Thesis (M.Sc.) - Cairo University - Faculty of Computers and Information - Department of Information Technology

This thesis handles the software aging problem, which may be prevented using some rejuvenation techniques for the system before the crash occurs. Several machine learning based techniques are used to detect when to rejuvenate the system, in order to ensure its reliability and availability. For this purpose several machine learning based techniques are tried and tested on a set of collected bank data. Feature selection and reduction is first applied to the collected data in order to avoid the curse of dimensionality problem. A set of experiments with different classifiers are run on the processed data. These include back-propagation neural networks, K-nearest neighbors, Naive Bayes classifier, support vector machine, quadratic discriminant analysis and linear discriminant analysis. The highest obtained accuracy with respect to some online banking data is 87% using the linear discriminant analysis method. These experimental results are very promising, proving that machine learning techniques are reliable in the field of software rejuvenation

Issued also as CD