Source-linked AI summary

Anomaly Detection for Resonant New Physics with Machine Learning

Jack H. Collins, Kiel Howe, Benjamin Nachman

arXiv:1805.02664v3hep-phhep-ex

TL;DR

The Standard Model is not a complete theory of nature, motivating searches for physics beyond it at the LHC. The paper presents a classifier-based, model-agnostic bump-hunt technique using auxiliary event information and demonstrates a 7σ local significance in a simulated dijet resonance search.

  • Problem

    The Standard Model is not a complete theory of nature, motivating continued searches for physics beyond the Standard Model at the LHC.

  • Method

    The method trains a classifier on auxiliary characteristics in signal and sideband regions with similar background properties, then uses its output to select signal-like events across the resonant-variable distribution.

  • Results

    7σ local significance emerged at the 0.2% classifier threshold, while no significant bumps were created in the signal-free test.

  • Takeaways & Limitations

    The technique extends bump hunting to resonant signals with limited prior knowledge and was demonstrated on a simulated all-hadronic LHC resonance search.

  • Takeaways & Limitations

    Performance can be limited by statistics, technical difficulties in training, and differences between background characteristics in signal and sideband regions.

Abstract

from arXiv · show

Despite extensive theoretical motivation for physics beyond the Standard Model (BSM) of particle physics, searches at the Large Hadron Collider (LHC) have found no significant evidence for BSM physics. Therefore, it is essential to broaden the sensitivity of the search program to include unexpected scenarios. We present a new model-agnostic anomaly detection technique that naturally benefits from modern machine learning algorithms. The only requirement on the signal for this new procedure is that it is localized in at least one known direction in phase space. Any other directions of phase space that are uncorrelated with the localized one can be used to search for unexpected features. This new method is applied to the dijet resonance search to show that it can turn a modest 2 sigma excess into a 7 sigma excess for a model with an intermediate BSM particle that is not currently targeted by a dedicated search.

Loading 1805.02664v3…