com.emphysic.myriad.core.data.roi

Class SVMROIFinder

java.lang.Object

com.emphysic.myriad.core.data.roi.SVMROIFinder

All Implemented Interfaces:

MLROIFinder, ROIFinder, ObjectMap, com.esotericsoftware.kryo.KryoSerializable

public class SVMROIFinder
extends java.lang.Object
implements MLROIFinder

SVMROIFinder - a Region Of Interest (ROI) finder based on an online Support Vector Machine (SVM). Created by ccoughlin on 5/28/17.

Constructor Summary

Constructors

Constructor and Description
SVMROIFinder() Default constructor.
SVMROIFinder(double C) Constructor for a linear kernel SVM.
SVMROIFinder(double Cp, double Cn) Constructor for a linear kernel SVM.
SVMROIFinder(smile.math.kernel.MercerKernel<double[]> kernel, double C) Constructor.
SVMROIFinder(smile.math.kernel.MercerKernel<double[]> kernel, double Cp, double Cn) Constructor.
SVMROIFinder(smile.classification.SVM model) Constructor

Method Summary

Modifier and Type	Method and Description
double	getC() Retrieves the soft penalty parameter.
double	getCn() Retrieves the soft penalty parameter for negative samples.
double	getCp() Retrieves the soft penalty parameter for positive samples.
int	getNumFeatures() Returns the number of features
java.util.Map<java.lang.String,java.lang.Object>	getObjectMap() Creates a map of the important fields for the instance, suitable for serialization.
long	getSerializationVersion() Returns the current version of the serialization format.
int	getVersion() Returns the current class version.
void	initCurrentVersion(java.util.Map<java.lang.String,java.lang.Object> objectMap) Initializes an instance with a current-version object graph.
boolean	isROI(Dataset dataset) Examine a dataset and return whether or not it seems to contain a region of interest (ROI)
boolean	isROI(double[] data) Examine an array of data and report whether it appears to contain a region of interest (ROI)
double	negativeClass() The numeric label assigned to negative samples i.e.
void	negativeClass(int negClass) Sets the negative (i.e.
double	positiveClass() The numeric label assigned to positive samples i.e.
void	positiveClass(int posClass) Sets the positive (i.e.
double	predict(Dataset data) Returns the predicted class/value of the sample
double	predict(double[] data) Returns the predicted class/value of the sample
void	train(double[][] X, int[] y) Trains the ROI finder on new data.

Methods inherited from class java.lang.Object

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

Methods inherited from interface com.emphysic.myriad.core.data.roi.ROIFinder

fromFile

Methods inherited from interface com.emphysic.myriad.core.data.util.ObjectMap

init, initPreviousVersion, initUnknownVersion, load, read, save, write

Constructor Detail

SVMROIFinder

public SVMROIFinder(smile.math.kernel.MercerKernel<double[]> kernel,
                    double Cp,
                    double Cn)

Constructor.

Parameters:

kernel - kernel function

Cp - soft margin penalty parameter for positive instances (0-1)

Cn - soft margin penalty parameter for negative instances (0-1)

SVMROIFinder

public SVMROIFinder(smile.math.kernel.MercerKernel<double[]> kernel,
                    double C)

Constructor.

Parameters:

kernel - kernel function

C - soft margin penalty parameter (0-1). The larger the penalty the more the model will try to avoid misclassification.

SVMROIFinder

public SVMROIFinder(double Cp,
                    double Cn)

Constructor for a linear kernel SVM.

Parameters:

Cp - soft margin penalty parameter for positive instances (0-1)

Cn - soft margin penalty parameter for negative instances (0-1)

SVMROIFinder

public SVMROIFinder(double C)

Constructor for a linear kernel SVM.

Parameters:

C - soft margin penalty parameter (0-1). The larger the penalty the more the model will try to avoid misclassification.

SVMROIFinder

public SVMROIFinder()

Default constructor. Creates a linear kernel SVM with a soft margin penalty parameter of 0.1.

