An Optimized Machine Learning Model for Detection and Classification of Supply Chain Attacks in Containerized Cloud Systems

Abstract

This research developed an enhanced machine-learning framework for classifying and detecting supply-chain attacks in containerized applications by integrating the Aquila Optimizer (AO) with the Extreme Gradient Boosting (XGBoost) algorithm. The research was motivated by the increasing security risks associated with container technologies such as Docker and Kubernetes, particularly across software supply chain stages including development, dependency management, and deployment. The NSL-KDD dataset and additional container-related datasets from GitHub repositories were utilized, with preprocessing steps including encoding, normalization, feature selection, and class balancing. The developed AO-XGBoost model optimizes hyperparameters and decision thresholds to improve detection capability. Experimental results show that the baseline XGBoost achieved an accuracy of 0.80 and an AUC-ROC of 0.85, while the developed AO-XGBoost model improved performance to 0.86 accuracy and 0.93 AUC-ROC. The optimized model also demonstrated significant improvements in precision, recall, and F1-score, indicating better balance and reduced false positives. These findings confirm that metaheuristic optimization enhances model generalization and effectiveness in high-dimensional cybersecurity datasets. The developed model provides a robust and scalable solution for detecting complex supply chain attacks in containerized environments, contributing to improved software security and resilience.