论文部分内容阅读
半监督聚类方法利用少量标记数据提高聚类算法的性能,已逐渐发展成为模式识别及相关领域的研究热点.文中首先综述了半监督聚类算法的一些新进展,包括基于约束的方法、基于距离的方法和基于距离与约束的融合方法.然后提出一种基于约束的半监督模糊C-means聚类算法.实验表明,该算法与传统的模糊C-means及半监督K-means方法相比,具有更好的聚类精度.
Semi-supervised clustering method has been developed into a hot spot in the field of pattern recognition and related fields by using a small amount of markup data to improve the performance of clustering algorithms.Firstly, some new developments of semi-supervised clustering algorithms are summarized, including the constraint-based method, Distance and the method based on the distance and constraint fusion.And then proposed a constraint-based semi-supervised fuzzy C-means clustering algorithm.Experiments show that this algorithm compared with the traditional fuzzy C-means and semi-supervised K-means method , With better clustering accuracy.