March 5, 2013 | City College
Dr. Arthur Szlam, assistant professor of mathematics at The City College of New York, has been awarded a prestigious Sloan Research Fellowship for 2013. Professor Szlam develops mathematics for cutting-edge applications in machine learning, a branch of artificial intelligence research focused on improving the abilities of computers to learn in a more human way.
The Alfred P. Sloan Foundation fellowships identify early-career scientists and scholars judged to be rising stars, whose achievements and independent scholarship demonstrate their potential to become leaders in their field. “The Sloan Research Fellows are the best of the best among young scientists,” said Dr. Paul Joskow, president of the Sloan Foundation. “If you want to know where the next big scientific breakthrough will come from, look to these extraordinary men and women.”
The Foundation added 126 researchers from the U.S. and Canada to its ranks this year. Each will receive $50,000 to further their research. A blue ribbon panel of three mathematicians judged Professor Szlam to be among the most exciting and promising researchers in the field, said a spokesman for the Sloan Foundation.
“It’s a huge prize and huge recognition,” said Professor Christian Wolf, chair of the mathematics department at City College. “We are very, very proud of having him in the department.”
Professor Szlam stands out for several reasons, one of which is his unusual approach, Professor Wolf explained. Applied mathematicians often begin with a particular mathematical technique and then find ways to apply it to solve problems in the real world – such as equipment failure or traffic routing.
“He does it the opposite way,” said Professor Wolf. “He looks at interesting applications and modifies existing mathematical tools – or invents the necessary mathematics – to solve these problems. Frequently he comes up something completely new. In the past year his research has really exploded.”
The mathematics for machine learning help a computer system discover how to analyze new types of data without being programmed with the specific steps to do so. Professor Szlam focuses on computer vision, allowing a computer to learn to distinguish and categorize objects in a collection of photos, for example, to allow precise image searching and sorting.
The field is still in its infancy, says Szlam, but the kinds of problems posed by computer vision and machine learning have spurred tremendous growth in mathematical tools.
“For most of the history of the universe – or the history of humans – the most interesting math came from physics,” Professor Szlam explained. “Now a lot of math is coming from machine learning and machine vision – this is a new place to get inspiration.”
This is also the math that could lead to so-called strong AI, the kind of artificial intelligence imagined in Ironman films, said Professor Szlam. “When Tony Stark is talking to his computer – if you’re interested in making that happen, the basic steps of machine vision come first.”
“The interesting thing here is that even if you don’t care about machine learning, it is inspiring really beautiful math,” he added.
Only six young faculty members from across The City University of New York have previously won the prize, created in 1955. Four of these hailed from City College at the time they won. In addition to Professor Szlam, two are current faculty: Professor of Mathematics Thea Pignataro (Sloan Fellow 1990) and Distinguished Professor of Science and Engineering – Physics, Robert Alfano (Sloan Fellow 1975). City College Professor of Mathematics Jay Jorgenson also won the prize in 1994, when he was at Yale University.
“The Sloan gives you a tremendous amount of breathing room and time to think about things,” said Professor Szlam. With that freedom he plans to attend various mathematics institutes for intensive research in the next year.
Of the problems he will tackle, he said, “The main goal for mathematics is to do something pretty – but it’s nice to have some goal, something to push you off.”
On the Internet: