Speaker
Marie Hein
(RWTH Aachen University)
Description
Weakly supervised methods have emerged as a powerful tool for model agnostic anomaly detection at the LHC. While remarkable performance has been achieved for specific sets of high-level input features, a further exploration of different input feature sets of various types will lead to more model agnostic and better performing setups. In this talk, we explore low-level features as well as some high-level features, including subjettiness based feature sets and energy flow polynomials.
Primary authors
Alexander Mück
Joep Geuskens
Lukas Lang
Marie Hein
(RWTH Aachen University)
Michael Krämer
(RWTH Aachen University)
Radha Mastandrea
