Instances instances)
      Function computes the log-likelihood value: -sum{1 to n}{ln(1+exp(-Beta*x(i)*y(i))}
      void computePenalty​(double[] betas, double[] hyperparameters)
      Skeleton function to compute penalty terms.
      double getLoglikelihood()  
      double getLogPosterior()  
      double getPenalty()  
      double update​(int j, Instances instances, double beta, double hyperparameter, double[] r, double deltaV)
      Interface for the update functions for different types of priors.
      • Methods inherited from class java.lang.Object

        equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
    • Constructor Detail

      • Prior

        public Prior()
    • Method Detail

      • update

        public double update​(int j,
                             Instances instances,
                             double beta,
                             double hyperparameter,
                             double[] r,
                             double deltaV)
        Interface for the update functions for different types of priors.
      • computelogLikelihood

        public void computelogLikelihood​(double[] betas,
                                         Instances instances)
        Function computes the log-likelihood value: -sum{1 to n}{ln(1+exp(-Beta*x(i)*y(i))}
        Parameters:
        betas -
        instances -
      • computePenalty

        public void computePenalty​(double[] betas,
                                   double[] hyperparameters)
        Skeleton function to compute penalty terms.
        Parameters:
        betas -
        hyperparameters -
      • getLoglikelihood

        public double getLoglikelihood()
        Returns:
        log-likelihood value.
      • getLogPosterior

        public double getLogPosterior()
        Returns:
        regularized log posterior value.
      • getPenalty

        public double getPenalty()
        Returns:
        penalty term.