Network Intrusion Detection with CNNs: A Comparative Study of Deep Learning and Machine Learning Models

The exponential increase in internet usage has led to a surge in cyber threats and network attacks, making them critical challenges in today’s digital landscape. Leveraging machine learning techniques has proven effective in detecting and addressing network intrusions. This study focuses on the application of deep learning, particularly convolutional neural networks (CNNs), for improving the detection and classification of intrusions within the NSL-KDD and UNSW-NB15 datasets. The performance of CNN models is compared against several widely-used machine learning algorithms, including Gaussian Naïve Bayes, Logistic Regression, K-Nearest Neighbors, Support Vector Machines (SVM), AdaBoost, XGBoost, CatBoost, and LightGBM. The experimental analysis reveals that CNN-based approaches consistently surpass traditional classifiers, offering enhanced accuracy and robustness in identifying cyber threats.