it.uniroma2.sag.kelp.learningalgorithm.classification.libsvm.solver

Class LibCSvmSolver

java.lang.Object

it.uniroma2.sag.kelp.learningalgorithm.classification.libsvm.solver.LibSvmSolver

it.uniroma2.sag.kelp.learningalgorithm.classification.libsvm.solver.LibCSvmSolver

All Implemented Interfaces:

BinaryLearningAlgorithm, LearningAlgorithm

Direct Known Subclasses:

BinaryCSvmClassification, EpsilonSvmRegression

public abstract class LibCSvmSolver
extends LibSvmSolver

This class implements the solver of the C-SVM quadratic problem described in [CC Chang & CJ Lin, 2011]. It is a Java porting of the library LIBSVM v3.17, written in C++.

Further details can be found in:

[CC Chang & CJ Lin, 2011] Chih-Chung Chang and Chih-Jen Lin. LIBSVM: A library for support vector machines. ACM Transactions on Intelligent Systems and Technology, 2:27:1-27:27, 2011.

and

http://www.csie.ntu.edu.tw/~cjlin/libsvm/

Author:

Danilo Croce

Nested Class Summary

Nested classes/interfaces inherited from class it.uniroma2.sag.kelp.learningalgorithm.classification.libsvm.solver.LibSvmSolver

LibSvmSolver.Pair

Field Summary

Fields inherited from class it.uniroma2.sag.kelp.learningalgorithm.classification.libsvm.solver.LibSvmSolver

active_set, active_size, alpha, alpha_status, cn, cp, doShrinking, eps, examples, G, G_bar, kernel, l, label, logIteration, p, QD, shrinkingIteration, TAU, unshrink, y

Constructor Summary

Constructors

Constructor and Description
LibCSvmSolver()
LibCSvmSolver(Kernel kernel, float cp, float cn)

Method Summary

Methods

Modifier and Type	Method and Description
protected boolean	be_shrunk(int i, float Gmax1, float Gmax2)
protected float	calculate_rho()
protected void	do_shrinking() Apply the shrinking step
protected float[]	getCSvmAlpha(Dataset trainingSet) Get the initial weight for the future Support Vectors
protected int	select_working_set(LibSvmSolver.Pair pair) Select the working set in each iteration.

Methods inherited from class it.uniroma2.sag.kelp.learningalgorithm.classification.libsvm.solver.LibSvmSolver

get_QD, get_Qij, getCn, getCp, getEps, getKernel, getLabel, getLabels, info, is_free, is_lower_bound, is_upper_bound, kernel, reconstruct_gradient, setC, setCn, setCp, setEps, setLabel, setLabels, solve, swap_index, swap, swap, swap, swap

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface it.uniroma2.sag.kelp.learningalgorithm.LearningAlgorithm

duplicate, getPredictionFunction, learn, reset, setPredictionFunction

Constructor Detail

LibCSvmSolver

public LibCSvmSolver(Kernel kernel,
             float cp,
             float cn)

Parameters:

kernel - The kernel function

cp - The regularization parameter for positive examples

cn - The regularization parameter for negative examples

LibCSvmSolver

public LibCSvmSolver()

Method Detail

select_working_set

protected int select_working_set(LibSvmSolver.Pair pair)

Description copied from class: LibSvmSolver

Select the working set in each iteration. This function returns the pair i,j such that
i: maximizes -y_i * grad(f)_i, i in I_up(\alpha)
j: minimizes the decrease of obj value
(if quadratic coefficeint <= 0, replace it with tau)
-y_j*grad(f)_j < -y_i*grad(f)_i, j in I_low(\alpha)

Specified by:

select_working_set in class LibSvmSolver

Parameters:

pair - The Q_ij to be evaluated

Returns:

1 if already optimal, return 0 otherwise

be_shrunk

protected boolean be_shrunk(int i,
                float Gmax1,
                float Gmax2)

calculate_rho

protected float calculate_rho()

Specified by:

calculate_rho in class LibSvmSolver

do_shrinking

protected void do_shrinking()

Description copied from class: LibSvmSolver

Apply the shrinking step

Specified by:

do_shrinking in class LibSvmSolver

getCSvmAlpha

protected float[] getCSvmAlpha(Dataset trainingSet)

Get the initial weight for the future Support Vectors

Parameters:

trainingSet - the training set

Returns: