Enhanced Network Security using LSTM-Based Autoencoder Models

Abstract

An intrusion detection system (IDS) is essential to protect the network from cyber threats. This paper uses network traffic statistics to develop an LSTM-based auto-encoder model for effective intrusion detection. Time dependence is captured by the model using the long short-term memory (LSTM) network, which can discriminate between normal and pathological activity. High accuracy, recall, and F1 scores are demonstrated in experiments from the NSL-KDD dataset, demonstrating the model's resilience in identifying network intrusions. The superiority of LSTM-based approaches is confirmed by comparison with traditional methods. Through the use of deep learning techniques, this research advances IDSs by highlighting their adaptability and scalability in dynamic network environments. The proposed model achieved the highest accuracy, precision, recall, and F1-score values of 99.9%, 99.9%, 99.9%, 99.7%, and 99.8%, respectively.