Density-ratio estimation
Since many existing distance-based clustering algorithms have a bias towards dense clusters, the proposed methods are used to “correct” this bias, based on the concept of “density-ratio”.
- modify a density-based clustering algorithm to do density-ratio based clustering by using its density estimator to compute density-ratio. The modified DBSCAN for density-ratio based clustering can be obtained from here.
- rescaling the given dataset to make equalise different clusters' densities, then a standard density-estimator can directly perform density-ratio estimation on the transformed dataset. The feature individual scaling method ReScale can be obtained from here, a multi-dimensional distance scaling method DScale can be obtained from here, and the latest data transformation method CDF-TS can be obtained from here.
Dr Ye Zhu
Senior Lecturer of Computer Science, IEEE Senior Member
My research works focus on the fields of clustering and anomaly detection.
Publications
CDF Transform-and-Shift: An effective way to deal with datasets of inhomogeneous cluster densities
CDF-TS is a preprocessing method to transform data and equalises the density of all clusters.
A Distance Scaling Method to Improve Density-Based Clustering
We propose a multi-dimensional scaling method, named DScale, which rescales based on the computed distance.
