Abstract
We report on the construction of a deep convolutional neural network that can reproduce the sensitivity of a matched-filtering search for binary black hole gravitational-wave signals. The standard method for the detection of well-modeled transient gravitational-wave signals is matched filtering. We use only whitened time series of measured gravitational-wave strain as an input, and we train and test on simulated binary black hole signals in synthetic Gaussian noise representative of Advanced LIGO sensitivity. We show that our network can classify signal from noise with a performance that emulates that of match filtering applied to the same data sets when considering the sensitivity defined by receiver-operator characteristics.
| Original language | English |
|---|---|
| Article number | 141103 |
| Number of pages | 6 |
| Journal | Physical Review Letters |
| Volume | 120 |
| Issue number | 14 |
| DOIs | |
| Publication status | Published - 6 Apr 2018 |
Keywords
- UKRI
- STFC
- ST/L000946/1
Access to Document
Fingerprint
Dive into the research topics of 'Matching matched filtering with deep networks for gravitational-wave astronomy'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver
}
In: Physical Review Letters, Vol. 120, No. 14, 141103, 06.04.2018.
Research output: Contribution to journal › Article › peer-review
TY - JOUR
T1 - Matching matched filtering with deep networks for gravitational-wave astronomy
AU - Gabbard, Hunter
AU - Williams, Michael
AU - Hayes, Fergus
AU - Messenger, Chris
N1 - Funding Information: We have demonstrated that deep learning, when applied to gravitational-wave time series data, is able to closely reproduce the results of a matched-filtering analysis in Gaussian noise. We employ a deep convolutional neural network with rigorously tuned hyperparameters and produce an output that returns a ranking statistic equivalent to the inferred probability that data contain a signal. Matched-filtering analyses are often described as the optimal approach for signal detection in Gaussian noise. By building a neural network that is capable of reproducing this optimality we answer a fundamental question regarding the applicability of neural networks for gravitational-wave data analysis. In practice, searches for transient signals in gravitational-wave data are strongly affected by non-Gaussian noise artefacts. To account for this, standard matched-filtering approaches are modified to include carefully chosen changes to the ranking statistic [55,56] together with the excision of poor quality data [57,58] . Our analysis represents a starting point from which a deep network can be trained on realistic non-Gaussian data. Since the claim of matched-filtering optimality is applicable only in the Gaussian noise case, there exists the potential for deep networks to exceed the sensitivity of existing matched-filtering approaches in real data. In this work we have presented results for BBH mergers; however, this method could be applied to other merger types, such as binary neutron star and neutron star-black hole signals. This supervised learning approach can also be extended to other well-modeled gravitational-wave