A Hybrid Loss for Multiclass and Structured Prediction

By Javen Qinfeng Shi, Mark D. Reid, Tibério S. Caetano, Anton van den Hengel, Zhenhua Wang. 2014

7 citations recorded by the source corpus.

We propose a novel hybrid loss for multiclass and structured prediction problems that is a convex combination of a 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 between the probabilities of the best label and the second best label. 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 least as well as-and often better than-both of its constituent losses on a variety of tasks, such as human action recognition. In doing so we also provide an empirical comparison of the efficacy of probabilistic and margin based approaches to multiclass and structured prediction.

Subfields: Artificial Intelligence, Optimization & Training, Statistical Learning Theory

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Optimization & Training

In Statistical Learning Theory

Citations in the corpus

On Structured Prediction Theory with Calibrated Convex Surrogate Losses (2017, neutral): “, 2001), Φlog(f ,y) := − log fy + log ∑ ŷ∈Y exp fŷ, and their hybrids (Hazan & Urtasun, 2010; Shi et al., 2015).”
On the Consistency of Ordinal Regression Methods (2014, methodology): “We will now apply the notion of F-consistency or parametric consistency of (Shi et al., 2015) to the threshold-based setting.”