A Machine Learning System to Enhance IT Security in Human Resource Management

With the increasing reliance of Human Resource (HR) management on digital systems, protection of sensitive employee data is critical to prevent unauthorised access. This research addresses the necessity of an advanced security system in detecting and mitigating potential security threats in HR operations. A machine learning (ML)-based system was thus designed and implemented, using the models support vector machines, recurrent neural networks, autoencoders, and random forests. The training was thus conducted utilising a mix of real-world and synthetic data, specifically including HR system logs as well as user activities and various security events, including login attempts and unauthorised access to some files. After feature engineering and preprocessing, all the models were evaluated with respect to the metrics of accuracy, precision, recall, and F1 score. Results indicated that Support Vector Machine (SVM) outperformed all models in terms of 98.54% accuracy with low latency, 15 ms, a throughput of 150 predictions per second, and 1.2 seconds for the detection time. These results point out that the SVM model is suitable for real-time applications, showing that it is efficient in detecting security breaches and scalable in large HR systems. Findings indicate that machine learning models, especially SVM, can be used to increase IT security in HR management, allowing for timely detection and mitigation of security threats.