targets such as the continuous emission from rapidly rotating nonaxisymmetric neutron stars [59] . Finally we mention the possibilities for parameter estimation [60] where in the simplest cases an output regression layer can return point estimates of parameter values. As was exemplified in the case of GW170817, rapid detection confidence coupled with robust and equally rapid parameter estimates is critical for gravitational-wave multimessenger astronomy. We acknowledge valuable input from the LIGO-Virgo Collaboration specifically from T. Dent, R. Reinhard, I. Siong Heng, M. Cavalgia, and the compact binary coalescence and machine-learning working groups. The authors also gratefully acknowledge the Science and Technology Facilities Council of the United Kingdom. C. M. is supported by the Science and Technology Research Council (Grant No. ST/L000946/1). [1] 1 B. P. Abbott ( LIGO Scientific Collaboration and Virgo Collaboration ) , Phys. Rev. Lett. 116 , 061102 ( 2016 ). PRLTAO 0031-9007 10.1103/PhysRevLett.116.061102 [2] 2 B. P. Abbott ( LIGO Scientific Collaboration and Virgo Collaboration ) , Phys. Rev. Lett. 116 , 241103 ( 2016 ). PRLTAO 0031-9007 10.1103/PhysRevLett.116.241103 [3] 3 B. P. Abbott ( LIGO Scientific Collaboration and Virgo Collaboration ) , Phys. Rev. Lett. 118 , 221101 ( 2017 ). PRLTAO 0031-9007 10.1103/PhysRevLett.118.221101 [4] 4 B. P. Abbott ( LIGO Scientific Collaboration and Virgo Collaboration ) , Phys. Rev. Lett. 119 , 161101 ( 2017 ). PRLTAO 0031-9007 10.1103/PhysRevLett.119.161101 [5] 5 B. P. Abbott ( LIGO Scientific Collaboration and Virgo Collaboration ) , Astrophys. J. 848 , L13 ( 2017 ). ASJOAB 1538-4357 10.3847/2041-8213/aa920c [6] 6 A. Goldstein , Astrophys. J. 848 , L14 ( 2017 ). ASJOAB 1538-4357 10.3847/2041-8213/aa8f41 [7] 7 V. Savchenko , Astrophys. J. 848 , L15 ( 2017 ). ASJOAB 1538-4357 10.3847/2041-8213/aa8f94 [8] 8 B. P. Abbott ( LIGO Scientific Collaboration, Virgo Collaboration, F. GBM, INTEGRAL, IceCube Collaboration, AstroSat Cadmium Zinc Telluride Imager Team, IPN Collaboration, The Insight-Hxmt Collaboration, ANTARES Collaboration, The Swift Collaboration, AGILE Team, The 1M2H Team, The Dark Energy Camera GW-EM Collaboration, the DES Collaboration, The DLT40 Collaboration, GRAWITA, GRAvitational Wave Inaf TeAm, The Fermi Large Area Telescope Collaboration, ATCA, A. Telescope Compact Array, ASKAP, A. SKA Pathfinder, Las Cumbres Observatory Group, OzGrav, DWF, AST3, CAASTRO Collaborations, The VINROUGE Collaboration, MASTER Collaboration, J-GEM, GROWTH, JAGWAR, C. NRAO, TTU-NRAO, NuSTAR Collaborations, Pan-STARRS, The MAXI Team, T. Consortium, KU Collaboration, N. Optical Telescope, ePESSTO, GROND, T. Tech University, SALT Group, TOROS, Transient Robotic Observatory of the South Collaboration, The BOOTES Collaboration, MWA, M. Widefield Array, The CALET Collaboration, IKI-GW Follow-up Collaboration, H. E. S. S. Collaboration, LOFAR Collaboration, LWA, L. Wavelength Array, HAWC Collaboration, The Pierre Auger Collaboration, ALMA Collaboration, Euro VLBI Team, Pi of the Sky Collaboration, The Chandra Team at McGill University, DFN, D. Fireball Network, ATLAS, H. Time Resolution Universe