<p>Identification of malicious attacks in network traffic is one of many important big data problems that have been approached from different directions, including machine learning. In this paper, it is used as an example for investigating the applicability of quantum machine learning to such problems. Particular focus is kept on the NISQ era, in which available computer sizes are typically fairly small. Instead of applying typical cutting and knitting techniques with their associated overhead cost, we use the trainable output layer inherent in many hybrid QML approaches to re-combine the results from a collection of smaller QVCs executed on different machines. It is compared to classical and established quantum approaches on four real-world datasets.</p>