A Novel Algorithm Based on Avoid Determining Noise Threshold in DENCLUE

This paper focuses on density-based clustering analysis. The determination of noise threshold set in DENCLUE is studied via analyzing several typical density-based clustering methods. An improved algorithm which does not use the noise threshold in DENCLUE is proposed based on the estimation of points in inner cluster. Compared to the original DENCLUE, smaller silhouette coefficients can be obtained from the proposed algorithm via experimental verification. Meanwhile, the noise in data sets can also be verified well in our method, which can be viewed as an improvement for applicability and performance of DENCLUE.