Summary
Clustering means putting items that are similar into groups without knowing their categories beforehand. Many methods need people to pick tricky settings like the number of groups or how close points should be to count as neighbors. The authors propose a new way called A-DLCC that figures out the groups automatically using a concept called local depth, which helps find points that best represent their neighborhoods. This method merges groups based on a careful measure from graph theory, so it stops merging at the right time and decides the number of clusters on its own. They tested A-DLCC on various data and found it gives clear groups without needing any number choices from users.
What this means in practice
- •For data science teams: Automatically identify natural groupings in complex datasets without setting parameters like cluster count, simplifying data analysis workflows.
- •For computer vision engineers: Cluster depth-based feature points adaptively to improve object grouping or scene segmentation tasks without manual parameter tuning.
Abstract
Clustering is an unsupervised learning technique that partitions unlabeled data into groups. Most existing methods require user-specified parameters, such as the number of clusters or neighborhood size. Conversely, we propose automatic depth-based local center clustering (A-DLCC), a fully data-driven method that eliminates numerical parameter tuning. A-DLCC uses the $β$-integrated local depth to identify stable exemplars, points consistently central across multiple locality levels, termed local centers, which are ranked by their representativeness. Each local center induces a group of similar points, with group-level similarity measured by a proposed nonparametric metric called group-level local similarity. To guide merging, we incorporate the bottleneck path idea from graph theory, which forms the basis of our adaptive merging criterion. Based on this criterion, we design a single agglomeration rule in which a group is either absorbed by a neighbor it reaches better than itself or bonded to a neighbor that both sides find more reachable than their own background, every merge being additionally required to be carried by a contact stronger than a configuration-model null expects. The rule automatically estimates the number of clusters and decides when to stop merging. Experiments on synthetic and real data show that A-DLCC produces interpretable clustering results without parameter tuning.