提出一种基于标签正负相关性的多标签类属特征学习方法(multi-label learning with label-specific features based on positive and negative label correlation,LIFTPNL)。基于k近邻的思想构建全局和局部的标签信息矩阵,根据此矩阵计...提出一种基于标签正负相关性的多标签类属特征学习方法(multi-label learning with label-specific features based on positive and negative label correlation,LIFTPNL)。基于k近邻的思想构建全局和局部的标签信息矩阵,根据此矩阵计算成对标签的正负相关性;对每个类别标签,基于属于相同和不同类簇的样本构建连接矩阵,联合该标签正负相关性计算样本相似度;采用谱聚类方法获得聚类中心,将原有特征转换成类属特征;通过二分类器得到分类结果。实验结果表明,所提算法优于多种多标签分类方法。展开更多
多标签特征选择能够有效去除冗余特征并提升分类精度,是解决“维数灾难”问题的有效方法.然而,已有的多标签特征选择算法是对所有标签选择出相同的特征,忽略了标签与特征之间的内在联系.事实上,每个标签都具有反映该标签特有属性的特征...多标签特征选择能够有效去除冗余特征并提升分类精度,是解决“维数灾难”问题的有效方法.然而,已有的多标签特征选择算法是对所有标签选择出相同的特征,忽略了标签与特征之间的内在联系.事实上,每个标签都具有反映该标签特有属性的特征,即类属特征.提出一种基于图拉普拉斯的多标签类属特征选择(multi-label label-specific feature selection based on graph Laplacian,LSGL)算法.对于每个类别标签,基于拉普拉斯映射获得数据的低维嵌入,再通过稀疏正则化获得数据空间到嵌入空间的投影矩阵,接着通过分析矩阵系数确定每个标签相应的类属特征,最后使用类属特征进行分类.在5个公共多标签数据集上的多标签特征选择与分类实验结果证明了所提算法的有效性.展开更多
针对忽视局部关系中的二阶标记关系问题,本文提出了一种基于全局和局部关系的类属特征多标记分类(global and lo⁃cal relationships based on multi⁃label classification algorithm with label⁃specific features,LFGML)算法。通过全...针对忽视局部关系中的二阶标记关系问题,本文提出了一种基于全局和局部关系的类属特征多标记分类(global and lo⁃cal relationships based on multi⁃label classification algorithm with label⁃specific features,LFGML)算法。通过全局关系的角度来获取类属特征,使用加权平均法计算每个实例的邻域信息,利用杰卡德相似度提取局部关系中的二阶标记关系。LFGML算法在10个多标记数据集Genbase、Medical、Arts、Health、Flags、Cal500、Yeast、Image、Education和Emotions进行了实验。结果表明,所提出的算法相对于其他对比算法在多标记分类中具有明显的的性能优势。展开更多
Finding channel sandbodies is an important task in oil and gas exploration due to the importance of fluvial reservoirs. It is difficult to describe fluvial reservoirs in detail owing to their frequent changes and seri...Finding channel sandbodies is an important task in oil and gas exploration due to the importance of fluvial reservoirs. It is difficult to describe fluvial reservoirs in detail owing to their frequent changes and serious intersections, as well as limitations of S/N ratio and seismic data resolution. Based on the Laohekou 3D data in Shengli Oilfield, we analyze the general characteristics of fluvial reservoirs in this area, from which we find that they are characterized by strong amplitudes on seismic profiles, high continuity on time slices, and low frequency in the frequency domain. In addition, a cluster of strong string-bead- like reflections was found after color processing and detailed interpretation. To understand this observation, we conduct forward modeling to explain the mechanism. This provides a new way to identify ancient channels in similar areas. By using the multi-attribute fusion and RGB display techniques, channel incision is more obvious and the characteristics of the channel structures are manifested much better. Finally, we introduce and apply multi-wavelet detection technology to identify weaker fluvial reservoir signals.展开更多
Classification of plume and spatter images was studied to evaluate the welding stability. A high-speed camera was used to capture the instantaneous images of plume and spatters during high power disk laser welding. Ch...Classification of plume and spatter images was studied to evaluate the welding stability. A high-speed camera was used to capture the instantaneous images of plume and spatters during high power disk laser welding. Characteristic parameters such as the area and number of spatters, the average grayscale of a spatter image, the entropy of a spatter grayscale image, the coordinate ratio of the plume centroid and the welding point, the polar coordinates of the plume centroid were defined and extracted. Karhunen-Loeve transform method was used to change the seven characteristics into three primary characteristics to reduce the dimensions. Also, K-nearest neighbor method was used to classify the plume and spatter images into two categories such as good and poor welding quality. The results show that plume and spatter have a close relationship with the welding stability, and two categories could be recognized effectively using K-nearest neighbor method based on Karhunen-Loeve transform.展开更多
Genus Eurydoxa Filipjev in China is reviewed and noted for the first time. Based on the present study, two species are recognized, including rhodopa Diakonoff and advena Filipjev. All available information for the spe...Genus Eurydoxa Filipjev in China is reviewed and noted for the first time. Based on the present study, two species are recognized, including rhodopa Diakonoff and advena Filipjev. All available information for the species is reviewed and provided.展开更多
文摘提出一种基于标签正负相关性的多标签类属特征学习方法(multi-label learning with label-specific features based on positive and negative label correlation,LIFTPNL)。基于k近邻的思想构建全局和局部的标签信息矩阵,根据此矩阵计算成对标签的正负相关性;对每个类别标签,基于属于相同和不同类簇的样本构建连接矩阵,联合该标签正负相关性计算样本相似度;采用谱聚类方法获得聚类中心,将原有特征转换成类属特征;通过二分类器得到分类结果。实验结果表明,所提算法优于多种多标签分类方法。
文摘多标签流形学习(multi-label manifold learning,ML^(2))基于特征流形构建标签流形,将标签逻辑值转换为实数值,能更好地反映标签相关性,提高分类性能.但是,ML^(2)与多数多标签分类方法一样,是基于数据的全部特征进行标签预测,没有考虑不同特征对不同类别标签的鉴别能力.因此,提出一种基于类属特征的多标签流形学习分类(label specific feature based multi-label manifold learning,LSF-ML^(2))方法.首先,利用标签数据优化类属特征重要度矩阵,确定类属特征子集;再将子集的特征流形映射到标签空间,使标签从离散型变为数值型;最后,通过多输出回归实现分类.实验结果表明,所提方法性能优于多种多标签分类方法.
