Unverified50% confidenceFactExact time
In the 2019 paper "How Powerful are Graph Neural Networks?", Xu et al. analyzed the theoretical upper bounds of different aggregation methods by comparing them with the Weisfeiler-Leman graph isomorphism test.
1
Sources
50%
Confidence
Long-term
Relevance
9/12/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedScarselli et al. published the original paper "The Graph Neural Network Model" in 2009.71% similarUnverifiedGraph neural network architectures such as GraphSAGE and GAT have demonstrated capabilities in scenarios including social network analysis and molecular property prediction.70% similarUnverified对抗样本概念由Szegedy等人于2013年在论文《Intriguing properties of neural networks》中正式提出70% similarUnverifiedNeural ODE was proposed by Chen Tianqi et al. in a NeurIPS 2018 Best Paper, with the core idea of treating the discrete layers of a residual network as the Euler discretization of a continuous dynamical system.69% similarUnverifiedThe phenomenon that optimization of over-parameterized neural networks easily converges to good solutions can be explained by Neural Tangent Kernel theory67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/914032API
curl https://kongchang.com/api/v1/knowledge/claims/914032MCP
get_claim(id=914032)