Survey, RIMAS, RATIR, and S. South Africa/MeerKAT ) , Astrophys. J. 848 , L12 ( 2017 ). ASJOAB 1538-4357 10.3847/2041-8213/aa91c9 [9] 9 B. P. Abbott ( LIGO Scientific Collaboration and Virgo Collaboration ) , Phys. Rev. Lett. 119 , 141101 ( 2017 ). PRLTAO 0031-9007 10.1103/PhysRevLett.119.141101 [10] 10 S. A. Usman , Classical Quantum Gravity 33 , 215004 ( 2016 ). CQGRDG 0264-9381 10.1088/0264-9381/33/21/215004 [11] 11 K. Cannon , R. Cariou , A. Chapman , M. Crispin-Ortuzar , N. Fotopoulos , M. Frei , C. Hanna , E. Kara , D. Keppel , L. Liao , S. Privitera , A. Searle , L. Singer , and A. Weinstein , Astrophys. J. 748 , 136 ( 2012 ). ASJOAB 1538-4357 10.1088/0004-637X/748/2/136 [12] 12 T. Dal Canton , A. H. Nitz , A. P. Lundgren , A. B. Nielsen , D. A. Brown , T. Dent , I. W. Harry , B. Krishnan , A. J. Miller , K. Wette , K. Wiesner , and J. L. Willis , Phys. Rev. D 90 , 082004 ( 2014 ). PRVDAQ 1550-7998 10.1103/PhysRevD.90.082004 [13] 13 D. A. Brown , I. Harry , A. Lundgren , and A. H. Nitz , Phys. Rev. D 86 , 084017 ( 2012 ). PRVDAQ 1550-7998 10.1103/PhysRevD.86.084017 [14] 14 T. D. Canton and I. W. Harry , arXiv:1705.01845 . [15] 15 I. W. Harry , B. Allen , and B. S. Sathyaprakash , Phys. Rev. D 80 , 104014 ( 2009 ). PRVDAQ 1550-7998 10.1103/PhysRevD.80.104014 [16] 16 P. Ajith , N. Fotopoulos , S. Privitera , A. Neunzert , N. Mazumder , and A. J. Weinstein , Phys. Rev. D 89 , 084041 ( 2014 ). PRVDAQ 1550-7998 10.1103/PhysRevD.89.084041 [17] 17 B. S. Sathyaprakash and S. V. Dhurandhar , Phys. Rev. D 44 , 3819 ( 1991 ). PRVDAQ 0556-2821 10.1103/PhysRevD.44.3819 [18] 18 A. Taracchini , A. Buonanno , Y. Pan , T. Hinderer , M. Boyle , D. A. Hemberger , L. E. Kidder , G. Lovelace , A. H. Mroué , H. P. Pfeiffer , M. A. Scheel , B. Szilágyi , N. W. Taylor , and A. Zenginoglu , Phys. Rev. D 89 , 061502 ( 2014 ). PRVDAQ 1550-7998 10.1103/PhysRevD.89.061502 [19] 19 S. Privitera , S. R. P. Mohapatra , P. Ajith , K. Cannon , N. Fotopoulos , M. A. Frei , C. Hanna , A. J. Weinstein , and J. T. Whelan , Phys. Rev. D 89 , 024003 ( 2014 ). PRVDAQ 1550-7998 10.1103/PhysRevD.89.024003 [20] 20 L. Blanchet , Living Rev. Relativity 17 , 2 ( 2014 ). 1433-8351 10.12942/lrr-2014-2 [21] 21 K. G. Arun , A. Buonanno , G. Faye , and E. Ochsner , Phys. Rev. D 84 , 049901(E) ( 2011 ). PRVDAQ 1550-7998 10.1103/PhysRevD.84.049901 [22] 22 A. Buonanno , B. R. Iyer , E. Ochsner , Y. Pan , and B. S. Sathyaprakash , Phys. Rev. D 80 , 084043 ( 2009 ). PRVDAQ 1550-7998 10.1103/PhysRevD.80.084043 [23] 23 C. K. Mishra , A. Kela , K. G. Arun , and G. Faye , Phys. Rev. D 93 , 084054 ( 2016 ). PRVDAQ 2470-0010 10.1103/PhysRevD.93.084054 [24] 24 A. Buonanno and T. Damour , Phys. Rev. D 59 , 084006 ( 1999 ). PRVDAQ 0556-2821 10.1103/PhysRevD.59.084006 [25] 25 F. Pretorius , Phys. Rev. Lett. 95 , 121101 ( 2005 ). PRLTAO 0031-9007 10.1103/PhysRevLett.95.121101 [26] 26 A. Krizhevsky , I. Sutskever , and G. E. Hinton , in Advances in Neural Information Processing Systems 25 , edited by F. Pereira , C. J. C. Burges , L. Bottou , and K. Q. Weinberger ( Curran Associates , Inc., 2012 ), pp. 1097–1105 . [27] 27 I. J. Goodfellow , J. Pouget-Abadie , M. Mirza , B. Xu , D. Warde-Farley , S. Ozair , A. Courville , and Y. Bengio , arXiv:1406.2661 . [28] 28 K. Simonyan and A. Zisserman , arXiv:1409.1556 . [29] 29 L.