Piecewise Training for Undirected Models
Charles Sutton, Andrew McCallum
For many large undirected models that arise in real-world applications, exact maximumlikelihood training is intractable, because it requires computing marginal distributions of the model. Conditional training is even more difficult, because the partition function depends not only on the parameters, but also on the observed input, requiring repeated inference over each training example. An appealing idea for such models is to independently train a local undirected classifier over each clique, afterwards combining the learned weights into a single global model. In this paper, we show that this piecewise method can be justified as minimizing a new family of upper bounds on the log partition function. On three natural-language data sets, piecewise training is more accurate than pseudolikelihood, and often performs comparably to global training using belief propagation.
PDF Link: /papers/05/p568-sutton.pdf
AUTHOR = "Charles Sutton
and Andrew McCallum",
TITLE = "Piecewise Training for Undirected Models",
BOOKTITLE = "Proceedings of the Twenty-First Conference Annual Conference on Uncertainty in Artificial Intelligence (UAI-05)",
PUBLISHER = "AUAI Press",
ADDRESS = "Arlington, Virginia",
YEAR = "2005",
PAGES = "568--575"