Elefant Components

Algorithm Components

Support Vector Machines (SVMs)

• One Class SVM (Novelty Detection)  
• Epsilon SVM Regression
• Nu SVM Regression
• Laplacian SVM Regression

• Nu SVM Classification
• C SVM Classification

• Online SVM
• Online SVM stepsize adaptation
• Online Nu SVM stepsize adaptation
• Stochastic gradient descent

Gaussian Processes (GPs)

• Standard Gaussian Regression
• Heteroscedastic Gaussian regression

• Multiclass GP
• Extreme multiclass GP
• Transductive Multiclass GP
• Inverse Transductive Multiclass GP

Kernel Hebbian Learning

Feature Selection

Other Components

Input/Output components:

• Batch and online data sets
• Sparse or dense matrices
• Random data generators
• Supported data formats: CSV, LibSVM, MATLAB, ARFF, etc.

Data preprocessors/filters

Model selection components

K-Fold Cross Validation

Data visualization

• Plotting 2-D data
• Plotting image data

Core Elefant Modules

Linear Algebra

• SciPy and NumPy
• PETSc + PyPETSc

Kernel Modules

• algorithms for fast computation of kernel matrices
• Linear, RBF, Dot Product, String, Multi-class

Optimizers and Solvers

Loopy Belief Propagation and Junction Tree algorithms

Cover Tree for calculating the nearest neighbor