Intrusion Detection Using Machine Learning




Janwari, Adnan Athar

Journal Title

Journal ISSN

Volume Title



Since the phenomenal growth in the usage of computer networks, concerns such as service availability, data integrity, and data confidentiality are becoming increasingly important. As a result, the network administrator must employ a variety of intrusion detection systems (IDS) to analyze traffic on the internet for unauthorized and hostile activity. The term "intrusion" refers to a malicious breach of security policy. As a result, an intrusion detection system analyzes the traffic passing via computer systems on a network to look for malicious activity and recognized threats and sends out warnings when it detects them. Machine learning algorithms are currently being widely used to develop efficient intrusion detection systems. Building an efficient intrusion detection system necessitates research into optimal ensemble methods. In this report, the CSE-CIC-IDS2018 dataset is utilized with eight network attacks and benign data. To identify the ten most important features, we have used Linear Discriminant Analysis (LDA). Naive Bayes, Random Forest (RF), and Decision Tree (DT) are Machine Learning (ML) techniques examined. The experiments are conducted with five-fold cross-validation utilizing the open-source software WEKA. The performance measures were used including execution time, accuracy, precision, F-measure, and recall. In terms of accuracy, precision, recall, and F-measure, the findings reveal that the Decision Tree surpasses the other methods.



Intrusion Detection, Machine Learning