Published Paper Details:

In today's world, networks are significant, and cyber security has emerged as an essential study field. An intrusion detection system (IDS), a significant cyber security method, keeps track of the condition of the network's software and hardware. Existing IDSs still confront hurdles in increasing detection accuracy, lowering false alarm rates, and identifying novel assaults, despite decades of research. Many academics have concentrated on building IDSs that employ machine learning approaches to overcome the difficulties mentioned above. In this paper we are evaluating the performance of various classical algorithms such as SVM, Random Forest and Naive Bayes to detect attacks on network using IDS datasets such KDD, NSL but this classical algorithms unable to predict dynamic (if attacker introduce new attacks with changes in attack parameter) attacks and need to be trained in advance to detect such attacks to overcome from this problem we are evaluating performance of Deep Neural Network (DNN) algorithm with dynamic attack signatures and detection accuracy of DNN shown to be better compare to all classical algorithms. Deep learning is a field of artificial intelligence that has an impressive results and is now a hotspot for study. This survey presents an IDS taxonomy that classifies and summarizes machine learning- and deep learning-based IDS literature using data objects as the primary dimension.