Uncertainty Quantification in Anomaly Detection with Cross-Conformal $p$-Values

arXiv:2402.16388v1 Announce Type: cross
Abstract: Given the growing significance of reliable, trustworthy, and explainable machine learning, the requirement of uncertainty quantification for anomaly detection systems has become increasingly important. In this context, effectively controlling Type I error rates ($\alpha$) without compromising the statistical power ($1-\beta$) of these systems can build trust and reduce costs related to false discoveries, particularly when follow-up procedures are expensive. Leveraging the principles of conformal prediction emerges as a promising approach for providing respective statistical guarantees …

abstract alpha anomaly anomaly detection arxiv become beta build context cs.lg detection error explainable machine learning machine machine learning power quantification significance statistical stat.ml systems trust trustworthy type uncertainty values