Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{200468,
        author = {Dhanashree Hiwarale and Satish Gujar},
        title = {Machine Learning-Based Intrusion Detection Model for Network Security},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {12},
        pages = {2471-2478},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=200468},
        abstract = {The rapid proliferation of digital infrastructure and internet-dependent services has intensified the need for robust cybersecurity mechanisms. Threat actors increasingly exploit vulnerabilities through sophisticated attack vectors including volumetric floods, covert access attempts, and reconnaissance-based intrusions. Conventional defence mechanisms that depend on pre-catalogued attack signatures face a fundamental limitation: they are incapable of recognising threats that fall outside their existing knowledge base. To overcome this bottleneck, this study presents an adaptive, intelligence-driven Intrusion Detection Framework (IDF) that autonomously learns distinguishing characteristics from historical network activity data.
The framework was developed and validated on the NSL-KDD benchmark corpus, a curated dataset widely adopted in cybersecurity research for its balanced and non-redundant structure. A rigorous data preparation workflow was employed, encompassing attribute transformation, value normalisation, and structural alignment of training and evaluation subsets. Three supervised classification algorithms, Logistic Regression, Decision Tree, and Random Forest were systematically applied to differentiate between legitimate and anomalous traffic. Comparative evaluation was performed across multiple dimensions including classification accuracy, predictive precision, sensitivity (recall), and the harmonic F1 measure.
Among the evaluated algorithms, Random Forest demonstrated superior generalisation capability and detection stability. To further strengthen real-world applicability, the framework incorporates SHAP (Shapley Additive Explanations) analysis, which offers post-hoc interpretability by quantifying the marginal contribution of each input variable to the classification outcome. This integration bridges the gap between model performance and decision transparency, yielding a framework that is simultaneously effective, computationally efficient, and operationally explainable},
        keywords = {Intelligent Intrusion Detection, Supervised Learning, Cyber Threat Classification, NSL-KDD Benchmark, Ensemble Classification, Explainable Artificial Intelligence, SHAP Interpretability},
        month = {May},
        }