K. K. Zharmakhan

HYBRID GAT-AUTOENCODER-BILSTM-ATTENTION ARCHITECTURE FOR MULTI CLASS CYBERATTACK DETECTION IN HETEROGENEOUS IOT NETWORKS

A novel hybrid deep learning architecture combining a Graph Attention Network (GAT), an Autoencoder (AE), a Bidirectional Long Short-Term Memory network (BiLSTM) and a Multi Head Self Attention (MHSA) is proposed into a single intrusion detection framework. The suggested traffic model integrates graph-based traffic rep...