A Consistent Regularization Approach for Structured Prediction
This work addresses structured prediction for machine learning applications, but appears incremental as it builds on existing regularization and surrogate loss techniques.
The authors tackled structured prediction problems by proposing a regularization approach that embeds structured outputs in a linear space, proving universal consistency and finite sample bounds while demonstrating practical usefulness in experiments.
We propose and analyze a regularization approach for structured prediction problems. We characterize a large class of loss functions that allows to naturally embed structured outputs in a linear space. We exploit this fact to design learning algorithms using a surrogate loss approach and regularization techniques. We prove universal consistency and finite sample bounds characterizing the generalization properties of the proposed methods. Experimental results are provided to demonstrate the practical usefulness of the proposed approach.