| Popis: |
Intrusion Detection Systems (IDS) are valuable toolsfor the proper identification and the timely response to potentialsecurity threats in a network, using traffic analysis and anomalousactivities detection.Traditional IDS rely on rule-based orsignature-based methods to detect known cyber attacks, but thesemethods often fail to detect novel ones. There has been a growinginterest recently, in using Machine Learning (ML) algorithms toenhance the detection capabilities of IDS. As a downturn, the datasets used by ML algorithms for IDS applications refers tonetwork logs which may contain sensitive information, resultingin privacy threats. To address this issue, Differential Privacy (DP)can be used to preserve the privacy of network logs, while stillallowing the ML algorithm to extract useful information from thedata. In this work we test the performance of four popular MLclassifiers (Gaussian Naive Bayes, Logistic Regression, SupportVector Machines, Random Forest Classifier) in the CIC-IDS2017dataset when a DP mechanism is added to each algorithm incomparison with the classical non-DP setting. |
