Abstract: An improved DBSCAN algorithm is presented, which is insensitive to input parameter by discovering connected clusters. The new algorithm produces better clustering results, while maintaining the high performance of the origin algorithm at the same time. The results of experiments demonstrate that the new algorithm outperforms OPTICS.

Key words: clustering, DBSCAN, sensitive to input parameter, data mining

摘要： 提出了一种新的DBSCAN改进算法，通过记录簇连接信息，能够有效地屏蔽输入参数敏感性，提高聚类结果的质量，同时保持了DBSCAN算法的高执行效率。测试结果表明新算法的性能较高。

关键词: 聚类, DBSCAN, 参数敏感, 数据挖掘