Package elki.outlier.svm

Class LibSVMOneClassOutlierDetection<V extends NumberVector>

java.lang.Object

elki.outlier.svm.LibSVMOneClassOutlierDetection<V>

Type Parameters:

V - vector type

All Implemented Interfaces:

Algorithm, OutlierAlgorithm

@Reference(authors="B. Sch\u00f6lkopf, J. C. Platt, J. Shawe-Taylor, A. J. Smola, R. C. Williamson",
           title="Estimating the support of a high-dimensional distribution",
           booktitle="Neural computation 13.7",
           url="https://doi.org/10.1162/089976601750264965",
           bibkey="DBLP:journals/neco/ScholkopfPSSW01")
public class LibSVMOneClassOutlierDetection<V extends NumberVector>
extends java.lang.Object
implements OutlierAlgorithm

Outlier-detection using one-class support vector machines.

Important note: from literature, the one-class SVM is trained as if 0 was the only counterexample. Outliers will only be detected when they are close to the origin in kernel space! In our experience, results from this method are rather mixed, in particular as you would likely need to tune hyperparameters. Results may be better if you have a training data set with positive examples only, then apply it only to new data (which is currently not supported in this implementation, it assumes a single-dataset scenario).

Reference:

B. Schölkopf, J. C. Platt, J. Shawe-Taylor, A. J. Smola, R. C. Williamson
Estimating the support of a high-dimensional distribution
Neural computation 13.7

Since:

0.7.0

Author:

Erich Schubert

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	LibSVMOneClassOutlierDetection.SVMKernel	Kernel functions.

Nested classes/interfaces inherited from interface elki.Algorithm

Algorithm.Utils

Field Summary

Fields

Modifier and Type	Field	Description
(package private) double	gamma	Gamma parameter (not for linear kernel)
protected LibSVMOneClassOutlierDetection.SVMKernel	kernel	Kernel function in use.
private static Logging	LOG	Class logger.
(package private) static libsvm.svm_print_interface	LOG_HELPER	Setup logging helper for SVM.
(package private) double	nu	Nu parameter.

Constructor Summary

Constructors

Constructor	Description
LibSVMOneClassOutlierDetection(LibSVMOneClassOutlierDetection.SVMKernel kernel, double nu, double gamma)	Constructor.

Method Summary

Modifier and Type	Method	Description
TypeInformation[]	getInputTypeRestriction()	Get the input type restriction used for negotiating the data query.
OutlierResult	run(Relation<V> relation)	Run one-class SVM.

Methods inherited from class java.lang.Object

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

Methods inherited from interface elki.outlier.OutlierAlgorithm

autorun

Field Detail

LOG

private static final Logging LOG

Class logger.

kernel

protected LibSVMOneClassOutlierDetection.SVMKernel kernel

Kernel function in use.

nu

double nu

Nu parameter.

gamma

double gamma

Gamma parameter (not for linear kernel)

LOG_HELPER

static final libsvm.svm_print_interface LOG_HELPER

Setup logging helper for SVM.

Constructor Detail

LibSVMOneClassOutlierDetection

public LibSVMOneClassOutlierDetection(LibSVMOneClassOutlierDetection.SVMKernel kernel,
                                      double nu,
                                      double gamma)

Constructor.

Parameters:

kernel - Kernel to use with SVM.

nu - Nu parameter

gamma - Gamma parameter

Method Detail

getInputTypeRestriction

public TypeInformation[] getInputTypeRestriction()

Description copied from interface: Algorithm

Get the input type restriction used for negotiating the data query.

Specified by:

getInputTypeRestriction in interface Algorithm

Returns:

Type restriction

run

public OutlierResult run(Relation<V> relation)

Run one-class SVM.

Parameters:

relation - Data relation

Returns:

Outlier result.