Intrusive Detection Techniques Utilizing Machine Learning, Deep Learning, and Anomaly-based Approaches
Mayur Rele, Dipti Patil
Abstract
This study illustrates how machine learning (ML), deep learning (DL), and anomaly-based techniques can be combined for network intrusion detection. The objective is to improve intrusion detection systems’ capability to identify known and novel network threats. While Random Forest is used for classification and feature selection in the ML portion, Convolutional Neural Networks (CNN) capture nuanced spatial patterns in network traffic data in the DL portion. Statistical analysis techniques are used to identify anomalies. Integrating the two approaches, the hybrid method enhances detection accuracy while minimizing false positives. Using real-world datasets, we demonstrate that the proposed method outperforms contemporary machine learning (ML), deep learning (DL), and anomaly-based methods. This hybrid method may greatly facilitate the proactive identification of hazards in the constantly evolving field of intrusion detection. It may be necessary to conduct additional research on optimization strategies and the hybrid model’s potential network-wide applicability.