Enhancing Gravitational Wave Detection Using Machine Learning and Ambient Noise Suppression

In this thesis, we present two separate deep learning pipelines for the detection and parameter estimation of astrophysical gravitational waves. In part one, we present a convolutional neural network, designed in the auto-encoder configuration that can detect and denoise gravitational waves from merging black hole binaries, orders of magnitude faster than the conventional matched-filtering based detection that is currently employed at advanced-LIGO (aLIGO) and the LVK (LIGO-VIRGO-KAGRA) network in general. The Neural-Network architecture is such that it learns from the sparse representation of data in the time-frequency domain and constructs a non-linear mapping function that maps this representation into two separate masks for signal and noise, facilitating the separation of the two, from raw data. This approach is the first of its kind to apply machine learning based gravitational wave detection/denoising from binary mergers in the 2D representation of gravitational wave data. We applied our formalism to the first gravitational wave event detected, GW150914, successfully recovering the signal at all three phases of coalescence at both aLIGO detectors. This method is further tested on the gravitational wave data from the second observing run (O2) of aLIGO, reproducing all binary black hole mergers detected in O2 at both detectors. The Neural-Net seems to have uncovered a pattern of ‘ringing’ after the ringdown phase of the coalescence, which is not a feature that is present in the conventional binary merger templates. This method can also interpolate and extrapolate between modeled templates and explore gravitational waves that are unmodeled and hence not present in the template bank of signals used in the matched-filtering detection pipelines. Faster and efficient detection schemes, such as this method, will be instrumental as ground based detectors reach their design sensitivity, likely to result in several hundreds of potential detections in a few months of observing runs. In part two, we present another deep learning based architecture using convolutional neural network to estimate the intrinsic parameter of the black holes from the observed raw gravi- tational wave data. This framework has the capability to estimate parameters of coalescing binaries, orders of magnitude faster than the conventional Bayesian analysis based param- eter inference employed at the LVK network. We also attempt to estimate the individual spin components of both black holes, which is often not possible using Bayesian inference. This machine learning based parameter inference scheme, to the best of our knowledge is the first of its kind that can make direct estimation of individual spin components of both black holes from raw detector data. In part three, we present an overview of Ambient Seismic Noise (ASN) and the importance of its mitigation in enhancing ground based gravitational wave detection. ASN is one of the biggest noise contributors below 20Hz in ground based gravitational wave detectors. Seismic Newtonian noise arising from gravity gradients created by seismic waves will become the limiting noise source at low frequencies for second generation gravitational wave detectors. Low frequency noise suppression will especially enhance gravitational wave detection from heavier coalescing astrophysical systems. In this part, we also present results from a series of seismic weight-drop experiments performed at the gravitational wave test detector site at Western Australia. This analysis would be one of many seismic experiments that can help characterize and study the detector location for better seismic isolation and noise suppression.