# Calendar

A new era of gravitational wave (GW) astronomy has begun with the recent Nobel-prize winning detections by LIGO. Current data analysis pipelines are, however, limited by the extreme computational costs of template-matching, thus facing significant delays and inability to detect all types of GW sources. I will start with an introduction to deep learning with artificial neural networks. I will then describe a highly scalable technique, based on two 1D convolutional neural networks, that I developed to resolve these issues, which allows real-time detection and parameter estimation of GW signals whose amplitudes are much weaker than the background noise. This uses data derived from high-performance physics simulations on supercomputers, including Blue Waters, to train artificial intelligence algorithms that exploit emerging hardware architectures such as deep-learning-optimized GPUs. I will also discuss my recent work on applying transfer learning and unsupervised clustering methods for classifying anomalous noise transients in spectrograms of LIGO data. I will conclude by discussing my ongoing research including new deep learning methods for denoising LIGO data with recurrent neural network auto-encoders and generative modeling of GW signals. These deep learning techniques for low-latency analysis of the raw big data collected by observational instruments can enable real-time gravitational wave and multimessenger astrophysics, which promises groundbreaking insights about the universe.

\n\nSPEAKER:Daniel George, University of Illinois at Urbana-Champaign

276 Loomis

false## Astrophysics, Gravitation and Cosmology Seminar: Deep Learning Techniques for Real-Time Gravitational Wave and Multimessenger Astrophysics

Speaker |
(sign-up)
Daniel George, University of Illinois at Urbana-Champaign |
---|---|

Date: | 11/8/2017 |

Time: | 12 p.m. |

Location: | 276 Loomis |

Event Contact: | Milton Ruiz ruizm@illinois.edu |

Sponsor: | Department of Physics |

Event Type: | Seminar/Symposium |

A new era of gravitational wave (GW) astronomy has begun with the recent Nobel-prize winning detections by LIGO. Current data analysis pipelines are, however, limited by the extreme computational costs of template-matching, thus facing significant delays and inability to detect all types of GW sources. I will start with an introduction to deep learning with artificial neural networks. I will then describe a highly scalable technique, based on two 1D convolutional neural networks, that I developed to resolve these issues, which allows real-time detection and parameter estimation of GW signals whose amplitudes are much weaker than the background noise. This uses data derived from high-performance physics simulations on supercomputers, including Blue Waters, to train artificial intelligence algorithms that exploit emerging hardware architectures such as deep-learning-optimized GPUs. I will also discuss my recent work on applying transfer learning and unsupervised clustering methods for classifying anomalous noise transients in spectrograms of LIGO data. I will conclude by discussing my ongoing research including new deep learning methods for denoising LIGO data with recurrent neural network auto-encoders and generative modeling of GW signals. These deep learning techniques for low-latency analysis of the raw big data collected by observational instruments can enable real-time gravitational wave and multimessenger astrophysics, which promises groundbreaking insights about the universe. |

To request disability-related accommodations for this event, please contact the person listed above, or the unit hosting the event.

## Event Mailing List

Receive Illinois Physics events weekly in your inbox.