在传统K-means算法中,初始簇中心选择的随机性,导致聚类结果随不同的聚类中心而不同。因此出现了很多簇中心的选择方法,但是很多已有的簇中心选择算法,其聚类结果受参数调节的影响较大。针对这一问题,提出了一种新的初始簇中心选择算法...在传统K-means算法中,初始簇中心选择的随机性,导致聚类结果随不同的聚类中心而不同。因此出现了很多簇中心的选择方法,但是很多已有的簇中心选择算法,其聚类结果受参数调节的影响较大。针对这一问题,提出了一种新的初始簇中心选择算法,称为WLV-K-means(weighted local variance K-means)。该算法采用加权局部方差度量样本的密度,以更好地发现密度高的样本,并利用改进的最大最小法,启发式地选择簇初始中心点。在UCI数据集上的实验结果表明,WLV-K-means算法不仅能够取得较好的聚类结果,而且受参数变化的影响较小,有更加稳定的表现。展开更多
In this paper, we propose a new method to removing the mixed Gaussian and salt-pepper noise based on wavelet. To estimate outlier, A scheme called max-min method is adopted after DWT. Experimental results show that th...In this paper, we propose a new method to removing the mixed Gaussian and salt-pepper noise based on wavelet. To estimate outlier, A scheme called max-min method is adopted after DWT. Experimental results show that this method is more effective than common image restoration methods, such as Median filter, center weighted median filter.展开更多
This paper describes analytical and numerical methods to analyze the steady state periodic response of an oscillator with symmetric elastic and inertia nonlinearity. A new implementation of the homotopy perturbation m...This paper describes analytical and numerical methods to analyze the steady state periodic response of an oscillator with symmetric elastic and inertia nonlinearity. A new implementation of the homotopy perturbation method （HPM） and an ancient Chinese method called the max-rain approach are presented to obtain an approximate solution. The major concern is to assess the accuracy of these approximate methods in predicting the system response within a certain range of system parameters by examining their ability to establish an actual （numerical） solution. Therefore, the analytical results are compared with the numerical results to illustrate the effectiveness and convenience of the proposed methods.展开更多
文摘在传统K-means算法中,初始簇中心选择的随机性,导致聚类结果随不同的聚类中心而不同。因此出现了很多簇中心的选择方法,但是很多已有的簇中心选择算法,其聚类结果受参数调节的影响较大。针对这一问题,提出了一种新的初始簇中心选择算法,称为WLV-K-means(weighted local variance K-means)。该算法采用加权局部方差度量样本的密度,以更好地发现密度高的样本,并利用改进的最大最小法,启发式地选择簇初始中心点。在UCI数据集上的实验结果表明,WLV-K-means算法不仅能够取得较好的聚类结果,而且受参数变化的影响较小,有更加稳定的表现。
文摘In this paper, we propose a new method to removing the mixed Gaussian and salt-pepper noise based on wavelet. To estimate outlier, A scheme called max-min method is adopted after DWT. Experimental results show that this method is more effective than common image restoration methods, such as Median filter, center weighted median filter.
文摘针对核模糊C-均值算法(kernel fuzzy C-means,KFCM)随机选择初始聚类中心而不能获得全局最优且在聚类中心较近或重合时易产生一致性聚类等问题,提出一种改进算法。改进算法在原目标函数中引入中心极大化约束项来调控簇间分离度,从而避免算法出现一致性聚类结果。利用磷虾群算法对基于新目标函数的KFCM算法进行优化,使算法不再依赖初始聚类中心,提高算法的稳定性。基于距离最大最小原则产生多组较优的聚类中心作为初始磷虾群体并在算法迭代过程中融合一种新的精英保留策略,从而确保算法收敛到全局极值;通过对个体随机扩散活动进行分段式Logistic混沌扰动,提高算法全局寻优能力。使用KDD Cup 99入侵检测数据进行仿真实验表明,改进算法具有更好的检测性能,解决了传统的聚类算法在入侵检测中稳定性差、检测准确率低的问题。
文摘This paper describes analytical and numerical methods to analyze the steady state periodic response of an oscillator with symmetric elastic and inertia nonlinearity. A new implementation of the homotopy perturbation method （HPM） and an ancient Chinese method called the max-rain approach are presented to obtain an approximate solution. The major concern is to assess the accuracy of these approximate methods in predicting the system response within a certain range of system parameters by examining their ability to establish an actual （numerical） solution. Therefore, the analytical results are compared with the numerical results to illustrate the effectiveness and convenience of the proposed methods.
