Intelligent System to Detect Software Defect in Critical Applications

Abstract

Intelligent systems are sophisticated machines that can sense and react to their surroundings. These systems investigate how these technologies interact with human users in constantly changing physical and social situations. Some of the applications of intelligent systems are traffic lights, smart meters, automobiles, digital television, and many more. In spite of it’s wide success, there are many software defects in these existing systems namely, system crashes, hangs, undefined behavior. Such defects are exploited by hackers for various security attacks. Many defects are discovered and addressed by various machine learning models. Hence, the prime focus of this article is to exhaustively review various software defects, methods to compare various approaches to address the detects. The article also compares various machine learning models (tree based gradient boosting, decision tree based gradient boosting, optimized distributed gradient boosting, Gaussian Naive Bayes, Multinomial Naive Bayes and Bernoulli Naive Bayes) on five PROMISE datasets including JM1, KC1, KC2, and PC1.