Leveraging Deep Neural Networks for Cloud-Based Network Traffic Anomaly Detection and Security Enhancement

Abstract

Network traffic anomaly detection plays a critical role in safeguarding cloud-based environments from emerging
cybersecurity threats. Traditional methods like Support Vector Machines (SVM) and Random Forest, though
effective in some cases, often struggle with detecting sophisticated and unknown anomalies in large, high-
dimensional datasets. This study proposes a cloud-based approach using Deep Neural Networks (DNN) to
improve the detection of both known and unknown network traffic anomalies. The DNN model is trained using a
dataset containing network traffic data with labeled instances of normal and malicious traffic, including features
such as packet size, protocol type, and flow duration. Various performance metrics, including accuracy, precision,
recall, F1-score, and ROC-AUC, are used to evaluate the model’s effectiveness. The results show that the DNN
outperforms traditional methods, achieving an accuracy of 95.2%, precision of 94.7%, recall of 96.0%, and ROC-
AUC of 0.98. These findings demonstrate the DNN's superior ability to detect network traffic anomalies with
higher precision and recall while minimizing false positives.