Comprehensive Detection of Known Attacks Using Integrated Datasets

Aouiche Chaima, Bolin Chen, Abdelaziz Aouiche, Bairong Shen, Rajeev K.Singla, Sahraoui Dhelim

TL;DR: the detection of know attacks from the integration of multiple datasets

Abstract: Cyberattacks, which are malicious attempts, are continuously increasing, leading to unauthorized data access, services disruptions, and network degradation. Efficient and proactive detection of these attacks is crucial to maintaining the confidentiality, integrity and availability of the digital environment. In this paper, we present an enhanced and comprehensive approach that cannot only detects known attacks but also identifies unknown ones through the integration of three up-to-date datasets and the implementation of sampling and feature selection techniques. To achieve this, we conducted experiments using two categories of methods: Machine Learning(ML), such as Naive Bayes (NB), Decision Trees (DT), Logistic Regression (LR), K-Nearest Neighbors (KNN), Random Forest (RF), XGBoost, and AdaBoost, and Deep Learning (DL) architectures, including Artificial Neural Networks (ANN), Deep Neural Networks (DNN), Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU), and Recurrent Neural Networks (RNN). ML models offers easy interpretability, while DL models excel at handling complex patterns. The results from the majority of models show promising accuracy rates, with 99\% for known attacks, significantly outperforming previous studies validating the effectiveness of our strategy.

Submission Number: 195