针对密度峰值聚类算法(Density Peaks Clustering,DPC)需要人为指定截断距离d c,以及局部密度定义简单和一步分配策略导致算法在复杂数据集上表现不佳的问题,提出了一种基于自然最近邻的密度峰值聚类算法(Density Peaks Clustering base...针对密度峰值聚类算法(Density Peaks Clustering,DPC)需要人为指定截断距离d c,以及局部密度定义简单和一步分配策略导致算法在复杂数据集上表现不佳的问题,提出了一种基于自然最近邻的密度峰值聚类算法(Density Peaks Clustering based on Natural Nearest Neighbor,NNN-DPC)。该算法无需指定任何参数,是一种非参数的聚类方法。该算法首先根据自然最近邻的定义,给出新的局部密度计算方法来描述数据的分布,揭示内在的联系;然后设计了两步分配策略来进行样本点的划分。最后定义了簇间相似度并提出了新的簇合并规则进行簇的合并,从而得到最终聚类结果。实验结果表明,在无需参数的情况下,NNN-DPC算法在各类数据集上都有优秀的泛化能力,对于流形数据或簇间密度差异大的数据能更加准确地识别聚类数目和分配样本点。与DPC、FKNN-DPC(Fuzzy Weighted K-nearest Density Peak Clustering)以及其他3种经典聚类算法的性能指标相比,NNN-DPC算法更具优势。展开更多
标记传播是使用最广泛的半监督分类方法之一。基于共识率的标记传播算法(Consensus Rate-based Label Propagation,CRLP)通过汇总多个聚类方法以合并数据各种属性得到的共识率来构造图。然而,CRLP算法与大多数基于图的半监督分类方法一...标记传播是使用最广泛的半监督分类方法之一。基于共识率的标记传播算法(Consensus Rate-based Label Propagation,CRLP)通过汇总多个聚类方法以合并数据各种属性得到的共识率来构造图。然而,CRLP算法与大多数基于图的半监督分类方法一样,在图中将每个标记样本视为同等重要,它们主要通过优化图的结构来提高算法的性能。事实上,样本不一定是均匀分布的,不同的样本在算法中的重要性也是不同的,并且CRLP算法容易受聚类数目和聚类方法的影响,对低维数据的适应性不足。针对这些问题,文中提出了一种基于加权样本和共识率的标记传播算法(Label Propagation Algorithm Based on Weighted Samples and Consensus-Rate,WSCRLP)。WSCRLP算法首先对数据集进行多次聚类,以探索样本的结构,并结合共识率和样本的局部信息构造图;然后为不同分布的标记样本分配不同的权重;最后基于构造的图和加权样本进行半监督分类。在真实数据集上的实验表明,WSCRLP算法对标记样本进行加权和构造图的方法可以显著提高分类准确率,在84%的实验中都优于对比方法。相比CRLP算法,WSCRLP算法不仅具有更好的性能,而且对输入参数具有鲁棒性。展开更多
OBJECTIVE TO investigate the neural protection of dehydrocostus lactone(DHL)against neuronal injury induced by oxygen and glucose deprivation/reperfusion(OGD/R)in differentiated PC12 cells.METHODS We used a cellular m...OBJECTIVE TO investigate the neural protection of dehydrocostus lactone(DHL)against neuronal injury induced by oxygen and glucose deprivation/reperfusion(OGD/R)in differentiated PC12 cells.METHODS We used a cellular model of 2 h of OGD and 24 h of reperfusion to mimic cerebral ischemia-reperfusion injury.Cell viability was used to reflect the degree of OGD/R-induced injury.Cells were treated with DHL during the reperfusion phase.Cell Counting Kit(CCK-8)and LDH assays were performed to determine the optimal dose of DHL and cell viability.Flow cytometry analysis and Monodansylcadaverine(MDC)staining were then conducted to detect apoptosis rate and autophagosome formation after OGD/R in PC12 cells.Immunofluorescence and Western blotting analyses were used to detect the expres⁃sion of proteins associated with autophagy and apoptosis.RESULTS OGD/R significantly decreased cell viability and increased apoptosis rate.The expression levels of autophagy-related proteins,namely,LC3 and Beclin-1,and apoptosisrelated proteins,namely,Bax and caspase-3 increased,but that of the anti-apoptosis Bcl-2 protein decreased.However,DHL attenuated OGD/R-induced neuronal injury through inhibition of apoptosis and autophagy properties by modulating au⁃tophagy-associated proteins(LC3 and Beclin-1)and apoptosis-modulating proteins(caspase-3 and Bcl-2/Bax).CONCLU⁃SION Our data provide an evidence for the neuroprotective effect of DHL against ischemic neuronal injury.Hence,DHL could be a promising candidate for treatment of ischemic stroke.展开更多
文摘针对密度峰值聚类算法(Density Peaks Clustering,DPC)需要人为指定截断距离d c,以及局部密度定义简单和一步分配策略导致算法在复杂数据集上表现不佳的问题,提出了一种基于自然最近邻的密度峰值聚类算法(Density Peaks Clustering based on Natural Nearest Neighbor,NNN-DPC)。该算法无需指定任何参数,是一种非参数的聚类方法。该算法首先根据自然最近邻的定义,给出新的局部密度计算方法来描述数据的分布,揭示内在的联系;然后设计了两步分配策略来进行样本点的划分。最后定义了簇间相似度并提出了新的簇合并规则进行簇的合并,从而得到最终聚类结果。实验结果表明,在无需参数的情况下,NNN-DPC算法在各类数据集上都有优秀的泛化能力,对于流形数据或簇间密度差异大的数据能更加准确地识别聚类数目和分配样本点。与DPC、FKNN-DPC(Fuzzy Weighted K-nearest Density Peak Clustering)以及其他3种经典聚类算法的性能指标相比,NNN-DPC算法更具优势。
文摘标记传播是使用最广泛的半监督分类方法之一。基于共识率的标记传播算法(Consensus Rate-based Label Propagation,CRLP)通过汇总多个聚类方法以合并数据各种属性得到的共识率来构造图。然而,CRLP算法与大多数基于图的半监督分类方法一样,在图中将每个标记样本视为同等重要,它们主要通过优化图的结构来提高算法的性能。事实上,样本不一定是均匀分布的,不同的样本在算法中的重要性也是不同的,并且CRLP算法容易受聚类数目和聚类方法的影响,对低维数据的适应性不足。针对这些问题,文中提出了一种基于加权样本和共识率的标记传播算法(Label Propagation Algorithm Based on Weighted Samples and Consensus-Rate,WSCRLP)。WSCRLP算法首先对数据集进行多次聚类,以探索样本的结构,并结合共识率和样本的局部信息构造图;然后为不同分布的标记样本分配不同的权重;最后基于构造的图和加权样本进行半监督分类。在真实数据集上的实验表明,WSCRLP算法对标记样本进行加权和构造图的方法可以显著提高分类准确率,在84%的实验中都优于对比方法。相比CRLP算法,WSCRLP算法不仅具有更好的性能,而且对输入参数具有鲁棒性。
基金National Natural Science Foundation of China(8166070081260679)Ningxia Col ege First-Class Discipline Construction Project(Chinese Medicine)Funded Project(NXYLXK2017A06)
文摘OBJECTIVE TO investigate the neural protection of dehydrocostus lactone(DHL)against neuronal injury induced by oxygen and glucose deprivation/reperfusion(OGD/R)in differentiated PC12 cells.METHODS We used a cellular model of 2 h of OGD and 24 h of reperfusion to mimic cerebral ischemia-reperfusion injury.Cell viability was used to reflect the degree of OGD/R-induced injury.Cells were treated with DHL during the reperfusion phase.Cell Counting Kit(CCK-8)and LDH assays were performed to determine the optimal dose of DHL and cell viability.Flow cytometry analysis and Monodansylcadaverine(MDC)staining were then conducted to detect apoptosis rate and autophagosome formation after OGD/R in PC12 cells.Immunofluorescence and Western blotting analyses were used to detect the expres⁃sion of proteins associated with autophagy and apoptosis.RESULTS OGD/R significantly decreased cell viability and increased apoptosis rate.The expression levels of autophagy-related proteins,namely,LC3 and Beclin-1,and apoptosisrelated proteins,namely,Bax and caspase-3 increased,but that of the anti-apoptosis Bcl-2 protein decreased.However,DHL attenuated OGD/R-induced neuronal injury through inhibition of apoptosis and autophagy properties by modulating au⁃tophagy-associated proteins(LC3 and Beclin-1)and apoptosis-modulating proteins(caspase-3 and Bcl-2/Bax).CONCLU⁃SION Our data provide an evidence for the neuroprotective effect of DHL against ischemic neuronal injury.Hence,DHL could be a promising candidate for treatment of ischemic stroke.
