anomaly detection, where data instances are discovered containing feature
patterns different from the majority, plays a fundamental role in various
applications. However, it is challenging for existing methods to handle the
scenarios where the instances are systems whose characteristic