SVM Parameter Optimization using Grid Search and Genetic Algorithm to Improve Classification Performance

Support Vector Machine Classifier Classification is a supervised learning technique which learns a function from training data set that consists of input features/attributes and categorical output [1]. Other than the wellknown classical data mining techniques such as naive Bayes, decision tree, rule...

Full description

Saved in:
Bibliographic Details
Published in:Telkomnika 2016-12, Vol.14 (4), p.1502
Main Authors: Syarif, Iwan, Prugel-Bennett, Adam, Wills, Gary
Format: Article
Language:English
Subjects:
Ant colony optimization
Biological activity
Biological evolution
Classification
Data mining
Decision trees
Genetic algorithms
Machine learning
Parameters
Rule induction
Support vector machines
Citations: Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Support Vector Machine Classifier Classification is a supervised learning technique which learns a function from training data set that consists of input features/attributes and categorical output [1]. Other than the wellknown classical data mining techniques such as naive Bayes, decision tree, rule induction, etc., Support Vector Machine (SVM) has gained more attention and has been adopted in data classification problems in order to find a good solution. Parameter Optimization Generally, most of machine learning algorithms will not achieve optimal results if their parameters are not being tuned properly. Parameter optimization can be very time consuming if done manually especially when the learning algorithm has many parameters [9, 10]. According to [11], the cross-validation technique can prevent the over-fitting problem. [...]there are more than three parameters but selecting more parameters and a large number of steps (or possible values of parameters) result in a huge number of combinations. [...]in our experiment we decided to make the range of C and ?? from 0.001 to 10,000. Parameter Optimization using Genetic Algorithm The GA which was firtsly proposed by John Holland in the 1975, is a method for solving optimization problems that is based on natural selection, the process that drives biological evolution. parameter combination within given ranges. The process was forced to stop after 2 weeks running. [...]grid search is very reliable only in low dimensional dataset with few parameters. Inducing decision trees with an ant colony optimization algorithm.
ISSN:1693-6930
2302-9293
DOI:10.12928/telkomnika.v14i4.3956