We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

hep-ph

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

High Energy Physics - Phenomenology

Title: A simple guide from Machine Learning outputs to statistical criteria

Abstract: In this paper we propose ways to incorporate Machine Learning training outputs into a study of statistical significance. We describe these methods in supervised classification tasks using a CNN and a DNN output, and unsupervised learning based on a VAE. As use cases, we consider two physical situations where Machine Learning are often used: high-$p_T$ hadronic activity, and boosted Higgs in association with a massive vector boson.
Comments: 32 pages, 15 figures
Subjects: High Energy Physics - Phenomenology (hep-ph)
Cite as: arXiv:2203.03669 [hep-ph]
  (or arXiv:2203.03669v1 [hep-ph] for this version)

Submission history

From: Veronica Sanz [view email]
[v1] Mon, 7 Mar 2022 19:07:06 GMT (1404kb,D)

Link back to: arXiv, form interface, contact.