Author

Umair Ahmed

Degree

Master of Science in Computer Science

Faculty / School

Faculty of Computer Sciences (FCS)

Department

Department of Computer Science

Date of Submission

2018-01-01

Advisor

Dr. Muhammad Sarim

Project Type

MSCS Survey Report

Abstract

There are millions of million (approx. 1012) galaxies thought in the universe. In the past decades, the challenging task was to see and photograph distant objects. It was very difficult to discover the galaxies and other heavenly bodies in the universe. The classification of galaxies at that time was not a big demand because of the very few images of galaxies in number we had. In a couple of decades, a very drastic change has occurred in the technology thus we have been able to photograph distant galaxies than in the past. With the advancement of technology, the discovering rate of galaxies is very fast. There has been a burst of galaxies is being found and we have huge data sets of images of different galaxies in the universe. To study further about galaxies and other heavenly bodies and their structure, composition and so. We need to identify the shape and classify galaxies properly into appropriate classes so that we can understand the origin and formation of galaxies and evaluation processes of the universe. Morphological Classification is a challenging task today. The term morphological we mean form and structure of the galaxy. Several challenges at the various level of classification - Big data handling, Computation power of hardware, noise and noisy elements in images, appropriate technique, or method to classify correctly, etc. make this problem harder to solve. In this review paper, we present a comprehensive survey of efforts in the past of a couple of decades to address the classification problem and focus on the recent machine learning methods are being used for classification of galaxies. There are various techniques of Artificial Intelligence for classification but some of the most recent techniques for galaxies classification such as Non Negative Matrix Factorization Method, Deep Convolution Neural Network, etc., will be discussed in detail and brief discussion on other most common techniques such as Artificial Neural Network (ANN), Decision Tree (DT), Navies Bayes and C4.5, etc. are also part of this review.

The full text of this document is only accessible to authorized users.

Share

COinS