[arXiv]score: 0.49
VISION Reframes Graph Few-Shot Node Classification as Fine-Tuning-Free Sequence Reasoning
May 26, 2026
VISION applies in-context learning to graph few-shot node classification via a context-aware network with role embeddings and dual-context fusion, eliminating task-specific fine-tuning at inference while leveraging unlabeled nodes.
cs.AI
HOW THIS AFFECTS YOU
●
researcherThe fine-tuning-free sequence reasoning formulation for graph few-shot learning is a direct adaptation of LLM in-context learning to GNNs, worth benchmarking against meta-learning baselines.