Geosemble Technologies to Support Department of Homeland Security Program. (03 November, 2006)

El Segundo, CA -- Geosemble Technologies, Inc., announced today that it has been selected to form part of a center located at the University of Southern California (USC), under a program from the Department of Homeland Security (DHS) to gather, integrate and extract knowledge from various kinds of data.

As part of a new DHS Research Center at USC, Geosemble Technologies will support the overall program goals of conducting research on advanced methods for information analysis and to develop computational technologies that contribute to securing the homeland. USC and three other University Affiliate Centers and their partners will collaborate with the Institute for Discrete Sciences (IDS), a joint project between DHS and several National Laboratories, led by Lawrence Livermore National Laboratory. In that capacity, Geosemble will be working to automatically extract and identify information from geospatial sources such as maps and aerial imagery. The focus will be on the identification of geospatial features of high interest to the government, including location names (e.g., cities, streets), population information, and temporal changes and relationships that signal unusual behavior of interest.

"For this government contract, the basic challenge is to automatically and accurately extract information from geospatial sources such as satellite images, aerial photographs and raster maps," said Cyrus Shahabi, Chief Technology Officer of Geosemble Technologies. ˇ§Towards this end, instead of applying pure image recognition techniques on the images, we will utilize and fuse other sources of geospatial datasets, including vector data sets, gazetteer point data, and digital elevation data, to help the image recognition process,ˇ¨ he said. ˇ§The methodology for solving these complex problems lies in underlying mathematical algorithms that rely on linear algebraic operations such as matrix transformations, as well as machine learning techniques such as Bayesian classifiers and support-vector-machines,ˇ¨ Shahabi explained.

Together with USC and other DHS University Affiliate Centers, Geosemble will be studying such topics as knowledge representation, natural language processing, text or information extraction, uncertainty quantification, and high-performance computing architectures. These results will be applied by DHS and the Science and Technology Directorate to address problems in information analysis, decision support, risk analysis, critical infrastructure protection, bioinformatics, and computational biology.

Additional users for Geosemble technology include military decision makers faced with vast amounts of maps and aerial images and insufficient processing capability, as well as commercial market users in the real estate industry, oil and gas, highway transportation, geological surveying, and other users of maps and aerial imagery.