A queuing model for distributed scheduling in IEEE 802.16 wireless mesh networks

The IEEE 802.16 standard for wireless broadband networks includes the mesh mode in its specifications, where network nodes interact to deliver packets from a client to a remote destination through intermediate nodes. This paper presents a study of the capacity of IEEE 802.16 wireless networks in mesh mode by using M/G/1/L queuing model that represents each network node by incorporating the features of the standard in order to calculate the average delay and throughput in the node. An iterative method integrates the calculation results at each node, obtaining the end‐to‐end delay from any node of the mesh to the Base Station. Because of multiple hops, a node far from the Base Station may have its flows damaged. To minimize this problem, we propose a criterion for a fair distribution of resources. We show the numerical results of the model which indicate a good fit when compared with simulation results. Copyright © 2013 John Wiley & Sons, Ltd.     

Vibration Modeling of PEM Fuel Cell for Prediction of Cell Number Effects by Experimental Data

One of the most important challenges in increasing the performance, reliability and lifetime of fuel cells is the mechanical load effects that occur on real applications. Therefore, the vibration model of fuel cell that predicts the behavior of various fuel cell layouts is very useful. The fuel cell is made up of different adjacent layers that may have semi opposite mechanical properties. This special structure leads to occurrence of non‐linear behavior of fuel cell under dynamic mechanical vibrations and so, a black box method is selected for modeling of its vibration behavior. In this study, the mechanical load experiments in various shape and axes were applied on five layouts of proposed fuel cell and the vibration of its body measure by some accelerometers. The NNARXM neural network is created and trained with the experimental data of three layouts of the fuel cell. Then, the prediction error of this neural network, validated with the two other experimental data of fuel cell layouts, by correlation coefficients and histogram of prediction errors. Neural network validation shows the well prediction of both untrained layout and suitable estimation for any desired layout.     

The Contributions of Blue-Green Infrastructure to Building Urban Climatic Resilience—Bibliometric Analysis Based on Co-citation Networks

Against the backdrop of global climate change and in regards of urban sustainable development, enhancing climate resilience has become a critical strategy in adapting climate change for urban areas, where blue-green infrastructure is considered an important means. Although existing studies mention that blue-green infrastructure (BGI) can promote urban resilience by increasing its own diversity, flexibility, redundancy, modularization, and decentralization, questions like where to promote, by what specific means to promote and to what extent it could promote to are still lack of scientific exploration, leading insufficient support for applying resilience theory into planning and design practice. This research recognizes the role of BGI in building climate resilience in the key fields of functioning–urban floods, sea level rise, and high temperature and heat waves–and summarizes that the common functioning mechanisms include the biophysical properties of BGI, forming modular units with other infrastructures of similar functions, and the reliance on networked structures to help the system restore its physical functions and social connections as quickly as possible after disturbances and attacks. This paper also analyzes possible obstacles that hinder the promotion of BGI solutions–the lack of data support to BGI functioning mechanism, the lack of comprehensive assessment on ecological-social-economic benefits, and the difficulty in gaining confidence from decision-makers and the public. Finally, this paper proposes countermeasures from aspects of theoretical development, planning practice, and implementation and management, in order to offer insights for building urban climate resilience.     

基于并行混合神经网络模型的短文本情感分析

针对传统的卷积神经网络（CNN）在进行情感分析任务时会忽略词的上下文语义以及CNN在最大池化操作时会丢失大量特征信息，从而限制模型的文本分类性能这两大问题，提出一种并行混合神经网络模型CA-BGA。首先，采用特征融合的方法在CNN的输出端融入双向门限循环单元（BiGRU）神经网络，通过融合句子的全局语义特征加强语义学习；然后，在CNN的卷积层和池化层之间以及BiGRU的输出端引入注意力机制，从而在保留较多特征信息的同时，降低噪声干扰；最后，基于以上两种改进策略构造出了并行混合神经网络模型。实验结果表明，提出的混合神经网络模型具有收敛速度快的特性，并且有效地提升了文本分类的F1值，在中文评论短文本情感分析任务上具有优良的性能。     

基于弱监督的改进型GoogLeNet在DR检测中的应用

针对糖尿病视网膜病变分级检测中标定样本少、多目标检测的问题，提出了一种基于改进型GoogLeNet的弱监督目标检测网络。首先，对GoogLeNet网络进行改进，去掉最后一个全连接层并保留检测目标的位置信息，添加全局最大池化层，以sigmoid交叉熵作为训练的目标函数以获得带有多种特征位置信息的特征图；然后，基于弱监督方法仅使用类别标签对网络进行训练；其次，设计一种连通区域算法来计算特征连通区域边界坐标集合；最后在待测图片中使用边界框定位病灶。实验结果表明，在小样本条件下，改进模型准确率达到了94.5%，与SSD算法相比，准确率提高了10%。改进模型实现了小样本条件下端到端的病变识别，同时该模型的高准确率保证了模型在眼底筛查中具有应用价值。     