SVMROIFinder

public SVMROIFinder(smile.classification.SVM model)

Constructor

Parameters:

model - SVM model

Method Detail

train

public void train(double[][] X,
                  int[] y)

Trains the ROI finder on new data. If the positive and negative labels have not been set and two unique values are found in y, the positive class is assumed to be the smaller of the two and the negative as the larger.

Specified by:

train in interface MLROIFinder

Parameters:

X - N examples with M features per example

y - N labels for the N examples in X

predict

public double predict(double[] data)

Description copied from interface: MLROIFinder

Returns the predicted class/value of the sample

Specified by:

predict in interface MLROIFinder

Parameters:

data - sample to predict

Returns:

predicted label / value

predict

public double predict(Dataset data)

Description copied from interface: MLROIFinder

Returns the predicted class/value of the sample

Specified by:

predict in interface MLROIFinder

Parameters:

data - sample to predict

Returns:

predicted label / value

positiveClass

public void positiveClass(int posClass)

Sets the positive (i.e. has ROI) label. If not set, will attempt to discover during training by finding the unique labels in the training set and taking the smallest value.

Parameters:

posClass - label of positive class

positiveClass

public double positiveClass()

Description copied from interface: MLROIFinder

The numeric label assigned to positive samples i.e. samples that contain ROI

Specified by:

positiveClass in interface MLROIFinder

Returns:

the value assigned to positive samples

negativeClass

public void negativeClass(int negClass)

Sets the negative (i.e. does not have ROI) label. If not set, will attempt to discover during training by finding the unique labels in the training set and taking the second smallest value.

Parameters:

negClass - label of negative class.

negativeClass

public double negativeClass()

Description copied from interface: MLROIFinder

The numeric label assigned to negative samples i.e. samples that do not contain ROI

Specified by:

negativeClass in interface MLROIFinder

Returns:

the value assigned to negative samples

getSerializationVersion

public long getSerializationVersion()

Description copied from interface: ObjectMap

Returns the current version of the serialization format. Increments when the serialization format changes, i.e. backwards compatibility is broken.

Specified by:

getSerializationVersion in interface ObjectMap

Returns:

serialization version

getVersion

public int getVersion()

Description copied from interface: ObjectMap

Returns the current class version. Class version increments when the class changes but doesn't necessarily break compatibility with serialization.

Specified by:

getVersion in interface ObjectMap

Returns:

class version

getObjectMap

public java.util.Map<java.lang.String,java.lang.Object> getObjectMap()

Description copied from interface: ObjectMap

Creates a map of the important fields for the instance, suitable for serialization.

Specified by:

getObjectMap in interface ObjectMap

Returns:

"object map" of the instance

initCurrentVersion

public void initCurrentVersion(java.util.Map<java.lang.String,java.lang.Object> objectMap)

Description copied from interface: ObjectMap

Initializes an instance with a current-version object graph.

Specified by:

initCurrentVersion in interface ObjectMap

Parameters:

objectMap - object graph for initialization

isROI

public boolean isROI(double[] data)

Description copied from interface: ROIFinder

Examine an array of data and report whether it appears to contain a region of interest (ROI)

Specified by:

isROI in interface ROIFinder

Parameters:

data - raw data to examine

Returns:

true if data appears to contain an ROI, false otherwise

isROI

public boolean isROI(Dataset dataset)

Description copied from interface: ROIFinder

Examine a dataset and return whether or not it seems to contain a region of interest (ROI)

Specified by:

isROI in interface ROIFinder

Parameters:

dataset - data to examine

Returns:

true if data appears to contain an ROI, false otherwise

getNumFeatures

public int getNumFeatures()

Returns the number of features

Returns:

number of features in the model's feature space

getCp

public double getCp()

Retrieves the soft penalty parameter for positive samples.

Returns:

soft penalty parameter, or 0 if not set.

getCn

public double getCn()

Retrieves the soft penalty parameter for negative samples.

Returns:

soft penalty parameter, or 0 if not set.

getC

public double getC()

Retrieves the soft penalty parameter.

Returns:

soft penalty parameter, or 0 if not set.