Enhancing Intrusion Detection Systems With a Novel Hybrid Learning-Based Framework

Abstract

The proliferation of cyber threats and the increasing sophistication of attacks have necessitated the development of robust Intrusion Detection Systems (IDS) to protect sensitive information and network infrastructures. Traditional IDS methods often struggle to effectively detect and respond to emerging threats due to their reliance on static rule-based approaches. To address these limitations, this paper presents a simplified and novel hybrid learning-based framework for strengthening IDS. The proposed framework integrates the strengths of two prominent Machine Learning (ML) techniques: Deep Learning (DL) and Ensemble Learning. DL models, specifically Convolutional Neural Networks (CNNs), are employed to extract high-level features from network traffic data, enabling the system to automatically learn and adapt to complex attack patterns. Ensemble learning is then utilized to combine multiple classifiers, leveraging the diversity of their decision boundaries, and enhancing the overall detection performance. The experimental results demonstrate that the suggested framework achieves superior detection accuracy, outperforming traditional rule-based approaches as well as standalone DL and ensemble learning models.