BU Today

Science & Tech

CAS’ Appavoo Nabs NSF CAREER Award

Exploring how computers can learn from past behavior

Jonathan Appavoo, Faculty Early Career Development Award National Science Foundation NSF,Boston University College of Arts and Sciences assistant professor of computer science, Rafik B. Hariri Institute for Computing and Computational Science & Engineering

Jonathan Appavoo says he’s excited about exploring “a different kind of computer that incorporates a biologically motivated mechanism into its inner workings.” Photo by Vernon Doucette

Jonathan Appavoo has long been fascinated by computers. “They somehow seemed to mimic and reflect our own abilities,” he says. As he studied them in greater depth, he realized “that while they are fascinating machines, their ‘human’ qualities are largely achieved by smoke and mirrors. That being said, their ability to be programmed to do arbitrary tasks over and over again at very high speeds is incredibly powerful.”

For his work exploring ways to teach computers how to learn from their past behavior, the College of Arts & Sciences assistant professor of computer science has been awarded a Faculty Early Career Development (CAREER) Award from the National Science Foundation (NSF). The award—valued at nearly $600,000—will support his research and teaching efforts in operating systems, in particular a project involving programmable smart machines.

The main challenge Appavoo seeks to address in his project is this: faster computers have enabled advances in science, commerce, and daily life. Unfortunately, computers have also become complex and more difficult to program efficiently, threatening further advances.

Perhaps, he posits, we can draw upon biologically inspired learning techniques in designing a new model of hybrid computer, a programmable smart machine, which learns from its past behavior to automatically improve its performance without the burden of more complex programming. Specifically, his work explores the addition of a smart memory to a computer, which gives it the ability to learn, store, and exploit patterns in past execution.

The father of two young children, Appavoo says he is keenly aware of “our limited human ability to be programmed and directed, but also of our incredible capacity to learn and improve based on experience.” In his research he is trying to answer the question of whether there is a way to combine these abilities to “yield machines that can both be programmed and also intrinsically improved by taking advantage of their size and experience.” In other words, he says, can you create a new kind of “programmable smart machine that improves with its size and experience?”

The $595,000, five-year NSF grant will allow him to “explore the possibility of constructing hybrid computing systems that behave as programmed, but transparently learn and automatically improve their operation by studying if and how a computer system can integrate learning mechanisms into their core execution model,” he says.

“The work Jon is doing in computer science is one more example of the leading-edge research that is going on in the College of Arts & Sciences,” says Virginia Sapiro, dean of Arts & Sciences. “I am excited to see Jon join the ranks of our young faculty members who have earned the distinction of winning the NSF Early Career Award.”

The CAREER Awards are given to support junior faculty “who exemplify the role of teacher-scholars through outstanding research, excellent education, and the integration of education and research within the context of the mission of their organizations,” according to the NSF.

Appavoo says the award will allow his research to move to the next stage. “This work began over a decade ago,” he says, “and it epitomizes the kind of exploration that drew me to science and has motivated my career to this point. It is very exciting to now have the resources to explore a different kind of computer that incorporates a biologically motivated mechanism into its inner workings.”

Jeremy Schwab can be reached at jschwab@bu.edu.

+ Comments

Post Your Comment

(never shown)