局部线性嵌入算法(locally linear embedding,LLE)是一种流形降维方法,在高维稀疏数据空间中,针对LLE不适合稀疏采样和欧氏距离公式的缺陷,研究该算法的扩展,引入核函数,并将样本映射到高维特征空间,核映射改善了样本的空间分布,改进的...局部线性嵌入算法(locally linear embedding,LLE)是一种流形降维方法,在高维稀疏数据空间中,针对LLE不适合稀疏采样和欧氏距离公式的缺陷,研究该算法的扩展,引入核函数,并将样本映射到高维特征空间,核映射改善了样本的空间分布,改进的LLE方法在适当选取近邻点个数情况下,可得到良好的效果。对从高维采样数据中恢复得到低维数据集,通过本文提出的离群数据假设,并结合本文给出的离群聚类方法对所得低维数据是否是离群数据进行判别。仿真文验的结果表明了该方法能够有效地发现高维数据集中的离群点,与此同时,该算法具有参数估计简单、参数影响不大等优点,该算法为离群点检测问题的机器学习提供了一条新的途径。展开更多
Absolute geostrophic currents in the North Pacific Ocean were calculated using P-vector method from newly gridded Argo profiling float data collected during 2004-2009. The meridional volume transport of geostrophic cu...Absolute geostrophic currents in the North Pacific Ocean were calculated using P-vector method from newly gridded Argo profiling float data collected during 2004-2009. The meridional volume transport of geostrophic currents differed significantly from the classical Sverdrup balance, with differences of 10×106 -20×106 m3 /s in the interior tropical Northwest Pacific Ocean. Analyses showed that errors of wind stress estimation could not explain all of the differences. The largest differences were found in the areas immediately north and south of the bifurcation latitude of the North Equatorial Current west of the dateline, and in the recirculation area of the Kuroshio and its extension, where nonlinear eddy activities were robust. Comparison of the geostrophic meridional transport and the wind-driven Sverdrup meridional transport in a high-resolution OFES simulation showed that nonlinear effects of the ocean circulation were the most likely reason for the differences. It is therefore suggested that the linear, steady wind-driven dynamics of the Sverdrup theory cannot completely explain the meridional transport of the interior circulation of the tropical Northwest Pacific Ocean.展开更多
文摘局部线性嵌入算法(locally linear embedding,LLE)是一种流形降维方法,在高维稀疏数据空间中,针对LLE不适合稀疏采样和欧氏距离公式的缺陷,研究该算法的扩展,引入核函数,并将样本映射到高维特征空间,核映射改善了样本的空间分布,改进的LLE方法在适当选取近邻点个数情况下,可得到良好的效果。对从高维采样数据中恢复得到低维数据集,通过本文提出的离群数据假设,并结合本文给出的离群聚类方法对所得低维数据是否是离群数据进行判别。仿真文验的结果表明了该方法能够有效地发现高维数据集中的离群点,与此同时,该算法具有参数估计简单、参数影响不大等优点,该算法为离群点检测问题的机器学习提供了一条新的途径。
基金Supported by the National Basic Research Program of China(973 Program)(No.2012CB956000)the National Natural Science Foundation of China(Nos.40888001,41176019)supported by KLOCAW1208
文摘Absolute geostrophic currents in the North Pacific Ocean were calculated using P-vector method from newly gridded Argo profiling float data collected during 2004-2009. The meridional volume transport of geostrophic currents differed significantly from the classical Sverdrup balance, with differences of 10×106 -20×106 m3 /s in the interior tropical Northwest Pacific Ocean. Analyses showed that errors of wind stress estimation could not explain all of the differences. The largest differences were found in the areas immediately north and south of the bifurcation latitude of the North Equatorial Current west of the dateline, and in the recirculation area of the Kuroshio and its extension, where nonlinear eddy activities were robust. Comparison of the geostrophic meridional transport and the wind-driven Sverdrup meridional transport in a high-resolution OFES simulation showed that nonlinear effects of the ocean circulation were the most likely reason for the differences. It is therefore suggested that the linear, steady wind-driven dynamics of the Sverdrup theory cannot completely explain the meridional transport of the interior circulation of the tropical Northwest Pacific Ocean.