Models of neural machine translation are often from a discriminative family of encoderdecoders that learn a conditional distribution of a target sentence given a source sentence. In this paper, we propose a variational model to learn this conditional distribution for neural machine translation: a variational encoderdecoder model that can be trained end-to-end. Different from the vanilla encoder-decoder model that generates target translations from hidden representations of source sentences alone, the variational model introduces a continuous latent variable to explicitly model underlying semantics of source sentences and to guide the generation of target translations. In order to perform efficient posterior inference and large-scale training, we build a neural posterior approximator conditioned on both the source and the target sides, and equip it with a reparameterization technique to estimate the variational lower bound. Experiments on both Chinese-English and EnglishGerman translation tasks show that the proposed variational neural machine translation achieves significant improvements over the vanilla neural machine translation baselines.

Implicit discourse relation recognition remains a serious challenge due to the absence of discourse connectives. In this paper, we propose a Shallow Convolutional Neural Network (SCNN) for implicit discourse relation recognition, which contains only one hidden layer but is effective in relation recognition. The shallow structure alleviates the overfitting problem, while the convolution and nonlinear operations help preserve the recognition and generalization ability of our model. Experiments on the benchmark data set show that our model achieves comparable and even better performance when comparing against current state-of-the-art systems.

No abstract

In the human fungal pathogen Candida albicans, environmental pH has profound effects on morphogenesis and response to extracellular pH is clearly relevant to the pathogenicity of this fungus. Yeast cells have evolved a complex network of mechanisms in response to the environmental pH and they often require the integration of the Rim101 and calcineurin/Crz1 signaling pathways. Ca(2+) burst is a common cellular response when cells are exposed to environmental stresses; therefore, in this study, we asked whether it follows the same case under alkaline stress and whether this calcium change is regulated by Rim101p and Crz1p. We confirmed the calcium influx was activated by KOH stimuli using a flow cytometry-based method, but it was obviously abolished in cells lacking MID1 or CCH1. We also found that alkaline pH-induced activation of the PHO89 promoter was blocked without the same gene; moreover, the response was Crz1p- and Rim101p-dependent. Finally, we investigated the regulation role of Rim101p and Crz1p in calcium influx through MID1, CCH1 and YVC1 genes, which were all downregulated in rim101Δ/Δ and crz1Δ/Δ mutants. The important role of calcium influx in the alkaline stress response and its regulation suggested a potential integration effect of Rim101 and Crz1 signaling pathways in C. albicans.

The maintenance of cellular calcium homeostasis is associated with cellular signalling transduction and the functions of many membrane compartments, especially endoplasmic reticulum (ER) function. ER-localized proteins that serve to maintain ER and cellular calcium homeostasis in Candida albicans are still unclear. In this study, Spf1, the putative C. albicans homologue of the Saccharomyces cerevisiae ER-localized P-type calcium ATPase ScSpf1, was investigated for its roles in cellular calcium homeostasis, hyphal development and virulence. We constructed an Spf1 null mutant which showed decreased vegetative growth rate and hypersensitivity to EGTA, high-level calcium and antifungal drugs. Similar to treatments of ER stress agents, deletion of SPF1 stimulated calcium influx in the presence of FK506, resulting in an increase in cellular calcium contents, and induced expression of the calcium-dependent response elements gene CCH1, which is essential for the cell calcium survival pathway. Moreover, the spf1 null mutant had defects in hyphal development and biofilm formation, and was severely attenuated in virulence. These findings provided phenotypic evidence supporting roles for Spf1 in the maintenance of cellular calcium homeostasis, ER stress responses, hyphal development, biofilm formation and virulence in C. albicans.

No abstract

Robust model-reference control for descriptor linear systems with structural parameter uncertainties is investigated. A sufficient condition for existing a model-reference zero-error asymptotic tracking controller is given. It is shown that the robust model reference control problem can be decomposed into two subproblems: a robust state feedback stabilization problem for descriptor systems subject to parameter uncertainties and a robust compensation problem. The latter aims to find three coefficient matrices which satisfy four matrix equations and simultaneously minimize the effect of the uncertainties to the tracking error. Based on a complete parametric solution to a class of generalized Sylvester matrix equations, the robust compensation problem is converted into a minimization problem with quadratic cost and linear constraints. A numerical example shows the effect of the proposed approach.

No abstract