TabClustPFN: A Prior-data Fitted Network for Tabular Data Clustering

TabClustPFN introduces a clustering Prior-data Fitted Network (PFN) that jointly infers cluster assignments and cardinality in one pass. It solves challenges like unknown cluster numbers and label switching by using a hybrid prior capturing real-world data geometry. The method achieves state-of-the-art performance on real benchmarks while maintaining fast inference.

Open the full episode

More from AI Radio