LIGO Document P2300230-v2
- Despite achieving sensitivities capable of detecting the extremely small amplitude of gravitational waves (GWs), LIGO and Virgo detector data contain frequent bursts of non-Gaussian transient noise, commonly known as ‘glitches’. Glitches come in various time-frequency morphologies, and they are particularly challenging when they mimic the form of real GWs. Given the higher expected event rate in the current observing run (O4), LIGO-Virgo GW event candidate validation requires increased levels of automation. Gravity Spy, a machine learning tool that successfully classified common types of LIGO and Virgo glitches in previous observing runs, has the potential to be restructured as a signal-vs-glitch classifier to distinguish between glitches and GW signals accurately. A signal-vs-glitch classifier used for automation must be robust and compatible with a broad array of background noise, new sources of glitches, and the likely occurrence of overlapping glitches and GWs. This dissertation presents GSpyNetTree, the Gravity Spy Convolutional Neural Network Decision Tree: a multi-CNN multi-label classifier using CNNs in a decision tree sorted via total GW candidate mass. Integrated into the LIGO-Virgo Data Quality Report, GSpyNetTree is one of the essential tools in assessing the necessity of glitch mitigation in O4. This thesis presents the development, building process, and results of GSpyNetTree: from its origin as a multi-class classifier based on Gravity Spy, to its current O4 status as a multi-label classifier. Finally, the performance of GSpyNetTree identifying data quality issues in the public O4 GW candidates published in GraceDB is evaluated, and new ways to improve the tool’s classifications are suggested.
DCC Version 3.4.2, contact