指数分布在嵌入马氏链构造中的应用

对于出现指数分布的过程 ,利用指数分布的无记忆性 ,结合实例 ,构造出了马氏更新过程 ,导出了其相伴的半马氏核和嵌入马氏链的转移阵     

一类非线性系统有界解的存在性

考虑非线性系统x′=A(t)x+f(t,x)有界解的存在性,其中线性系统x′=A(t)x满足指数型二分性.在f(t,x)关于x不满足Lipschitz条件的情况下,应用Leray-Shauder不动点定理和Arzela-Ascoli定理给出一个有界连续解存在的充分条件.即若f(t,x)∶R×Rn→Rn连续;存在常数m>0及R+=[0,∞)上的连续递增函数g(t)满足limt→∞(g(t))/t=0,使得|f(t,x)|≤m+g(|x|),(t,x)∈R×Rn,则该系统x′=A(t)x+f(t,x)存在有界连续解.     

一类具有脉冲和时滞的细胞神经网络系统的反周期解

通过微分不等式分析技巧，得出了一类具有脉冲和时滞的细胞神经网络系统反周期解存在性和全局指数稳定性的新的充分条件，推广和改进了已有文献的相关结果。     

生存寿命的联合预测模型及其应用

基于模糊示例学习与指数回归理论分别得到患者生存寿命的预测模型,并对2种模型进行优化组合,提出联合预测模型,其预测结果可为患者选择治疗方案提供必要的信息.     

Engineering education as a complex system

This paper presents a theoretical basis for cultivating engineering education as a complex system that will prepare students to think critically and make decisions with regard to poorly understood, ill-structured issues. Integral to this theoretical basis is a solution space construct developed and presented as a benchmark for evaluating problem-solving orientations that emerge within students’ thinking as they progress through an engineering curriculum. It is proposed that the traditional engineering education model, while analytically rigorous, is characterised by properties that, although necessary, are insufficient for preparing students to address complex issues of the twenty-first century. A Synthesis and Design Studio model for engineering education is proposed, which maintains the necessary rigor of analysis within a uniquely complex yet sufficiently structured learning environment.     

Masculinity,educational achievement and social status: a social network analysis

This study utilises a quantitative case study social network approach to explore the connection between masculinity and scholastic achievement in two secondary, all-boys schools in Australia. In both schools two social networks representing social status are explored: the friendship network as a measure of status that includes emotional investment, and the influence network as a measure of status without emotion. Results demonstrate that for both schools the friendship network is not associated with academic achievement, though boys with similar levels of academic application cluster together. However, for the influence network there are effects related to achievement, application and attitudes toward an anti-academic/masculinity connection, as too toward ethno-cultural background. The results differ by school and show the importance of local context. However, similarities in the schools show that masculine identities may be engaged in the social interactions of students, ones which engage masculine stereotypes in public but not in private.     

双参数指数分布族参数的经验贝叶斯估计的渐近性

在LINEX损失函数下,讨论双参数指数分布位置参数的Bayes估计。假设样本是iid,利用概率密度函数的核估计方法,构造边缘分布的概率密度估计,按照参数的Bayes估计形式,提出参数的经验Bayes(EB)估计函数,在一定的条件下可以证明所提出的这个经验Bayes估计函数是渐近最优的,并获得其收敛速度,文尾举例说明满足定理条件的参数的先验分布是存在的。     

混合指数分布定数结尾时参数估计的Monte Carlo EM加速算法

应用Monte Carlo EM加速算法对于单参数混合指数分布在定数结尾场合的参数估计问题做了较深入的探讨,做出了模拟试验。     

基于用户画像和视频兴趣标签的个性化推荐

【目的/意义】用户画像深刻地描述了视频用户的个体和群体行为特征,为视频的个性化推荐服务提供参 考。【方法/过程】通过文本挖掘对爬取的视频、用户及其观影数据分析,构建单个用户画像,并通过K-Means和LDA 模型对用户聚类并提取主题,挖掘群体用户特征。基于用户画像和时间指数衰减的视频兴趣标签,并结合视频喜 爱度和协同过滤,进行视频推荐。【结果/结论】考虑时间指数衰减的个性化推荐,提高了系统对用户兴趣的感知。 结合视频喜爱度和协同过滤,推荐视频评分达0.87,有助于提高用户对网站的忠诚度和活跃度。【创新/局限】基于用 户生成内容的文本挖掘结果,进行单个和群体用户画像,并创新性采用时间指数衰减构建用户视频兴趣标签,以捕 获用户兴趣的变化。由于网络爬虫的限制,实验数据量有一定的局限性,且特征提取兴趣范围有限。     

广义指数分布在TFR模型中的参数估计

广义指数分布是应用非常广泛的一种分布,近年对该分布的讨论主要是常规寿命试验数据的统计分析方法,研究重点是参数的点估计.但基于不完全样本、应用TFR模型、探讨广义指数分布在步加试验中的参数估计的文献却很少见.对此,利用EM算法给出了参数估计的显性表达式,并通过数据模拟说明了估计方法的可行性.