Loss function to optimise signal significance in particle physics

TL;DR

This paper introduces a new loss function tailored for particle physics that directly optimizes the significance metric, leading to improved signal detection efficiency in collider searches.

Contribution

The paper presents a novel surrogate loss function specifically designed to optimize signal significance, demonstrating its effectiveness over traditional cross-entropy loss in particle physics classification tasks.

Findings

Higher signal efficiency with the new loss compared to cross-entropy

Decision boundaries adapt to process cross sections

Potential to enhance collider search sensitivity

Abstract

We construct a surrogate loss to directly optimise the significance metric used in particle physics. We evaluate our loss function for a simple event classification task using a linear model and show that it produces decision boundaries that change according to the cross sections of the processes involved. We find that the models trained with the new loss have higher signal efficiency for similar values of estimated signal significance compared to ones trained with a cross-entropy loss, showing promise to improve sensitivity of particle physics searches at colliders.