Reproduce “TabClustPFN: A Prior-data Fitted Network for Tabular Data Clustering” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Tianqi Zhao, Guanyang Wang, Yan Shuo Tan, Qiong Zhang
TabClustPFN jointly clusters records and estimates cluster count for unlabeled tables. It is pretrained on synthetic tasks and uses iterative prototype–data refinement for assignments, with a separate network selecting count. On a curated benchmark of 44 real-world datasets, its automatic-count variant has median ARI rank 8.5, versus 10.5 for silhouette-selected KMeans. The benchmark is limited to moderate-sized datasets.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.