Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 16: GRID Computing / Experimentelle Methoden I

T 16.3: Talk

Monday, March 19, 2018, 16:30–16:45, Z6 - SR 2.006

Systematic Uncertainties In Machine Learning Based Analyses — Raphael Friese, Günter Quast, Roger Wolf, Sebastian Wozniewski, and •Stefan Wunsch — Institut für Experimentelle Teilchenphysik, Karlsruher Institut für Technologie

During the last years, the field of machine learning became more and more important, also in high-level data analyses in particle physics. In the next years the published results of the LHC experiments will more and more rely on these methods. An essential part of such analyses is the proper estimation of the contributing uncertainties. On the other hand, up to date, profound studies of the effects of systematic uncertainties in the usage of modern machine learning methods are still missing. This talk proposes possible approaches to identify and propagate systematic uncertainties to the final result in machine learning based analyses.