This paper presents an efficient implementation of linearised lattice minimum Bayes-risk decoding using weighted finite state transducers. We introduce transducers to efficiently count lattice paths containing n-grams and use these to gather the required statistics. We show that these procedures can be implemented exactly through simple transformations of word sequences to sequences of n-grams....

We consider bandit problems involving a large (possibly infinite) collection of arms, in which the expected reward of each arm is a linear function of an r-dimensional random vector Z ∈ Rr, where r ≥ 2. The objective is to minimize the cumulative regret and Bayes risk. When the set of arms corresponds to the unit sphere, we prove that the regret and Bayes risk is of order Θ(r √ T ), by establis...

The Pitman closeness ordering of statistical decision rules is considered in the general framework of comparative risk functions. The notions of admis-sibility and of the Bayes rule in this setting (C{admissibility and C{Bayes rule) are introduced. C{admissibility of C{Bayes procedures is demonstrated along with the geometric interpretation of these properties. It is shown that for general comp...

A new procedure for learning cost-sensitive SVM classifiers is proposed. The SVM hinge loss is extended to the cost sensitive setting, and the cost-sensitive SVM is derived as the minimizer of the associated risk. The extension of the hinge loss draws on recent connections between risk minimization and probability elicitation. These connections are generalized to costsensitive classification, i...

Minimum Bayes Risk (MBR) has been used as a decision rule for both singlesystem decoding and system combination in machine translation. For system combination, we argue that common MBR implementations are actually not correct, since probabilities in the hypothesis space cannot be reliably estimated. These implementations achieve the effect of consensus decoding (which may be beneficial in its o...

A new variant of AdaBoost is applied to a Minimum Bayes Risk discriminative training procedure that directly aims at reducing Word Error Rate for Automatic Speech Recognition. Both techniques try to improve the discriminative power of a classifier and we show that can be combined together to yield even better performance on a small vocabulary continuous speech recognition task. Our results also...

A two-factor experiment with interaction between factors wherein observations follow an Inverse Gaussian model is considered. Analysis of the experiment is approached via an empirical Bayes procedure. The conjugate family of prior distributions is considered. Bayes and empirical Bayes estimators are derived. Application of the procedure is illustrated on a data set, which has previously been an...