Can TabPFN Compete with GNNs for Node Classification via Graph Tabularization?

2025-12-09

AI Summary

Graph node classification typically relies on graph neural networks (GNNs) or large language models, which require architecture-specific design and extensive training. Method: This paper proposes TabPFN-GN, the first framework to systematically reformulate node classification as a tabular learning task—leveraging the zero-shot generalization capability of the pre-trained tabular foundation model TabPFN. It constructs transferable graph-to-tabular representations by jointly encoding node attributes, structural topology, positional embeddings, and neighborhood-smoothed features, enabling end-to-end zero-shot inference without graph-specific architectures. Contribution/Results: TabPFN-GN establishes a novel paradigm for adapting general-purpose tabular models to graph data, eliminating dependence on GNNs or LLMs. Experiments across 12 benchmark datasets demonstrate that it matches state-of-the-art GNNs on homophilic graphs and consistently outperforms them on heterophilic graphs, validating the effectiveness and robustness of cross-domain zero-shot transfer.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Foundation models pretrained on large data have demonstrated remarkable zero-shot generalization capabilities across domains. Building on the success of TabPFN for tabular data and its recent extension to time series, we investigate whether graph node classification can be effectively reformulated as a tabular learning problem. We introduce TabPFN-GN, which transforms graph data into tabular features by extracting node attributes, structural properties, positional encodings, and optionally smoothed neighborhood features. This enables TabPFN to perform direct node classification without any graph-specific training or language model dependencies. Our experiments on 12 benchmark datasets reveal that TabPFN-GN achieves competitive performance with GNNs on homophilous graphs and consistently outperforms them on heterophilous graphs. These results demonstrate that principled feature engineering can bridge the gap between tabular and graph domains, providing a practical alternative to task-specific GNN training and LLM-dependent graph foundation models.

Problem

• Reformulates graph node classification as tabular learning problem
• Introduces TabPFN-GN to transform graph data into tabular features
• Evaluates performance against GNNs on homophilous and heterophilous graphs

Innovation

• Transforms graph data into tabular features
• Enables direct node classification without graph-specific training
• Uses feature engineering to bridge tabular and graph domains