Deep Learning Based Intrusion Detection System using Gated Recurrent Neural Network with Sigmoid Activation Function
N. Nathiya, Aruna S K, S J Nikalyaa, C. Kavya · 2024
In recent years, Intrusion Detection System (IDS) significant progress has been made in the area of computer networks and the internet. This brings up significant security concerns. Since the number of cases of piracy has increased and numerous modern systems have been hacked, it is now crucial to develop information security technologies that can identify new attacks. An Intrusion Detection System (IDS) that uses Machine Learning (ML) and Deep Learning (DL) to find network anomalies is one of the most important information security technologies. Despite decades of development, existing IDSs still face difficulties in reducing false alarm rates, increasing detection accuracy, and identifying unknown attacks. The primary goal of this project is to use DL-based, advanced intrusion detection systems with high network performance to identify unknown attacks in order to combat these issues. We implement a Gated Recurrent Neural Network with Sigmoid Activation Function (GRN2SAF) method for intrusion detection. Initially, we collected the intrusion detection dataset from the online repository. Next, we pre-process the dataset to normalize the dataset using Z-score normalization method. Then, the Chi-square Mutual Feature Selection (C2MFS) algorithm is used for dimensionality reduction. Finally, the proposed GRN2SAF algorithm is used to categorize the attacks in the network. The experimental results show that the proposed approach based on the GRN2SAF has high accuracy in intrusion detection.