CURTAINs for your Sliding Window: Constructing Unobserved Regions by Transforming Adjacent Intervals

TL;DR

CURTAINs is a neural network-based method that constructs background templates for LHC data analysis, improving anomaly detection sensitivity by transforming side-band data into the signal region in a model-independent way.

Contribution

This paper introduces CURTAINs, a novel invertible neural network technique for data-driven background estimation in bump hunts at the LHC, capable of training on smaller data ranges.

Findings

Matches performance of leading approaches in anomaly detection

Requires smaller invariant mass ranges for training

Fully data-driven and model-independent

Abstract

We propose a new model independent technique for constructing background data templates for use in searches for new physics processes at the LHC. This method, called CURTAINs, uses invertible neural networks to parametrise the distribution of side band data as a function of the resonant observable. The network learns a transformation to map any data point from its value of the resonant observable to another chosen value. Using CURTAINs, a template for the background data in the signal window is constructed by mapping the data from the side-bands into the signal region. We perform anomaly detection using the CURTAINs background template to enhance the sensitivity to new physics in a bump hunt. We demonstrate its performance in a sliding window search across a wide range of mass values. Using the LHC Olympics dataset, we demonstrate that CURTAINs matches the performance of other leading approaches which aim to improve the sensitivity of bump hunts, can be trained on a much smaller range of the invariant mass, and is fully data driven.