Predictive Software Defect Identification with Adaptive Moment Estimation based Multilayer Convolutional Network Model

Abstract

The practice of predicting software errors in quality assurance now experiences a significant advancement through AI automation. The system uses natural language processing together with data analytics and machine learning techniques to examine historical records in order to make defect identification. The existing defectidentification models struggle with various challenges that stem from noisy data and class imbalanced datasets and complex pattern recognition tasks because their performance deteriorates. In this research develop a new model Adaptive Moment Estimation based Convolutional Neural network –Multi Layer Perception model (Adam based CNN-MLP), for defect identification as it combines CNN features extraction power with adaptive MLP identification capabilities. The system extracts essential data points from unprocessed information while developing general skills through its ability to detect intricate patterns between software faults. The CNN segment first extracts spatial patterns from the data before the MLP component uses identification abilities to analyze high-level dependencies. The combination of AI advanced features creates an optimal solution which enables efficient and accurate scaling of software defect identification while expanding AI quality management capabilities in software engineering