The scientists developed a smart anomaly detector using a type of artificial intelligence called Reinforcement Learning. This detector can spot unusual patterns in data without needing specific rules or thresholds. It learns and improves over time, getting better at detecting anomalies as it gains experience. In tests, the detector showed promising results in finding anomalies in network data.