Deep-learning jets with uncertainties and more
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 fi...
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Dokumenttyp: | Article (Journal) |
| Sprache: | Englisch |
| Veröffentlicht: |
16-01-2020
|
| In: |
SciPost physics
Year: 2020, Jahrgang: 8, Heft: 1, Pages: 1-25 |
| ISSN: | 2542-4653 |
| DOI: | 10.21468/SciPostPhys.8.1.006 |
| Online-Zugang: | Resolving-System, Volltext: https://doi.org/10.21468/SciPostPhys.8.1.006 Verlag: https://scipost.org/10.21468/SciPostPhys.8.1.006 
|
| Verfasserangaben: | Sven Bollweg, Manuel Haussmann, Gregor Kasieczka, Michel Luchmann, Tilman Plehn and Jennifer Thompson |
| Zusammenfassung: | 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. |
|---|---|
| Beschreibung: | Gesehen am 27.02.2020 |
| Beschreibung: | Online Resource |
| ISSN: | 2542-4653 |
| DOI: | 10.21468/SciPostPhys.8.1.006 |