You are here

Machine Learning Methods to Catalog Sources from Diverse, Widely Distributed Sensors

Award Information
Agency: Department of Defense
Branch: Air Force
Contract: FA9453-19-P-0684
Agency Tracking Number: F19A-012-0065
Amount: $149,927.00
Phase: Phase I
Program: STTR
Solicitation Topic Code: AF19A-T012
Solicitation Number: 19.A
Solicitation Year: 2019
Award Year: 2019
Award Start Date (Proposal Award Date): 2019-07-23
Award End Date (Contract End Date): 2020-07-23
Small Business Information
2839 Paces Ferry Road Suite 1160
Atlanta, GA 30339
United States
DUNS: 961914884
HUBZone Owned: No
Woman Owned: No
Socially and Economically Disadvantaged: Yes
Principal Investigator
 Zhigang Peng
 Professor of Geophysics
 (404) 894-0231
Business Contact
 Dr. Ash Thakker
Phone: (770) 803-3001
Research Institution
 Georgia Institute of Technology
 Timothy Gehret Timothy Gehret
311 Ferst Drive
Atlanta, GA 30332
United States

 (404) 594-0950
 Nonprofit College or University

The USAF seek innovative technology solutions to develop automated algorithms to process variety of different sensor data from around the world, using Machine Learning methods and classify different seismic events from possible nuclear tests. In Phase I of this proposal, a research team from Global Technology Connection Inc. and our university partner Georgia Tech (RI) plans to work together to evaluate the performance of existing ML algorithms such as CNNs to distinguish between tectonic earthquakes, conventional chemical blasts and nuclear explosions. We would first build a labeled seismic dataset that will include different types of seismic events, and then explore multiple deep-learning methods to this dataset to compare their performance We will be leveraging our university partners’ unique understanding gained from several ongoing projects related to ML detection of seismic events in tectonically active regions to detect and picking seismic phases and classifying different types of earthquakes and tremors around the world and our knowledge of building ML based Prognostic Self- learning Algorithms. This applies for the inverse of nuclear blasts where energy is quite extreme, but in a very short packet of time. We will develop working prototype in Phase II and aggressive commercialization strategy for Phase III.

* Information listed above is at the time of submission. *

US Flag An Official Website of the United States Government