The direct outcome of our work is a new spectral regularization method for max-margin structured prediction. Our experiments confirm that our proposed ...

The direct outcome of our work is a new spectral regularization method for max-margin structured prediction. Our exper- iments confirm that our proposed ...

Abstract. We frame max-margin learning of latent variable structured prediction models as a convex opti- mization problem, making use of scoring func-.

Table of values Ñ must correspond to valid IO-OOM. 2. Regularizer over table Ñ must correspond to #states of IO-OOM. 3. Recover parameters of A from this ...

The direct outcome of our work is a new spectral regularization method for max-margin structured prediction. Our experiments confirm that our proposed ...

The direct outcome of our work is a new spectral regularization method for max-margin structured prediction. Our experiments confirm that our proposed ...

1. Obtain H optimizing L, via matrix completion techniques (Balle & Mohri 2012). 2. Recover A from H using the spectral method of (Hsu, Kakade & Zhang 2009).

The direct outcome of our work is a new spectral regularization method for max-margin structured prediction. Our experiments confirm that our proposed ...

The direct outcome of our work is a new spectral regularization method for max-margin structured prediction. Our experiments confirm that our proposed ...

Ariadna Quattoni, Borja Balle, Xavier Carreras, Amir Globerson: Spectral Regularization for Max-Margin Sequence Tagging. ICML 2014: 1710-1718.