Building a Tax Predictive Model Based on the Cloud Neural Network

Tax is very important to the whole country, so a scientific tax predictive model is needed. This paper introduces the theory of the cloud model. On this basis, it presents a cloud neural network, and analyzes the main factors which influence the tax revenue. Then if proposes a tax predictive model based on the cloud neural network. The model combines the strongpoints of the cloud model and the neural network. The experiment and simulation results show the effectiveness of the algorithm in this paper.

Job Scheduling for Cloud Computing Using Neural Networks

Cloud computing aims to maximize the benefit of distributed resources and aggregate them to achieve higher throughput to solve large scale computation problems. In this technology, the customers rent the resources and only pay per use. Job scheduling is one of the biggest issues in cloud computing. Scheduling of users’ requests means how to allocate resources to these requests to finish the tasks in minimum time. The main task of job scheduling system is to find the best resources for user’s jobs, taking into consideration some statistics and dynamic parameters restrictions of users’ jobs. In this research, we introduce cloud computing, genetic algorithm and artificial neural networks, and then review the literature of cloud job scheduling. Many researchers in the literature tried to solve the cloud job scheduling using different techniques. Most of them use artificial intelligence techniques such as genetic algorithm and ant colony to solve the problem of job scheduling and to find the optimal distribution of resources. Unfortunately, there are still some problems in this research area. Therefore, we propose implementing artificial neural networks to optimize the job scheduling results in cloud as it can find new set of classifications not only search within the available set.

Flatness predictive model based on T-S cloud reasoning network implemented by DSP

The accuracy of present flatness predictive method is limited and it just belongs to software simulation. In order to improve it, a novel flatness predictive model via T-S cloud reasoning network implemented by digital signal processor(DSP) is proposed. First, the combination of genetic algorithm(GA) and simulated annealing algorithm(SAA) is put forward, called GA-SA algorithm, which can make full use of the global search ability of GA and local search ability of SA. Later, based on T-S cloud reasoning neural network, flatness predictive model is designed in DSP. And it is applied to 900 HC reversible cold rolling mill. Experimental results demonstrate that the flatness predictive model via T-S cloud reasoning network can run on the hardware DSP TMS320 F2812 with high accuracy and robustness by using GA-SA algorithm to optimize the model parameter.

CT-CloudDetect:用于遥感卫星云检测的混合模型

云检测是在遥感卫星云图中检测云的任务。近年来,人们提出了基于深度学习的云检测方法,并取得了良好的性能。然而,现有的基于深度学习的云检测模型大多还是基于卷积神经网络(convolutional neural network,CNN),由于卷积运算的固有局部性,难以捕获长距离依赖关系。针对上述问题,文章提出一个基于CNN和ViT(Vision Transformer)的混合型云检测模型,并提出一种基于CNN和ViT的编码器,使网络具备捕捉局部和全局信息的能力。为了更好地融合语义和尺度不一致的特征,提出了一个双尺度注意力融合模块,通过注意力机制有选择地融合特征。此外,提出了轻量级路由解码器,该解码器通过路由结构降低模型复杂度。在3个公开云检测数据集上对模型进行了评估。大量实验表明,所提出的模型具有比现有模型更好的性能。

Point Cloud Classification Network Based on Graph Convolution and Fusion Attention Mechanism

The classification of point cloud data is the key technology of point cloud data information acquisition and 3D reconstruction, which has a wide range of applications. However, the existing point cloud classification methods have some shortcomings when extracting point cloud features, such as insufficient extraction of local information and overlooking the information in other neighborhood features in the point cloud, and not focusing on the point cloud channel information and spatial information. To solve the above problems, a point cloud classification network based on graph convolution and fusion attention mechanism is proposed to achieve more accurate classification results. Firstly, the point cloud is regarded as a node on the graph, the k-nearest neighbor algorithm is used to compose the graph and the information between points is dynamically captured by stacking multiple graph convolution layers;then, with the assistance of 2D experience of attention mechanism, an attention mechanism which has the capability to integrate more attention to point cloud spatial and channel information is introduced to increase the feature information of point cloud, aggregate local useful features and suppress useless features. Through the classification experiments on ModelNet40 dataset, the experimental results show that compared with PointNet network without considering the local feature information of the point cloud, the average classification accuracy of the proposed model has a 4.4% improvement and the overall classification accuracy has a 4.4% improvement. Compared with other networks, the classification accuracy of the proposed model has also been improved.

