In this work, we address the test-time adaptation challenge in graph neural networks (GNNs), focusing on overcoming the limitations in flexibility and generalization inherent in existing data-centric approaches. To this end, we propose a novel research problem, test-time graph neural dataset search, which seeks to learn a parameterized test-time graph distribution to enhance the inference performance of unseen test graphs on well-trained GNNs. Specifically, we propose a generative Projection based test-time Graph Neural Dataset Search method, named PGNDS, which maps the unseen test graph distribution back to the known training distribution through a generation process guided by well-trained GNNs. The proposed PGNDS framework consists of three key modules: (1) dual conditional diffusion for GNN-guided generative projection through test-back-to-training distribution mapping; (2) dynamic search from the generative sampling space to select the most expressive test graphs; (3) ensemble inference to aggregate information from original and adapted test graphs. Extensive experiments on real-world graphs demonstrate the superior ability of our proposed PGNDS for improved test-time GNN inference.

Graph neural networks (GNNs) often struggle to perform well when applied to new, unseen graphs due to distribution shifts in the data. Our paper introduces PGNDS, a new method that adapts GNNs at test time without modifying the model. Rather than fine-tuning the model, PGNDS automatically transforms new graphs to resemble the training data, enabling better performance on shifted distributions. This improves prediction accuracy across tasks and provides a practical, data-centric solution for deploying GNNs in dynamic, real-world scenarios.