文摘多标签特征选择能够有效去除冗余特征并提升分类精度,是解决“维数灾难”问题的有效方法.然而,已有的多标签特征选择算法是对所有标签选择出相同的特征,忽略了标签与特征之间的内在联系.事实上,每个标签都具有反映该标签特有属性的特征,即类属特征.提出一种基于图拉普拉斯的多标签类属特征选择(multi-label label-specific feature selection based on graph Laplacian,LSGL)算法.对于每个类别标签,基于拉普拉斯映射获得数据的低维嵌入,再通过稀疏正则化获得数据空间到嵌入空间的投影矩阵,接着通过分析矩阵系数确定每个标签相应的类属特征,最后使用类属特征进行分类.在5个公共多标签数据集上的多标签特征选择与分类实验结果证明了所提算法的有效性.
文摘针对忽视局部关系中的二阶标记关系问题,本文提出了一种基于全局和局部关系的类属特征多标记分类(global and lo⁃cal relationships based on multi⁃label classification algorithm with label⁃specific features,LFGML)算法。通过全局关系的角度来获取类属特征,使用加权平均法计算每个实例的邻域信息,利用杰卡德相似度提取局部关系中的二阶标记关系。LFGML算法在10个多标记数据集Genbase、Medical、Arts、Health、Flags、Cal500、Yeast、Image、Education和Emotions进行了实验。结果表明,所提出的算法相对于其他对比算法在多标记分类中具有明显的的性能优势。
基金sponsored by The Science and Technology Research Project,Shengli Oilfield (Grant No. YKW1002)
文摘Finding channel sandbodies is an important task in oil and gas exploration due to the importance of fluvial reservoirs. It is difficult to describe fluvial reservoirs in detail owing to their frequent changes and serious intersections, as well as limitations of S/N ratio and seismic data resolution. Based on the Laohekou 3D data in Shengli Oilfield, we analyze the general characteristics of fluvial reservoirs in this area, from which we find that they are characterized by strong amplitudes on seismic profiles, high continuity on time slices, and low frequency in the frequency domain. In addition, a cluster of strong string-bead- like reflections was found after color processing and detailed interpretation. To understand this observation, we conduct forward modeling to explain the mechanism. This provides a new way to identify ancient channels in similar areas. By using the multi-attribute fusion and RGB display techniques, channel incision is more obvious and the characteristics of the channel structures are manifested much better. Finally, we introduce and apply multi-wavelet detection technology to identify weaker fluvial reservoir signals.
基金Project (51175095) supported by the National Natural Science Foundation of ChinaProjects (10251009001000001,9151009001000020) supported by the Natural Science Foundation of Guangdong Province,ChinaProject (20104420110001) supported by the Specialized Research Fund for the Doctoral Program of Higher Education of China
文摘Classification of plume and spatter images was studied to evaluate the welding stability. A high-speed camera was used to capture the instantaneous images of plume and spatters during high power disk laser welding. Characteristic parameters such as the area and number of spatters, the average grayscale of a spatter image, the entropy of a spatter grayscale image, the coordinate ratio of the plume centroid and the welding point, the polar coordinates of the plume centroid were defined and extracted. Karhunen-Loeve transform method was used to change the seven characteristics into three primary characteristics to reduce the dimensions. Also, K-nearest neighbor method was used to classify the plume and spatter images into two categories such as good and poor welding quality. The results show that plume and spatter have a close relationship with the welding stability, and two categories could be recognized effectively using K-nearest neighbor method based on Karhunen-Loeve transform.
基金This study was carried out under the financial support by KOSEF (Korea Science & Engineering Foundation) with the program of 'Korea and China Young Scientist Exchange Program' (2002 2003).
文摘Genus Eurydoxa Filipjev in China is reviewed and noted for the first time. Based on the present study, two species are recognized, including rhodopa Diakonoff and advena Filipjev. All available information for the species is reviewed and provided.
