svm_main.cc File Reference

Include dependency graph for svm_main.cc:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Functions
void	DoSvmNormalize (Dataset *dataset)
	Data Normalization.
void	GenerateArtificialDataset (Dataset *dataset)
	Generate an artificial data set.
int	LoadData (Dataset *dataset, String datafilename)
	Load data set from data file.
int	main (int argc, char *argv[])
	Multiclass SVM classification/ SVM regression - Main function.
Variables
const fx_module_doc	svm_main_doc
const fx_entry_doc	svm_main_entries_doc []

---

Detailed Description

Author:

Hua Ouyang

This file contains main routines for performing 0. multiclass SVM classification (one-vs-one method is employed). 1. SVM regression (epsilon-insensitive loss i.e. epsilon-SVR). 2. one-class SVM (TODO)

It provides four modes: "cv": cross validation; "train": model training "train_test": training and then online batch testing; "test": offline batch testing.

Please refer to README for detail description of usage and examples.

See also:

svm.h

smo.h

Definition in file svm_main.cc.

---

Function Documentation

void DoSvmNormalize

(

Dataset *

dataset

)

Data Normalization.

Parameters:

the dataset to be normalized

Definition at line 99 of file svm_main.cc.

References linalg__private::AddTo(), GenVector< T >::CopyValues(), fx_default_param(), fx_param_bool(), GenVector< T >::Init(), GenMatrix< T >::Init(), GenVector< T >::length(), Dataset::matrix(), linalg__private::MulInit(), linalg__private::MulTransBInit(), Dataset::n_features(), Dataset::n_points(), linalg__private::Scale(), la::ScaleRows(), GenVector< T >::SetZero(), la::TransposeInit(), and Dataset::WriteCsv().

Referenced by LoadData().

void GenerateArtificialDataset

(

Dataset *

dataset

)

Generate an artificial data set.

Parameters:

the dataset to be generated

Definition at line 165 of file svm_main.cc.

References fx_param_double(), fx_param_int(), Dataset::OwnMatrix(), and data::Save().

Referenced by LoadData().

int LoadData	(	Dataset *	dataset,
		String	datafilename
	)

Load data set from data file.

If data file not exists, generate an artificial data set.

Parameters:

		the dataset
		name of the data file to be loaded

Definition at line 211 of file svm_main.cc.

References DoSvmNormalize(), fx_param_bool(), fx_param_exists(), fx_param_str_req(), GenerateArtificialDataset(), and Dataset::InitFromFile().

Referenced by main().

int main	(	int	argc,
		char *	argv[]
	)

Multiclass SVM classification/ SVM regression - Main function.

Parameters:

		argc
		argv

Definition at line 240 of file svm_main.cc.

References SVM< TKernel >::BatchPredict(), fx_done(), fx_param_int_req(), fx_param_str_req(), fx_root, fx_submodule(), SVM< TKernel >::Init(), GeneralCrossValidator< TLearner >::Init(), SVM< TKernel >::InitTrain(), LoadData(), SVM< TKernel >::LoadModelBatchPredict(), and GeneralCrossValidator< TLearner >::Run().

---

Variable Documentation

const fx_module_doc svm_main_doc

Initial value:

 {
  svm_main_entries_doc, NULL,
  "These are the implementations for Support Vector Machines (SVM), including Multiclass classification, Regression, and One Class SVM)\n"
}

Definition at line 89 of file svm_main.cc.