Machine Learning and its application towards the search for Gravitational waves
Invited Talk, The International Congress on Industrial and Applied Mathematics (ICIAM), Valencia, Spain
Talks and presentations
This is a map of all of the most recent talks which I have given over the last several years. You can find further details about those talks below.
See a map of all the places I've given a talk!
Matching Matched Filtering Using Deep Learning
Talk, SUPA Annual Gathering, Glasgow, UK
CNNs and GANs applied towards the search for gravitational waves
Talk, Annual ScotDist Meeting, Glasgow, UK
Gravitational wave data analysis with machine learning
Talk, Edinburgh MOST Meeting, Edinburgh, UK
Gravitational wave data analysis with machine learning
Talk, Gravitational-wave Excellencethrough Alliance Training Kick-off Meeting, Beijing, China
An overview of LIGO-Virgo machine learning applications for gravitational wave astronomy
Talk, Machine Learning in Science and Engineering conference, Pittsburgh, PA
Matching Matched Filtering with Deep Networks in Gravitational-wave Astronomy
Talk, LVC March Meeting, Sonoma, CA
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.
Machine Learning and Gravitational-wave Astronomy
Outreach Talk, 7 Minutes of Science, Glasgow, UK
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.
Genetic Programming Applied to Glitch Classification at LIGO
Talk, Computing in HighEnergy AstroParticle Research Conference, Columbus, OH
Quality improvement of interferometric data collected by gravitational-wave detectors such as Advanced LIGO and Virgo is mission critical for the success of gravitational-wave astrophysics. Gravitational-wave detectors are sensitive to a variety of disturbances of non-astrophysical origin with characteristic frequencies in the instrument band of sensitivity. Removing nonastrophysical artifacts that corrupt the data stream is crucial for increasing the number and statistical significance of gravitational-wave detections and enabling refined astrophysical interpretations of the data. Machine learning has proved to be a powerful tool for analysis of massive quantities of complex data in astronomy and related fields of study. We present two machine learning methods, based on random forest and genetic programming algorithms, that can be used to determine the origin of non-astrophysical transients in the LIGO detectors. We use two classes of transients with known instrumental origin that were identified during the first observing run of Advanced LIGO to show that the algorithms can successfully identify the origin of non-astrophysical transients in real interferometric data and thus assist in the mitigation of instrumental and environmental disturbances in gravitational-wave searches. While the datasets described in this paper are specific to LIGO, and the exact procedures employed were unique to the same, the random forest and genetic programming code bases and means by which they were applied as a dual machine learning approach are completely portable to any number of instruments in which noise is believed to be generated through mechanical couplings, the source of which is not yet discovered.
A low-latency Glitch Classification Algorithm Based in Waveform Morphology for Advanced LIGO
Talk, April APS Meeting, Salt Lake City, UT
The Virgo and LIGO detectors aim at detecting gravitational wave (GW) signals produced by astrophysical sources. The burst group, in which the LAL group is involved, is looking for poorly-modeled and short-duration signals. To achieve this, members of the LAL group built search pipelines which are sensitive as well as robust against a large variety of detector glitches. The main obstacle to a GW burst detection is the data quality which is polluted by a great amount of detector glitch noise which mimics genuine burst signals. The LAL group is currently developing new tools to understand and characterize the detector’s glitch noise with a low latency. Mr. Gabbard helped characterize and improve one of these new tools (Omicron event trigger generator). Additionally, he helped to integrate this newly acquired information directly in burst search pipelines. He also assisted in Integrating these developments into an online-based analysis architecture and contributed to the development of new improvement ideas.
The Birth of Gravitational-wave Astronomy
Invited Talk, Mississippi State University SPS, Starkville, MS
An outreach talk to the MS State University SPS group on graviational wave astronomy.
A low-latency Glitch Classification Algorithm Based in Waveform Morphology for Advanced LIGO
Talk, Mississippi Academy of Sciences Meeting, Hattiesburg, MS
Through a fellowship funded by the National Science Foundation (NSF) Mr. Gabbard applied programming and mathematical skills for the analysis of gravitational wave data. Mr. Gabbard presented a novel and efficient machine learning algorithm for classification of signals that arise in gravitational wave channels of the Laser Interferometer Gravitational Wave Observatory (LIGO). Using data from LIGO’s sixth science run (S6), a new glitch classification algorithm based mainly on the morphology of the waveform as well as several other parameters (signal-to-noise ratio (SNR), duration, bandwidth, etc.) was developed.This was done using two novel methods, Kohonen Self Organizing Feature Maps (SOM), and discrete wavelet transform coefficients. The study showed the feasibility of utilizing SOMs to display a multidimensional trigger set in a low-latency two dimensional format.
Characterization of the Omicron Trigger Generator and Transient analysis of aLIGOData
Poster, April APS Meeting, Baltimore, Maryland
The Virgo and LIGO detectors aim at detecting gravitational wave (GW) signals produced by astrophysical sources. The burst group, in which the LAL group is involved, is looking for poorly-modeled and short-duration signals. To achieve this, members of the LAL group built search pipelines which are sensitive as well as robust against a large variety of detector glitches. The main obstacle to a GW burst detection is the data quality which is polluted by a great amount of detector glitch noise which mimics genuine burst signals. The LAL group is currently developing new tools to understand and characterize the detector’s glitch noise with a low latency. Mr. Gabbard helped characterize and improve one of these new tools (Omicron event trigger generator). Additionally, he helped to integrate this newly acquired information directly in burst search pipelines. He also assisted in Integrating these developments into an online-based analysis architecture and contributed to the development of new improvement ideas.