N-gram posterior probability confidence measures for statistical machine translation: An empirical study

We report an empirical study of n-gram posterior probability confidence measures for statistical machine translation (SMT). We first describe an efficient and practical algorithm for rapidly computing n-gram posterior probabilities from large translation word lattices. These probabilities are shown to be a good predictor of whether or not the n-gram is found in human reference translations, motivating their use as a confidence measure for SMT. Comprehensive n-gram precision and word coverage measurements are presented for a variety of different language pairs, domains and conditions. We analyze the effect on reference precision of using single or multiple references, and compare the precision of posteriors computed from k-best lists to those computed over the full evidence space of the lattice. We also demonstrate improved confidence by combining multiple lattices in a multi-source translation framework. © 2012 The Author(s).

Probability-Matching Predictors for Extreme Extremes

A location- and scale-invariant predictor is constructed which exhibits good probability matching for extreme predictions outside the span of data drawn from a variety of (stationary) general distributions. It is constructed via the three-parameter {\mu, \sigma, \xi} Generalized Pareto Distribution (GPD). The predictor is designed to provide matching probability exactly for the GPD in both the extreme heavy-tailed limit and the extreme bounded-tail limit, whilst giving a good approximation to probability matching at all intermediate values of the tail parameter \xi. The predictor is valid even for small sample sizes N, even as small as N = 3. The main purpose of this paper is to present the somewhat lengthy derivations which draw heavily on the theory of hypergeometric functions, particularly the Lauricella functions. Whilst the construction is inspired by the Bayesian approach to the prediction problem, it considers the case of vague prior information about both parameters and model, and all derivations are undertaken using sampling theory.

System combination and score normalization for spoken term detection

Spoken content in languages of emerging importance needs to be searchable to provide access to the underlying information. In this paper, we investigate the problem of extending data fusion methodologies from Information Retrieval for Spoken Term Detection on low-resource languages in the framework of the IARPA Babel program. We describe a number of alternative methods improving keyword search performance. We apply these methods to Cantonese, a language that presents some new issues in terms of reduced resources and shorter query lengths. First, we show score normalization methodology that improves in average by 20% keyword search performance. Second, we show that properly combining the outputs of diverse ASR systems performs 14% better than the best normalized ASR system. © 2013 IEEE.

Efficient decoding with generative score-spaces using the expectation semiring

State-of-the-art speech recognisers are usually based on hidden Markov models (HMMs). They model a hidden symbol sequence with a Markov process, with the observations independent given that sequence. These assumptions yield efficient algorithms, but limit the power of the model. An alternative model that allows a wide range of features, including word- and phone-level features, is a log-linear model. To handle, for example, word-level variable-length features, the original feature vectors must be segmented into words. Thus, decoding must find the optimal combination of segmentation of the utterance into words and word sequence. Features must therefore be extracted for each possible segment of audio. For many types of features, this becomes slow. In this paper, long-span features are derived from the likelihoods of word HMMs. Derivatives of the log-likelihoods, which break the Markov assumption, are appended. Previously, decoding with this model took cubic time in the length of the sequence, and longer for higher-order derivatives. This paper shows how to decode in quadratic time. © 2013 IEEE.

Estimation of the detection probability for Yangtze finless porpoises (Neophocaena phocaenoides asiaeorientalis) with a passive acoustic method

Yangtze finless porpoises were surveyed by using simultaneous visual and acoustical methods from 6 November to 13 December 2006. Two research vessels towed stereo acoustic data loggers, which were used to store the intensity and sound source direction of the high frequency sonar signals produced by finless porpoises at detection ranges up to 300 m on each side of the vessel. Simple stereo beam forming allowed the separation of distinct biosonar sound source, which enabled us to count the number of vocalizing porpoises. Acoustically, 204 porpoises were detected from one vessel and 199 from the other vessel in the same section of the Yangtze River. Visually, 163 and 162 porpoises were detected from two vessels within 300 m of the vessel track. The calculated detection probability using acoustic method was approximately twice that for visual detection for each vessel. The difference in detection probabilities between the two methods was caused by the large number of single individuals that were missed by visual observers. However, the sizes of large groups were underestimated by using the acoustic methods. Acoustic and visual observations complemented each other in the accurate detection of porpoises. The use of simple, relatively inexpensive acoustic monitoring systems should enhance population surveys of free-ranging, echolocating odontocetes. (C) 2008 Acoustical Society of America.

Probability of Sediment Incipient Motion Under Complex Flows

Presented in this paper is a mathematical model to calculate the probability of the sediment incipient motion, in which the effects of the fluctuating pressure and the seepage are considered. The instantaneous bed shear velocity and the pressure gradient on the bed downstream of the backward-facing step flow are obtained according to the PIV measurements. It is found that the instantaneous pressure gradient on the bed obeys normal distribution. The probability of the sediment incipient motion on the bed downstream of the backward-facing step flow is given by the mathematical model. The predicted results agree well with the experiment in the region downstream of the reattachment point while a large discrepancy between the theory and experiment is seen in the region near the reattachment point. The possible reasons for this discrepancy are discussed.