Document Details

Document Type : Thesis 
Document Title :
New Machine Learning Approaches to Improve Software Bug Prediction
أساليب جديدة في تعلم الآلة لتحسين التنبؤ بالأخطاء البرمجية
 
Subject : Faculty of Computing and Information Technology 
Document Language : Arabic 
Abstract : Predicting bugs of a software system is an important research problem in software engineering. By predicting software bugs correctly, developers can accelerate the testing process for locating source code components containing bugs and hence reduce the time associated with software maintenance and development, thereby improving the software development cycle. In today’s software industry, plethora of software engineering environments has led to substantial amounts of data stored in repositories. Mining such data is a challenging task. A key goal of this thesis is to deliver reliable machine learning tools to software developers, who would use these tools as prediction calculators to identify bugs in software systems. To accomplish this goal, I develop new machine learning approaches combining an unsupervised technique with feature selection and supervised learning techniques. The supervised learning algorithms include support vector machines, random forests, neural networks, and deep neural networks. Experimental results on various bug prediction datasets demonstrate that my machine learning approaches generate higher performance results with statistical significance when compared against existing baseline approaches. 
Supervisor : Dr. Abdullah Algarni 
Thesis Type : Master Thesis 
Publishing Year : 1440 AH
2019 AD 
Co-Supervisor : Dr. Turki Turki 
Added Date : Monday, August 19, 2019 

Researchers

Researcher Name (Arabic)Researcher Name (English)Researcher TypeDr GradeEmail
عماد نبيل كائنKaen, Emad NabilResearcherMaster 

Files

File NameTypeDescription
 44865.pdf pdf 

Back To Researches Page