Despite a lot of research efforts devoted in recent years, how to efficiently learn long-term dependencies from sequences still remains a pretty challenging task. As one of the key models for sequence learning, recurrent neural network (RNN) and its variants such as long short term memory (LSTM) and gated recurrent unit (GRU) are still not powerful enough in practice. One possible reason is that they have only feedforward connections, which is different from the biological neural system that is typically composed of both feedforward and feedback connections. To address this problem, this paper proposes a biologically-inspired deep network, called shuttleNet. Technologically, the shuttleNet consists of several processors, each of which is a GRU while associated with multiple groups of hidden states. Unlike traditional RNNs, all processors inside shuttleNet are loop connected to mimic the brain's feedforward and feedback connections, in which they are shared across multiple pathways in the loop connection. Attention mechanism is then employed to select the best information flow pathway. Extensive experiments conducted on two benchmark datasets (i.e UCF101 and HMDB51) show that we can beat state-of-the-art methods by simply embedding shuttleNet into a CNN-RNN framework.
Whether crop phenology changes are caused by change in managements or by climate change belongs to the category of problems known as detection-attribution. Three type of rice (early, late and single rice) in China show an average increase in Length of Growing Period (LGP) during 1991-2012: 1.0 +/- 4.8 day/decade ( standard deviation across sites) for early rice, 0.2 +/- 4.5 day/decade for late rice and 2.0 +/- 6.0 day/decade for single rice, based on observations from 141 long-term monitoring stations. Positive LGP trends are widespread, but only significant (P <0.05) at 25% of early rice, 22% of late rice and 38% of single rice sites. We developed a Bayes-based optimization algorithm, and optimized five parameters controlling phenological development in a process-based crop model (ORCHIDEE-crop) for discriminating effects of managements from those of climate change on rice LGP. The results from the optimized ORCHIDEE-crop model suggest that climate change has an effect on LGP trends dependent on rice types. Climate trends have shortened LGP of early rice (-2.0 +/- 5.0 day/decade), lengthened LGP of late rice (1.1 +/- 5.4 day/decade) and have little impacts on LGP of single rice (-0.4 +/- 5.4 day/decade). ORCHIDEEcrop simulations further show that change in transplanting date caused widespread LGP change only for early rice sites, offsetting 65% of climate change induced LGP shortening. The primary drivers of LGP change are thus different among the three types of rice. Management are predominant driver of LGP change for early and single rice. This study shows that complex regional variations of LGP can be reproduced with an optimized crop model. We further suggest that better documenting observational error and management practices can help reduce large uncertainties existed in attribution of LGP change, and future rice crop modelling in global/regional scales should consider different types of rice and variable transplanting, dates in order to better account impacts of management and climate change. (C) 2016 Elsevier B.V. All rights reserved.
Context: Maternal hypertensive disorders during pregnancy are suggested to affect obesity risk in offspring. However, little is known about the prospective association of rise in maternal blood pressure within normal range during pregnancy with this risk for obesity. Objective: To clarify the associations of diastolic and systolic blood pressure during pregnancy among normotensive women with the risk for obesity in offspring. Design: Prospective cohort study. Setting: Southeast China. Participants: Up to 2013, a total of 88,406 mother-child pairs with anthropometric measurements of offspring age 4 to 7 years were included in the present analysis. Main Outcome Measures: Overweight/obesity risk in offspring. Results: Among normotensive women, second- and third-trimester diastolic and systolic blood pressures were positively associated with risk for overweight/obesity in offspring: odds ratios per 10-mm Hg higher second- and third-trimester diastolic blood pressure were 1.05 [95% confidence interval (CI), 1.01 to 1.09] and 1.05 (95% CI, 1.02 to 1.10), respectively, and for systolic blood pressure were 1.08 (95% CI, 1.05 to 1.11) and 1.06 (95% CI, 1.03 to 1.09). Each 10-mm Hg greater rise in blood pressure between first and third trimesters was associated with a higher risk for offspring overweight/obesity: diastolic, 1.06 (95% CI, 1.01 to 1.10); systolic, 1.05 (95% CI, 1.02 to 1.07). Among all women (combining normotensive and hypertensive women), maternal hypertension in the second and third trimesters was associated with 49% and 14% higher risks for overweight/obesity in offspring, respectively. Conclusions: These results suggest that rise in maternal blood pressure during pregnancy and hypertension during pregnancy, independent of maternal body size before pregnancy, are risk factors for offspring childhood obesity.
