Leave-One-Out-, Bootstrap- and Cross-Conformal Anomaly Detectors
This work addresses the need for reliable anomaly detection systems in low-data regimes, offering incremental advancements over existing conformal prediction methods.
The paper tackled the problem of uncertainty quantification in anomaly detection by proposing leave-one-out, bootstrap, and cross-conformal methods to control Type I error rates without compromising statistical power, demonstrating that these methods achieve a practical compromise between statistical and computational efficiency with quantified improvements across various classifiers and datasets.
The requirement of uncertainty quantification for anomaly detection systems has become increasingly important. In this context, effectively controlling Type I error rates ($α$) without compromising the statistical power ($1-β$) of these systems can build trust and reduce costs related to false discoveries. The field of conformal anomaly detection emerges as a promising approach for providing respective statistical guarantees by model calibration. However, the dependency on calibration data poses practical limitations - especially within low-data regimes. In this work, we formally define and evaluate leave-one-out-, bootstrap-, and cross-conformal methods for anomaly detection, incrementing on methods from the field of conformal prediction. Looking beyond the classical inductive conformal anomaly detection, we demonstrate that derived methods for calculating resampling-conformal $p$-values strike a practical compromise between statistical efficiency (full-conformal) and computational efficiency (split-conformal) as they make more efficient use of available data. We validate derived methods and quantify their improvements for a range of one-class classifiers and datasets.