DNA甲基化作为一种重要的表观遗传修饰,其甲基化水平被发现与疾病的发生发展密切相关,对其进行聚类分析有希望发现新的疾病亚型并建立有效的疾病预测预后方法。传统的聚类分析方法之一模糊C-均值(FCM:Fuzzy C-means)适用于特征空间呈球...DNA甲基化作为一种重要的表观遗传修饰,其甲基化水平被发现与疾病的发生发展密切相关,对其进行聚类分析有希望发现新的疾病亚型并建立有效的疾病预测预后方法。传统的聚类分析方法之一模糊C-均值(FCM:Fuzzy C-means)适用于特征空间呈球形或椭球形分布的场景,缺乏普适性。而Illumina Golden Gate平台通过计算基因的各甲基化位点的甲基化百分比描述其甲基化程度,其值位于(0,1)之间,服从混合贝塔分布,不能直接采用FCM进行聚类分析。鉴于此,本文提出基于KL特征测度的KL-FCM聚类算法,采用各样本间的K-L距离作为样本划分时的度量准则。最后,本文基于KL-FCM算法实现IRIS测试数据集和基因的DNA甲基化水平数据的聚类分析。实验结果表明该方法可以以更低的计算负荷获得优于k-均值(k-means)和传统FCM的分类效果。展开更多
The research of robot localization aims at accuracy, simplicity and robustness. This article improves the performance of particle filters in robot localization via the utilization of novel adaptive technique. The prop...The research of robot localization aims at accuracy, simplicity and robustness. This article improves the performance of particle filters in robot localization via the utilization of novel adaptive technique. The proposed algorithm introduces probability retracing to initialize particle sets, uses consecutive window filtering to update particle sets, and refreshes the size of particle set according to the estimation state. Extensive simulations show that the proposed algorithm is much more effective than the traditional particle filters. The proposed algorithm successfully solves the nonlinear, non-Gaussian state estimation problem of robot localization.展开更多
文摘DNA甲基化作为一种重要的表观遗传修饰,其甲基化水平被发现与疾病的发生发展密切相关,对其进行聚类分析有希望发现新的疾病亚型并建立有效的疾病预测预后方法。传统的聚类分析方法之一模糊C-均值(FCM:Fuzzy C-means)适用于特征空间呈球形或椭球形分布的场景,缺乏普适性。而Illumina Golden Gate平台通过计算基因的各甲基化位点的甲基化百分比描述其甲基化程度,其值位于(0,1)之间,服从混合贝塔分布,不能直接采用FCM进行聚类分析。鉴于此,本文提出基于KL特征测度的KL-FCM聚类算法,采用各样本间的K-L距离作为样本划分时的度量准则。最后,本文基于KL-FCM算法实现IRIS测试数据集和基因的DNA甲基化水平数据的聚类分析。实验结果表明该方法可以以更低的计算负荷获得优于k-均值(k-means)和传统FCM的分类效果。
文摘The research of robot localization aims at accuracy, simplicity and robustness. This article improves the performance of particle filters in robot localization via the utilization of novel adaptive technique. The proposed algorithm introduces probability retracing to initialize particle sets, uses consecutive window filtering to update particle sets, and refreshes the size of particle set according to the estimation state. Extensive simulations show that the proposed algorithm is much more effective than the traditional particle filters. The proposed algorithm successfully solves the nonlinear, non-Gaussian state estimation problem of robot localization.