Machine Learning Algorithms for FPGA Implementation in Software Engineering Applications

This research investigates the FPGA implementation of lightweight machine learning algorithms for software engineering applications, focusing on software defect prediction. Models, including Decision Trees and Shallow Neural Networks, were developed and evaluated on a real-world dataset. Software simulation metrics such as accuracy, precision, recall, and F1-score were analyzed, alongside FPGA synthesis results involving latency, power consumption, and resource usage. Experimental results demonstrate that Shallow Neural Networks offer superior prediction accuracy, while Decision Trees provide more efficient hardware deployment. The study highlights a complete design flow from software model training to FPGA implementation, supporting real-time, energy-efficient software quality monitoring.
Keywords: FPGA, Software Defect Prediction, Machine Learning, High-Level Synthesis, Decision Tree, Shallow Neural Network