This study presents the correlational paraconsistent machine (CPM), a tool for anomaly detection that incorporates unsupervised models for traffic characterization and principles of paraconsistency, to inspect irregularities at the network traffic flow level.