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).

Effects of damming on population sustainability of Chinese sturgeon, Acipenser sinensis: evaluation of optimal conservation measures

The numbers of spawning sites for Chinese sturgeon have been drastically reduced since the construction of the Gezhouba Dam across the Yangtze River. This dam has blocked migration of Chinese sturgeon to their historic spawning ground causing a significant decline of the Chinese sturgeon population. We conducted a VORTEX population viability analysis to estimate the sustainability of the population and to quantify the efficiency of current and alternative conservation procedures. The model predicted the observed decline of Chinese sturgeon, resulting from the effect of the Gezhouba Dam. These simulations demonstrated the potential interest of two conservation measures: increasing spawning area and reducing predation on sturgeon eggs. The simulations also demonstrated that the actual restocking program is not sufficient to sustain sturgeon population as the artificial reproduction program induce the loss of more wild mature adults that the recruitment expected by the artificial reproduction.

A reversible infinite HMM using normalised random measures

Copyright 2014 by the author(s). We present a nonparametric prior over reversible Markov chains. We use completely random measures, specifically gamma processes, to construct a countably infinite graph with weighted edges. By enforcing symmetry to make the edges undirected we define a prior over random walks on graphs that results in a reversible Markov chain. The resulting prior over infinite transition matrices is closely related to the hierarchical Dirichlet process but enforces reversibility. A reinforcement scheme has recently been proposed with similar properties, but the de Finetti measure is not well characterised. We take the alternative approach of explicitly constructing the mixing measure, which allows more straightforward and efficient inference at the cost of no longer having a closed form predictive distribution. We use our process to construct a reversible infinite HMM which we apply to two real datasets, one from epigenomics and one ion channel recording.

Production planning of multi-stage multi-option seru production systems with sustainable measures

With the concerns over climate change and the escalation in worldwide population, sustainable development attracts more and more attention of academia, policy makers, and businesses in countries. Sustainable manufacturing is an inextricable measure to achieve sustainable development since manufacturing is one of the main energy consumers and greenhouse gas contributors. In the previous researches on production planning of manufacturing systems, environmental factor was rarely considered. This paper investigates the production planning problem under the performance measures of economy and environment with respect to seru production systems, a new manufacturing system praised as Double E (ecology and economy) in Japanese manufacturing industries. We propose a mathematical model with two objectives minimizing carbon dioxide emission and makespan for processing all product types by a seru production system. To solve this mathematical model, we develop an algorithm based on the non-dominated sorting genetic algorithm II. The computation results and analysis of three numeral examples confirm the effectiveness of our proposed algorithm. © 2014 Elsevier Ltd. All rights reserved.

Detecting a set of entanglement measures in an unknown tripartite quantum state by local operations and classical communication

We propose a more general method for detecting a set of entanglement measures, i.e., negativities, in an arbitrary tripartite quantum state by local operations and classical communication. To accomplish the detection task using this method, three observers do not need to perform partial transposition maps by the structural physical approximation; instead, they only need to collectively measure some functions via three local networks supplemented by a classical communication. With these functions, they are able to determine the set of negativities related to the tripartite quantum state.

Measures for describing the size distribution of sediments

Numerous measures are used in the literature to describe the grain-size distribution of sediments. Consideration of these measures indicates that parameters computed from quartiles may not be as significant as those based on more rigorous statistical concepts. In addition, the lack of standardization of descriptive measures has resulted in limited application of the findings from one locality to another. The use of five parameters that serve as approximate graphic analogies to the moment measures commonly employed in statistics is recommended. The parameters are computed from five percentile diameters obtained from the cumulative size-frequency curve of a sediment. They include the mean (or median) diameter, standard deviation, kurtosis, and two measures of skewness, the second measure being sensitive to skew properties of the "tails" of the sediment distribution. If the five descriptive measures are listed for a sediment, it is possible to compute the five percentile diameters on which they are based (phi 5 , phi 16 , phi 50 , phi 84 , and phi 95 ), and hence five significant points on the cumulative carve of the sediment. This increases the value of the data listed for a sediment in a report, and in many cases eliminates the necessity of including the complete mechanical analysis of the sediment. The degree of correlation of the graphic parameters to the corresponding moment measures decreases as the distribution becomes more skew. However, for a fairly wide range of distributions, the first three moment measures can be ascertained from the graphic parameters with about the same degree of accuracy as is obtained by computing rough moment measures.

Impacts of water conservancy and soil conservation measures on annual runoff in the Chaohe River Basin during 1961-2005

National Key Technology RD Program [2006BAD03A02]

The Impact of Environmental Non-Tariff Measures on China's Agricultural Exports

This article assesses the impact of environmental non-tariff measures on China's agricultural exports. It uses two measures to show which countries make most intensive use of environmental non-tariff measures and which agricultural products exported by China are the most affected. Environmental non-tariff measures were most prevalence in Malaysia, Vietnam, Philippines, Mexico, Japan, Russian, USA, Germany, and Netherland. Labor-intensive agricultural products, such as fish, vegetables, and fruit were most seriously affected.