Page 32 - GCN, June/July 2018

Industry Insight
BY BEN CONKLIN
How artificial intelligence is transforming geospatial intelligence
Artificial intelligence has improved by leaps and bounds since IBM’s chess- playing Deep Blue defeated reigning world champion Garry Kasparov in 1997. But that early face-off il- lustrated machine learning in its nascence: A computer makes sense of data it is given, finds patterns and crafts solutions based on the presented scenarios.
Today’s systems have access to infinitely more data from which they can uncover relationships and predict outcomes without pre-existing empirical models.
The new frontier is turning geographic
data into deep location intelligence. Enabling applications to understand relationships in that data is the key to addressing some of the most pressing threats facing the geopolitical world today. And the people at the forefront of those challenges are in the intelligence community.
In recent years, advances in AI algorithms for object recognition and broad-area searches have removed much of the inherent uncer- tainty from geospatial intel- ligence decisions made in the field. Combining AI and intelligence information can allow us to move beyond simple recognition and into intelligent models for alert-
ing and notification.
The scale, complexity and
pace of modern conflicts demand fresh approaches to delivering intelligence capabilities. Location intel- ligence provides the key to understanding situations better because of the deeper insights that real-time spa- tial analytics enables. For instance, if an intelligence analyst tracking border- crossing smugglers has
a mass of point locations received through observa- tions, sensors and reports, a
second of a video to unlock valuable data, but AI can extract that information and locate it in space and time so that it can be connected with other relevant data. And it can do that faster
and more effectively than humans.
Algorithms can also ana- lyze and learn from imagery and location data based
on known patterns to help organize and categorize the information analysts need and care about most. For instance, if specific types
the U.S. military, political or economic interests or to citizens abroad — can be as- sessed the instant that data is processed.
Intelligence analysts
have a specific set of needs, and each of those needs contains a component of location. The integration
of location intelligence
into the GEOINT analysts’ toolkit allows for a more holistic and contextual understanding of the data they see every day. Applying cutting-edge AI technology
32 GCN JUNE/JULY 2018 • GCN.COM
Combining AI and intelligence information can allow us to move beyond simple recognition and into intelligent models for alerting and notification.
spatial query could iden- tify the relevant data and pinpoint smugglers’ possible routes. The analyst can then visualize the relationships between the people, groups and objects and determine the smugglers’ likely course of action.
However, GEOINT ana- lysts still must collect much of their intelligence from unconventional sources. Although such data has a location component, it often must be extracted from its source and scrubbed before it can be analyzed. Humans can’t examine every pixel
of an image or watch every
of structures are commonly used for storage or manu- facturing, they can be easily codified, understood and ultimately identified by pattern recognition filters in machine learning algorithms.
One of the most ground- breaking influences that AI can have on the GEOINT an- alyst’s day-to-day activities is giving the detection and reporting of foreign threats a real-time edge. Algorithms are allowing the GEOINT community to model indica- tions and warning scenarios so that information regard- ing a threat — whether to
to spatial analytics creates a smarter GEOINT capability — a definite edge.
The intelligence com- munity must make sense of mountains of geospatial in- formation quickly and turn that into actionable intel- ligence for decision-makers. A technology that performs the time-consuming and de- tailed function of classifying data frees GEOINT analysts to make the cognitive con- nections that only trained human intelligence experts can understand. •
— Ben Conklin is defense and intelligence industry manager at Esri.