When: October 19, 2016, 12:30 PM
Location: 3rd Floor Orchard View Room , Discovery Building
Contact: 608-316-4401, firstname.lastname@example.org
Speeding up Machine Learning using Graphs and Codes
Mixed-integer convex optimization problems are convex problems with the additional (non-convex) constraints that some variables may take only integer values. Despite the past decades’ advances in algorithms and technology for both mixed-integer *linear* and *continuous, convex* optimization, mixed-integer convex optimization problems have remained relatively more challenging and less widely used in practice. In this talk, we describe our recent algorithmic work on mixed-integer convex optimization which has yielded advances over the state of the art, including the globally optimal solution of open benchmark problems. Based on our developments, we have released Pajarito, an open-source solver written in Julia and accessible from popular optimization modeling frameworks. Pajarito is immediately useful for solving challenging mixed combinatorial-continuous problems arising from engineering and statistical applications.
Disciplined Convex Programming (DCP) is an algebraic modeling concept proposed by Grant, Boyd, and Ye. DCP underlies the successful software package CVX and more recent implementations CVXPY and Convex.jl. In this tutorial, we will present DCP through theory and software, demonstrating its usefulness for anyone interested in modeling and solving nonlinear optimization problems. Although DCP was envisioned strictly for convex optimization, we will review very recent work by different researchers on how ideas based on DCP can help solve nonconvex problems.
Note: Attendees should bring their laptops if they want to gain hands-on experience.
I am a fifth-year Ph.D. candidate in Operations Research at MIT advised by Juan Pablo Vielma. I received my B.S. in Applied Mathematics and M.S. in Statistics from the University of Chicago in 2011. After graduating, I spent a year as a researcher at Argonne National Laboratory before starting at MIT.
My research interests span diverse areas of mathematical optimization, with a unifying theme of developing new methodologies for large-scale optimization drawing from motivating applications in renewable energy. I have published work in chance constrained optimization, mixed-integer conic optimization, robust optimization, stochastic programming, algebraic modeling, automatic differentiation, numerical linear algebra, and parallel computing techniques for large-scale problems.
In 2012, Iain Dunning and I (later joined by Joey Huchette) started developing JuMP, an open-source algebraic modeling language for optimization. Since then, JuMP has been used for teaching in at least 10 universities and by numerous researchers and companies worldwide. I am a co-founder of the JuliaOpt organization which has brought together early adopters in academia and industry with the goal of developing high quality open-source software for optimization in Julia. I’m always interested to hear of who’s using JuMP, so please get in touch.
The weekly SILO seminar series is made possible through the generous support of the 3M Company and its Advanced Technology Group
SILO is a lecture series with speakers from the UW faculty, graduate students or invited researchers that discuss mathematical related topics. The seminars are organized by WID’s Optimization research group.
SILO’s purpose is to provide a forum that helps connect and recruit mathematically-minded graduate students. SILO is a lunch-and-listen format, where speakers present interesting math topics while the audience eats lunch.