Experts at GEOINT Foreword discuss analysis-as-a-service and non-traditional GEOINT
The next series was titled “Perspectives on Analysis-as-a-Service/non-traditional GEOINT.”
Steven Brumby, co-founder and chief science advisor of Descartes Labs, said a geospatial analysis revolution is underway, based on three things: the availability of data, the fact that we can now access supercomputing for pennies on the dollar compared to previous eras, and the fact that we now have algorithms to process the data.
“The sad fact of most imagery analysis,” Brumby said, “ is that nobody really wants pixels. They want answers.”
Brumby discussed the automated process of producing maps and that machine learning can leverage the human-generated maps to produce new maps. He anticipates a future task for analysts will be to assess and clean up maps that are 95-99 percent complete when they emerge from the algorithm.
David Gauthier, director of the National Geospatial-Intelligence Agency’s (NGA) office of Strategic Operations, said in the early days of GEOINT, there were three ways to access capabilities: build them, buy them, and share them. Today, he explained, with an explosion of capabilities in the commercial market, the focus is: lead, broker, and differentiate.
Gauthier said one of the best services to broker is analysis, because it costs less than inventing infrastructure, provides more access to non-traditional GEOINT data, offers more innovative analytic techniques, and allows more time for generating insights. Gauthier called on audience members to help NGA solve analysis challenges such as data quality control, perception and bias, and incoherence such as noise and negative gain.
In the future, Gauthier said, “We have to stretch the fabric of the discipline. We have to pull ourselves up continuously—Invent, innovate, so we can truly differentiate what we’re inventing, building on the government side.”
Finally, Matthew Chwastek, director of product management for Orbital Insight’s public sector, talked about extracting and delivering analysis, whether it’s a poverty map of Sri Lanka or a product that analyzes Wal-Mart’s sales based on the number of cars in its parking lots. He said the confluence of machine learning advances, cloud computing, and satellite imagery has made it possible to understand the world in a different way.
He noted that the explosion of data means an even more important role for automation. “We did the calculations, and if every satellite that’s scheduled to go into space by 2022 [does], you’ll need 8 million people working full time every day looking at pixels.”
Given this explosion of data, it’s fortunate that many everyday tasks can be automated, Chwastek said, so the IC doesn’t have to hire humans to analyze it all.
“I don’t know about the audience here, but I don’t think that’s affordable for [anyone].”
Collaboration and planning integral to best serve the GEOINT Community