-C. Chen , G. Papandreou , I. Kokkinos , K. Murphy , and A. L. Yuille , arXiv:1412.7062 . [30] 30 M. D. Zeiler and R. Fergus , arXiv:1311.2901 . [31] 31 C. Szegedy , W. Liu , Y. Jia , P. Sermanet , S. Reed , D. Anguelov , D. Erhan , V. Vanhoucke , and A. Rabinovich , arXiv:1409.4842 . [32] 32 R. Zhang , P. Isola , and A. A. Efros , arXiv:1603.08511 . [33] 33 A. Karpathy and L. Fei-Fei , arXiv:1412.2306 . [34] 34 I. Kononenko , Artif. Intell. Med. 23 , 89 ( 2001 ). AIMEEW 0933-3657 10.1016/S0933-3657(01)00077-X [35] 35 M. Pirooznia , J. Y. Yang , M. Q. Yang , and Y. Deng , BMC Genomics 9 , S13 ( 2008 ). 1471-2164 10.1186/1471-2164-9-S1-S13 [36] 36 N. Mukund , S. Abraham , S. Kandhasamy , S. Mitra , and N. S. Philip , Phys. Rev. D 95 , 104059 ( 2017 ). PRVDAQ 2470-0010 10.1103/PhysRevD.95.104059 [37] 37 M. Zevin , S. Coughlin , S. Bahaadini , E. Besler , N. Rohani , S. Allen , M. Cabero , K. Crowston , A. K. Katsaggelos , S. L. Larson , T. K. Lee , C. Lintott , T. B. Littenberg , A. Lundgren , C. sterlund , J. R. Smith , L. Trouille , and V. Kalogera , Classical Quantum Gravity 34 , 064003 ( 2017 ). CQGRDG 0264-9381 10.1088/1361-6382/aa5cea [38] 38 D. George , H. Shen , and E. A. Huerta , arXiv:1706.07446 . [39] 39 D. George and E. Huerta , Phys. Lett. B 778 , 64 ( 2018 ). PYLBAJ 0370-2693 10.1016/j.physletb.2017.12.053 [40] 40 T. Gebhard , N. Kilbertus , G. Parascandolo , I. Harry , and B. Schölkopf , in Workshop at the 31st Conference on Neural Information Processing Systems (NIPS), Long Beach, CA ( 2017 ). [41] 41 I. Goodfellow , Y. Bengio , and A. Courville , Deep Learning ( MIT Press , Cambridge, MA, 2016 ). [42] 42 B. P. Abbott , R. Abbott , T. D. Abbott , M. R. Abernathy , F. Acernese , K. Ackley , C. Adams , T. Adams , P. Addesso , R. X. Adhikari , Living Rev. Relativity 19 , 1 ( 2016 ). 1433-8351 10.1007/lrr-2016-1 [43] 43 S. Husa , S. Khan , M. Hannam , M. Pürrer , F. Ohme , X. J. Forteza , and A. Bohé , Phys. Rev. D 93 , 044006 ( 2016 ). PRVDAQ 2470-0010 10.1103/PhysRevD.93.044006 [44] 44 S. Khan , S. Husa , M. Hannam , F. Ohme , M. Pürrer , X. J. Forteza , and A. Bohé , Phys. Rev. D 93 , 044007 ( 2016 ). PRVDAQ 2470-0010 10.1103/PhysRevD.93.044007 [45] 45 B. P. Abbott ( LIGO Scientific Collaboration and Virgo Collaboration ) , Phys. Rev. X 6 , 041015 ( 2016 ). PRXHAE 2160-3308 10.1103/PhysRevX.6.041015 [46] 46 B. Allen , W. G. Anderson , P. R. Brady , D. A. Brown , and J. D. E. Creighton , Phys. Rev. D 85 , 122006 ( 2012 ). PRVDAQ 1550-7998 10.1103/PhysRevD.85.122006 [47] 47 Y. Lecun , L. Bottou , Y. Bengio , and P. Haffner , Proc. IEEE 86 , 2278 ( 1998 ). IEEPAD 0018-9219 10.1109/5.726791 [48] 48 Y. LeCun , L. Bottou , G. B. Orr , and K. R. Müller , in Neural Networks: Tricks of the Trade , edited by G. B. Orr and K.