Accurate Surrogate Amplitudes with Calibrated Uncertainties

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bahl, Henning, Elmer, Nina, Favaro, Luigi, Haußmann, Manuel, Plehn, Tilman, Winterhalder, Ramon
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908613132419072
author Bahl, Henning
Elmer, Nina
Favaro, Luigi
Haußmann, Manuel
Plehn, Tilman
Winterhalder, Ramon
author_facet Bahl, Henning
Elmer, Nina
Favaro, Luigi
Haußmann, Manuel
Plehn, Tilman
Winterhalder, Ramon
contents Neural networks for LHC physics have to be accurate, reliable, and controlled. Using neural surrogates for the prediction of loop amplitudes as a use case, we first show how activation functions are systematically tested with Kolmogorov-Arnold Networks. Then, we train neural surrogates to simultaneously predict the target amplitude and an uncertainty for the prediction. We disentangle systematic uncertainties, learned by a well-defined likelihood loss, from statistical uncertainties, which require the introduction of Bayesian neural networks or repulsive ensembles. We test the coverage of the learned uncertainties using pull distributions to quantify the calibration of cutting-edge neural surrogates.
format Preprint
id arxiv_https___arxiv_org_abs_2412_12069
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Accurate Surrogate Amplitudes with Calibrated Uncertainties
Bahl, Henning
Elmer, Nina
Favaro, Luigi
Haußmann, Manuel
Plehn, Tilman
Winterhalder, Ramon
High Energy Physics - Phenomenology
Neural networks for LHC physics have to be accurate, reliable, and controlled. Using neural surrogates for the prediction of loop amplitudes as a use case, we first show how activation functions are systematically tested with Kolmogorov-Arnold Networks. Then, we train neural surrogates to simultaneously predict the target amplitude and an uncertainty for the prediction. We disentangle systematic uncertainties, learned by a well-defined likelihood loss, from statistical uncertainties, which require the introduction of Bayesian neural networks or repulsive ensembles. We test the coverage of the learned uncertainties using pull distributions to quantify the calibration of cutting-edge neural surrogates.
title Accurate Surrogate Amplitudes with Calibrated Uncertainties
topic High Energy Physics - Phenomenology
url https://arxiv.org/abs/2412.12069