A Hybrid AI Approach for Anomaly Detection in Cybersecurity: Integrating Deep Learning with Statistical Models

Abstract

Cybersecurity threats are evolving in complexity, necessitating advanced detection mechanisms to safeguard digital infrastructures. This paper proposes a hybrid AI approach for anomaly detection, combining deep learning techniques with traditional statistical models to enhance threat detection accuracy. Using real-world cybersecurity datasets, we demonstrate how convolutional neural networks (CNNs) and recurrent neural networks (RNNs) can effectively detect anomalies when integrated with statistical methods such as Bayesian inference and Markov models. Our results indicate that this hybrid approach significantly reduces false positives and enhances detection rates compared to standalone methods. The findings contribute to the development of robust AI-driven security solutions for modern cyber threats.