A Modified SOFM Method for Point Cloud Segmentation in Reverse Engineering

The purpose of reverse engineering is to convert a large point cloud into a CAD model. In reverse engineering, the key issue is segmentation, i.e. studying how to subdivide the point cloud into smaller regions, where each of them can be approximated by a single surface. Segmentation is relatively simple, if regions are bounded by sharp edges and small blends; problems arise when smoothly connected regions need to be separated. In this paper, a modified self-organizing feature map neural network （SOFM） is used to solve segmentation problem. Eight dimensional feature vectors （3-dimensional coordinates, 3-dimensional normal vectors, Gaussian curvature and mean curvature） are taken as input for SOFM. The weighted Euclidean distance measure is used to improve segmentation result. The method not only can deal with regions bounded by sharp edges, but also is very efficient to separating smoothly connected regions. The segmentation method using SOFM is robust to noise, and it operates directly on the point cloud. An examples is given to show the effect of SOFM algorithm.

EFFECTS OF A CLOUD FILTERING METHOD FOR FENGYUN-3C MICROWAVE HUMIDITY AND TEMPERATURE SOUNDER MEASUREMENTS OVER OCEAN ON RETRIEVALS OF TEMPERATURE AND HUMIDITY

For Microwave Humidity and Temperature sounder(MWHTS) measurements over the ocean, a cloud filtering method is presented to filter out cloud-and precipitation-affected observations by analyzing the sensitivity of the simulated brightness temperatures of MWHTS to cloud liquid water, and using the root mean square error(RMSE)between observation and simulation in clear sky as a reference standard. The atmospheric temperature and humidity profiles are retrieved using MWHTS measurements with and without filtering by multiple linear regression(MLR),artificial neural networks(ANN) and one-dimensional variational(1DVAR) retrieval methods, respectively, and the effects of the filtering method on the retrieval accuracies are analyzed. The numerical results show that the filtering method can improve the retrieval accuracies of the MLR and the 1DVAR retrieval methods, but have little influence on that of the ANN. In addition, the dependencies of the retrieval methods upon the testing samples of brightness temperature are studied, and the results show that the 1DVAR retrieval method has great stability due to that the testing samples have great impact on the retrieval accuracies of the MLR and the ANN, but have little impact on that of the 1DVAR.

A Survey of Cloud Computing Detection Techniques against DDoS Attacks

A Distributed Denial of Service Attack (DDoS) is an attack in which multiple systems compromised by a Trojan are maliciously used to target a single system. The attack leads to the denial of a certain service on the target system. In a DDoS attack, both the target system and the systems used to perform the attack are all victims of the attack. The compromised systems are also called Botnets. These attacks occur on networked systems, among them the cloud computing facet. Scholars have tried coming up with separate mechanisms for detecting and preventing such attacks long before they occur. However, as technology progresses in advancement so do the attack mechanisms. In cloud computing, security issues affect various stakeholders who plan on cloud adoption. DDoS attacks are such serious concerns that require mitigation in the cloud. This paper presents a survey of the various mechanisms, both traditional and modern, that are applied in detecting cloud-based DDoS attacks.

Development of Crisis Management Models Combined with Cloud Computing

Three monitor models of enterprise crisis were introduced,i.e.,the monitoring model of enterprise crisis based on intelligent Meta search,the enterprise crisis management model based on artificial neural network and the combined early-warning model.Combined with the advantages of cloud computing,the prominent crisis management models are improved and more efficient,comprehensive and accurate in enterprise crisis management.Through the empirical study of the models,cloud computing makes the early warning structures of enterprise crisis tend to be more simple and efficient,cloud computing can effectively enhance the recognition ability and learning ability of the crisis management,and cloud computing can keep data information updating and realize the dynamic management of enterprise joint early-warning.At the same time,according to the comparative analysis and the experimental result,the crisis management models based on cloud computing also need some improvements.

智能云平台异构数据库协同检索算法研究

智能云平台是一种集成各种资源和功能的高效计算平台,可以为用户提供灵活的数据存储和高效的数据检索服务.随着信息技术飞速发展,异构数据库中数据呈爆炸式增长.为了提升网络异构数据库检索效果,提出智能云平台异构数据库协同检索算法.构建智能云平台异构数据库,均衡异构数据库中的节点能耗;排序云平台多源异构数据,预处理异构数据;建立以索引库为核心的检索服务引擎,利用神经网络提取多源异构数据特征,实现异构数据库的匹配检索.测试结果表明,所提算法查准率为96%,查全率为94%,数据丢失量仅为1.由此证明,所提方法有效提高了网络异构数据库检索效果.

