Scalable Uncertainty Quantification for Black-Box Density-Based Clustering
This addresses the problem of reliable uncertainty estimation in clustering for practitioners dealing with complex data, though it appears incremental as it builds on existing paradigms and methods.
The paper tackles uncertainty quantification in density-based clustering by combining the martingale posterior paradigm with density-based clustering to propagate uncertainty from density estimates to clustering structure, achieving scalability to high-dimensional and irregularly shaped data through neural density estimators and GPU-friendly parallel computation.
We introduce a novel framework for uncertainty quantification in clustering. By combining the martingale posterior paradigm with density-based clustering, uncertainty in the estimated density is naturally propagated to the clustering structure. The approach scales effectively to high-dimensional and irregularly shaped data by leveraging modern neural density estimators and GPU-friendly parallel computation. We establish frequentist consistency guarantees and validate the methodology on synthetic and real data.