求翻译：对复杂流形结构的数据集进行聚类是模式识别中常见的任务，也是一项具有挑战性的工作，主要难点在于对多流形分布的样本点难以度量相似度和确定聚类中心个数。为了解决这一难题，首先提出一种基于流形距离的度量方法判断样本点间的近邻关系；其次，综合考虑数据集全局和局部的空间流形分布，定义各样本点的局部密度；再次根据各样本点局部密度大小及其与其他样本点局部密度的关系，定义聚类中心判别准则；最后基于分级的示例判决策略实现对聚类中心数的自动确定和聚类中心的自动选择。实验结果表明，对于存在密度多变、多尺度和噪声重叠特性的数据集，相对于已有算法能够有效提高聚类精度。是什么意思？

Manifold structure of complex data sets clustering is common in pattern recognition tasks , but also a challenging task , the main difficulty lies in the distribution of sample points manifolds difficult to measure the number of similarity and determine the cluster centers

The complex flow-shaped structure of the data set to be clustered is pattern recognition in the common task, and it is also a challenging work, the main difficulty lies in the multi-stream-shaped distribution of the sample is difficult to measure similarity and determine together such a center. In o

Carries on the cluster to the complex manifold structure data set is in the pattern recognition the common duty, also is one item has the challenging work, the main difficulty lies in the sample point which distributes to the multi-manifolds to measure the similarity and the determination cluster ce

For complex data set cluster structure the manifold is the common task of pattern recognition, is a challenging task, the main difficulty of the manifold distribution similar to the sample point is difficult to measure and determine the number of cluster centers. To solution this a problem, first pr

正在翻译，请等待...