基于局部敏感哈希算法和神经网络学习的跨媒体检索方法

为了提高跨媒体检索的效率,可行的方法是降低数据集中不相关内容的比例。采用局部敏感哈希算法将图像数据映射到汉明空间并利用神经网络学习将文本数据映射到汉明空间的哈希函数,提出一种可以显著提高数据集中相关文件比例的高效跨媒体检索方法。实验结果表明,提出的方法能够有效去除数据集中的不相关内容,相对于已有的跨媒体检索方法,其有效提高了检索效率与准确率。     

网络异构信息的张量分解聚类方法

提出基于张量分解的聚类算法,能够同时处理网络中多类型、多语义关系的异构信息。网络信息体系中的各种异构信息被建模为一个多维张量,异构信息之间丰富的语义关系建模为张量中的元素。提出有效的张量分解方法,将不同类型的信息对象一次性划分到不同的簇中。在人工合成的数据集和真实数据集上的实验结果表明:该聚类方法可以很好地处理网络信息体系中的异构信息聚类问题,并且性能优于现有的聚类方法。     

The Environment as a Factor in Small Wars*

This research paper analyzes the relationship between small wars, insurgency, and the natural environment. Existing literature and data are organized into four behavioral patterns: the resource-based wars accounts for the fight over natural resources; the warfare ecology paradigm refers to non-premeditated damage in preparation for as well as during and after conflicts; the environment as a target discusses intended attacks on the ecosystem; and the insurgency–climate intersection pattern denotes a deviation in climate change that increases the frequency of intergroup violence. The main premise is that small wars emerge when the ecosystem becomes a political asset.     

作战设计——制胜未来之道

一流军队设计战争。作战设计是战争设计的内核。作战设计发端于科技革命和战略需求,形成于战争设计工程,实化为作战概念、作战构想、作战方案等,核心是作战体系设计和作战方式设计,基本遵循是战争制胜机理。作战设计是改变未来战争"游戏规则"的重要手段,是连接战略需求与作战能力的桥梁,是探求未来战争是什么样、未来作战怎么打、打赢战争需要什么能力的根本途径,成为开启全面建成世界一流军队的新引擎。     

Shots per casualty: an indicator of combat efficiency for the first Australian task force in South Vietnam

ABSTRACT

In combat, the ratio of shots fired per casualty inflicted can provide a measure of the combat effectiveness of a force. The shots per casualty ratio achieved by the 1st Australian Task Force in Vietnam is shown to change according to factors including marksmanship, tactics and combat type. While, over the course of the campaign, 1ATF fired an increasing number of shots to achieve a casualty, this is explained by improvements in the quality of Viet Cong and People’s Army small arms. Australian Task Force and US Army shots per casualty ratios are briefly compared..     

数据库中关联规则的并行/分布式采掘技术

高效的并行/分布式采掘方法是数据库中关联规则发现技术的重要的研究方向,首先给出了关联规则的并行/分布式采掘的相关描述,针对关联规则并行/分布式采掘的基本思想,分析算法中可以改进效能的环节,提出生成更少候选集、局部剪枝、减少通讯量和优化扫描等提高性能和降低复杂度的方法。     

Markovian models of three‐on‐one combat involving a hidden defender

In this paper, Markovian models of three‐on‐one stochastic firefights between ground‐based weapon systems are developed. These models address a common scenario of interest to the military, but one which has been much neglected in analytic combat models, that of combat between a hidden defender and an exposed attacking force. Each combatant must detect an opponent before commencing their firing cycle, a task which is considerably more difficult for the attacker. In the models developed here, the defender detects the exposed attacking group after an exponentially distributed time interval, while each attacker has a fixed probability of detecting the defender via the flash signature produced after each shot fired by him. The utility of the approach is demonstrated by investigating what impact the introduction of a coordinated gun‐laying system for the attacking force might have, a system made possible by battlefield digitization. The method used here allows models to be developed incrementally. This and other advantages of the Markovian approach are discussed. © 2002 Wiley Periodicals, Inc. Naval Research Logistics 49: 627–646, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/nav.10041     

多目标的分布式协同进化MDO算法

通过引入非优超排序和排挤的多目标处理机制 ,将分布式协同进化MDO算法的能力扩展到多目标的多学科设计优化问题。多目标的分布式协同进化MDO算法在保持各学科充分自治和各学科并行设计优化协同的基础上 ,通过一次运行即可获得具有良好分布的多个Pareto最优解 ,逼近整个Pareto最优前沿。应用于导弹气动 /发动机 /控制三学科两目标设计优化问题 ,与约束法计算结果的对比表明算法能够有效逼近该问题的Pareto最优前沿 ,为设计决策提供了丰富的信息     

基于共进化的多任务分配与调度

在考虑多种约束关系基础上 ,提出一种并行与分布式系统中多约束关系的任务分配与调度的共同进化遗传算法。仿真试验结果表明所给算法比传统单种遗传算法更能有效地进行多任务分配与调度 ,具有一定的工程价值     

基于仿真推演的导弹作战实验框架分析与设计

针对导弹武器论证设计高成本、高复杂度的问题,本文梳理了导弹作战实验设计服务的多种功能需求,阐述了导弹作战实验设计的数理方法;为构建具有较好柔性的导弹作战实验仿真框架,针对导弹系统多个性能指标,辨析了各类导弹作战实验仿真变量;针对作战实验中的环境和装备特性,梳理了导弹作战实验模型体系,提出了仿真新模型开发框架;结合导弹作战实验案例,构建了实验设计优化方法,提出了作战实验检验设计方案;从运行实验出发,阐述了设定作战想定、建立控制机制、开展结果分析等仿真环节的具体设计。通过这些工作,形成了一种支撑导弹作战实验的柔性仿真框架,为进一步开展基于仿真推演的导弹作战实验提供参考。