Detecting anomalies is one fundamental aspect of a safety-critical software
system, however, it remains a long-standing problem. Numerous branches of works
have been proposed to alleviate the complication and have demonstrated their
efficiencies. In particular, self-supervised learning