Machine Learning Advancements for Surveillance: Cybersecurity and Behavior Risk Forecasting

Abstract

The increasing integration of surveillance systems in various domains has amplified the importance of ensuring cybersecurity and predicting behavioral risk factors. This paper presents an innovative approach, the "Risk and Anomaly Detection via Deviation Analysis and Ranges" (RADAR) algorithm, designed to enhance cybersecurity and enable behavioral risk factor prediction within surveillance systems using machine learning techniques. RADAR is applied to a dataset from a Behavioral Risk Factor Surveillance System (BRFSS), a critical component of public health monitoring. The algorithm comprises several steps, including data preprocessing, Normalization and a dynamic threshold for anomaly detection based on deviation analysis. Furthermore, an ensemble method is employed to identify instances as anomalies if they deviate from multiple attributes simultaneously. In addition to anomaly detection, RADAR contributes to behavioral risk prediction by integrating machine learning classifiers. It allows the system to predict behavioral risk factors based on the data, providing valuable insights for surveillance purposes. Experimental results demonstrate the effectiveness of RADAR in enhancing cybersecurity by identifying anomalies and enabling accurate behavioral risk factor prediction. The algorithm showcases its ability to evaluate and select the best-performing classifiers, thereby enhancing the reliability of surveillance systems.