A Comparative Study of Logistic Regression, SVM, and ANN for Intrusion Detection Using the NSL-KDD Dataset
M Dinesh Reddy · 2025
This paper explores the effectiveness of classical and neural machine learning models in automating intrusion detection within cybersecurity systems. Using the publicly available NSL-KDD dataset, we benchmark three models-Support Vector Machines (SVM), Logistic Regression, and a feedforward Artificial Neural Network (ANN)-to evaluate their performance in binary classification of network intrusions. The dataset is preprocessed and normalized to ensure consistency across models. Each model is trained and evaluated using precision, recall, and F1-score metrics, providing a fair and grounded comparison. While deep learning models are often considered superior, this study emphasizes the viability of simpler algorithms in low resource or rapid deployment scenarios. Our findings suggest that, when properly tuned, even basic ML models can perform competitively in certain cybersecurity use-cases, validating the practicality of lightweight solutions in real-world systems.