Brute-force Detection Using Ensemble Classification

Abstract

Traditional brute-force is a dictionary-based attack that tries to unlock an authentication process in service. This type of brute force can be applied in web and SSH services, and brute-force XSS injects JavaScript code. In this paper, we explore four types of ensemble classifiers using CIC-CSE-IDS 2018 to determine which yields the highest accuracy, recall, precision, and F1 in detecting three types of brute force. The first step of the research is to normalise the dataset with the tanH operator. The second step is to train the single classifier to determine three types of single classifiers combined as ensemble classifiers. The last step is predicting and comparing the results of four ensemble classifiers. The stacking algorithm achieves the best test result that reaches 94.87%, 99.94%, 98.82%, and 99.37% for accuracy, precision, recall, and F1, respectively.