一种用于淮河上游日径流预测的增强型LSTM模型

径流预报对防洪具有重要意义。然而,由于径流过程的复杂性和随机性,对日径流量进行准确预测是困难的,尤其是对峰值径流量的预测。为了解决这一问题,本研究提出了一种用于径流预测的增强型长短期记忆(LSTM)模型,其中引入了新的损失函数并集成了特征提取器。设计了峰值误差tanh(peak error tanh,PET)和峰值误差swish(peak error swish,PES)两个损失函数,增强了峰值径流预测的重要性,弱化了正常径流预测的权重。为每个气象站建立由3个LSTM网络组成的特征提取器,目的是提取每个气象站输入数据的时间特征。以中国淮河上游为例,利用增强型LSTM模型对1960—2016年的日径流量进行了预测。结果表明,改进后的LSTM模型表现良好,在验证期内(2005年11月至2016年12月),Nash-Sutcliffe效率(NSE)系数在0.917-0.924之间,优于广泛使用的集总水文模型(Australian Water Balance model(AWBM)、Sacramento、Sim Hyd和Tank模型)和数据驱动模型(人工神经网络(ANN)、支持向量回归(SVR)和门控循环单元(GRU))。以PES为损失函数的增强型LSTM对洪水极端径流的预测效果最好,平均NSE为0.873。此外,海拔较高的气象站降水对径流预测的贡献比最近的气象站更大。该研究为流域日径流预测提供了有效工具,有利于流域防洪和水安全管理。

Integration of data-intensive,machine learning and robotic experimental approaches for accelerated discovery of catalysts in renewable energy-related reactions

Technological advancements in recent decades have greatly transformed the field of material chemistry.Juxtaposing the accentuating energy demand with the pollution associated,urgent measures are required to ensure energy maximization,while reducing the extended experimental time cycle involved in energy production.In lieu of this,the prominence of catalysts in chemical reactions,particularly energy related reactions cannot be undermined,and thus it is critical to discover and design catalyst,towards the optimization of chemical processes and generation of sustainable energy.Most recently,artificial intelligence(AI)has been incorporated into several fields,particularly in advancing catalytic processes.The integration of intensive data set,machine learning models and robotics,provides a very powerful tool in modifying material synthesis and optimization by generating multifarious dataset amenable with machine learning techniques.The employment of robots automates the process of dataset and machine learning models integration in screening intermetallic surfaces of catalyst,with extreme accuracy and swiftness comparable to a number of human researchers.Although,the utilization of robots in catalyst discovery is still in its infancy,in this review we summarize current sway of artificial intelligence in catalyst discovery,briefly describe the application of databases,machine learning models and robots in this field,with emphasis on the consolidation of these monomeric units into a tripartite flow process.We point out current trends of machine learning and hybrid models of first principle calculations(DFT)for generating dataset,which is integrable into autonomous flow process of catalyst discovery.Also,we discuss catalyst discovery for renewable energy related reactions using this tripartite flow process with predetermined descriptors.

White Matter Abnormalities in Major Depression Biotypes Identified by Diffusion Tensor Imaging

Identifying data-driven biotypes of major depressive disorder(MDD) has promise for the clarification of diagnostic heterogeneity. However, few studies have focused on white-matter abnormalities for MDD subtyping. This study included 116 patients with MDD and118 demographically-matched healthy controls assessed by diffusion tensor imaging and neurocognitive evaluation.Hierarchical clustering was applied to the major fiber tracts, in conjunction with tract-based spatial statistics, to reveal white-matter alterations associated with MDD.Clinical and neurocognitive differences were compared between identified subgroups and healthy controls. With fractional anisotropy extracted from 20 fiber tracts, cluster analysis revealed 3 subgroups based on the patterns of abnormalities. Patients in each subgroup versus healthy controls showed a stepwise pattern of white-matter alterations as follows: subgroup 1(25.9% of patient sample),widespread white-matter disruption;subgroup 2(43.1% of patient sample), intermediate and more localized abnormalities in aspects of the corpus callosum and left cingulate;and subgroup 3(31.0% of patient sample),possible mild alterations, but no statistically significant tract disruption after controlling for family-wise error. The neurocognitive impairment in each subgroup accompanied the white-matter alterations: subgroup 1, deficits in sustained attention and delayed memory;subgroup 2, dysfunction in delayed memory;and subgroup 3, no significant deficits. Three subtypes of white-matter abnormality exist in individuals with major depression, those having widespread abnormalities suffering more neurocognitive impairments, which may provide evidence for parsing the heterogeneity of the disorder and help optimize typespecific treatment approaches.

Electrochemical performances of NiO/Ni2N nanocomposite thin film as anode material for lithium ion batteries

Despite the high specific capacities,the practical application of transition metal oxides as the lithium ion battery(LIB)anode is hindered by their low cycling stability,severe polarization,low initial coulombic efficiency,etc.Here,we report the synthesis of the NiO/Ni2N nanocomposite thin film by reactive magnetron sputtering with a Ni metal target in an atmosphere of 1 vol.% O2 and 99 vol.%N2.The existence of homogeneously dispersed nano Ni2N phase not only improves charge transfer kinetics,but also contributes to the one-off formation of a stable solid electrolyte interphase(SEI).In comparison with the NiO electrode,the NiO/Ni2N electrode exhibits significantly enhanced cycling stability with retention rate of 98.8%(85.6%for the NiO electrode)after 50 cycles,initial coulombic efficiency of 76.6%(65.0%for the NiO electrode)and rate capability with 515.3 mA·h·g^−1(340.1 mA·h·g^−1 for the NiO electrode)at 1.6 A·g^−1.

Location Prediction on Trajectory Data: A Review

Location prediction is the key technique in many location based services including route navigation, dining location recommendations, and traffic planning and control, to mention a few. This survey provides a comprehensive overview of location prediction, including basic definitions and concepts, algorithms, and applications. First, we introduce the types of trajectory data and related basic concepts. Then, we review existing location-prediction methods, ranging from temporal-pattern-based prediction to spatiotemporal-pattern-based prediction. We also discuss and analyze the advantages and disadvantages of these algorithms and briefly summarize current applications of location prediction in diverse fields. Finally, we identify the potential challenges and future research directions in location prediction.