The genetic variants near the Melanocortin-4 receptor gene (MC4R), a key protein regulating energy balance and adiposity, have been related to obesity and glucose metabolism. We aimed to assess whether the MC4R genotype affected longitudinal changes in body weight and glucose metabolism biomarkers among women with prior gestational diabetes mellitus (GDM). The MC4R genotype, postpartum weight reduction, and glycemic changes between after delivery and pregnancy were assessed in a cohort of 1208 Chinese women who had experienced GDM. The adiposity-increasing allele (C) of the MC4R variant rs6567160 was associated with greater postpartum increase of HbA1c (beta = 0.08%; P = 0.03) and 2-hour OGTT glucose concentrations (beta = 0.25 mmol/L; P = 0.02). In addition, we found an interaction between the MC4R genotype and postpartum weight reduction on changes in fasting plasma glucose (P-interaction = 0.03). We found that the MC4R genotype was associated with postpartum glycemic changes; and the association with fasting glucose were significantly modified by postpartum weight reduction in women who had experienced GDM.
Nanothick metallic transition metal dichalcogenides such as VS2 are essential building blocks for constructing next-generation electronic and energy-storage applications, as well as for exploring unique physical issues associated with the dimensionality effect. However, such two-dimensional (2D) layered materials have yet to be achieved through either mechanical exfoliation or bottom-up synthesis. Herein, we report a facile chemical vapor deposition route for direct production of crystalline VS2 nanosheets with sub-10 nm thicknesses and domain sizes of tens of micrometers. The obtained nanosheets feature spontaneous superlattice periodicities and excellent electrical conductivities (∼3 × 103 S cm–1), which has enabled a variety of applications such as contact electrodes for monolayer MoS2 with contact resistances of ∼1/4 to that of Ni/Au metals, and as supercapacitor electrodes in aqueous electrolytes showing specific capacitances as high as 8.6 × 102 F g–1. This work provides fresh insights into the delicate structure–property relationship and the broad application prospects of such metallic 2D materials.
The microbial community diversity in anaerobic-, anoxic- and oxic-biological zones of a conventional Carrousel oxidation ditch system for domestic wastewater treatment was systematically investigated. The monitored results of the activated sludge sampled from six full-scale WWTPs indicated that Proteobacteria, Chloroflexi, Bacteroidetes, Actinobacteria, Verrucomicrobia, Acidobacteria and Nitrospirae were dominant phyla, and Nitrospira was the most abundant and ubiquitous genus across the three biological zones. The anaerobic-, anoxic-and oxic-zones shared approximately similar percentages across the 50 most abundant genera, and three genera (i.e. uncultured bacterium PeM15, Methanosaeta and Bellilinea) presented statistically significantly differential abundance in the anoxic-zone. Illumina high-throughput sequences related to ammonium oxidizer organisms and denitrifiers with top50 abundance in all samples were Nitrospira, uncultured Nitrosomonadaceae, Dechloromonas, Thauera, Denitratisoma, Rhodocyclaceae (norank) and Comamonadaceae (norank). Moreover, environmental variables such as water temperature, water volume, influent ammonium nitrogen, influent chemical oxygen demand (COD) and effluent COD exhibited significant correlation to the microbial community according to the Monte Carlo permutation test analysis (p < 0.05). The abundance of Nitrospira, uncultured Nitrosomonadaceae and Denitratisoma presented strong positive correlations with the influent/effluent concentration of COD and ammonium nitrogen, while Dechloromonas, Thauera, Rhodocyclaceae (norank) and Comamonadaceae (norank) showed positive correlations with water volume and temperature. The established relationship between microbial community and environmental variables in different biologically functional zones of the six representative WWTPs at different geographical locations made the present work of potential use for evaluation of practical wastewater treatment processes.