Data ultrametricity and clusterability
This addresses the challenge of clustering massive datasets with high computational costs for users, though it appears incremental as it builds on ultrametric concepts.
The paper tackles the problem of determining if a dataset is clusterable for efficient partitioning into well-differentiated groups, proposing a novel ultrametric-based approach that evaluates clusterability and generates the sub-dominant ultrametric of the dissimilarity.
The increasing needs of clustering massive datasets and the high cost of running clustering algorithms poses difficult problems for users. In this context it is important to determine if a data set is clusterable, that is, it may be partitioned efficiently into well-differentiated groups containing similar objects. We approach data clusterability from an ultrametric-based perspective. A novel approach to determine the ultrametricity of a dataset is proposed via a special type of matrix product, which allows us to evaluate the clusterability of the dataset. Furthermore, we show that by applying our technique to a dissimilarity space will generate the sub-dominant ultrametric of the dissimilarity.