Deep-Learning Jets with Uncertainties and More
Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:LMOQSD7Trecord.jsonopen to challenge →
read the original abstract
Bayesian neural networks allow us to keep track of uncertainties, for example in top tagging, by learning a tagger output together with an error band. We illustrate the main features of Bayesian versions of established deep-learning taggers. We show how they capture statistical uncertainties from finite training samples, systematics related to the jet energy scale, and stability issues through pile-up. Altogether, Bayesian networks offer many new handles to understand and control deep learning at the LHC without introducing a visible prior effect and without compromising the network performance.
This paper has not been read by Pith yet.
Forward citations
Cited by 4 Pith papers
-
Unbinned extraction of $\gamma$ from $B\to DK$ with normalizing flows
Normalizing flows trained on D decay data create a continuous unbinned model of Dalitz plot amplitudes, allowing extraction of gamma from B decay data with successful recovery of injected values in Monte Carlo tests.
-
Uncovering Hidden Systematics in Neural Network Models for High Energy Physics
Neural networks for HEP tasks can be fooled at significant rates by subtle perturbations inside uncertainty envelopes, revealing hidden systematics not captured by conventional methods.
-
Optimizing The Cut And Count Method In Phenomenological Studies
An iterative ranking-based optimization of cut-and-count using MadAnalysis5 enhances signal-background separation and discovery reach for singly charged Higgs in the Two Higgs Doublet Model.
-
Amplitude Uncertainties Everywhere All at Once
Compares ensemble, Bayesian, and evidential regression approaches for uncertainty quantification in amplitude surrogates and shows they detect localized training data issues.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.