Optimized Processing of Data Privacy Threats Through CNN and XGBoost: An Analytical Approach with Scientific Repositories

This research addresses the critical challenge of securing sensitive information by leveraging machine learning to detect data privacy threats. In this study systematically evaluates and compares the performance of CNN and XGBoost classifier later to optimized with the advanced hyperparameter tuning framework. This robust preprocessing pipeline, including privacypreserving noise, was implemented to ensure data integrity. The results demonstrate a clear performance hierarchy, that an optimized XGBoost model achieving a superior classification accuracy that significantly outperforms than others. The analysis of feature importances from the optimized model provides a unique and interpretable to identifying the most influential features driving the model's decisions. These findings underscore the potential of combining powerful boosting algorithms with modern optimization techniques to build highly effective and insightful solutions for data privacy protection