Machine Learning based Classifier and Feature Extractor for Intrusion Detection

Cyberattacks, or attacks on computer networks, now affect practically every device linked to the Internet on a daily basis. Unseen Cyberattacks create a huge problem in these devices. Thus, some methods or techniques tried to be employed that can encounter unseen attack along with known attacks. ML and DL based IDS are being used to encounter them. This research will show a variety of methods used to protect against various types of cyberattacks. After being applied to datasets, the classification algorithms are compared in accordance with precision, accuracy, recall, F1 score. The classification algorithms used in the paper are the LR, Naïve Bayes and KNN as classifier and PCA as feature extractor. Importance of feature extraction is also presented in the paper by taking performance of specified classifier for different feature values in account. Deep learning classification algorithms have been critically reviewed using specified performance criteria.