Novel Framework for Anomaly Detection Using Machine Learning Technique on CIC-IDS2017 Dataset
Richa Singh, Gaurav Srivastav
Abstract
There are various deep learning-based IDS techniques are implemented in large scale. Intrusion detection systems are critical components for protecting ICT infrastructure (IDSs). Keeping this in mind, solid solution is required for different types of new attacks and complexity control. Deep learning and machine learning is widely used to handle high dimensional, complex type data. The IDS detects and attracts various attack types such as known, unknown, and zero-day attacks using unsupervised machine learning techniques. To detect threats without prior knowledge, a framework has been designed that uses the concept of One Class SVM (OCSVM) and active learning. The CIC-IDS2017 dataset was used to test the performance of the framework and compare the result with UNSW-NB15 and KDD cup 99 dataset. The final output shows that this framework gives better performance than other.