Mandated defence offsets: can they ever deliver?

Defence offsets are elements of defence procurement deals additional to the primary content. Offsets are usually expected to yield technological or industrial benefits to the purchasing country (e.g. countertrade, technology transfers, or additional jobs) and military buyers often require suppliers to make offsets available “cost-free.” The authors argued previously that such strategies achieve little of value to buyers that lack market power and are unnecessary otherwise, since purchasers with the market power to extract more value for money from foreign suppliers can do so anyway. This article also focuses on the supply side of offset deals. The USA is the world's largest defence offsets supplier but the US government opposes offsets demands as economically inefficient and trade distorting. Even if offsets are inefficient and trade distorting, they may still benefit a materiel-exporting country such as the USA as they may induce exports and create associated benefits for the offsets provider.     

微信:网络舆论战的新利器

微信以惊人的速度发展,成为时下最热门的社交信息平台和绝大多数手机用户的一种生活方式。微信是舆论生成和扩散的重要平台,其多元化的信息、多渠道的传播以及创新的特点,为网络舆论战带来了全新的机遇。把握微信的信息传播特点,有效针对特定对象人群,采取积极主动的方式开展网络舆论宣传与自我防御,使微信成为舆论战发展的有力工具,是新时期开展网络舆论战的必然选择。     

The German wars of liberation 1807–1815: The restrained insurgency

In the Age of Napoleon, ‘small wars’ and ‘revolutionary war’ were closely connected. There were, however, different strands of this phenomenon: speaking professionally, conservative officers condemned small wars as an irregular regression to previous less disciplined forms of warfare. The Prussian state continually tried to discipline and regulate spontaneous risings. Yet the irregular character of small wars offered the opportunities for a less complex way of fighting, thus enabling the arming of the ‘people’ to fight. Individual undertakings, such as Ferdinand von Schill's doomed campaign in 1809, were designed to spark off a general popular uprising. But they were cheered by many and supported by few. Meanwhile, Neidhardt von Gneisenau conceived guerrilla-style Landsturm home-defence forces, which were designed for an irregular people's war. These concepts were put into practice in the ‘war of freedom’ – or ‘war of liberation’ – in 1813. Eventually both the mobilisation and the tactics remained regular, however, despite the emphatic appeal to a national ‘people's war’.     

求解面向进攻的武器-目标分配问题的蚁群算法

面向进攻的武器-目标分配问题是军事运筹学研究中的重要课题,旨在制定合理的打击策略以最大程度摧毁敌方目标。采用一种融合局部搜索和信息素控制的蚁群算法,兼顾控制解的局部收敛速度和全局收敛质量。在解的构造过程中直接处理约束条件,提高生成解的可行性,并大大缩小了搜索空间,提高了算法效率。通过采用多种算法对不同规模的武器-目标分配问题进行实验,结果表明改进的蚁群算法在收敛速度和求解质量上表现优异。     

Approximations and empirics for stochastic war equations

The article develops a theorem which shows that the Lanchester linear war equations are not in general equal to the Kolmogorov linear war equations. The latter are time‐consuming to solve, and speed is important when a large number of simulations must be run to examine a large parameter space. Run times are provided, where time is a scarce factor in warfare. Four time efficient approximations are presented in the form of ordinary differential equations for the expected sizes and variances of each group, and the covariance, accounting for reinforcement and withdrawal of forces. The approximations are compared with “exact” Monte Carlo simulations and empirics from the WWII Ardennes campaign. The band spanned out by plus versus minus the incremented standard deviations captures some of the scatter in the empirics, but not all. With stochastically varying combat effectiveness coefficients, a substantial part of the scatter in the empirics is contained. The model is used to forecast possible futures. The implications of increasing the combat effectiveness coefficient governing the size of the Allied force, and injecting reinforcement to the German force during the Campaign, are evaluated, with variance assessments. © 2005 Wiley Periodicals, Inc. Naval Research Logistics, 2005.     

政府适度负债规模的研究——从欧洲主权债务危机到中国地方政府负债风险谈起

在现代经济环境下政府倘若仅依靠自身财力,既无外债又无内债的发展经济已显不足;但凡经济发展较快的国家无不举债,依靠举债借助外力不仅能促进且能加快经济发展。然而近年西欧爆发主权债务危机,我国中央地方两级负债暗藏隐性风险,地方政府融资平台风险甚是暴露,也从另一个角度警示政府负债是一把双刃剑。而根据自身实力与未来发展预期适度负债才是最佳选择,负债不足与过度负债都是不可取的。本文通过对政府适度负债标准的研究,为各级政府提出一个适度负债标准或控制临界点,同时也为各级政府根据此标准对自身负债水平提出相应的预警及对策,从而保持经济的稳定发展。     

中文分词与命名实体识别的联合学习

将卷积结构引入循环神经网络,从而构建卷积循环神经网络。以此为基础,研究构建了面向中文分词与实体识别联合学习的序列标注模型。该模型依托卷积循环神经网络构建特征编码层,实现中文字序列局部空间特征和长距离时序依赖特征的联合提取;依托改进的循环神经网络构建标签解码层,实现标签序列长距离时序依赖的有效建模;依托统一的分词与实体识别序列标注模式实现分词信息与实体信息的联合学习,避免传统流水线法的误差传播问题。在人民日报语料和微软标注语料上的实验结果显示,该框架较传统统计模型和神经网络模型有显著的性能提升,尤其是在识别字数较多的命名实体时,其效果明显优于其他方法。     

一个监控式局部自主机器人控制器的实现与分析

本文介绍了一个严格监督控制（Ｓｕｐｅｒｖｉｓｏｒｙｃｏｎｔｒｏｌ）定义下的局部自主机器人控制器ＧＫＤ３的设计和实现，并从监督控制理论的角度，对ＧＫＤ３的体系结构和功能进行了分析和讨论。     

基于像素局部对比度的直线特征检测算法*

本文提出了基于像素局部对比度和全局虚警抑制的直线段检测算法。本文算法分为四步：第一步,计算像素点的梯度以及局部对比度;第二步,将方向大致相同的像素点分成直线支撑区域(像素方向为像素梯度方向的顺时针垂直方向),并对支撑区域进行矩形近似;第三步,在直线支撑区域内拟合直线段,计算直线参数;第四步,根据直线支撑区域信息验证直线段,抑制虚警。实验表明本文算法在复杂条件下,特别是光照影响条件下,具有更好的鲁棒性和效率。