Graph Attention
Graph attention mechanisms enhance graph neural networks by selectively weighting the importance of connections between nodes, improving information propagation and representation learning. Current research focuses on applying graph attention networks (GATs) to diverse problems, including predicting gene-disease links, analyzing brain activity, and optimizing vehicle routing, often incorporating advanced architectures like transformers and recurrent networks to handle temporal data. This approach leads to improved accuracy and interpretability in various domains, offering significant advancements in fields ranging from drug discovery to cognitive science and autonomous systems.
Papers
Less is More: Hop-Wise Graph Attention for Scalable and Generalizable Learning on Circuits
Chenhui Deng, Zichao Yue, Cunxi Yu, Gokce Sarar, Ryan Carey, Rajeev Jain, Zhiru Zhang
Dual Graph Attention based Disentanglement Multiple Instance Learning for Brain Age Estimation
Fanzhe Yan, Gang Yang, Yu Li, Aiping Liu, Xun Chen
GATGPT: A Pre-trained Large Language Model with Graph Attention Network for Spatiotemporal Imputation
Yakun Chen, Xianzhi Wang, Guandong Xu
Out-of-Distribution Generalized Dynamic Graph Neural Network with Disentangled Intervention and Invariance Promotion
Zeyang Zhang, Xin Wang, Ziwei Zhang, Haoyang Li, Wenwu Zhu