Heavier tail likelihoods for robustness against data outliers; Applications to the analysis of Gravitational Wave data

5 Apr 2023, 17:00
20m
Main Auditorium (Conference Centre “Karolos Papoulias”)

Main Auditorium

Conference Centre “Karolos Papoulias”

Speaker

Argyro Sasli (Aristotle University of Thessaloniki)

Description

In recent years we have been witnesses of the blooming of Gravitational Wave Astronomy. In the near future, with the more advanced, as well as the space-based detectors coming online, it is expected to detect events originating from compact binary objects at much higher rates. One of the future data analysis challenges, is performing robust statistical analyses in the presence of detector noise transients, or non-stationarities, which might originate from astrophysical sources. In this work, we propose a heavier-tailed likelihood filter based on the Hyperbolic distribution. We discuss the advantages of this formulation, after applying it to examples taken from synthetic datasets.

Primary author

Argyro Sasli (Aristotle University of Thessaloniki)

Co-authors

Dr Nikolaos Karnesis (Aristotle University of Thessaloniki) Nikolaos Stergioulas (Aristotle University of Thessaloniki)

Presentation materials