Max-margin Markov networks

DMCA

Max-margin Markov networks (2003)

Cached

Download Links

Other Repositories/Bibliography

CiteULike
DBLP

by Ben Taskar , Carlos Guestrin , Daphne Koller

Citations:

603 - 15 self

BibTeX

@INPROCEEDINGS{Taskar03max-marginmarkov,
    author = {Ben Taskar and Carlos Guestrin and Daphne Koller},
    title = {Max-margin Markov networks},
    booktitle = {},
    year = {2003},
    publisher = {MIT Press}
}

OpenURL

Abstract

In typical classification tasks, we seek a function which assigns a label to a single object. Kernel-based approaches, such as support vector machines (SVMs), which maximize the margin of confidence of the classifier, are the method of choice for many such tasks. Their popularity stems both from the ability to use high-dimensional feature spaces, and from their strong theoretical guarantees. However, many real-world tasks involve sequential, spatial, or structured data, where multiple labels must be assigned. Existing kernel-based methods ignore structure in the problem, assigning labels independently to each object, losing much useful information. Conversely, probabilistic graphical models, such as Markov networks, can represent correlations between labels, by exploiting problem structure, but cannot handle high-dimensional feature spaces, and lack strong theoretical generalization guarantees. In this paper, we present a new framework that combines the advantages of both approaches: Maximum margin Markov (M 3) networks incorporate both kernels, which efficiently deal with high-dimensional features, and the ability to capture correlations in structured data. We present an efficient algorithm for learning M 3 networks based on a compact quadratic program formulation. We provide a new theoretical bound for generalization in structured domains. Experiments on the task of handwritten character recognition and collective hypertext classification demonstrate very significant gains over previous approaches. 1

Keyphrases

max-margin markov network collective hypertext classification problem structure lack strong theoretical generalization guarantee maximum margin markov many real-world task single object kernel-based method high-dimensional feature structured domain efficient algorithm multiple label new framework cannot handle high-dimensional feature space high-dimensional feature space compact quadratic program formulation strong theoretical guarantee typical classification task probabilistic graphical model much useful information kernel-based approach many task new theoretical bound support vector machine markov network significant gain previous approach handwritten character recognition structured data

DMCA