See: Description
Interface | Description |
---|---|
Classifier<T> |
Classifier interface that provides training and evaluation methods.
|
KernelSVM<T> |
Interface for SVM algorithms using a non-linear kernel (dual optimization mainly)
|
MKL<T> |
Interface for Multiple Kernel Classes.
|
Class | Description |
---|---|
DoubleLLSVM |
Locally Linear SVM, as described in:
Locally Linear Support Vector Machines L'ubor Ladicky, Philip H.S. |
DoublePegasosSVM |
Linear SVM classifier on arrays of double using the PEGASOS algorithm.
|
DoubleQNPKL |
Implementation of the QNPKL solver.
Original java code |
DoubleSAG |
Linear SVM using the SAG algorithm:
"A Stochastic Gradient Method with an Exponential Convergence Rate for Strongly-Convex Optimization with Finite Training Sets", Nicolas Le Roux, Mark Schmidt and Francis Bach. |
DoubleSGD |
Linear SVM classifier using stochastic gradient descent algorithm
|
DoubleSGDQN |
Linear SVM classifier using SGDQN algorithm.
|
GradMKL<T> |
MKL algorithm using a naive gradient descent.
|
LaSVM<T> |
Kernel SVM classifier implementing the LaSVM algorithm
|
LaSVMI<T> |
Kernel SVM classifier implementing LaSVM-I algorithm
|
NystromLSSVM<T> |
This classifier is a fast approximate SVM classifier using the Nystrom kernel
for project non-linearly the samples in a subspace in which a linear
classifier is learned.
|
ParzenClassifier<T> |
Classification tool using a Parzen window
|
SDCA<T> |
SDCA svm algorithm from Shalev-Shwartz.
|
SequentialMKL<T> | |
SimpleMKL<T> |
Implementation of the SimpleMKL solver.
Java conversion of the original matlab code. |
SMOSVM<T> |
SVM classifier using SMO algorithm
|
TSMKL<T> |
Implementation of the Two-Stages MKL solver by Abhishek Kumar et al.
This is a original implementation using the tools available in JKernelMachines, and not a Java conversion of the original matlab code. |
Provides basic class for classifiers, and some well known implementations.
Features SVM dual algorithms (SMO, LaSVM), MKL (SimpleMKL, Grad MKL), fast linear SVM (Pegasos, SGDQN, SGD) and Product Kernel Learning (QNPKL).