Visions of planetary discovery and strange new moons are part of a new exhibit, "The Heritage of...
$2.2 Million Grant Calls for Designing Computer Software to Predict the Unpredictable
University of Arizona Professor Jerzy Rozenblit has received a $2.2 million grant to design computer software that will analyze volatile political and military situations.
The software will predict the actions of paramilitary groups, ethnic factions, terrorists and criminal groups, while aiding commanders in devising strategies for stabilizing areas before, during and after conflicts.
It also will have many civilian applications in finance, law enforcement, epidemiology and the aftermath of natural disasters, such as hurricane Katrina.
The Asymmetric Threat Response and Analysis Project, known as ATRAP, is a massively complex set of computer algorithms (mathematical procedures) that sift through millions of pieces of data, considering many factors including social, political, cultural, military and media influences, said Rozenblit, who holds the Raymond J. Oglethorpe Endowed Chair in the Electrical and Computer Engineering Department at the UA.
The software can handle data loads that would overwhelm human analysts, while dispassionately exploring actions and behaviors based solely on the data, sidestepping human cultural biases that might prematurely rule out unorthodox or seemingly bizarre courses of action.
Actions Sometimes Defy Logic
Since the end of the Cold War, our opponents have behaved in ways that defy what we would consider normal logic, pursuing actions that we find almost inconceivable, said Rozenblit, who heads the Electrical and Computer Engineering department at UA. Predicting these asymmetric behaviors is difficult and further complicated by the massive amounts of intelligence data available.
ATRAP will use sophisticated computational methods based on game theory, co-evolution and genetic development models to find solutions that make sense in illogical times.
Genetic algorithms analyze situations in an evolutionary context, where actions with the highest “fitness factor” (chance of achieving the greatest success) gravitate toward one another, produce offspring and eventually rise to the top.
Co-evolutionary algorithms analyze how the actions of one group affect the other groups and how those other groups adapt, or co-evolve, in response to the changing situation. For instance, if one group becomes more influential in an area where ethnic factions are vying for supremacy, the other groups will respond in ways that will try to make that first faction less influential, Rozenblit said.
The algorithms are designed to recognize links and patterns within the data and to find connections, much as an investigative reporter might do when examining financial records - but on a vastly more complex and detailed scale.
Predicting the Unpredictable
“The computer can look at very, very complex data sets that as an individual or even as a group of individuals, you could never analyze,” said Brian Ten Eyck, ATRAP project manager and associate director for research support in ECE. “The computer can bring the patterns and connections to the surface and can predict scenarios that might never occur to human analysts.”
Deep Blue, the first computer program to beat a world chess champion, is an example of how ATRAP can respond to changing factors, Ten Eyck explained. “Every time its opponent made a move, Deep Blue recalculated all the possibilities and likely courses of action, eventually settling on the fittest move that would achieve its goal of winning the game.”
However, chess is not an exact analogy because only two players are involved and the end goal is for one player to win.
In unstable areas, winning often means establishing an environment in which the factions co-exist in a win-win situation or at least in an equilibrium in which there are no rewards, and some penalties, for disturbing the status quo, Rozenblit said.
“Deep Blue is a good analogy because it illustrates the complexity of the problems, but in chess you have a finite court and a well-defined set of operations,” Rozenblit added. “Therefore, a move constitutes a valid move.
But what we’re dealing with now is a world with no rules, with infinite possibilities and moves that defy logic, such as total disregard for the basic instinct of self preservation.”|
Quick Response is Vital
Ultimately, the software program will be designed to display data in graphical, 3-D and other forms that can be quickly grasped, allowing decision makers to rapidly respond to changing situations, Rozenblit said.
In managing conflicts such as those that occurred in Kosovo or Somalia in the 1990s, commanders will need to respond quickly. “In those situations, we don’t have two months to figure things out,” Rozenblit said. “So the second part of our project involves harnessing massively parallel computing architectures to do computations very rapidly.”
Parallel computing, which relies on several large computers working on portions of a problem simultaneously, will allow commanders to rapidly analyze millions of data points from intelligence reports.
Students and Local Contractors Benefit
While the software ultimately could save millions of lives, it’s immediately benefiting local companies and students in the short term.
Rozenblit plans to outsource some parts of the project to local contractors because the UA part of the research doesn’t involve working with classified data. “There’s nothing about our part of the project that’s classified,” Rozenblit said. “We’re an open, academic institution and it’s difficult for us to be involved in classified work. So we need contractors to handle the classified parts of the project.”
ATRAP research also is giving students valuable skills. “There is a dire need for engineers with expertise in this area and our graduate students and undergraduates are in great demand,” Rozenblit said. One student who recently graduated with a bachelor’s degree after doing research related to ATRAP was hired at an annual starting salary of $90,000.
The ATRAP software is being developed in collaboration with the Army Battle Command Battle Laboratory at Ft. Huachuca, Ariz.
While ATRAP can also address many complex, non-military situations that require analysis of complex data and balancing the desires of competing factions, its military application is equally concerned with conflict avoidance.
“The goal is to handle conflict areas in a manner that leads to stability and support so war is not necessary,” Rozenblit said. “That’s the philosophy behind much of the ATRAP effort.”