Software Reliability Optimization using Soft Computing and Evolutionary Algorithms

Abstract

Software systems play a pivotal role in modern society, powering critical applications across various domains. Ensuring that these software systems are reliable is extremely important to prevent major failures, financial losses, and bad user experiences. Software reliability is one of the essential factors of quality in software engineering, like other quality attributes such as functionality, usability, maintainability, performance, serviceability, and documentation, etc. Over the last few years, several software reliability models have been developed. Many software reliability growth models (SRGMs) have been presented to estimate product reliability growth over the last three decades, ensuring software quality and analyzing software reliability. The field of software reliability prediction has seen substantial advancements in recent years, particularly with the integration of soft computing techniques and optimization algorithms. However, a notable research gap exists in addressing the correlation between these metrics or exploring the most optimal combinations while optimizing for multiple objectives in software reliability prediction. newline newlineSoft Computing is a group of Artificial Intelligence and bio-inspired algorithms. Artificial Intelligence contains many subdomains such as Machine Learning, Deep Learning, Reinforcement Learning, Natural Language Processing, Computer Vision, etc. Machine Learning algorithms are highly computational and learn patterns from data. This work aims to bridge this gap by introducing a novel approach called the Correlation Constrained Multi-Objective Evolutionary Optimization Algorithm (CCMOEO) for software reliability prediction. CCMOEO proves to be an effective optimization technique for estimating the parameters of popular growth models that encompass reliability. To maximize classification accuracy, our proposed method mitigates modeling uncertainties by integrating multiple metrics with multiple objective functions. newline newlineDifferent datasets, such as Eclipse Equinox Lucy in Mylyn and P