Scholarlink

For distributed computing, there is an urgent need to solve the problem of device placement, so we propose an innovative graph neural network framework NEAGMamba. It employs a decoupling and importance score evaluation mechanism to prevent excessive smoothing of node encodings during the aggregation process. In addition, a new Graph Mamba Network (Gmamba) is designed to replace the traditional GNN, capable of extracting node characteristics and capturing remote dependencies. Specifically, the features of each node are decoupled into two parts: discriminant features and aggregation features. At the same time, the proposed GMamba is used to obtain the aggregation features. Finally, the importance of adjacent nodes is evaluated by the importance score to realize the integration of adjacent information. The experimental results showed that the execution time was improved by 13.55% using NEAGMamba compared to Placeto. NEAGMamba performed 8.18% better than GraphSAGE and 1.22% better than P-GNN. In terms of computation time, NEAGMamba increased by 96.96% compared with Placeto, 97.34% with GraphSAGE and 96.86% with P-GNN.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Novel Device Placement Approach with Neighbor Effect Aware Graph Mamba Networks

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Hao Shu,
  • Wangli Hao,
  • Meng Han,
  • Fuzhong Li

摘要

For distributed computing, there is an urgent need to solve the problem of device placement, so we propose an innovative graph neural network framework NEAGMamba. It employs a decoupling and importance score evaluation mechanism to prevent excessive smoothing of node encodings during the aggregation process. In addition, a new Graph Mamba Network (Gmamba) is designed to replace the traditional GNN, capable of extracting node characteristics and capturing remote dependencies. Specifically, the features of each node are decoupled into two parts: discriminant features and aggregation features. At the same time, the proposed GMamba is used to obtain the aggregation features. Finally, the importance of adjacent nodes is evaluated by the importance score to realize the integration of adjacent information. The experimental results showed that the execution time was improved by 13.55% using NEAGMamba compared to Placeto. NEAGMamba performed 8.18% better than GraphSAGE and 1.22% better than P-GNN. In terms of computation time, NEAGMamba increased by 96.96% compared with Placeto, 97.34% with GraphSAGE and 96.86% with P-GNN.