Lattice-Based Minimum Error Rate Training Using Weighted Finite-State Transducers with Tropical Polynomial Weights
نویسندگان
چکیده
Minimum Error Rate Training (MERT) is a method for training the parameters of a loglinear model. One advantage of this method of training is that it can use the large number of hypotheses encoded in a translation lattice as training data. We demonstrate that the MERT line optimisation can be modelled as computing the shortest distance in a weighted finite-state transducer using a tropical polynomial semiring.
منابع مشابه
EM training of finite-state transducers and its application to pronunciation modeling
Recently, finite-state transducers (FSTs) have been shown to be useful for a number of applications in speech and language processing. FST operations such as composition, determinization, and minimization make manipulating FSTs very simple. In this paper, we present a method to learn weights for arbitrary FSTs using the EM algorithm. We show that this FST EM algorithm is able to learn pronuncia...
متن کاملMultilingual syllabification using weighted finite-state transducers
This paper describes an approach to syllabification that has been incorporated into the English and German text-to-speech systems at Bell Labs. Implemented as a weighted finite-state transducer, the syllabifier is easily integrated – via mathematical composition – into the finite-state based text analysis component of the textto-speech system. The weights are based on frequencies of onset, nucl...
متن کاملConstruction algorithms for polynomial lattice rules for multivariate integration
We introduce a new construction algorithm for digital nets for integration in certain weighted tensor product Hilbert spaces. The first weighted Hilbert space we consider is based on Walsh functions. Dick and Pillichshammer calculated the worst-case error for integration using digital nets for this space. Here we extend this result to a special construction method for digital nets based on poly...
متن کاملEfficient Path Counting Transducers for Minimum Bayes-Risk Decoding of Statistical Machine Translation Lattices
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....
متن کاملEm Training of Finite-sta and Its Application to Pronu
Recently, finite-state transducers (FSTs) have been shown to be useful for a number of applications in speech and language processing. FST operations such as composition, determinization, and minimization make manipulating FSTs very simple. In this paper, we present a method to learn weights for arbitrary FSTs using the EM algorithm. We show that this FST EM algorithm is able to learn pronuncia...
متن کامل