Towards Parameter Robust Hierarchical Clustering

Density-based clustering is a statistical learning technique that aims at finding high density regions in the data separated by low density regions, finding applications in virtually all fields of knowledge. Hierarchical density-based clustering goes one step beyond and finds a hierarchy of density-based clusters at different density levels according to a user-defined density smoothing parameter. In this project, we intend to investigate how a set of density-based clustering hierarchies with regard to different values of the density smoothing parameter can be used to build a single and easy-to-analyze hierarchical organization of a dataset. Our main motivation comes from the observation that, in some cases, using one single parameter value is not enough to find all the cluster structures in the data. TO BE CONT’D

Faculty Supervisor:

Joerg Sander

Student:

Partner:

University of Newcastle

Discipline:

Computer science

Sector:

Education

University:

University of Alberta

Program: