半监督聚类在入侵检测中的应用研究     

周志平  庄金莲  陈佳丽 《南平师专学报》2013,(5):56-60

随着网络的快速发展,入侵检测系统生成的告警信息越来越多,聚类技术广泛的应用于处理告警信息.针对传统的K-Means算法易陷入局部最优,提出一种改进半监督聚类算法ISC.从数据集中抽取若干正常与异常样本分别采用层次聚类算法分别计算作为初始质心辅助K-Means算法进行聚类.实验结果表明,与现有相关算法相比,该算法具有更高的攻击检测率以及更低的误报率.  相似文献   

Adult students' achievement goal orientations and evaluations of the learning environment: a person-centred longitudinal analysis     

Antti-Tuomas Pulkka  Markku Niemivirta 《Educational Research and Evaluation》2013,19(4):297-322

This study focused on the stability and change in students' achievement goal orientations and whether the students' perceptions of the learning environment vary as a function of their achievement goal orientations. Participants were 169 students of the Finnish National Defense University. The students' goal orientations and their evaluations of the learning environment were assessed twice, approximately four months apart. Four groups of students were identified based on their achievement goal orientation profiles, and 60% of the participants displayed identical profiles in both measurements. Students with different goal orientation profiles differed in their evaluations of the quality of pedagogical materials, effort and attainment, and participation. Students oriented towards an increase of competence and mastery or towards normative performance and success were most positive in their evaluations. Students oriented towards avoidance of effort or display of incompetence gave less positive ratings. The relation between motivation and perceptions of learning environment is discussed.  相似文献   

Growth Mixture Models Outperform Simpler Clustering Algorithms When Detecting Longitudinal Heterogeneity,Even With Small Sample Sizes     

Daniel P. Martin  Timo von Oertzen 《Structural equation modeling》2013,20(2):264-275

The expansion in the number of people entering higher education has resulted in a substantial increase in the proportion of students enrolling in nontraditional modes, such as part-time study. This study examined the question of whether part-time study curtails the development of the types of intellectual capabilities needed for a knowledge-based economy, because the students would have markedly less exposure to a stimulating campus environment than their full-time counterparts. Graduates from discrete full- and part-time programs from 1 university in Hong Kong completed a survey seeking perceptions of the nurturing of a range of capabilities, together with measures of teacher-student relationships and type of teaching experienced. Two hypotheses were tested by structural equation modeling: (a) the same mechanism for capability development operated for full- and part-time modes and (b) the principal element of the mechanism was the nature of teaching and the quality of teacher-student interaction. Hypothesis 1 was supported because configural invariance between hypothesized models for capability development between the 2 modes was found. Hypothesis 2 was also supported because the models showed that the principal influence on capability development came from teaching for understanding, through promoting active learning experiences and the degree and quality of teacher-student interaction.  相似文献   

大学生数学史教育现状的灰色模糊评价模型     

徐婷 《海南广播电视大学学报》2013,(1):149-154

采用灰色模糊聚类模型,研究大学生数学史教育现状。结果表明:学生对数学史认识比较高,掌握的数学史知识比较差,运用数学史能力更差。  相似文献   

K-means聚类算法研究   总被引：2，自引：0，他引：2  

华婷婷 《黄山高等专科学校学报》2013,(5):17-19

聚类分析是模式识别中一个重要研究领域,是一种将数据划分或分组处理的重要手段和方法.K-means算法是聚类算法中主要算法之一,文章介绍了K-means聚类算法的实现方法,并对其优点和缺点进行了详细的分析.  相似文献   

葡萄与葡萄酒质量的综合评价   总被引：1，自引：0，他引：1  

朱家明  张晓芳  王强  胡继俊 《通化师范学院学报》2013,34(6):8-12

从体现酒品质量的感官评价和酿酒葡萄与葡萄酒的理化指标两方面入手,运用t检验、主成分分析法、模糊c均值聚类分析、多元线性回归分别建立显著性差异检验、主成分、聚类分析、质量影响和论证等模型,并运用SPSS、Matlab7软件,得出两组评酒员的评价结果差异性和可信度,酿酒葡萄的分级,酿酒葡萄与葡萄酒的理化指标之间的联系,最后分析并论证了葡萄、葡萄酒的理化指标与葡萄酒质量的关系.  相似文献   

