image: Richard Veras, Ph.D.
Credit: University of Oklahoma/Travis Caperton.
NORMAN, OKLA. – In an increasingly data-saturated world, computing infrastructure innovations are needed to make sense of new types of information. Richard Veras, a professor in the School of Computer Science at the University of Oklahoma, has received a National Science Foundation Faculty Early Career Development Program (CAREER) award to develop such an innovation by creating more efficient infrastructure for the computation of sparse and irregular data.
Big data – datasets that are challenging to manage using traditional processing tools due to size and complexity, such as social media connections or human interactions for epidemiology research – have increased in availability over the past 20 to 25 years.
“We have more data than we know what to do with, and there are a lot of interesting questions that we want to be able to ask from the data,” said Veras.
But for the past four or five decades, he notes, computers have been built similarly, in a way that favors crunching dense, regular computations. Consequently, the types of algorithms needed to study the questions asked of sparse, irregular datasets are extremely computationally heavy.
“We want to look at data that is different than what we’ve historically been looking at it, and we want to look at it in a different way, too. But hardware has been designed for a specific type of data. The question is how to reconcile these things.” said Veras. “How do we build better software and hardware for the type of data we’re looking at and the type of problems we have?”
To solve this problem, Veras will develop a new coding language, the Graph Structure Descriptor Language, and a software program that takes a high-level representation of a problem described in a coding language afnd translates it into machine code. He will create the language to describe the shape and structure of these irregular data and the infrastructure necessary to translate that language into efficient code.
Veras hopes that the infrastructure he creates will be embedded into the machinery researchers are already using to work with big data.
In connection with this grant, OU will create a new degree concentration in high-performance computing in computer science. Veras is passionate about creating research opportunities for undergraduate students, the types of opportunities that he had that set him on a successful career in research.
The core of the program will grow around a parallel programming course that Veras teaches. He hopes to expand the concept to two additional courses, one theory-based and the other a capstone course. The capstone course will offer students hands-on opportunities to make a difference in the OU research community.
“We’ll solicit problems from across the university and put students on them to do performance engineering to make these applications more efficient for the labs throughout OU,” said Veras.
Though the concentration is in the very early stages, connecting with industry partners and career preparation will be important parts of the program. The goal, says Veras, is more than working with code; it’s getting students jobs.
“By the time you’re in your junior or senior year and decide that research is interesting, you’re late to the game,” said Veras, who started his experience with research as a college freshman. “I’m very much a proponent of making research available to students as early as we can.”