MF-CNN: Hybrid Classification Technique for Intrusion Detection System
Abstract
The Internet's popularity has grown fast, increasing the amount of sensitive data transmitted and addressed over the Internet. As a consequence of these developments, the number of attacks on critical Internet information increases year after year. One of the most significant risks to the Internet is an intrusion. To overcome the shortcomings of intrusion detection systems, such as poor accuracy, a large number of false alarms, and a sluggish response time, a variety of methods and approaches have been developed. This paper suggested a hybrid machine learning method for network intrusion detection based on a mix of Mayfly and Convolutional Neural Network classification. This research focuses on lowering the rate of false-positive alarms, lowering the rate of false-negative alarms, and enhancing the detection rate. The proposed approach made use of the NSL-KDD dataset. Some steps on the dataset have been done to improve classification performance. The classification was carried out by a Mayfly and Convolutional Neural Network. The proposed hybrid machine learning method has a higher positive detection rate and a lower false alarm rate after training and testing.