Joseph Moore

This paper evaluates unsupervised temporal learning models, specifically recurrent autoencoders, for real-time anomaly detection in vulnerable IEC-61850 GOOSE networks, demonstrating that the GRU model offers the best balance of accuracy and low latency.