Deep learning approach for ids using dnn for network anomaly detection

Abstract

With the astonishing development of the Internet and its applications in the last decade, cyberattacks are changing quickly, and the necessity of protection for communication network has improved tremendously. As the primary defense, the intrusion detection system plays a crucial role in making sure the network security. Key to intrusion detection system is actually to determine a variety of attacks effectively as well as to adjust to a constantly changing threat scenario. DNN or Deep Neural Network on NSL-KDD dataset for effective detection of an attack. Firstly, the dataset was preprocessed and normalized and then fed to the DNN algorithm to create a model. For testing purpose, entire dataset of NSL-KDD was used. Finally, to analyze the accuracy and precision of the DNN model, we use accuracy and precision matrices. The proposed DNN-based strategy enhances network anomaly detection and opens new analysis gateway for intrusion detection systems.