Paper Title
HYBRID TRANSFORMER-BASED INTRUSION DETECTION USING LOGBERT AND ANOMALY TRANSFORMER ON NETWORK TRAFFIC DATA

Abstract
Network intrusion detection is of critical importance for the protection of modern communication systems against the increasing sophistication of cyberattacks. It suggests a hybrid deep learning-based intrusion detection system that uses the combined potential of Log-BERT and Anomaly Transformer using a feature fusion method to attain improved intrusion detection and classification accuracy. The proposed system utilizes the ability of the Log-BERT model to learn the contextual information from the network traffic data and the ability of the Anomaly Transformer to identify the time anomalies in the network traffic patterns. The proposed system is validated using the UNSW-NB15 dataset, which consists of various types of attacks including DoS, Exploits, Fuzzers, Reconnaissance, etc. Furthermore, a lightweight web-based monitoring system is proposed for the proposed intrusion detection system using the Streamlit library. Keywords - Intrusion Detection System, Log-BERT, Anomaly Transformer, Hybrid Fusion Model, Deep Learning, Network Security, Streamlit.