A recent breakthrough by Tsinghua researchers introduces AutoGFM, an adaptive graph neural architecture search framework that dramatically improves the performance and generalization of graph foundation models across diverse tasks and domains, as validated by extensive ICML‑2025 experiments.
This article introduces PaSca, a scalable graph neural architecture search framework based on the SGAP paradigm, detailing its problem motivation, methodological innovations, extensive experiments on scalability and prediction performance, and summarizing its open‑source release and impact on large‑scale graph learning.
The paper presents PaSca, a scalable graph neural architecture search system that introduces a new SGAP modeling paradigm, a 150,000‑structure design space, and an automated multi‑objective search engine, achieving high scalability and strong predictive performance on real‑world large‑scale graph tasks.
This paper proposes an intra‑ensemble strategy that trains multiple sub‑networks within a single neural network using random training operations, width‑depth variations, and parameter sharing, achieving diverse models and improved performance comparable to traditional ensembles while adding only marginal parameter overhead.
