Searching for Anomalies with Foundation Models

TL;DR

This paper investigates the use of large foundation models for anomaly detection in high-energy physics data, revealing challenges in background estimation and encouraging further analysis of potential signals.

Contribution

It provides a comprehensive analysis of foundation models applied to CMS data, highlighting their potential and limitations in anomaly detection tasks.

Findings

Background estimation fits well in validation regions

Model struggles to accurately predict the signal region

Further scrutiny of anomalies is needed

Abstract

Foundation models have the potential to extend the discovery reach for anomaly detection searches. When studying the large OmniLearned foundation model on data from the CMS experiment, unexpected behavior was observed in a mass sideband. The purpose of this paper is to perform a full analysis, including a complete background estimate, on the phase space picked out by the large model. We find that the background estimation describes the data well in validation regions, but is unable to accurately model the signal region. We invite further scrutiny of these events and our methods.