ZEUS: Zero-shot Embeddings for Unsupervised Separation of Tabular Data

2025-05-15

AI Summary

This paper addresses the challenges in unsupervised clustering of tabular data—namely, inconsistent cross-dataset similarity distributions and the absence of supervision—which render deep clustering methods highly sensitive to hyperparameters and unstable in performance. To this end, we propose the first zero-shot unsupervised embedding framework specifically designed for tabular data. Our method leverages latent-variable priors to generate synthetic data for pretraining and integrates self-supervised representation decomposition. Crucially, it requires neither ground-truth labels nor target-domain fine-tuning, directly producing high-quality embeddings and clustering assignments. Extensive experiments demonstrate that our approach matches or surpasses state-of-the-art traditional and deep clustering methods in clustering accuracy, while achieving significantly faster inference. Moreover, it is entirely hyperparameter-free and plug-and-play, exhibiting strong generalization across diverse tabular datasets and practical applicability in real-world scenarios.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Computer Vision: Representation Learning for Vision

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Clustering tabular data remains a significant open challenge in data analysis and machine learning. Unlike for image data, similarity between tabular records often varies across datasets, making the definition of clusters highly dataset-dependent. Furthermore, the absence of supervised signals complicates hyperparameter tuning in deep learning clustering methods, frequently resulting in unstable performance. To address these issues and reduce the need for per-dataset tuning, we adopt an emerging approach in deep learning: zero-shot learning. We propose ZEUS, a self-contained model capable of clustering new datasets without any additional training or fine-tuning. It operates by decomposing complex datasets into meaningful components that can then be clustered effectively. Thanks to pre-training on synthetic datasets generated from a latent-variable prior, it generalizes across various datasets without requiring user intervention. To the best of our knowledge, ZEUS is the first zero-shot method capable of generating embeddings for tabular data in a fully unsupervised manner. Experimental results demonstrate that it performs on par with or better than traditional clustering algorithms and recent deep learning-based methods, while being significantly faster and more user-friendly.

Problem

• Clustering tabular data lacks dataset-independent similarity metrics
• Unsupervised deep learning clustering suffers from unstable hyperparameter tuning
• Zero-shot embeddings for tabular data require no per-dataset training

Innovation

• Zero-shot learning for tabular data clustering
• Pre-training on synthetic latent-variable datasets
• Self-contained model without fine-tuning