LaplacianInferenceMethod.h

Go to the documentation of this file.

/*
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 3 of the License, or
 * (at your option) any later version.
 *
 * Copyright (C) 2012 Jacob Walker
 *
 * Code adapted from Gaussian Process Machine Learning Toolbox
 * http://www.gaussianprocess.org/gpml/code/matlab/doc/
 *
 */

#ifndef CLAPLACIANINFERENCEMETHOD_H_
#define CLAPLACIANINFERENCEMETHOD_H_

#include <shogun/lib/config.h>
#ifdef HAVE_EIGEN3
#ifdef HAVE_LAPACK

#include <shogun/regression/gp/InferenceMethod.h>

namespace shogun
{

class CLaplacianInferenceMethod: public CInferenceMethod
{

public:

    /*Default Constructor*/
    CLaplacianInferenceMethod();

    /* Constructor
     * @param kernel covariance function
     * @param features features to use in inference
     * @param mean mean function
     * @param labels labels of the features
     * @param model Likelihood model to use
]    */
    CLaplacianInferenceMethod(CKernel* kernel, CFeatures* features,
            CMeanFunction* mean, CLabels* labels, CLikelihoodModel* model);

    /*Destructor*/
    virtual ~CLaplacianInferenceMethod();

    virtual float64_t get_negative_marginal_likelihood();

    virtual CMap<TParameter*, SGVector<float64_t> >
    get_marginal_likelihood_derivatives(
            CMap<TParameter*, CSGObject*>& para_dict);

    virtual SGVector<float64_t> get_alpha();


    virtual SGMatrix<float64_t> get_cholesky();

    virtual SGVector<float64_t> get_diagonal_vector();

    virtual const char* get_name() const
    {
        return "LaplacianInferenceMethod";
    }

    /*Get the gradient
     *
     * @return Map of gradient. Keys are names of parameters, values are
     * values of derivative with respect to that parameter.
     */
    virtual CMap<TParameter*, SGVector<float64_t> > get_gradient(
            CMap<TParameter*, CSGObject*>& para_dict)
            {
        return get_marginal_likelihood_derivatives(para_dict);
            }

    /*Get the function value
     *
     * @return Vector that represents the function value
     */
    virtual SGVector<float64_t> get_quantity()
            {
        SGVector<float64_t> result(1);
        result[0] = get_negative_marginal_likelihood();
        return result;
            }

    /*Get tolerance for Newton Iterations
     *
     * @return Tolerance for Newton Iterations
     */
    virtual float64_t get_newton_tolerance() {
        return m_tolerance;}

    /*Set tolerance for Newton Iterations
     *
     * @param Tolerance for Newton Iterations
     */
    virtual void set_newton_tolerance(float64_t tol) {
        m_tolerance = tol;}

    /*Get tolerance for Brent's Minimization Method
     *
     * @return tolerance for Brent's Minimization Method
     */
    virtual float64_t get_minimization_tolerance() {
        return m_opt_tolerance;}

    /*Set tolerance for Brent's Minimization Method
     *
     * @param tolerance for Brent's Minimization Method
     */
    virtual void set_minimization_tolerance(float64_t tol) {
        m_opt_tolerance = tol;}

    /*Get max iterations for Brent's Minimization Method
     *
     * @return max iterations for Brent's Minimization Method
     */
    virtual int32_t get_minimization_iterations() {
        return m_max;}

    /*Set max iterations for Brent's Minimization Method
     *
     * @param max iterations for Brent's Minimization Method
     */
    virtual void set_minimization_tolerance(int32_t itr) {
        m_max = itr;}

    /*Get max Newton iterations
     *
     * @return max Newton iterations
     */
    virtual int32_t get_newton_iterations() {
        return m_max_itr;}

    /*Set max Newton iterations
     *
     * @param max Newton iterations
     */
    virtual void set_newton_tolerance(int32_t itr) {
        m_max_itr = itr;}



protected:
    virtual void update_alpha();
    virtual void update_chol();
    virtual void update_train_kernel();
    virtual void update_all();

private:

    void init();

private:

    void check_members();

    /*Amount of tolerance for Newton's Iterations*/
    float64_t m_tolerance;

    /*Amount of tolerance for Brent's Minimization Method*/
    float64_t m_opt_tolerance;

    /*Max iterations for Brent's Minimization Method*/
    float64_t m_max;

    /*Max Newton Iterations*/
    index_t m_max_itr;

    /*Kernel Matrix*/
    SGMatrix<float64_t> temp_kernel;

    /*Eigen version of alpha vector*/
    SGVector<float64_t> temp_alpha;

    /*Function Location*/
    SGVector<float64_t> function;

    /*Noise Matrix*/
    SGVector<float64_t> W;

    /*Square root of W*/
    SGVector<float64_t> sW;

    /*Eigen version of means vector*/
    SGVector<float64_t> m_means;

    /*Derivative of log likelihood with respect
     * to function location
     */
    SGVector<float64_t> dlp;

    /*Second derivative of log likelihood with respect
     * to function location
     */
    SGVector<float64_t> d2lp;

    /*Third derivative of log likelihood with respect
     * to function location
     */
    SGVector<float64_t> d3lp;

    /*log likelihood*/
    float64_t lp;

};

}
#endif // HAVE_EIGEN3
#endif // HAVE_LAPACK

#endif /* CLAPLACIANINFERENCEMETHOD_H_ */