Using t-SNE for characterizing glitches in LIGO detectors

TL;DR

This paper introduces a method combining Q-transform and t-SNE to analyze and classify glitches in LIGO detector data, aiding in distinguishing noise transients from true gravitational wave signals.

Contribution

The study presents a novel approach integrating t-SNE with Q-transform data for effective glitch characterization and tracking in gravitational wave data analysis.

Findings

Effective glitch classification achieved

Parameter influence on classification analyzed

Outlier transients tracked over a week

Abstract

Glitches are non-Gaussian noise transients originating from environmental and instrumental sources that contaminate data from gravitational wave detectors. Some glitches can even mimic gravitational wave signals from compact object mergers, which are the primary targets of terrestrial observatories. In this study, we present a method to analyze noise transients from the LIGO observatories using Q-transform information combined with t-Distributed Stochastic Neighbor Embedding (t-SNE). We implement classification techniques, examine the influence of parameters on glitch classification, and conduct a week-long daily analysis to track outlier transients over time.