基于GA-BP对叶片点云数据修补及逆向建模

为了获得某型号发动机叶片的三维建模及其相关数据,给逆向设计提供一些零件的三维建模尺寸。由于获得建模的三维点云数据时必然会产生孔洞,文中使用基于Genetic Algorithm(基因遗传算法)优化的Back Propagation(反向传播)神经网络(又称GA-BP神经网络)作为一个回归预测算法,来对产生的散乱点云孔洞加以修复和点云处理,最后再通过Geomagic Wrap对某型号发动机叶片进行逆向建模。通过对比BP神经网络可知,GA-BP修补孔洞的误差明显降低,满足对三维模型精度较高的要求,可应用于逆向工程。

基于点云反射特性的前方道路附着系数估计方法研究

路面附着系数是影响自动驾驶系统决策控制策略的重要因素。为实现对道路附着系数前瞻性的高精度感知,本文基于车载激光雷达设计了一种新的路面附着系数估计方法。首先采集了干燥柏油路面、混凝土路面、湿滑柏油路面、结冰路面和积雪路面构建道路数据集;基于使用布料模拟滤波和RANSAC算法进行了道路点云提取、基于高斯滤波去除反射率异常噪点;根据点云反射率随距离和入射角变化的规律将路面划分为不同区域分别提取特征;基于深度神经网络构建了道路识别模型,并基于采集数据集进行了训练,最后基于路面材质和峰值附着系数的统计经验确定了前方道路的附着系数。测试结果表明,本文提出的算法道路类型辨识精度超过99.3%,算法平均运行周期55ms,可实现实时高精度的路面峰值附着系数估计。

基于深度学习的自动驾驶场景3D目标检测方法

针对传统PV-RCNN在点云上采样效率低下和采样精度存在偏差等问题,提出了一种基于PV-RCNN改进的3D目标检测方法。更改关键点采样策略,使得有限的关键点可以更加地聚集在proposal区域范围内,更多的编码有效前景点特征来用于后面的proposal refinement,有效产生更具有代表性的关键点。用局部特征聚合的VectorPool聚合模块取代体素集抽象和ROI网格池化模块中的集合抽象,更高效的针对稀疏和不规则点云数据进行编码。在KITTI数据集上对算法验证,结果表明:行人鸟瞰图检测,困难级别检测精度提升较为显著,达到了10.46%,整体帧率提升为33.74%,文中的方法拥有更好的检测性能。

基于纹理特征的深度学习云和云阴影检测

针对云和云阴影检测过程中存在边界不准确以及易与地表混淆等问题,构建一种融合纹理特征模块的卷积神经网络模型对Landsat 8遥感图像进行云和云阴影检测。引入基于统计特性的纹理特征模块进行纹理特征的提取和学习,在训练过程采用焦点损失函数削弱样本不均衡带来的影响。实验结果表明,该模型细化了云和云阴影的边界等纹理细节,减少了云和云阴影的误检和漏检现象,提高了云和云阴影的检测精度。

基于视觉图像与激光点云融合的交通标志快速识别方法

交通标志对车辆交通起到重要作用和意义,而智能交通中交通标志识别由于标志特征提取效果差,导致识别率低、识别时间长,因此,提出一种新的基于视觉图像与激光点云融合的交通标志快速识别方法。采用双边滤波方法预处理原始激光点云数据;通过归一化处理得到视觉图像激光点云融合的目标空间激光点云位置测距数值;通过测距值获取目标图像位置,归一化处理交通标志视觉图像,引入k均值聚类算法二聚类处理图像,采用制作的切割模板切割图像感兴趣区域,提取交通标志图像的深度特征,结合卷积神经网络二次过滤特征,重新标定二次过滤后的特征,最终利用卷积神经网络模型实现交通标志快速识别;经实验对比证明,采用所提方法提取各个类型交通标志特征的提取效果较好,并且识别率达到89.74%,识别时间仅为13.1 s,干扰下识别时间最高仅为15.1 s,验证了该方法可以快速且准确识别各个类型的交通标志。

定性与定量信息相结合预测金品位的方法研究

将改进云模型和改进RBF神经网络相结合,提出了一种预测矿石中金品位的模型。先利用DS证据理论和云模型将定性信息定量化,再采用量子粒子群算法和RBF神经网络完成矿石中金品位预测。结果表明:该模型的均方根误差为0.0092,最大误差为0.0161,相关系数为0.9402,可较好保留定性信息特性,金品位预测效果较好。

