Data science has transformed fields from biology to astronomy, and social networks to politics, influencing most aspects of modern life.
To transform data science, a multidisciplinary team of researchers at the University of Wisconsin-Madison wants to return to the fundamentals.
The group includes 14 researchers from engineering, mathematics, statistics and computer sciences in creating the Institute for Foundations of Data Science. The new institute will play a key role in the future of data science, developing fundamental techniques for handling increasingly massive data sets in shorter times.
A $1.5 million grant from the National Science Foundation Transdisciplinary Research in Principles of Data Science (TRIPODS) initiative supports the new institute, which includes several data-science experts from the Department of Electrical and Computer Engineering (ECE) at UW-Madison.
The institute is among 12 NSF-funded TRIPODS projects at 14 institutions and falls under the NSF “Harnessing the Data Revolution,” initiative—first of the agency’s 10 “big ideas” it deems essential for future investment.
The UW-Madison institute will have three research themes. ECE professors Robert Nowak and Dimitris Papailiopoulos, and ECE and math professor Nigel Boston, are leading projects under the algebra and optimization in data science theme. ECE professor Rebecca Willett and statistics professor Michael Newton are leading projects in the graphs and networks in data science theme, while ECE professor Po-Ling Loh and computer sciences professor Jerry Zhu are leading projects in the data acquisition theory and methods theme. (Papailiopoulos and Loh also are affiliated with the Grainger Institute for Engineering at UW-Madison.) “All these projects bring together researchers from math, statistics, and computer science,” says Willett. “Our electrical and computer engineering research in data science integrates all three of these areas, and hence we are heavily represented in the interdisciplinary project.”
Widespread applications of data science depend on fundamental mathematical and statistical tools for acquiring, handling and solving mountains of data—and with those theoretical tools in place, researchers then can extract meaning from huge quantities of data.
“Underlying all of what goes on in data science are these algorithms and formulations and models,” says Stephen Wright, a professor of computer sciences at UW-Madison and a leader of the institute. “These theoretical foundations are what makes the data revolution possible.”
Researchers in the institute will play a key role in the future of data science, developing fundamental techniques for handling increasingly massive data sets in shorter times. “Data is accelerating the pace of scientific discovery and innovation,” says Jim Kurose, NSF assistant director for computer and information science and engineering. “These new TRIPODS projects will help build the theoretical foundations of data science that will enable continued data-driven discovery and breakthroughs across all fields of science and engineering.”
The UW–Madison proposal also was one of three selected for co-funding from the new, interdisciplinary NSF “Convergence” program—which aims to integrate knowledge, techniques and expertise from many disciplines to address key scientific and societal challenges.
The institute will be housed within the Wisconsin Institute for Discovery at UW-Madison—itself an interdisciplinary organization—as part of its new data science hub, where it will contribute to the hub’s applied and systems-oriented components.
The institute will not be restricted to the 14 UW–Madison faculty involved in the proposal, but rather, will collaborate with colleagues at UW-Madison and other institutions.
Researchers in the institute also plan to enrich graduate programs in data science and foster outreach to industrial partners with interests in fundamental data science research.
The institute could evolve into a larger phase-two NSF institute—a goal of its researchers as they rally around activity they see as critical to the future of data science.
“You can’t have developments in data science without having continuing attention to the fundamentals,” says Wright. “There is a lot of excitement about the algorithmic space, the theoretical space and the mathematical underpinnings, and that’s going to drive a lot of the future of data science.”