The Optimization Arithmetic of K-means Clustering Based on Indirect Feature Weight Learning

摘要：

The performance of K-means clustering algorithm depended on the selection of distance metrics, there was a problem with Dimension Trap. Using the feature learning parameter can solve this problem, but the choice of feature learning was difficult, so the improper choice of feature learning would affect the convergence speed of clustering algorithm, even leading to non-convergence. In regard to the choice of feature learning, a new clustering method is discussed. The method of feature learning Indirect Feature Weight Learning automatically is adopted to protect more rapid convergence and improve the clustering performance. The result in testing data in typical UCI machine learning repository indicate that these measures have improved clustering performance.

关键词： learning rate similarity metrics Indirect Feature Weight Learning gradient-descent technique

作者: Bin Zeng Chao Luo Wei Zhao Benyue Chen

作者单位: School of Information Engineering Zhejiang Forestry University Linan, Zhejiang, China Tianmu College Zhejiang Forestry University Linan, Zhejiang, China Center of Modern Educational Technology Zhejiang Forestry University Linan ,Zhejiang,China School of Environmental Science Zhejiang Forestry University Linan, Zhejiang, China

会议类型: 国际会议

会议名称: 2010 International Conference on Computer and Communication Technologies in Agriculture Engineering(计算机与通信技术在农业工程国际会议 CCTAE 2010)

会议地点: 成都

会议语种:英文

页码: 243-246

在线出版日期: 2010-06-12（万方平台首次上网日期，不代表论文的发表时间）