OpEffiRes Net: Optimization based EfficientNetB0-ResNet50 and correlation based feature selection for software...

Area of Science:

• Computer Science
• Software Engineering
• Artificial Intelligence

Background:

• Software failures pose significant risks to system reliability and operational continuity.
• Accurate Software Failure Prediction (SFP) is crucial for proactive maintenance and defect mitigation.
• Existing prediction methods often face challenges in accuracy and efficiency.

Purpose of the Study:

• To develop an advanced hybrid Deep Learning (DL) strategy for enhanced Software Failure Prediction (SFP).
• To introduce a novel feature selection algorithm (DE-COA) and a hybrid prediction model (OpEffiResNet).
• To improve the accuracy and reliability of software defect prediction.

Main Methods:

• Data pre-processing including min-max normalization.
• Feature selection using the proposed Double Exponential-Coati Optimization Algorithm (DE-COA).
• Hybrid prediction model combining EfficientNetB0 and ResNet50 (OpEffiResNet).
• Evaluation using 5-fold cross-validation.

Main Results:

• The proposed OpEffiResNet model achieved high performance metrics.
• Average accuracy of 98.8% ± 0.6, sensitivity of 99.4% ± 0.4, specificity of 99.0% ± 0.5, precision of 98.8% ± 0.7, and F1-score of 98.7% ± 0.6.
• Demonstrated significant improvement in software defect prediction.

Conclusions:

• The developed hybrid DL framework effectively enhances software defect prediction.
• The OpEffiResNet model offers accurate, efficient, and reliable software fault predictions.
• Future work includes validation on diverse datasets and comparison with other advanced methods.