A Robust Adaptive Intrusion Detection System using Hybrid Deep Learning
Parthiban Aravamudhan, Thirunavukkarasu Kanimozhi · 2022 International Conference on Computer Communication and Informatics (ICCCI) · 2022
Network attacks have increased dramatically over the last 25 years. Network intrusions are increasing enormously due to rise in communication and technology paradigm. With the advancement in the field of computer network technology, new security issues in the network emerge on an everyday basis creating more and tougher to ignore them. Attackers create new types of attack every day. In-order to detect these types of attack, identification of attack is more important. Traditional firewalls alone may not be sufficient to detect the modern attacks. This work describes a hybrid intrusion recognition system centered on deep learning that is capable to detect efficiently and precisely the network threats and intrusions. This method is designed by combining deep CNN (Convolutional Neural Networks), RCNN (Region Based Convolution Neural Network) and GBR (Gradient Boost Regression) which has the capability of detecting different network intrusions. The dataset used in this work is considered from the NIDS dataset V.10 2017 from the Kaggle website. The outcomes demonstrate that the recommended algorithm is enhanced compared to the previous work. The proposed method have increased accuracy and less time consumption compared with prior research.