TRAP@NCI

Characteristics Behind the Selection of Base Classifiers in Multiple Classifier System

Tomer, Vikas (2016) Characteristics Behind the Selection of Base Classifiers in Multiple Classifier System. Masters thesis, Dublin, National College of Ireland.

[img]
Preview
PDF (Master of Science)
Download (1MB) | Preview

Abstract

The pace of generating data in all areas is extremely high. This pace has been mounting the pressure on data scientists to make the advancement continuously, in the selection of machine learning algorithms which are very specific and precise. As there is a range of classifier families available, it is hard to choose which classifier or family of classifiers is appropriate for a particular dataset. This difficulty hinders the advancement of the Multiple Classifier System (MCS). A specified system from several machine learning algorithms will not only be time saving, but also will give more definite and detailed results. There are many articles and a lot of research available in the literature which discuss this problem. However the job of selecting the best classifier from several, for a particular classification task on any dataset, is still incomplete. There is no concrete solution available for this complex problem in the collected works. This project focuses on selecting the appropriate base classifiers in the Multiple Classifier System for a particular dataset. The primary motive of this project is to make some helpful contribution in this direction by finding out those characteristics which decide the selection method of base classifiers. This project advocates for the absence of some certain classifier in the base layer of the MCS, as the absence of these classifier increase the accuracy of pridiction in the MCS. Furthermore, this project corroborate the efficacy of Vote Meta learner more than the Stacking Meta learner.

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: CAOIMHE NI MHAICIN
Date Deposited: 27 Jan 2017 16:24
Last Modified: 27 Jan 2017 16:24
URI: http://trap.ncirl.ie/id/eprint/2524

Actions (login required)

View Item View Item