Conditional Random Fields and Support Vector Machines: A Hybrid Approach

Details Conditional Random Fields and Support Vector Machines: A Hybrid Approach Conditional Random Fields and Support Vector Machines: A Hybrid Approach

2010-09-17

arxiv.org/abs/1009.3346v1

Computer Science

Learning

16 pages, 3 figures

Scientific paper

We propose a novel hybrid loss for multiclass and structured prediction problems that is a convex combination of log loss for Conditional Random Fields (CRFs) and a multiclass hinge loss for Support Vector Machines (SVMs). We provide a sufficient condition for when the hybrid loss is Fisher consistent for classification. This condition depends on a measure of dominance between labels - specifically, the gap in per observation probabilities between the most likely labels. We also prove Fisher consistency is necessary for parametric consistency when learning models such as CRFs. We demonstrate empirically that the hybrid loss typically performs as least as well as - and often better than - both of its constituent losses on variety of tasks. In doing so we also provide an empirical comparison of the efficacy of probabilistic and margin based approaches to multiclass and structured prediction and the effects of label dominance on these results.

Affiliated with

Also associated with

No associations

Rating

Conditional Random Fields and Support Vector Machines: A Hybrid Approach does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.

If you have personal experience with Conditional Random Fields and Support Vector Machines: A Hybrid Approach, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Conditional Random Fields and Support Vector Machines: A Hybrid Approach will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFWR-SCP-O-441676