Talks and Abstracts
Moon Duchin, Associate Professor of Mathematics at Tufts University
"Math, Race, and Voting Rights"
Many U.S. elections are carried out in districted plurality elections-- whoever gets the most votes in a certain geographical terrain wins! This system isn't ideal for securing representation for minorities, whether we're talking about Latino voters in Chicago or Republicans in Massachusetts. Bringing computation, statistics, and geometry together can create new tools and interventions in data science for civil rights. I'll describe some recent modeling projects as well as new research directions.
Alyssa Goodman, Robert Wheeler Wilson Professor of Applied Astronomy, Harvard University
"Glue-ing Together the Universe"
Astronomers have a long history of visualization. Going back only as far as Galileo, discoveries were made using sketches of celestial objects moving over time. Today, Astronomy inquiries can, and often do, make use of petabytes of data at once. Huge surveys are analyzed statistically to understand tiny fluctuations that hint at the fundamental nature of the Universe, and myriad data sets, from telescopes across the globe and in space are brought together to solve problems ranging from the nature of black holes to the structure of the Milky Way to the origins of planets like Earth. In this talk, I will summarize the state of partnerships between astronomical, physical, and computational approaches to gleaning insight from combinations of scientific and information visualization in Astrophysics. In particular, I will discuss how the “glue” linked-view visualization environment (glueviz.org), developed originally to facilitate high-dimensional data exploration in Astronomy and Medicine, can be extended to many other fields of data-driven inquiry. In addition, I will explain how the current open-source, plug & play, approach to software facilitates the combination of powerful programs and projects such as glue, WorldWide Telescope, ESASky, OpenSpace and the Zooniverse Citizen Science platform. Throughout the talk, I will emphasize the commonalities amongst many fields of science that rely on high-dimensional data. I will highlight our team's recent work on “The Radcliffe Wave” as a great example of a discovery enabled by data science and visualization.
Shirley Liu, Professor of Biostatistics and Computational Biology at the Department of Data Sciences at Dana-Farber Cancer Institute and Harvard University
Asuman Ozdaglar, Distinguished Professor of Engineering, Department Head, MIT, EECS
"Robustness in Machine Learning and Optimization: A Minmax Approach"
Jessica Stauth, Managing Director, Fidelity Labs
"Putting the Science back in Data Science: applying basic research principles to modern data science business problems"
It’s been over 7 years since DJ Patil declared “Data Scientist” the Sexiest Job of the 21st Century in the Harvard Business Review. What have we learned in the first decade of the data scientist? In this talk we will touch of some of the big wins and the existential challenges that still face the field of data science. Specifically we will review the case for hypothesis-driven science in the increasingly discovery-driven world of machine learning data science.