jsat.outlier

Class LinearOCSVM

java.lang.Object

jsat.outlier.LinearOCSVM

All Implemented Interfaces:

java.io.Serializable, Outlier

public class LinearOCSVM
extends java.lang.Object
implements Outlier

This class implements the One-Class SVM (OC-SVM) algorithm for outlier detection. This implementation works only in the primal or "linear" space. As such it works best when the data is sparse and high dimensional. If your data is dense and low dimensional, you may get better results by first applying a non-linear transformation to the data. See:

• Schölkopf, B., Williamson, R., Smola, A., Shawe-Taylor, J., & Platt, J. (1999). Support Vector Method for Novelty Detection. In Advances in Neural Information Processing Systems 12 (pp. 582–588). Denver, CO.
• Manevitz, L. M., & Yousef, M. (2001). One-class Svms for Document Classification. J. Mach. Learn. Res., 2, 139–154. Retrieved from http://dl.acm.org/citation.cfm?id=944790.944808

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
LinearOCSVM()

Method Summary

Modifier and Type	Method and Description
void	fit(DataSet d, boolean parallel)
double	getV()
double	score(DataPoint x) Returns an unbounded anomaly/outlier score.
void	setV(double v)

Methods inherited from class java.lang.Object

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

Methods inherited from interface jsat.outlier.Outlier

fit, isOutlier

Constructor Detail

LinearOCSVM

public LinearOCSVM()

Method Detail

setV

public void setV(double v)

getV

public double getV()

fit

public void fit(DataSet d,
                boolean parallel)

Specified by:

fit in interface Outlier

score

public double score(DataPoint x)

Description copied from interface: Outlier

Returns an unbounded anomaly/outlier score. Negative values indicate the input is likely to be an outlier, and positive values that the input is likely to be an inlier.

Specified by:

score in interface Outlier

Returns: