M. Thangaraj

Work place: Department of Computer Science, Madurai Kamaraj University, Madurai, 625021, India

E-mail: thangaraj-dcs@mkuniversity.org

Website:

Research Interests: Big Data Analytics

Biography

Dr. M. Thangaraj, Department of Computer Science, School of Information Technology, Madurai Kamaraj University, India. M Thangaraj received his post-graduate degree in Computer Science from Alagappa University, Karaikudi, M.Tech. degree in Computer Science from Pondicherry University and Ph.D. degree in Computer Science from Madurai Kamaraj University, Madurai, TN, South India in 2006. He is now the professor of Computer Science Department at Madurai Kamaraj University. He is an active researcher in Big Data Analytics, Social Media Analytics, Wireless Sensor Networks and has published more than 150 papers in Journals and Conference Proceedings. 

Author Articles
Intelligent Autoencoder with LSTM based Intrusion Detection and Recommender System

By V. G. Aishvarya Shree M. Thangaraj

DOI: https://doi.org/10.5815/ijcnis.2025.05.04, Pub. Date: 8 Oct. 2025

With the swift growth of digital networks and information in both public and private sectors, it is essential to deal with the considerable threat that network attacks pose to data integrity and confidentiality. Consequently, there is a pressing requirement for the establishment of effective mechanisms to detect and provide recommendations for addressing intrusion attacks. In this paper, we propose a semantic-based intrusion detection system that aims to improve performance by incorporating semantic representations consisting of feature groups and their associated weights, leading to the creation of a weighted knowledge graph. The weights of the features are determined using sparse autoencoders. From these weights, the most significant features are normalized to a specific range. This approach comprises a combination of a Deep Auto Encoder (AE) and Long Short-Term Memory (LSTM) networks for intrusion detection. Furthermore, the ensemble method of Extreme Gradient Boosting (XGBoost) is used to identify and recommend high-probability attack scenarios. The dataset used to evaluate is the CSE-CIC-IDS dataset. Performance metrics such as accuracy, precision, recall, false positive rate, receiver operating characteristic metrics, loss, and error rate are used to measure the performance, and the results show the approach demonstrates substantial improvements in detection accuracy, minimizing false positives, enhancing reliability, and outperforming existing models. The combination of semantic knowledge, deep learning, and ensemble learning ensures a proactive and adaptive cybersecurity framework.

[...] Read more.
Other Articles