一篇新论文探讨了不同循环图神经网络（RGNN）模型的表达能力，特别关注收敛型、输出收敛型和停止型RGNN。研究表明，在无向图上，收敛型RGNN与分级双模拟不变停止型RGNN的表达能力相当，而输出收敛型RGNN的表达能力至少与之相当。该研究引入了一种“交通灯”协议，以解决在用收敛型RGNN模拟停止型RGNN时出现的去同步化挑战，从而回答了该领域的一个开放性问题。 AI
影响 阐明了RGNN变体的理论表达能力限制，可能指导未来基于图的AI研究。
排序理由 分析RGNN模型理论性质的学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2604.25551v1 Announce Type: new Abstract: Recurrent Graph Neural Networks (RGNNs) extend standard GNNs by iterating message-passing until some stopping condition is met. Various RGNN models have been proposed in the literature. In this paper, we study three such models: con…
Recurrent Graph Neural Networks (RGNNs) extend standard GNNs by iterating message-passing until some stopping condition is met. Various RGNN models have been proposed in the literature. In this paper, we study three such models: converging RGNNs, where all vertex representations …