[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117558-en":3,"doc-seo-117558-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},117558,2336464648746,"Skyler","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","Genetic-Evolutionary Graph Neural Networks - A Paradigm for Improved Graph Representation Learning","Message-passing graph neural networks are widely used for learning from graph-structured data, yet iterative message passing often leads to over-smoothing, where node embeddings become increasingly similar and lose representational power. This work attributes the root cause to insufficient diversity among generated embeddings and proposes improving diversity during the embedding-generation process. A genetic-evolutionary graph neural network models each layer as an evolutionary process using crossover and mutation to avoid embedding similarity. Experiments on six benchmark datasets show substantial gains and new state-of-the-art results for graph representation learning.","Genetic-Evolutionary Graph Neural Networks: A Paradigm for Improved Graph Representation Learning  \nHaimin Zhang [haimin.zhang@uts. edu. au](haimin.zhang@uts. edu. au)  \nSchool of Electrical and Data Engineering University of Technology Sydney  \nMin Xu [min.xu@uts. edu. au](min.xu@uts. edu. au)  \nSchool of Electrical and Data Engineering University of Technology Sydney  \nReviewed on OpenReview: [https: // openreview. net/ forum? id= qzYTklXVAB](https: // openreview. net/ forum? id= qzYTklXVAB)  \nAbstract  \nMessage-passing graph neural networks have become the dominant framework for learning over graphs. However, empirical studies continually show that message-passing graph neural networks tend to generate over-smoothed representations for nodes after iteratively applying message passing. This over-smoothing problem is a core issue that limits the representational capacity of message-passing graph neural networks. We argue that the fundamental problem with over-smoothing is a lack of diversity in the generated embeddings, and the problem could be reduced by enhancing the embedding diversity in the embedding generation process. To this end, we propose genetic-evolutionary graph neural networks, a new paradigm for graph representation learning inspired by genetic algorithms. We view each layer of a graph neural network as an evolutionary process and develop operations based on crossover and mutation to prevent embeddings from becoming similar to one another, thus enabling the model to generate improved graph representations. The proposed framework is well-motivated, as it directly draws inspiration from genetic algorithms for preserving population diversity. We experimentally validate the proposed framework on six benchmark datasets on different tasks. The results show that our method significantly advances the performance of current graph neural networks, resulting in new state-of-the-art results for graph representation learning on these datasets.  \n1 Introduction  \nGraphs are a general data structure for representing and analyzing complex relationships among entities. Many real-world systems, such as social networks, molecular structures, communication networks, can be modeled using graphs. It is essential to develop intelligent models for uncovering the underlying patterns and interactions within these graph-structured systems. Recent years have seen an enormous body of studies on learning over graphs. The studies include graph foundation models, geometry processing and deep graph embedding. These advances have produced new state-of-the-art or human-level results in various domains, including recommender systems (Zhang et al., 2024), chemical synthesis (Xie et al., 2024), and 2D and 3D vision tasks (Chen et al., 2024; Kim et al., 2023) .  \nGraph neural networks have emerged as a dominant framework for learning from graph-structured data. The development of graph neural network models can be motivated from different approaches. The fundamental graph neural networks have been derived as a generalization of convolutions to non-Euclidean data (Bruna et al., 2014), as well as by analogy to classic graph isomorphism tests (Hamilton et al., 2017) . Regardless of the motivations, the defining feature of the graph neural network framework is that it utilizes a form of message passing wherein messages are exchanged between nodes and updated using neural networks  \n(Hamilton, 2020) . During each graph neural network layer, the model aggregates features from a node’s local neighbourhood and then updates the node’s representation according to the aggregated information.  \nMessage passing is at the heart of current graph neural networks. However, this paradigm of message passing also has major limitations. Theoretically, it is connected to the Weisfeiler-Lehman (WL) isomorphism test as well as to simple graph convolutions. The representational capacity of message-passing graph neural networks is inherently bounded by the WL isomorphism tes","cbCaidNFGuL2HawK","https://ap.wps.com/l/cbCaidNFGuL2HawK","pdf",741030,1,20,"English","en",105,"# Introduction\n## Graphs and graph neural networks\n## Message passing and its limitations\n## Over-smoothing and embedding diversity\n# Genetic-evolutionary graph neural networks","[{\"question\":\"What problem do message-passing graph neural networks face?\",\"answer\":\"They often suffer from over-smoothing, where node representations become overly similar after many message-passing iterations, reducing the model’s ability to learn meaningful differences.\"},{\"question\":\"How does the proposed method address over-smoothing?\",\"answer\":\"It argues the issue stems from limited diversity in generated embeddings and enhances embedding diversity during layer-wise generation by introducing genetic-inspired crossover and mutation operations.\"},{\"question\":\"How is each graph neural network layer treated in the genetic-evolutionary framework?\",\"answer\":\"Each layer is viewed as an evolutionary process, where operations based on crossover and mutation are used to prevent embeddings from becoming too alike.\"}]","Genetic-Evolutionary Graph Neural Networks - A Paradigm for Improved Graph Representation Learning | PDF",1785676968,50,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"genetic-evolutionary-graph-neural-networks-a-paradigm-for-improved-graph-representation-learning","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/genetic-evolutionary-graph-neural-networks-a-paradigm-for-improved-graph-representation-learning/117558/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-02",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem do message-passing graph neural networks face?","Question",{"text":75,"@type":76},"They often suffer from over-smoothing, where node representations become overly similar after many message-passing iterations, reducing the model’s ability to learn meaningful differences.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the proposed method address over-smoothing?",{"text":80,"@type":76},"It argues the issue stems from limited diversity in generated embeddings and enhances embedding diversity during layer-wise generation by introducing genetic-inspired crossover and mutation operations.",{"name":82,"@type":73,"acceptedAnswer":83},"How is each graph neural network layer treated in the genetic-evolutionary framework?",{"text":84,"@type":76},"Each layer is viewed as an evolutionary process, where operations based on crossover and mutation are used to prevent embeddings from becoming too alike.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,114,119,122,126,129,133],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":29,"slug":113},6,"Technology","technology",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":21,"slug":125},9,"Religion & Spirituality","religion-spirituality",{"id":21,"doc_module":4,"doc_module_name":46,"category_name":127,"show_sort_weight":21,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":46,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":46,"category_name":135,"show_sort_weight":106,"slug":136},19,"General","general"]