-R. Müller ( Springer Berlin Heidelberg , Berlin, Heidelberg, 1998 ), pp. 9–50 . [49] 49 M. Abadi , A. Agarwal , P. Barham , E. Brevdo , Z. Chen , C. Citro , G. S. Corrado , A. Davis , J. Dean , M. Devin , arXiv:1603.04467 . [50] 50 T. Dozat , ICLR Workshop, San Juan, Puerto Rico ( 2016 ). [51] 51 S. Babak , Phys. Rev. D 87 , 024033 ( 2013 ). PRVDAQ 1550-7998 10.1103/PhysRevD.87.024033 [52] 52 S. Babak , R. Balasubramanian , D. Churches , T. Cokelaer , and B. S. Sathyaprakash , Classical Quantum Gravity 23 , 5477 ( 2006 ). CQGRDG 0264-9381 10.1088/0264-9381/23/18/002 [53] 53 A. Nitz , Pycbc software ( 2017 ). [54] We are limited to a minimal false alarm probability of ∼ 10 - 4 due to the limited number of testing samples used. [55] 55 B. Allen , Phys. Rev. D 71 , 062001 ( 2005 ). PRVDAQ 1550-7998 10.1103/PhysRevD.71.062001 [56] 56 A. H. Nitz , T. Dent , T. D. Canton , S. Fairhurst , and D. A. Brown , Astrophys. J. 849 , 118 ( 2017 ). ASJOAB 1538-4357 10.3847/1538-4357/aa8f50 [57] 57 B. P. Abbott ( LIGO Scientific Collaboration and Virgo Collaboration ) , Classical Quantum Gravity 35 , 065010 ( 2018 ). CQGRDG 0264-9381 10.1088/1361-6382/aaaafa [58] 58 B. P. Abbott , Classical Quantum Gravity 33 , 134001 ( 2016 ). CQGRDG 0264-9381 10.1088/0264-9381/33/13/134001 [59] 59 B. P. Abbott ( LIGO Scientific Collaboration and Virgo Collaboration ) , arXiv:1707.02669 . [60] 60 D. George and E. A. Huerta , Phys. Rev. D 97 , 044039 ( 2018 ). PRVDAQ 2470-0010 10.1103/PhysRevD.97.044039 Publisher Copyright: © 2018 authors. Published by the American Physical Society.
PY - 2018/4/6
Y1 - 2018/4/6
N2 - We report on the construction of a deep convolutional neural network that can reproduce the sensitivity of a matched-filtering search for binary black hole gravitational-wave signals. The standard method for the detection of well-modeled transient gravitational-wave signals is matched filtering. We use only whitened time series of measured gravitational-wave strain as an input, and we train and test on simulated binary black hole signals in synthetic Gaussian noise representative of Advanced LIGO sensitivity. We show that our network can classify signal from noise with a performance that emulates that of match filtering applied to the same data sets when considering the sensitivity defined by receiver-operator characteristics.
AB - We report on the construction of a deep convolutional neural network that can reproduce the sensitivity of a matched-filtering search for binary black hole gravitational-wave signals. The standard method for the detection of well-modeled transient gravitational-wave signals is matched filtering. We use only whitened time series of measured gravitational-wave strain as an input, and we train and test on simulated binary black hole signals in synthetic Gaussian noise representative of Advanced LIGO sensitivity. We show that our network can classify signal from noise with a performance that emulates that of match filtering applied to the same data sets when considering the sensitivity defined by receiver-operator characteristics.
KW - UKRI
KW - STFC
KW - ST/L000946/1
UR - http://www.scopus.com/inward/record.url?scp=85045022506&partnerID=8YFLogxK
UR - https://eprints.gla.ac.uk/159638/
U2 - 10.1103/PhysRevLett.120.141103
DO - 10.1103/PhysRevLett.120.141103
M3 - Article
C2 - 29694122
AN - SCOPUS:85045022506
SN - 0031-9007
VL - 120
JO - Physical Review Letters
JF - Physical Review Letters
IS - 14
M1 - 141103
ER -