基于深度学习的跳频信号识别

针对跳频信号分选存在人工提取参数特征具有复杂性的问题，提出了一种基于深度学习的识别方法。首先对跳频信号进行短时傅里叶变换，得到二维的时频矩阵；接着提取信号的轮廓特征，构造三维矩阵作等高线图，并对等高线图进行预处理；最后把预处理后的等高线图输入到卷积神经网络中进行训练、测试，进而实现分类识别。仿真结果表明，在不需要复杂的人工提取参数特征的基础上，在分选率为100〖WT《Times New Roman》〗%〖WTBZ〗时，所提方法经裁剪处理下的信噪比为-15 dB，比支持向量机和传统K-Means聚类算法都低10 dB。实测数据的算法验证表明，所提方法能够将大疆精灵4Pro、hm无人机、司马航模X8HW以及大疆悟2这四类无人机正确分类。     

东南亚国家社交商务与非本土节日消费行为分析——以中国春节为例

在世界范围内,中国春节越来越受消费者欢迎,特别是东南亚消费者会在社交商务中参加中国春节活动。然而,现实中却存在一个矛盾的现象,即一些东南亚消费者春节期间热衷于购买中国商品,而春节过后却抵制中国商品。实证研究发现,东南亚消费者并未完全从态度上偏好中国春节和中国商品,只是出于社交商务的需要才参与中国春节和购买中国商品。东南亚消费者可区分为文化杂食型和节日身份型两个细分市场。其中,文化杂食型消费者包容并喜欢多元文化,会长期偏好中国节日和中国商品;节日身份型消费者仅在中国节日期间会参与中国节日和购买中国商品。这样的结论对指导中国企业实践具有重要价值,首先,为更好地响应“一带一路”倡议,应树立中国品牌意识,在“一带一路”沿线国家培育偏好中国文化和产品的朋友圈,转变消费者态度;其次,为更好地实施“走出去”战略,应在国际市场进行文化杂食和节日身份的细分,对文化杂食型消费者以文化为纽带培育其长期稳定的消费习惯,对节日身份型消费者通过营销策划刺激其消费;其三,为成功进入东南亚市场,对在东南亚运营的社交商务,应侧重培育消费者亲中国文化与产品的态度,并增加其使用社交商务平台的愉悦感。     

一种基于微波雷达回波信号的车型分类方法

车辆类型识别方法是智能交通系统的关键技术之一。利用深度学习的高维特征泛化学习能力，将改进的LeNet-5卷积神经网络用于基于交通微波雷达的大小车型分类识别。首先，以雷达触发前的N帧信号为基础，对雷达的回波信号进行分析并构建数据集；然后，分析LeNet-5卷积神经网络的特点；最后提出一种改进的LeNet-5卷积神经网络。实验结果表明，与传统的支持向量机方法相比，所提方法能够智能学习大小车的雷达时频信号特征，大小车型识别准确率达到97%以上，可为交通场景下的车型识别研究提供新的技术途径。     

Chinese multinationals on the New Silk Route: Managing political risk by branding the nation

Though Chinese multinational enterprises (CMNEs) have brought investment and opportunities to countries along the Belt and Road Initiative, there has been scepticism regarding the content and context of their investment. On the one hand, infrastructural development contributes to GDP formation and enhances the efficiency of productive inputs where international trade enhances technology and export development, which jointly provides a path to economic growth. On the other hand, critical issues such as debt sustainability and national sovereignty among countries that have gone through the decolonization process mean that CMNEs need to grasp the implication of political risk when investing. This article focuses on South and Southeast Asia, where China has historical socio‐economic relationship, and proposes a nation branding model combining tradition and modernity which can be the way forward for CMNEs to mitigate political risk in relation to the Belt and Road Initiative (BRI) investment. Nation branding of BRI could be seen as a continuity of China's unfinished business in globalization that has preceded the modern polity; nevertheless, there is a need to communicate a coherent and authentic message that reflects the reality of business operations.     

