MulticlassSDCA

java.lang.Object
- fr.lip6.jkernelmachines.classifier.multiclass.MulticlassSDCA<T>

All Implemented Interfaces:

Classifier<T>, KernelSVM<T>, MulticlassClassifier<T>, java.lang.Cloneable
```
public class MulticlassSDCA<T>
extends java.lang.Object
implements MulticlassClassifier<T>, KernelSVM<T>
```
This is a straight forward extension of SDCA svm algorithm from Shalev-Shwartz to multiclass using a multiclass loss function.

Stochastic Dual Coordinate Ascent Methods for Regularized Loss Minimization,
Shai Shalev-Shwartz, Tong Zhang
JMLR, 2013.

Author:

picard

Constructor Summary

Constructors
Constructor and Description

MulticlassSDCA(Kernel<T> k)

Constructors
Constructor and Description
`MulticlassSDCA(Kernel<T> k)`

Method Summary

All Methods Instance Methods Concrete Methods
Modifier and Type	Method and Description
`Classifier<T>`	`copy()` Creates and returns a copy of this object.
`double[]`	`getAlphas()` Tells the weights of training samples
`double`	`getC()` Tells the hyperparameter C
`double`	`getConfidence(T t)` Tells the confidence associated with the predicted class
`java.util.Map<java.lang.Integer,java.lang.Double>`	`getConfidences(T t)` Tells the confidences for all classes
`double`	`getE()` return the number of epochs
`Kernel<T>`	`getKernel()` Tells the current Kernel.
`double[][]`	`getMulticlassAlphas()` Returns the matrix of dual variables in the order [sample, class]
`void`	`setC(double c)` Sets the hyperparameter C
`void`	`setE(double e)` Sets the number of epochs
`void`	`setKernel(Kernel<T> kernel)` Sets the kernel to use as similarity measure
`void`	`train(java.util.List<TrainingSample<T>> l)` Replace the current training list and train the classifier
`double`	`valueOf(T e)` Computes the category of the provided example

Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

- Constructor Detail
  - MulticlassSDCA
```
public MulticlassSDCA(Kernel<T> k)
```
- Method Detail
  - train
```
public void train(java.util.List<TrainingSample<T>> l)
```
    Description copied from interface: Classifier
    
    Replace the current training list and train the classifier
    
    Specified by:
    
    train in interface Classifier<T>
    
    Parameters:
    
    l - list of training samples
  - valueOf
```
public double valueOf(T e)
```
    Description copied from interface: Classifier
    
    Computes the category of the provided example
    
    Specified by:
    
    valueOf in interface Classifier<T>
    
    Parameters:
    
    e - example
    
    Returns:
    
    >0. if e belongs to the category, <0. if not.
  - copy
```
public Classifier<T> copy()
                   throws java.lang.CloneNotSupportedException
```
    Description copied from interface: Classifier
    
    Creates and returns a copy of this object.
    
    Specified by:
    
    copy in interface Classifier<T>
    
    Throws:
    
    java.lang.CloneNotSupportedException
    
    See Also:
    
    Object.clone()
  - getConfidence
```
public double getConfidence(T t)
```
    Description copied from interface: MulticlassClassifier
    
    Tells the confidence associated with the predicted class
    
    Specified by:
    
    getConfidence in interface MulticlassClassifier<T>
    
    Parameters:
    
    t - the sample to evaluate
    
    Returns:
    
    the confidence
  - getConfidences
```
public java.util.Map<java.lang.Integer,java.lang.Double> getConfidences(T t)
```
    Description copied from interface: MulticlassClassifier
    
    Tells the confidences for all classes
    
    Specified by:
    
    getConfidences in interface MulticlassClassifier<T>
    
    Parameters:
    
    t - the sample to evaluate
    
    Returns:
    
    the vector of confidences for all classes
  - getKernel
```
public Kernel<T> getKernel()
```
    Description copied from interface: KernelSVM
    
    Tells the current Kernel.
    
    Specified by:
    
    getKernel in interface KernelSVM<T>
    
    Returns:
    
    the current kernel
  - setKernel
```
public void setKernel(Kernel<T> kernel)
```
    Description copied from interface: KernelSVM
    
    Sets the kernel to use as similarity measure
    
    Specified by:
    
    setKernel in interface KernelSVM<T>
    
    Parameters:
    
    kernel - the kernel function
  - getC
```
public double getC()
```
    Description copied from interface: KernelSVM
    
    Tells the hyperparameter C
    
    Specified by:
    
    getC in interface KernelSVM<T>
    
    Returns:
    
    the hyperparameter C
  - setC
```
public void setC(double c)
```
    Description copied from interface: KernelSVM
    
    Sets the hyperparameter C
    
    Specified by:
    
    setC in interface KernelSVM<T>
    
    Parameters:
    
    c - the hyperparameter C
  - getE
```
public double getE()
```
    return the number of epochs
    
    Returns:
  - setE
```
public void setE(double e)
```
    Sets the number of epochs
    
    Parameters:
    
    e -
  - getAlphas
```
public double[] getAlphas()
```
    Description copied from interface: KernelSVM
    
    Tells the weights of training samples
    
    Specified by:
    
    getAlphas in interface KernelSVM<T>
    
    Returns:
    
    an array of double representing the weights
  - getMulticlassAlphas
```
public double[][] getMulticlassAlphas()
```
    Returns the matrix of dual variables in the order [sample, class]
    
    Returns:

Class MulticlassSDCA<T>

Constructor Summary

Method Summary

Methods inherited from class java.lang.Object

Constructor Detail

MulticlassSDCA

Method Detail

train

valueOf

copy

getConfidence

getConfidences

getKernel

setKernel

getC

setC

getE

setE

getAlphas

getMulticlassAlphas