碳酸性侵蚀与AAR作用下纳米混凝土的耐久性

地铁混凝土处于地下空间,容易受到地下水的碳酸性侵蚀；碱集料反应 (AAR)是一种严重的混凝土耐久性问题,既难以发现又难以修补,由两者共同作用引起的混凝土耐久性降低严重影响地铁隧道的正常使用.为研究纳米材料对地铁混凝土在碳酸性侵蚀和AAR共同作用下耐久性的影响,在普通混凝土中掺入适量纳米SiO₂和纳米Fe₂O₃,利用自行研制的碳酸性侵蚀试验箱进行试验,采用碳酸性侵蚀深度、膨胀率和声速作为测试指标来评价纳米混凝土在碳酸性侵蚀和AAR共同作用下的耐久性.试验结果表明:掺入纳米颗粒后,混凝土的膨胀率和侵蚀深度有了明显降低,而声速有了明显提升,说明纳米混凝土的耐久性优于普通混凝土；在182 d龄期时,掺量为2%的纳米SiO₂混凝土耐久性改善最明显,侵蚀深度和膨胀率最小,声速最大且声速下降幅度最小；其次是掺量为1%的纳米Fe₂O₃混凝土.由于纳米颗粒特殊的物理化学性质,改善了混凝土内部的微观结构和孔溶液的化学组成,使碳酸性侵蚀和碱集料反应共同作用下混凝土的耐久性得到了提高.     

Characterisation of the Dynamic Interactions between Complex N-Glycans and Human CD22

CD22 (Siglec-2) is a B-cell surface inhibitory protein capable of selectively recognising sialylated glycans, thus dampening autoimmune responses against self-antigens. Here we have characterised the dynamic recognition of complex-type N-glycans by human CD22 by means of orthogonal approaches including NMR spectroscopy, computational methods and biophysical assays. We provide new molecular insights into the binding mode of sialoglycans in complex with h-CD22, highlighting the role of the sialic acid galactose moieties in the recognition process, elucidating the conformational behaviour of complex-type N-glycans bound to Siglec-2 and dissecting the formation of CD22 homo-oligomers on the B-cell surface. Our results could enable the development of additional therapeutics capable of modulating the activity of h-CD22 in autoimmune diseases and malignancies derived from B-cells.     

Bridging the Gap: Plant-Endophyte Interactions as a Roadmap to Understanding Small-Molecule Communication in Marine Microbiomes

Probing the composition of the microbiome and its association with health and disease states is more accessible than ever due to the rise of affordable sequencing technology. Despite advances in our ability to identify members of symbiont communities, untangling the chemical signaling that they use to communicate with host organisms remains challenging. In order to gain a greater mechanistic understanding of how the microbiome impacts health, and how chemical ecology can be leveraged to advance small-molecule drug discovery from microorganisms, the principals governing communication between host and symbiont must be elucidated. Herein, we review common modes of interkingdom small-molecule communication in terrestrial and marine environments, describe the differences between these environments, and detail the advantages and disadvantages for studies focused on the marine environment. Finally, we propose the use of plant-endophyte interactions as a stepping stone to a greater understanding of similar interactions in marine invertebrates, and ultimately in humans.     

公民水素养评价表征因素模型构建与分析

针对我国水资源紧缺、水生态损害、水环境污染等相关水问题,以提升全民爱水惜水节水护水的意识和能力为目标,从水素养概念出发,通过扎根理论质性研究方法对公民水素养评价的表征因素进行筛选和提炼,最终得出包含水知识、水态度、水行为为主范畴的10个表征因素并运用解释结构模型(ISM)深入剖析各因素之间的内在联系。在此基础上建立了3级多层次递阶结构模型,进一步精确描述系统各要素两两之间的直接关系。结果表明:公民水素养评价的表征因素自上而下形成了表象层、中间层和根源层等3个层级,可直观、清晰地反映表征因素之间的递阶、层级关系;水态度在各表征因素中具有调节作用以及各表征因素之间存在内在逻辑关系。研究成果为提高公民水素养水平提供一定的科学依据和政策参考。     

基于条件信息卷积生成对抗网络的图像识别

传统的图像识别方法需要大量有标签样本进行训练，且模型训练难以达到稳定。针对这些问题，结合条件生成网络和信息最大化生成网络的结构优势建立了条件信息卷积生成网络（C-Info-DCGAN）。模型增加图像的类别信息和潜在信息作为输入数据，然后利用Q网络去更好地发挥类别信息和潜在信息对训练的引导作用，并且利用深度卷积网络来加强对图像特征的提取能力。实验结果表明，该方法能够加快模型训练收敛速度，并有效提高图像识别的准确率。     

