Max-Margin Infinite Hidden Markov Models
Aonan Zhang, Jun Zhu, Bo Zhang · 2014
Infinite hidden Markov models (iHMMs) are nonparametric Bayesian extensions of hidden Markov models (HMMs) with an infinite number of states. Though flexible in describing sequen-tial data, the generative formulation of iHMMs could limit their discriminative ability in sequen-tial prediction tasks. Our paper introduces max-margin infinite HMMs (M2iHMMs), new infinite HMMs that explore the max-margin principle for discriminative learning. By using the theory of Gibbs classifiers and data augmentation, we de-velop efficient beam sampling algorithms with-out making restricting mean-field assumptions or truncated approximation. For single variate clas-sification, M2iHMMs reduce to a new formula-tion of DP mixtures of max-margin machines. Empirical results on synthetic and real data sets show that our methods obtain superior perfor-mance than other competitors in both single vari-ate classification and sequential prediction tasks.