Glitch Classification for Gravitational Wave Interferometry Using Machine Learning

Gravitational waves are disturbances in the curvature of space-time generated by accelerated masses that propagate as waves outward from their source at the speed of light. A deep and analytical study of instrumental responses in the ecosystem of environmental noise is required for their optimal characterization. Hence of pertinent interest in this paper is the study of anomalous non Gaussian noise transients called ‘Glitches'. These are short duration instrumental artifacts that can pollute the classification space. This paper tries to analyze appropriate machine learning techniques using a novel ‘divide and conquer’ based pipeline to classify glitches so that these can be subtracted from gravitational wave detector data for efficient detection and classification of gravitational wave ripples.