视频侦查中人像智能分析应用及算法优化

人像智能分析指的是对视频或录像中的人像进行结构化和可视化分析,对目标人物进行性别、年龄、发型等特征的智能识别,这项技术在视频侦查中有极高的应用价值。人像识别早期的算法是通过人工提取特征,通过学习低级视觉特征来针对不同属性进行分类学习,这种基于传统方法的模型表现常常不尽如人意。在计算机视觉领域,通过海量图像数据学习的神经网络比传统方法有更丰富的信息量和特征可以被提取。文章尝试通过深度学习技术训练神经网络模型对行人进行检测和识别,对于衣着不同的行人进行智能识别,具有更好的鲁棒性,提升了视频人像识别的准确率,拓展了人工智能技术在身份识别领域的应用。     

Muddling between science and engineering: an epistemic strategy for developing human factors and ergonomics as a hybrid discipline

Human Factors and Ergonomics (HFE) recognises itself as a design-driven, systemic and scientific discipline geared towards well-being and performance. Being a scientific discipline and design-oriented requires that the epistemic basis of science and design/engineering be fully comprehended. In interdisciplinary research where these two viewpoints meet, there are often dilemmas posed in terms of knowledge construction and labelling of activity. Therefore, this article scrutinises these two orientations and addresses the differences and commonalities, using case studies from engineering and psychological science (both constituents of HFE). Based on these insights, a way forward is suggested in terms of (1) a reflexive engagement with epistemic concepts and methods; (2) finding a conceptual space for balancing and bridging the science-engineering divide; (3) comprehending ‘design-thinking/design knowledge’ and not treating it as an application of science; (4) providing emphasis on problem formulation and practices of HFE focusing on developing them in systemic terms.     

Selection of risk response actions with consideration of secondary risks

Secondary risk in project risk management refers to the risk that arises as a direct result of implementing a risk response action (RRA). It is important for project managers (PMs) to consider the effects caused by the secondary risks in the process of RRA selection. The purpose of this paper is to propose an optimization method to address the problem of selecting risk response actions (RRAs) with consideration of secondary risk which is seldom considered in the existing studies. The optimization model aims to minimize the total risk costs with time constraint being placed on the project makespan. By solving the model, an optimal set of RRAs along with the earliest start time for each activity can both be obtained. The results show that secondary risk plays an important role in the process of RRA selection. Project managers should allocate more budget for responding the project risk when the secondary risk is considered, and consider all factors relating to both time and cost so as to select appropriate RRAs to mitigate primary risk and secondary risk.     

小尺度交通信号灯的检测与状态识别

交通信号灯识别包括检测和状态识别，在智能交通系统中发挥重要作用。基于YOLOv3算法提出了一种交通信号灯检测与状态识别模型。针对交通信号灯相较于交通场景中其他目标具有尺度小的特性进行了算法的设计：降低骨干网络的下采样倍率以增加小尺度目标的特征描述能力；通过增大特征图的尺度来改进多尺度特征融合；引入广义交并比作为检测任务的损失函数来改进目标边界框的回归效果。同时，根据交通信号灯本身的特性，使用颜色和形状约束的方法对信号灯进行状态识别和类别验证。最后在公开的Bosch交通信号灯数据集上和实际的城区道路进行了实验验证。实验结果表明，所提出的算法能够提升交通信号灯识别的精度和召回率，识别准确率可以达到90%左右。     

Convolutional neural network with adaptive inferential framework for skeleton-based action recognition

In the task of skeleton-based action recognition, CNN-based methods represent the skeleton data as a pseudo image for processing. However, it still remains as a critical issue of how to construct the pseudo image to model the spatial dependencies of the skeletal data. To address this issue, we propose a novel convolutional neural network with adaptive inferential framework (AIF-CNN) to exploit the dependencies among the skeleton joints. We particularly investigate several initialization strategies to make the AIF effective with each strategy introducing the different prior knowledge. Extensive experiments on the dataset of NTU RGB+D and Kinetics-Skeleton demonstrate that the performance is improved significantly by integrating the different prior information. The source code is available at: https://github.com/hhe-distance/AIF-CNN.