Convex cost closure and Markov Random Fields problems: Applications and fastest-possible algorithms

Neyman Seminar
Mar 1, 2017 4:00pm to 5:00pm
1011 Evans Hall
Happening As Scheduled
Many problems in fitting observations while satisfying rank order constraints, occur in contexts of learning, Lipschitz regularization and Isotonic regression (with or without fused Lasso). All these problems can be abstracted as a convex cost closure problem which is to minimize the cost of deviating from the observations while satisfying rank order constraints. Any feasible solution that...
Dorit S. Hochbaum, IEOR dept UC Berkeley