May 16 – 18, 2025
College of Management, National Formosa University 國立虎尾科技大學第三校區文理暨管理大樓
Asia/Taipei timezone
The ASROC2025 Program is Now Available!

Investigating Relations Between Denoised GW Structures and the Performance of ML-based GW Analysis

May 17, 2025, 5:00 PM
15m
Room CMA0103 (College of Management, National Formosa University)

Room CMA0103

College of Management, National Formosa University

Speaker

Yu-Chiung Lin (National Tsing Hua University)

Description

We investigated the relationship between the GW denoiser’s signal recovery and the prediction of the ML-based CBC analysis by training BBH, NSBH, and BNS binary classifiers on denoised strain data. We found that the GW detector can make confident detections when the signal recovery, measured by overlap, is larger than 0.2 for BBH, NSBH, and 0.1 for BNS. The results are consistent with our statements in the previous work. We also trained several regressors to investigate the relationship between the signal recovery and the parameter estimation results. Our work provides new insight into evaluating and developing future GW denoisers.

Section Cosmology

Primary author

Yu-Chiung Lin (National Tsing Hua University)

Presentation materials