基于多尺度多任务卷积神经网络的人群计数

在智能监控领域，实现人群计数具有重要价值，针对人群尺度不一、人群密度分布不均及遮挡等问题，提出一种多尺度多任务卷积神经网络（MMCNN）进行人群计数的方法。首先提出一种新颖的自适应人形核生成密度图描述人群信息，消除人群遮挡影响；其次通过构建多尺度卷积神经网络解决人群尺度不一问题，以多任务学习机制同时估计密度图及人群密度等级，解决人群分布不均问题；最后设计一种加权损失函数，提高人群计数准确率。在UCF_CC_50和World Expo'10数据库上进行了评估，验证了自适应人形核的有效性。实验结果表明：所提算法比Sindagi等的方法（SINDAGI V A，PATEL V M.CNN-based cascaded multi-task learning of high-level prior and density estimation for crowd counting.Proceedings of the 2017 14th IEEE International Conference on Advanced Video and Signal Based Surveillance.Piscataway，NJ：IEEE，2017：1-6）在UCF_CC_50数据库上平均绝对误差（MAE）数值和均方误差（MSE）数值分别降低约1.7和45；与Zhang等的方法（ZHANG Y，ZHOU D，CHEN S，et al.Single-image crowd counting via multi-column convolutional neural network.Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition.Washington，DC：IEEE Computer Society，2016：589-597）相比，在World Expo'10数据库上所提算法的MAE值降低约1.5，且在真实公共汽车数据库上仅0~3人的计数误差，表明其实用性较强。     

结合深度学习与相关反馈的遥感图像检索

目的 针对基于内容的图像检索存在低层视觉特征与用户对图像理解的高层语义不一致、图像检索的精度较低以及传统的分类方法准确度低等问题，提出一种基于卷积神经网络和相关反馈支持向量机的遥感图像检索方法。方法 通过对比度受限直方图均衡化算法对遥感图像进行预处理，限制遥感图像噪声的放大，采用自学习能力良好的卷积神经网络对遥感图像进行多层神经网络的监督学习提取丰富的图像特征，并将支持向量机作为基分类器，根据测试样本数据到分类超平面的距离进行排序得到检索结果，最后采用相关反馈策略对检索结果进行重新调整。结果 在UC Merced Land-Use遥感图像数据集上进行图像检索实验，在mAP（mean average precision）精度指标上，当检索返回图像数为100时，本文方法比LSH（locality sensitive Hashing）方法提高了29.4%，比DSH（density sensitive Hashing）方法提高了37.2%，比EMR（efficient manifold ranking）方法提高了68.8%，比未添加反馈和训练集筛选的SVM（support vector machine）方法提高了3.5%，对于平均检索速度，本文方法比对比方法中mAP精度最高的方法提高了4倍，针对复杂的遥感图像数据，本文方法的检索效果较其他方法表现出色。结论 本文提出了一种以距离评价标准为核心的反馈策略，以提高检索精度，并采用多距离结合的Top-k排序方法合理筛选训练集，以提高检索速度，本文方法可以广泛应用于人脸识别和目标跟踪等领域，对提升检索性能具有重要意义。     

A novel Intrusion Detection System for Vehicular Ad Hoc Networks (VANETs) based on differences of traffic flow and position

Vehicle ad hoc networks (VANETs) have attracted great interests from both industry and academia, but a number of issues, particularly security, have not been readily addressed. Intrusion Detection System (IDS) as one of the most important approaches to protect network security has been studied adequately in previous literatures. However, the performance of IDSs still needs to be improved to adapt the scenario of VANETs which are very fast moving and highly dynamic. In this paper, we propose a novel IDS that is able to be appropriately used in the wireless and dynamic networks, like VANETs. It mainly contains a novel feature extraction algorithm and a classifier based on an improved growing hierarchical self-organizing map (I-GHSOM) for IDS in VANETs. The proposed feature extraction algorithm is used to quickly extract distinct features from vehicle messages for IDS’s training and test. In the proposed algorithm, two key features including the differences of traffic flow and of position are extracted. The former feature is calculated according to the range of the distance between vehicles, while both a voting filter mechanism and a semi-cooperative mechanism are designed to get the latter feature. Furthermore, in the I-GHSOM-based classifier, for quickly attaining precise classification results, two novel mechanisms (relabeling and recalculating mechanisms) are proposed to relabel the units of GHSOM and check whether the balance of GHSOM structure is broken or not. Simulation results show that the proposed IDS is better than others in the measurement of accuracy, stability, processing efficiency and message scales.     

基于Kohonen神经网络算法的医学图像分割

长期以来,在医学图像处理方面,图像分割一直是该项工作中的重点,同时也是其中的难点。所谓的图像分割实质上就是遵照相关的原则对图像进行分割,使其被分为几个部分的过程。它是以图像测量、配准、融合和三维重建作为基础,所以在临床医学研究中起着举足轻重的作用。文中研究的Kohonen聚类神经网络算法是以VC++为基础,并在此基础上进行了一定的优化,为图像分割的具体应用提供基础条件,提高效果,这对医生临床诊断具有十分重要的意义。