arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2302.12432cs.LGcs.AI
收藏

具有可学习与最优多项式基的图神经网络

Graph Neural Networks with Learnable and Optimal Polynomial Bases

Renmin University of China(中国人民大学);Peng Cheng Laboratory(鹏城实验室)

Yuhe Guo, Zhewei Wei

AI总结:

针对多项式滤波器图神经网络中多项式基的选择问题,提出FavardGNN和OptBasisGNN两个谱GNN模型,分别从数据中学习正交多项式基并为给定图计算最优基,实验证明其有效性。

摘要:

多项式滤波器是一类图神经网络,通常使用预定的多项式基并从训练数据中学习系数。已有观察表明,模型的有效性高度依赖于多项式基的性质。由此引出两个自然且基本的问题:我们能否从训练数据中学习合适的基?能否为给定图和节点特征确定最优基?本文提出两个谱GNN模型给出了肯定回答。首先,受Favard定理启发,提出FavardGNN模型,从所有可能的正交基空间中学习多项式基。其次,研究Wang & Zhang (2022)中看似不可解的最优多项式基定义,提出OptBasisGNN模型,为给定图结构和图信号计算最优基。大量实验证明了所提模型的有效性。代码见https://github.com/yuziGuo/FarOptBasis。

英文摘要:

Polynomial filters, a kind of Graph Neural Networks, typically use a predetermined polynomial basis and learn the coefficients from the training data. It has been observed that the effectiveness of the model is highly dependent on the property of the polynomial basis. Consequently, two natural and fundamental questions arise: Can we learn a suitable polynomial basis from the training data? Can we determine the optimal polynomial basis for a given graph and node features? In this paper, we propose two spectral GNN models that provide positive answers to the questions posed above. First, inspired by Favard's Theorem, we propose the FavardGNN model, which learns a polynomial basis from the space of all possible orthonormal bases. Second, we examine the supposedly unsolvable definition of optimal polynomial basis from Wang & Zhang (2022) and propose a simple model, OptBasisGNN, which computes the optimal basis for a given graph structure and graph signal. Extensive experiments are conducted to demonstrate the effectiveness of our proposed models. Our code is available at https://github.com/yuziGuo/FarOptBasis.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