Accurate and robust methods for direct background estimation in resonant anomaly detection

TL;DR

This paper demonstrates that resonant anomaly detection methods can be used to directly estimate background expectations without fitting, improving sensitivity and avoiding background sculpting in bump hunt searches.

Contribution

It introduces a novel approach to directly estimate background in resonant anomaly detection, bypassing traditional fitting procedures and enhancing detection capabilities.

Findings

Background estimation via anomaly detection is effective.

Avoids background sculpting by not fitting invariant mass.

Improves significance at high background rejection rates.

Abstract

Resonant anomaly detection methods have great potential for enhancing the sensitivity of traditional bump hunt searches. A key component of these methods is a high quality background template used to produce an anomaly score. Using the LHC Olympics R&D dataset, we demonstrate that this background template can also be repurposed to directly estimate the background expectation in a simple cut and count setup. In contrast to a traditional bump hunt, no fit to the invariant mass distribution is needed, thereby avoiding the potential problem of background sculpting. Furthermore, direct background estimation allows working with large background rejection rates, where resonant anomaly detection methods typically show their greatest improvement in significance.