基于传感器网络的分布式遗传k-means聚类算法研究     

洪月华  朱景锋 《南宁职业技术学院学报》2013,(3):79-82

传统的集中式聚类算法不适宜对传感器网络的分布式数据进行聚类,用遗传进化机制对传统k-means的分布式聚类算法进行优化,可得出遗传k-means聚类算法。遗传k-means聚类算法即在传感器网络中sink节点传送随机选取的初始k个簇心到各个传感器节点,在这些节点上分别用遗传k-means聚类算法将本地的数据划分到距离最近的簇,然后将簇信息在无线传感器网络里通过路由逐层上传合并汇聚到sink节点,计算k个簇心的平均值,再往下传送k个簇心,反复迭代更新直至聚类目标函数值达到最小为止。实验表明,遗传k-means聚类算法的聚类效果较好,收敛速度较快。  相似文献   

基于FCM聚类及抖动半调的数码迷彩生成方法     

朱映辉  江玉珍  李建忠 《韩山师范学院学报》2013,(3):33-37

数码迷彩因其不规则、不平滑的边缘特征而具备更强的伪装适应性．为快速有效地生成实用型数码迷彩,在传统迷彩基础上,提出一种基于误差扩散（E—D）抖动半调的数码迷彩生成算法．利用像素的三基色分量值作为特征值对迷彩原图作模糊C均值（FCM）聚类分析以提取迷彩主色,结合图像缩放、Gaussian滤波等图像处理,运用E—D抖动半调方法将传统迷彩图转换成数码迷彩图．实验证明,生成图既保持了迷彩原图的颜色分布及纹理特征,也具备了不规则的、锯齿状的边缘特征,有效地提高了迷彩的伪装隐蔽性能．  相似文献   

基于C-OWA算子的业主驱动下BIM应用效益灰色聚类评价     

董娜  胡琳璐 《科技管理研究》2019,(4)

随着BIM的推广普及和深化应用,构建一套科学可行的BIM效益评价体系对推动BIM落地和效益发挥具有非常重要的现实意义。本文在文献综述的基础上,从BIM应用现状和业主方项目管理目标的视角出发,从七个维度建立了业主方驱动下的BIM效益评价指标,运用C-OWA算子赋权法和灰色系统理论进行评价,提出了业主方BIM应用效益评价指标体系,为BIM应用的前期决策和后期评价提供参考,促进BIM效益的真正发挥。  相似文献   

网络舆情衍进指数构建与实证分析   总被引：1，自引：0，他引：1  

黄微  朱镇远  许烨婧  孙悦 《图书情报工作》2019,63(20):26-33

[目的/意义]提出和构建网络舆情衍进指数，以描述网络舆情演化过程中常衍生出新的子话题的现象，对于舆情预警、预测具有重要的理论及实践意义。[方法/过程]以文本聚类结果和文本聚类有效性为依据，提出网络舆情衍进的判别标准和舆情衍进指数的构建过程，并以"教科书老赖"这一事件作为样本数据进行实证分析。[结果/结论]所构建的舆情衍进速率指数可以用于描述舆情衍进。在突发期阶段话题舆情衍进指数最高，此后逐渐下降，这一阶段的舆情衍进最为剧烈，子话题的出现呈现爆发性增长；舆情衍进指数在舆情蔓延期内出现阶梯式下降，此后保持为负值，舆情的子话题开始逐渐减少，舆情内容本身由发散转为收敛；进入消散期后，子话题数量趋于稳定。作为舆情衍进速率的测度和舆情衍进的判别方式，舆情衍进指数为舆情监管和舆情预警提供了全新的角度。  相似文献