NeurIPS
·
2022
Abstract
This paper proposes generalizations of CWOLA and SALAD, which exploit multiple reference datasets to improve performance in resonant anomaly detection, and provides finite-sample guarantees to go beyond existing asymptotic analyses.



