(please see the main page for schedule information)
The modeling of the geographic distribution of a plant or animal species is a critical problem in conservation biology: to save a threatened species, one first needs to know where it prefers to live, and what its requirements are for survival. We propose the use of maximum-entropy methods for this challenging problem, a set of decades-old techniques that happen to fit this problem very cleanly and effectively. We prove theoretically that maxent methods enjoy strong theoretical performance guarantees. We also describe a number of experiments comparing maxent with a standard distribution-modeling tool, called GARP, for several different animal species. In our study, maxent is substantially superior to the standard method, performing well with fairly few examples. We also show that maxent models can be easily interpreted by human experts, a property of considerable practical importance.This talk includes joint work with Steven Phillips, Robert Anderson and Miroslav Dudik.
Robert Schapire received his ScB in math and computer science from Brown University in 1986, and his SM (1988) and PhD (1991) from MIT under the supervision of Ronald Rivest. After a short post-doc at Harvard, he joined the technical staff at AT&T Labs (formerly AT&T Bell Laboratories) in 1991 where he remained for eleven years. At the end of 2002, he became a Professor of Computer Science at Princeton University. His awards include the 1991 ACM Doctoral Dissertation Award as well as the 2003 Goedel Prize. His main research interest is in theoretical and applied machine learning.