基于经验移植的自主空战对抗学习方法

现有的机器学习方法大多是交互式的学习模式,这类方法在训练过程非常依赖与环境之间的交互数据。空战对抗任务是一种奖励非常稀疏的训练任务,智能体在学习开始的很长一段时间内,都在探索能够获得奖励的动作。如果每一个新的任务都重新训练,是非常浪费计算资源的。因此,设计了一种基于经验移植的学习方法,使得经过训练的智能体能够将知识分享给新的智能体,提高其在新任务上的学习效率。首先,借鉴人类通过经验进行快速学习的现象,构建了基于经验移植的学习的模型;其次,兼顾知识分享和新任务的特征,明确了经验的内涵,建立了"知识+任务→经验"的融合认知方式;再次,设计了借鉴学习方法,将外部经验与任务相结合,进而转化为新个体的知识;最后,使用经验适用度作为筛选指标,分析了经验适用度对借鉴学习效率的影响,确定了执行借鉴学习的筛选边界。新个体通过借鉴学习后能够获得关于新任务的初步知识,在新任务中更快地找到能够获得奖励的动作策略,从而提升在新的任务中的学习速度。     

Expression of p53-regulated genes in human cultured lymphoblastoid TSCE5 and WTK1 cell lines after spaceflight in a frozen state

The 53 kDa tumor suppressor protein p53 is generally thought to contribute to the genetic stability of cells and to protect cells from DNA damage through the activity of p53-centered signal transduction pathways. To clarify the effect of space radiation on the expression of p53-dependent regulated genes, gene expression profiles were compared between two human cultured lymphoblastoid cell lines: one line (TSCE5) has a wild-type p53 gene status, and the other line (WTK1) has a mutated p53 gene status. Frozen human lymphoblastoid cells were stored in a freezer in the International Space Station (ISS) for 133 days. Gene expression was analyzed using DNA chips after culturing the space samples for 6 h on the ground after their return from space. Ground control samples were also cultured for 6 h after being stored in a frozen state on the ground for the same time period that the frozen cells were in space. p53-Dependent gene expression was calculated from the ratio of the gene expression values in wild-type p53 cells and in mutated p53 cells. The expression of 50 p53-dependent genes was up-regulated, and the expression of 94 p53-dependent genes was down-regulated after spaceflight. These expression data identified genes which could be useful in advancing studies in basic space radiation biology. The biological meaning of these results is discussed from the aspect of gene functions in the up- and down-regulated genes after exposure to low doses of space radiation.     

基于AES算法的FPGA实现技术研究

AES加密标准是2000年10月美国国家标准与技术研究所(NIST)提出来的新型加密算法标准,用来取代上一代的DES数据加密标准.基于AES(Advanced Encryption Standard)加密标准算法,研究其在FPGA设计中的应用.通过对AES基本原理的介绍,并通过对整体体系结构的设计,进行verilog代码编写,然后进行仿真验证,将其与理论计算结果进行了比较.通过对设计仿真综合,最后将Verilog语言的RTL级代码在FPGA进行原型验证,测试结果符合设计要求.     

构建廊坊市政府新型服务管理体系——知识管理

本文试着以廊坊市政府知识管理体系建设为切入点,作为打开市政府知识管理大门的钥匙,在发展完善知识管理理论体系、开拓知识库应用领域的同时,对促进政府适应知识经济发展、向知识型政府转变有一定的启发和指导作用。     

知识联盟企业价值链整合与竞争优势关系研究

通过企业价值链尤其是企业外部价值链的整合来提高企业的竞争优势是企业界的热门话题,企业外部价值链整合的途径有很多,企业通过参与知识联盟,与联盟伙伴一起进行知识和技术的共享能够有效的整合企业的外部价值链。知识联盟可以有效促进联盟企业的知识共享,提高联盟伙伴的创新能力,降低研发风险。通过参与知识联盟进行价值链整合,企业可以获得差异化竞争优势、规模化经济效应以及协同经济效应。     

大型复杂装备研发成本控制专家系统

根据大型复杂装备研发周期长、耗资大、风险大的特点,引入前馈控制机制,建立了大型复杂装备研发项目成本控制专家系统的体系结构,同时以该体系结构为主体,给出专家系统的知识表示及推理机的一种实现方法,最终使得知识库和推理机得以独立实现.实现后的系统以知识库和专家系统相结合的方式对装备研发成本进行实时监控和预测,对于输入的问题,系统经过推理后给出相应的参考建议,为成本管理智能化提供可行依据.由于知识库和推理机是相对独立实现,因此该系统相对传统专家系统来说具有更好的可维护性和可靠性.     

测控资源调度管理系统的构造与实现

针对我国航天测控实际问题的特点和现状,基于知识发现提出了测控资源动态调度算法的框架,在此基础上,基于数据仓库建立了测控资源动态调度管理系统的体系结构,并采用面向对象(Ob ject-O riented,O-O)技术,分析和设计了测控资源动态调度管理系统,最后阐明了该系统的优点。实际应用表明,该方法使我国“一网管多星”的效率得到极大的提高。     

基于专家系统的机场停机位自动分配系统研究

民航机场停机位分配是民航机场运营组织的关键环节.根据专家系统的基本原理,设计出了基于规则的机场停机位自动分配系统,并对该系统的知识库和推理机的工作机理进行了较深入探讨,以期为民航机场停机位分配提供先进的技术手段.     

支持工艺设计的制造知识类型及层次模型

针对工艺设计人员对大量制造知识的需求问题,分析了制造知识与工艺设计的关系,按照对工艺设计的支持作用,将制造知识分为制造资源类、制造对象类和制造过程类三大类别,建立了包括零件层、属性层和数据库层的制造知识结构层次模型,以制造设备为例,采用IDEF1X方法建立了支持工艺设计的制造知识表达模型.     

Egocentric frames of reference used for the retrieval of survey knowledge learned by map and navigation

Two experiments are reported that use a ``point-to-unseen-targets' task to study the role of egocentric reference frames in the retrieval of survey knowledge learned from either studying a map or navigating an environment. In Experiment 1, performance was generally consistent with the hypothesis that map knowledge is retrieved using a frame of reference centered on the eye, characterized by (a) a fixed orientation in a ``frontal representational plane' and (b) equal access to spatial relations both in front of, or above, and behind, or below, a right-left retrieval axis. The results of Experiment 2 were consistent with the hypothesis that environment knowledge is retrieved within a frame of reference centered on the body, characterized by (a) flexible orientation within a ``transverse representational plane' and (b) privileged access to spatial relations located in front of the right-left retrieval axis in representational space. Both types of knowledge function as if they preserve information about the Euclidean angles connecting elements in physical space.