Investigating Deep Neural Networks for Gravitational Wave Detection in Advanced LIGO Data

Since the first detection of a Gravitational Wave (GW) in September 2015 at the Laser Interferometer Gravitational-Wave Observatory (LIGO), it was unclear if the Einstein's Theory (E=MC2) was true and if GWs existed. A huge investment in highly advanced optical and electronic equipment was taken to build huge detectors for the possibility to record something which was, until then, only a theory. The LIGO detectors are so special because they can detect a change in length by one ten-thousandth the width of a proton. These constructions are huge and need big financial and technological investments to achieve the precisions. We evaluated Deep Neural Networks (DNNs), Convolutional Neural Networks (CNNs), Temporal Convolutional Networks (TCNs), Gated Recurrent Units (GRUs) and Long Short-Term Memory (LSTM) on the performance of detecting GWs in a very noisy signal from the LIGO detectors. Higher precision in GW detection will help the LIGO team to more quickly classify GW events in recorded data streams and help to reduce the time span between measurement and detection of a GW. The reduced time is crucial for learning more about GWs because the telescopes and satellites (for example the Hubble Space Telescope) need time to change direction and locate exactly the root of the GW. Every second saved in detection of the GW can lead to more seconds of recorded data from the observatories. During our research we found out that we could achieve good results with the Temporal Convolutional Networks (TCNs), which had in our implementation improved precision compared to the original matched-filtering and the CNN of D. George and E. A. Huerta [1]. The time for detection is thereby a mixture between the time for the detector from the trained TCN or CNN and the time for the scientist of LIGO to double check the wave data and evaluate the situation with advanced signal processing. The accuracy of detection between the TCN and CNN is drastically different because a TCN is 14% more accurate than the one from the CNN, which leads to less false detections for the researchers.

[1]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[2]  Yoshua Bengio,et al.  Gradient-based learning applied to document recognition , 1998, Proc. IEEE.

[3]  Erin Kara,et al.  TOWARD EARLY-WARNING DETECTION OF GRAVITATIONAL WAVES FROM COMPACT BINARY COALESCENCE , 2011, 1107.2665.

[4]  Yoshua Bengio,et al.  Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation , 2014, EMNLP.

[5]  Neil J. Cornish,et al.  Bayeswave: Bayesian inference for gravitational wave bursts and instrument glitches , 2014, 1410.3835.

[6]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[7]  Matthew West,et al.  The PyCBC search for gravitational waves from compact binary coalescence , 2015, 1508.02357.

[8]  Vicky Kalogera,et al.  SYSTEMATIC ERRORS IN LOW-LATENCY GRAVITATIONAL WAVE PARAMETER ESTIMATION IMPACT ELECTROMAGNETIC FOLLOW-UP OBSERVATIONS , 2016, 1601.02661.

[9]  Von Welch,et al.  Reproducing GW150914: The First Observation of Gravitational Waves From a Binary Black Hole Merger , 2016, Computing in Science & Engineering.

[10]  B. P. Abbott,et al.  Publisher’s Note: Observing gravitational-wave transient GW150914 with minimal assumptions [Phys. Rev. D 93 , 122004 (2016)] , 2016 .

[11]  R. Bonnand,et al.  Observing gravitational-wave transient GW150914 with minimal assumptions , 2016 .

[12]  Daniel George,et al.  Deep Neural Networks to Enable Real-time Multimessenger Astrophysics , 2016, ArXiv.

[13]  Li Fei-Fei,et al.  Measuring Depression Symptom Severity from Spoken Language and 3D Facial Expressions , 2018, ArXiv.

[14]  Vladlen Koltun,et al.  An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling , 2018, ArXiv.

[15]  Leo C. Stein,et al.  The binary black hole explorer: on-the-fly visualizations of precessing binary black holes , 2018, Classical and Quantum Gravity.