Dortmund 2021 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
T: Fachverband Teilchenphysik
T 21: Data analysis, Information technology I
T 21.2: Vortrag
Montag, 15. März 2021, 16:15–16:30, Tu
Searching for new physics with anomaly detection — •Manuel Sommerhalder1, Tobias Lösche1, Gregor Kasieczka1, David Shih2, and Anna Hallin2 — 1Universität Hamburg — 2Rutgers University
Most analyses looking for new physics, such as beyond Standard Model searches at the LHC, rely on a specific signal hypothesis for selecting relevant data points. However, recently there is an increased interest in developing model-independent selection criteria with the aim of generically gaining sensitivity to unthought-of phenomena. A potential solution to this problem lies in anomaly detection, which consists of various techniques that quantify how anomalous each data point (or group of data points) is with respect to the entire data set, thus yielding an additional measure to identify signal candidates.
We present our recent developments in using anomaly detection to search for new physics via the concrete example of resonance searches at the LHC. The primary focus is our application of Anomaly Detection with Density Estimation (ANODE) using normalizing flows. Building up on the previous state-of-the-art performance achieved by ANODE, we highlight further advancements to improve its performance and useability.