May 15 – 17, 2026
College of Hakka Studies at NYCU, Zhubei, Hsinchu County 國立陽明交通大學客家學院(竹北六家校區)
Asia/Taipei timezone

Rapid Gravitaional Wave Detector Background Estimation using Self-supervised Learning

May 16, 2026, 4:45 PM
15m
International Conference Hall, College of Hakka Studies, NYCU 國立陽明交通大學客家文化學院國際會議廳

International Conference Hall, College of Hakka Studies, NYCU 國立陽明交通大學客家文化學院國際會議廳

Speaker

Yu-Chiung Lin (National Tsing Hua University)

Description

Estimating the background noise power is essential for gravitational wave (GW) analysis. Typically, people compute the power spectral density (PSD) of the noise background using the Welch method, which is a sliding average over a long period of data, usually spanning tens to thousands of seconds. However, the detector strain data is non-stationary and sometimes glitchy due to the detector's bad status. This can make the estimated noise background not representative of the data we are analyzing. In such cases, people can either manually select a relatively clean segment or use the Monte Carlo method to estimate the noise PSD. Still, the former requires additional manpower for the search, and the latter requires Monte Carlo sampling and is computationally expensive. In this work, we developed a machine-learning method to estimate the PSD from the data we analyzed, without accounting for prior data quality.

Participate the oral/poster presentation award competition No

Authors

Yu-Chiung Lin (National Tsing Hua University) Albert Kong

Presentation materials

There are no materials yet.