```html Network Intrusion Detection Using Machine and Deep Learning: A Comprehensive Review of Detection Models, Evaluation Strategies, and Emerging Trends| IJICCS

International Journal of Intelligent Communication and Computer Science

Peer Reviewed Scholarly Journal for Advanced Research

Network Intrusion Detection Using Machine and Deep Learning: A Comprehensive Review of Detection Models, Evaluation Strategies, and Emerging Trends

Anushka Singh1, Upasana Dugal2

1 Research Scholar, Department of Computer Science and Engineering, Pranveer Singh Institute of Technology, Kanpur, Uttar Pradesh, India

2 Department of Computer Science and Engineering, Pranveer Singh Institute of Technology, Kanpur, Uttar Pradesh, India

India

Corresponding Author: Anushka Singh

Email: anushkasingh.res85@gmail.com

Abstract

Network Intrusion Detection Systems (NIDS) are essential for protecting modern communication networks from increasingly sophisticated and evolving cyberattacks. The rapid expansion of cloud computing, Internet of Things (IoT), edge computing, and interconnected services has created complex network environments where conventional security mechanisms may not provide sufficient protection. This review presents a comprehensive analysis of machine learning (ML) and deep learning (DL) approaches for intelligent network intrusion detection. Classical ML techniques, including decision trees, support vector machines, random forests, k-nearest neighbours, and ensemble methods, are examined in terms of their detection capability, feature requirements, and computational efficiency. The review further investigates DL architectures, including convolutional neural networks, recurrent neural networks, long short-term memory networks, autoencoders, hybrid models, and Transformer-based approaches for learning complex traffic patterns. Major NIDS datasets, including KDD-based datasets, UNSW-NB15, CICIDS, Bot-IoT, TON_IoT, IoT-23, MQTT-IoT-IDS2020, and Edge-IIoTset, are discussed with emphasis on attack diversity, class imbalance, and evaluation limitations. Key performance metrics and benchmarking practices are also examined. The review identifies major challenges, including dataset bias, zero-day attacks, concept drift, encrypted traffic, adversarial attacks, explainability, privacy, computational constraints, and cross-dataset generalization. Finally, emerging directions such as continual learning, federated learning, explainable AI, graph learning, edge-cloud intelligence, and Transformer-based detection are discussed. The review provides a structured perspective on current NIDS research and future requirements for robust, adaptive, efficient, and deployment-oriented intrusion detection systems.

Keywords

NIDS; Machine Learning; Deep Learning; Intrusion Detection; Vision Transformer; Anomaly Detection

Full Text

The complete article is available in PDF format.

View / Download PDF

How to Cite

Anushka Singh, and Upasana Dugal "Network Intrusion Detection Using Machine and Deep Learning: A Comprehensive Review of Detection Models, Evaluation Strategies, and Emerging Trends" International Journal of Intelligent Communication and Computer Science, vol. 4, no. 1, 2026, pp. 22-61.

Copyright and Licensing

Beginning with Volume 4, articles published in IJICCS are licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).

The article may be shared and adapted for non-commercial purposes, provided appropriate credit is given to the original author(s) and IJICCS.

View CC BY-NC 4.0 License

```