基于图的点云研究综述

点云的处理、传输、语义分割等是3维计算机视觉领域重要的分析任务.现如今,图神经网络和图结构在点云研究方面的有效性已被证实,基于图的点云(graph-based point cloud,GPC)研究不断涌现.因此,一种统一的研究角度、框架和方法论亟待形成.系统性梳理了GPC研究的各种应用场景,包括配准、降噪、压缩、表示学习、分类、分割、检测等任务,概括出GPC研究的一般性框架,提出了一条覆盖当前GPC全域研究的技术路线.具体来说,给出了GPC研究的分层概念范畴,包括底层数据处理、中层表示学习、高层识别任务;综述了各领域中的GPC模型或算法,包括静态和动态点云的处理算法、有监督和无监督的表示学习模型、传统或机器学习的GPC识别算法;总结了其中代表性的成果及其核心思想,譬如动态更新每层特征空间对应的最近邻图、分层以及参数共享的动态点聚合模块,结合图划分和图卷积提高分割精度;对比了模型性能,包括总体精度(overall accuracy,OA)、平均精度(mean accuracy,mAcc)、平均交并比(mean intersection over union,mIoU);在分析比较现有模型和方法的基础上,归纳了GPC目前面临的主要挑战,提出相应的研究问题,并展望未来的研究方向.建立的GPC研究框架具有一般性和通用性,为后续研究者从事GPC这个新型交叉领域研究提供了领域定位、技术总结及宏观视角.点云研究的出现,是探测器硬件技术长足进步后应运而生的结果;点云研究的现状表明在理论和实践之间存在一些挑战,一些关键问题还有待解决.同时,点云研究的发展将推动人工智能进入新的时代.

基于加权组合算法的点云孔洞修补

为了对无人机航测数据中的点云孔洞进行修补,将最小二乘支持向量机算法LS-SVM和遗传算法GA优化的反向传播神经网络算法BP进行线性组合,构建一种加权组合模型,用于散乱点云数据中的孔洞修补。通过两种修补方法的误差进行两者的加权组合,建立出与两种修补方法误差相关的加权组合模型,并将加权组合模型的修补结果与单一使用最小二乘支持向量机、遗传算法优化的BP神经网络两种修补方法的修补结果进行残差和内外符合精度的比较与分析。结果表明:采用加权组合模型得到的点云修补结果内外符合精度较高,且具有更强的稳定性,为无人机获取的点云数据提供了一种有效的孔洞修补方法。

子图匹配和强化学习增强的三维点云配准

针对低质量三维点云数据配准精度不足、效率低的问题,为了实现低质量点云的精确、快速配准,提出一种基于子图匹配和强化学习的点云配准方法.首先将三维点云配准转化为一系列离散的刚性变换连续作用结果,利用强化学习策略训练一个端到端的模型以迭代输出刚性变换动作;然后对于模型架构,采用双流主干网络分别提取源点云与目标点云的局部特征信息,设计交叉图注意力模块将源点云图和目标点云图中的相似节点关联起来,使用带选通向量的加权实现图节点的聚合,分别获取源点云图与目标点云图的全局特征表示;最后融合源点云图与目标点云图的全局特征,基于融合特征预测离散的刚性变换动作.强化学习策略的引入显著提高了点云配准算法的泛化性,在加入交叉图注意力模块后,点云配准的精度及效率也进一步被提升.在ModelNet40和ScanObjectNN这2个公共基准数据集上与最新的点云配准方法 ReAgent进行实验的结果表明,所提方法能够将旋转误差的均方差数值降低至少0.16,各向同性旋转误差数值也降低至少0.16,有效地提升低质量点云配准的精度.

基于人工神经网络算法的多相云雾爆轰毁伤效应预测模型

云雾爆轰毁伤范围的预测是云爆武器造成大规模毁伤的研究基础,但云雾爆轰后超压场分布规律与燃料浓度的依赖关系未知,制约了对于云爆毁伤范围的预测。因此,针对2种最为常用的多相云雾燃料,采用最小自由能法计算得到了液相燃料完全以液滴或蒸汽形式存在的多相云雾发生理想爆轰的CJ参数,并通过拟合,得到体JWL状态方程参数。在此基础上计算得到了不同浓度和状态的多相云雾理想爆轰造成的超压场,并借助人工神经网络获得了浓度范围在0.03~0.30 kg/m 3的气固两相和气液固三相云雾场爆轰峰值超压随比例距离衰减规律的代理模型并预测不同毁伤等级对应的毁伤比例半径随燃料浓度的变化,得到毁伤比例半径最大的最优浓度。研究结果表明:云雾区中液相燃料以液滴或蒸汽形式存在对云雾爆轰参数,产物JWL状态方程参数,与云爆爆轰后超压场分布规律的影响都比较微弱(<1.5%);在0.03~0.18 kg/m 3的燃料浓度范围内,Ⅰ级~Ⅲ级毁伤比例半径的最大和最小值分别相差21%、19%、6%,因此大装药结构形成的云雾场爆轰后,Ⅰ级和Ⅱ级毁伤半径与燃料浓度的依赖性更强。