Local integration and co‐evolution of internationalizing Chinese firms

This study investigates the integration of internationalizing Chinese firms into local host markets. We explore the market‐driven investment of a new wave of Chinese private and local state‐owned firms in Australia since 2012, which has replaced the initial large‐scale investment in resources by central state‐owned enterprises. Using an in‐depth analysis of nine Chinese firms operating in various sectors of the Australian market, we argue that market integration, adaptation, and bilateral institution‐building through co‐evolution and empowerment of local subsidiaries of Chinese multinational enterprises results in entrepreneurial autonomy and characterizes a new generation of Chinese investors. We propose that Chinese multinational subsidiaries have transferred domestic practices to the Australian market and have reconfigured domestic and host market resources to gain a competitive advantage in their original investment industry and new industries. Our study advances middle‐range theory building and provides a practical understanding of the strengths and weaknesses of Chinese investors, their potential to disrupt local markets, and their responsiveness to market‐oriented institutional guidance. The results of this study suggest that the bilateral institution‐building and resource reconfiguration capabilities of Chinese enterprises can be transferred to other developed and developing markets, including Belt and Road Initiative countries.     

地方人大中的地域代表现象探析——基于五省市人大代表建议的内容分析

本文试图探析在中国地方人大中是否存在地域性代表及其相关因素。通过收集和分析2017年国内五省市人大收到的近四千条代表建议,发现在地方人大代表的履职行为中,一定范围内存在从整体资源中为选区争取利益的现象。进一步回归分析的结果显示:来自经济发展水平相对较低的地区的代表,更倾向于提出地域性建议;由代表联名提出的建议更可能是地域性建议;担任地方党政领导职务的省市人大代表,在提出建议时更倾向于提出地域性建议。上述实证发现对中国地方人大地域性代表的实现机制提供了初步启发性证据,也有助于丰富我们对非竞选体制下的政治代表性和回应性机制的理解。     

关于改进个人银行账户分类管理问题的分析

随着互联网和移动通讯技术的发展,支付领域风险范围、扩散速度、溢出效应程几何级数增长。为有效防范网络新型违法犯罪,保护人民群众财产安全和合法权益,中国人民银行相继发布了多项制度条例,全面推进个人账户分类管理。本文结合对商业银行账户分类管理专项调查结果,分析分类管理中存在的问题及原因,在此基础上,提出进一步改进个人银行账户分类管理的措施和建议。     

深度学习方法在作物遥感分类中的应用和挑战

[目的]准确估算作物的面积和分布对粮食安全至关重要。与传统的机器学习方法相比,深度学习具有多种优势,如端到端训练、可迁移性。为有效利用高时空数据进行作物识别提供了新的机遇。已有多种模型被应用于作物分类任务中,针对不同的分类任务,如何有效地选择模型,并对其进行训练和使用已成为关键问题。[方法]文章回顾了利用深度学习模型对作物分类的主要研究。N维卷积神经网络(N-D CNN)(N=1、2、3)和递归神经网络(RNN)已被有效用于作物分类任务。长短期记忆RNN(LSTM RNN)和门控循环单元RNN(GRU RNN)是RNN的变体,解决了随着时间序列增加RNN出现的梯度消失或爆炸问题。此外,还有研究使用CNN和RNN(我们称为RCNN)的混合模型对作物进行分类。该文首先阐述了使用深度学习方法进行作物制图的背景和意义,并介绍了CNN和RNN模型结构。然后回顾了一些典型的研究,包括模型的结构、遥感数据源、数据处理方法和分类精度。最后,总结了使用深度学习方法进行作物分类的挑战以及现有解决方案的局限性。[结果](1)1-D CNN可用于提取时间特征,或时间+光谱特征,分类效果良好;2-D CNN已被广泛应用于单时相数据的空间特征提取,分类精度依赖于数据源;3-D CNN应用较少,但具有很大的潜力,尤其是时间+空间维度的特征提取;(2)相同条件下(架构、数据源、研究区域、类别),LSTM RNN和GRU RNN分类效果通常高于普通RNN,而前两者的效果差距不大,但GRU RNN训练时间较短;(3)CNN+RNN混合模型(RCNN)用RNN比3-D CNN更适合提取时间特征。这主要是由于RNN建立了对序列数据的长期依赖,而3-D CNN卷积核是局部计算的。[结论]通过分析,认为深度学习技术是作物遥感分类的有效工具。此外,与其他模型相比,RCNN,3-D CNN和GUR RNN具有更大的潜力。     

Social media bots and stock markets

This study examines the link between information spread by social media bots and stock trading. Based on a large sample of tweets mentioning 55 companies in the FTSE 100 composites, we find significant relations between bot tweets and stock returns, volatility, and trading volume at both daily and intraday levels. These results are also confirmed by an event study of stock response following abnormal increases in the volume of tweets. The findings are robust to various specifications, including controlling for traditional news channel, alternative measures of volatility, information flows in pretrading hours, and different measures of sentiment.