home
past talks
resources
conferences
|
|
|
Machine Learning for Combinatorial Optimization
Robert Moll
UMass
Abstract
Combinatorial or discrete optimization, both for deterministic and stochastic problems, is a vital area of study at the boundary between mathematics and computer science. Its applications extend from management science to engineering. In this talk I will discuss three applications -- graph coloring, rectangle layout, and a stochastic partitioning and routing problem -- that illustrate how learning can interact with and improve the performance of more traditional approaches to optimization problems Then I'll present some broad, "blue-sky" ideas about where this and related work might or should lead.
Back to ML Lunch home
|