Part of Advances in Neural Information Processing Systems 19 (NIPS 2006)
Samuel Gross, Olga Russakovsky, Chuong B., Serafim Batzoglou
We consider the problem of training a conditional random field (CRF) to maximize per-label predictive accuracy on a training set, an approach motivated by the principle of empirical risk minimization. We give a gradient-based procedure for minimizing an arbitrarily accurate approximation of the empirical risk under a Hamming loss function. In experiments with both simulated and real data, our optimization procedure gives significantly better testing performance than several current approaches for CRF training, especially in situations of high label noise.