A Novel Classification System for Faster Assessment of IDS Alerts Using Convolutional Neural Network
Satoshi Kimura, Hiroyuki Inaba · 2023
The Intrusion Detection System (IDS) is designed to detect cyberattacks. However, given the vast number of alerts generated by IDS for different registered attack signatures, there's a growing demand for a method to quickly assess the status of IDS alerts. In this study, we propose a novel approach where we create image data, termed “composite block diagram”, based on the alert counts for each signature in IDS. These composite block diagrams are labeled with the corresponding alert detection time as the ground truth. We trained the pairs of composite block diagrams and alert detection times by using Convolutional Neural Network (CNN). The results indicate that our proposed method can recognize patterns in the composite block diagrams corresponding to the detection times with an accuracy of 75.44% on unseen data.