Please use this identifier to cite or link to this item:
Title: Design of an Efficient Classifier for Bioinformatics Application
Authors: Garg, Deepika
Supervisor: Mishra, Amit
Keywords: Machine Learning;Bioinformatics;Ensemble;Genetic Algorithms;Classifiers Optimization
Issue Date: 2-Sep-2017
Abstract: In today’s integrated world, solutions for problems are inter disciplinary by nature. Soft computing proves to be potent for obtaining solutions accurately, and quickly. Moreover, the combination or hybrid of one of more methodology has resulted into the new class of system called Ensemble Models or Hybrid models. This has resulted in creating a classifier with better efficiency used for the design of intelligent systems. The study employed various machine learning techniques for the prediction of diseases. Particularly, the study applies the several data mining techniques: Decision tree, Neural Networks, Support Vector Machine, Linear Regression, Linear Discriminant Analysis, Naïve Bayes and k-nearest neighbor. The study deals with the focus on improving the accuracy of classification of machine learning algorithms. Undoubtedly, Support Vector Machine has provided the better results as compare to the other techniques in classification. In this thesis, various parameters of SVM have been exploited in order to get the best results. The proposed research converges on the hybrid of heterogeneous classifiers for disease prediction. The performance of a classifier is judged by two parameters namely Classification accuracy and Simulation Time. Another effort has been put up to optimize the classification technique such as SVM using Genetic algorithm in order to get the best fit value. Many techniques have been applied on various diseases dataset. One such technique employed SVM which considered ovarian cancer features for the optimization process.
Description: Master of Engineering -Wireless Communication
Appears in Collections:Masters Theses@ECED

Files in This Item:
File Description SizeFormat 
4812.pdf1.76 MBAdobe PDFThumbnail

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.