CODEWITHSUNDEEP
javaeclipsesvmlibsvm
samia
samia

Reputation: 21

How to start a project for SVM-based classification in Eclipse?

I'm extremely new to SVM. I need to classifier dataset with SVM in eclipse and I have found an article that explains how to do it by using SVM algorithm. Is it correct to use Weka to solve my problem?

Upvotes: 1

Views: 407

Answers (1)

MWiesner
MWiesner

Reputation: 9043

To start developing software for SVM-based classification, WEKA is a good starting point. The WEKA toolkit is, however, quite heavy in developer terms as it includes implementations for many other machine learning algorithms. Moreover, WEKA is GPL-licensed which might have implications on the license of the own project. Still, it is well-documented and and a very robust library as such.

If you would like to use only SVMs for your project there is a lightweight alternative to it. Give zlibsvm on top of libsvm a try. zlibsvm is a lightweight OO-oriented adapter to the somewhat oldskool original libsvm Java implementation.

Simply create a new Java-project. Next, you can add zlibsvm to your fresh Java project via this Maven pom declaration

<dependency>
    <groupId>de.hs-heilbronn.mi</groupId>
    <artifactId>zlibsvm-core</artifactId>
    <version>1.2</version>
</dependency>

or download it via Maven central here: zlibsvm-api-1.2.jar and zlibsvm-core-1.2.jar. Put both files in the classpath of the project.

You will also have to add libsvm.jar contained in the latest release ZIP file to the classpath of your project.

Examples and a short introduction on how to use zlibsvm can be found on the project's GitHub page or in the zlibsvm example project.

Hope it helps.

Upvotes: 1

Related Questions