Skip to main content

NCSA research assistant wins SC17 Graduate Student Research Competition

Out of the flurry of activity that encompassed the annual SC17 conference in Denver, Colorado on November 12-17, one of the most notable achievements from the National Center for Supercomputing Applications (NCSA) came from Daniel George, a research assistant at NCSA. George, whose research combines deep neural networks with the search for gravitational waves, took home first place in the Association for Computing Machinery (ACM)’s Graduate Student Research Competition.

Out of 58 posters entered in the ACM competition, Daniel’s, titled “Deep Learning with HPC Simulations for Extracting Hidden Signals: Detecting Gravitational Waves” was selected as the best not only for it’s combining of deep learning and cosmology, but also because of the truly cutting-edge nature, practicality and demand for his research. In order to create a robust system for identifying gravitational waves, George combined deep learning algorithms, black hole simulations, the Einstein Toolkit and the Laser Interferometer Gravitational Wave Observatory (LIGO)’s gravitational wave data analysis.

By combining all of these resources, George (With the help of NCSA’s Eliu Huerta) was able to introduce Deep Filtering, a type of end-to-end time-series signal processing which combines two deep convolutional neural networks to rapidly detect and estimate parameters of signals, in this case, signals from gravitational waves caused by black hole collisions. This, in turn, allows researchers to pinpoint gravitational waves among background noise, increasing accuracy in wave identification.

Not only is this filtering method more accurate, but it’s also more efficient. When compared to previous conventional machine learning methods of wave identification, George and Huerta’s method allows real-time analysis of raw data, and will improve LIGO’s ability to identify gravitational waves as they hit Earth, instead of relying on slower analysis techniques.

Learn more about Daniel George’s award-winning research here.

Disclaimer: Due to changes in website systems, we've adjusted archived content to fit the present-day site and the articles will not appear in their original published format. Formatting, header information, photographs and other illustrations are not available in archived articles.

Back to top