Integrated Features by Administering the Support Vector Machine of Translational Initiations Sites in Alternative Polymorphic Context

Nanna Suryana, Herman and Burairah, Hussin (2012) Integrated Features by Administering the Support Vector Machine of Translational Initiations Sites in Alternative Polymorphic Context. International Journal Bioautomation, 16 (1). pp. 13-22. ISSN 1314-2321

PDF bioautomation.pdf - Published Version Restricted to Registered users only Available under License Creative Commons Attribution. Download (208kB)

Abstract

Many algorithms and methods have been proposed for classification problems in bioinformatics. In this study, the discriminative approach in particular support vector machines (SVM) is employed to recognize the studied TIS patterns. The applied discriminative approach is used to learn about some discriminant functions of samples that have been labelled as positive or negative. After learning, the discriminant functions are employed to decide whether a new sample is true or false. In this study, support vector machines (SVM) is employed to recognize the patterns for studied translational initiation sites in alternative weak context. The method has been optimized with the best parameters selected; c = 100, E = 10-6 and ex = 2 for non linear kernel function. Results show that with top 5 features and non linear kernel, the best prediction accuracy achieved is 95.8%. J48 algorithm is applied to compare with SVM with top 15 features and the results show a good prediction accuracy of 95.8%. This indicates that the top 5 features selected by the IGR method and that are performed by SVM are sufficient to use in the prediction of TIS in weak contexts.

Item Type:	Article
Uncontrolled Keywords:	Kernel function, Algorithms, Non linear kernel, IGR method.
Subjects:	Q Science > Q Science (General)
Divisions:	Faculty of Information and Communication Technology > Department of Software Engineeering
Depositing User:	Burairah Hussin
Date Deposited:	21 May 2012 08:33
Last Modified:	30 Sep 2021 05:17
URI:	http://eprints.utem.edu.my/id/eprint/3170
Statistic Details:	View Download Statistic

Actions (login required)

View Item