Gangda Deng, Hongkuan Zhou, Rajgopal Kannan, Viktor Prasanna
{"title":"嗜异性条件下的图神经网络个性化范围学习","authors":"Gangda Deng, Hongkuan Zhou, Rajgopal Kannan, Viktor Prasanna","doi":"arxiv-2409.06998","DOIUrl":null,"url":null,"abstract":"Heterophilous graphs, where dissimilar nodes tend to connect, pose a\nchallenge for graph neural networks (GNNs) as their superior performance\ntypically comes from aggregating homophilous information. Increasing the GNN\ndepth can expand the scope (i.e., receptive field), potentially finding\nhomophily from the higher-order neighborhoods. However, uniformly expanding the\nscope results in subpar performance since real-world graphs often exhibit\nhomophily disparity between nodes. An ideal way is personalized scopes,\nallowing nodes to have varying scope sizes. Existing methods typically add\nnode-adaptive weights for each hop. Although expressive, they inevitably suffer\nfrom severe overfitting. To address this issue, we formalize personalized\nscoping as a separate scope classification problem that overcomes GNN\noverfitting in node classification. Specifically, we predict the optimal GNN\ndepth for each node. Our theoretical and empirical analysis suggests that\naccurately predicting the depth can significantly enhance generalization. We\nfurther propose Adaptive Scope (AS), a lightweight MLP-based approach that only\nparticipates in GNN inference. AS encodes structural patterns and predicts the\ndepth to select the best model for each node's prediction. Experimental results\nshow that AS is highly flexible with various GNN architectures across a wide\nrange of datasets while significantly improving accuracy.","PeriodicalId":501032,"journal":{"name":"arXiv - CS - Social and Information Networks","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Learning Personalized Scoping for Graph Neural Networks under Heterophily\",\"authors\":\"Gangda Deng, Hongkuan Zhou, Rajgopal Kannan, Viktor Prasanna\",\"doi\":\"arxiv-2409.06998\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Heterophilous graphs, where dissimilar nodes tend to connect, pose a\\nchallenge for graph neural networks (GNNs) as their superior performance\\ntypically comes from aggregating homophilous information. Increasing the GNN\\ndepth can expand the scope (i.e., receptive field), potentially finding\\nhomophily from the higher-order neighborhoods. However, uniformly expanding the\\nscope results in subpar performance since real-world graphs often exhibit\\nhomophily disparity between nodes. An ideal way is personalized scopes,\\nallowing nodes to have varying scope sizes. Existing methods typically add\\nnode-adaptive weights for each hop. Although expressive, they inevitably suffer\\nfrom severe overfitting. To address this issue, we formalize personalized\\nscoping as a separate scope classification problem that overcomes GNN\\noverfitting in node classification. Specifically, we predict the optimal GNN\\ndepth for each node. Our theoretical and empirical analysis suggests that\\naccurately predicting the depth can significantly enhance generalization. We\\nfurther propose Adaptive Scope (AS), a lightweight MLP-based approach that only\\nparticipates in GNN inference. AS encodes structural patterns and predicts the\\ndepth to select the best model for each node's prediction. Experimental results\\nshow that AS is highly flexible with various GNN architectures across a wide\\nrange of datasets while significantly improving accuracy.\",\"PeriodicalId\":501032,\"journal\":{\"name\":\"arXiv - CS - Social and Information Networks\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-09-11\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - CS - Social and Information Networks\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2409.06998\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Social and Information Networks","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.06998","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Learning Personalized Scoping for Graph Neural Networks under Heterophily
Heterophilous graphs, where dissimilar nodes tend to connect, pose a
challenge for graph neural networks (GNNs) as their superior performance
typically comes from aggregating homophilous information. Increasing the GNN
depth can expand the scope (i.e., receptive field), potentially finding
homophily from the higher-order neighborhoods. However, uniformly expanding the
scope results in subpar performance since real-world graphs often exhibit
homophily disparity between nodes. An ideal way is personalized scopes,
allowing nodes to have varying scope sizes. Existing methods typically add
node-adaptive weights for each hop. Although expressive, they inevitably suffer
from severe overfitting. To address this issue, we formalize personalized
scoping as a separate scope classification problem that overcomes GNN
overfitting in node classification. Specifically, we predict the optimal GNN
depth for each node. Our theoretical and empirical analysis suggests that
accurately predicting the depth can significantly enhance generalization. We
further propose Adaptive Scope (AS), a lightweight MLP-based approach that only
participates in GNN inference. AS encodes structural patterns and predicts the
depth to select the best model for each node's prediction. Experimental results
show that AS is highly flexible with various GNN architectures across a wide
range of datasets while significantly improving accuracy.