Uncertainty Aware Deep Learning for Particle Accelerators

TL;DR

This paper explores uncertainty-aware deep learning models, specifically Deep Gaussian Process Approximation, to improve prediction reliability in particle accelerators by detecting and managing uncertain or out-of-distribution inputs.

Contribution

It introduces the application of DGPA methods for uncertainty estimation in accelerator systems, enhancing prediction confidence and robustness.

Findings

DGPA effectively detects uncertain predictions in accelerator data.

Uncertainty-aware models improve reliability of beam predictions.

Application to SNS and FNAL accelerators demonstrates practical benefits.

Abstract

Standard deep learning models for classification and regression applications are ideal for capturing complex system dynamics. However, their predictions can be arbitrarily inaccurate when the input samples are not similar to the training data. Implementation of distance aware uncertainty estimation can be used to detect these scenarios and provide a level of confidence associated with their predictions. In this paper, we present results from using Deep Gaussian Process Approximation (DGPA) methods for errant beam prediction at Spallation Neutron Source (SNS) accelerator (classification) and we provide an uncertainty aware surrogate model for the Fermi National Accelerator Lab (FNAL) Booster Accelerator Complex (regression).