Institute for Foundations of Data Science

2017 - presentpresent

As data continues to accumulate at an ever increasing rate, so does the need for powerful and novel methods to extract information from data, in a form that is useful to individuals, society, researchers, and commerce. Research into the fundamentals of data science brings people with expertise in mathematics, statistics, and theoretical computer science together in a concerted transdiciplinary effort to explore new approaches to the formulation and solution of problems in data analysis. A previous generation of researchers at UW-Madison made foundational contributions to data science, in such areas as kernel learning, splines, and experimental design. The current generation of faculty is carrying forward this tradition. During the past decade, researchers across campus have done important work in diverse aspects of fundamental data science, as well as in applications to numerous areas of domain science, engineering, and medicine. This group now comes together to propose a new Institute for Foundations of Data Science (IFDS) at UW-Madison. Building on previous work, and pursuing new goals sparked at the interfaces of mathematics, statistics, and theoretical computer science, IFDS aims to produce excellent research and to epitomize the possibilities of collaborative approach to investigating fundamental issues in data science. IFDS will be a prototype for a larger Phase-II institute, to be organized and motivated along similar lines, but expanding to involve other researchers in theoretical data science and possibly other institutions. IFDS will be organized around three research themes, selected because they encompass problems at the cutting edge of fundamental data-science research and because they have the potential to benefit maximally from collaborations among researchers in mathematics, statistics, and theoretical computer sciences. These themes are Algebra and Optimization in Data Science, Graphs and Networks in Data Science, and Data Acquisition Theory and Methods. Several specific topics are proposed within each theme; each topic has a leader and a transdisciplinary research team of IFDS senior personnel and students. All topics represent areas of significant current interest in data science, because of their intrinsic fundamental import and their wide applicability. Advances in fundamental data science quickly translate into practice because of the enormous commercial impact of the area. IFDS research will be no exception to this rule. There will be impacts too on scientific and medical applications of data science, not least because of close contacts between IFDS researchers and data-related centers and research groups in these areas at UW-Madison. The majority of IFDS’s budget is for training of junior researchers in fundamental data science. Strategies for deploying these funds will maximize their impact on graduate training in data science at UWMadison, and will help put in place a new generation of researchers with the ability to collaborate across disciplines. Workshops will bring together IFDS personnel and outside researchers, and will help to set research priorities for the wider community in fundamental data science. Outreach to the Madison community will build on previous activities of IFDS personnel and will be facilitated by IFDS’s locus in the Discovery Building.