In-Depth Analysis of Anomaly Detection in Software Defined Infrastructures Leveraging Hybrid ML and DL Models

Detecting anomalies is crucial to assuring Software-Defined Infrastructures' (SDI) stability, security, and performance. A performance-centric evaluation of Machine Learning (ML) and Deep Learning (DL) models developed specifically for anomaly detection in SDI systems. The paper compares widely used Support Vector Machines (SVM), Random Forest, and Isolation Forest machine learning (ML) strategies, to advanced DL, Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) networks, hybrid CNN-LSTM models, and Auto encoders. Focusing on key performance parameters such as accuracy, precision, recall, F1-score, and computational efficiency, this study explores SDI anomaly detection in scenarios including network traffic analysis, resource utilization monitoring, and threat identification. This paper does an in-depth comparative analysis to identify the strengths, limits, and tradeoffs of ML and DL algorithms in dealing with complex, high-dimensional, and dynamic datasets typical in SDI. The findings shed light on model adequacy for certain anomaly detection tasks, enabling a thorough understanding of their applicability and operational restrictions. Furthermore, the article identifies future research initiatives focused at improving model scalability, lowering false positive rates, and boosting real-time detection capabilities to protect SDI from emerging threats and vulnerabilities.