21.9.2007 HIIT Seminar: Janey Yu

HIIT seminars in fall 2007 will be held in hall **B222** of Exactum,
on Fridays starting at 10:15 a.m. Coffee available from 10.

Sep 21:
  Janey Yu
  Discriminative Training for Structured Predictions: An Efficient
  Optimization Method

We consider structured prediction problems with a parametrized linear
prediction function. Corresponding to such type of problems are many
applications: sentence alignment, image segmentation, and HMM, for
instance. Our focus is on solving the associated parameter optimization
problems in the large margin type of discriminative training. We propose
a dual optimization approach which uses the restricted simplicial
decomposition method to optimize a reparametrized dual problem. Our
reparametrization reduces the dimension of the space of the dual
function to one that is linear in the number of parameters and training
examples, and hence independent of the dimensionality of the prediction
outputs. This in conjunction with simplicial decomposition makes our
approach efficient. We discuss the connections of our approach with
related earlier works, and we show its advantages.

Joint work with Juho Rousu.

Last updated on 20 Sep 2007 by Teija Kujala - Page created on 21 Sep 2007 by Teija Kujala