Enhanced Software Defect Prediction using Quantum Hamiltonian Generative Adversarial Network for Improved Software Performance Reliability

The development of software systems with high quality and decreased maintenance costs depends on Software Defect Prediction (SDF). The current dataset qualities combined with feature redundancies make confident fault detection difficult through conventional methods. Current defect prediction techniques display multiple deficits through inaccurate results along with many false alarms and dataset bias. A high-performance quantum-inspired method must be researched in working to create an accurate precision model. The project will create the Enhanced Software Defect Prediction using Quantum Hamiltonian Generative Adversarial Network for Improved Software Performance Reliability (DBO-QH-GAN) model. Several software metrics need to be retrieved from the PROMISE datasets before this process can be initiated. The second step includes Fuzzy K- Top Matcher (FK- TM) functionality that removes undesirable noise factors such as redundant features and missing values as well as outliers. The Botox Optimization (BotO) algorithm selects five essential metrics that encompass cyclomatic complexity as well as Halstead measures and coupling and code churn and historical defect density. These preprocessed set of characteristics are fed to a Quantum Hamiltonian Generative Adversarial Network (QH-GAN) which identifies the defects from probabilistic thresholding and thereafter optimized using Duck Bevy Optimization (DBO) to achieve the optimal possible performance. Accuracy of the model was 99.2%, precision was 99.4%, and recall was 99.1 %. The DBO-QH-GAN model sets new standards for defect prediction by providing an improved enhancement of reliability along with decreased maintenance cost.