FuseSampleAgg: One-Pass Neighborhood Estimation for Budgeted Knowledge-Graph Refresh and Validation
FuseSampleAgg: 预算知识图谱刷新与验证的单遍邻域估计
机构 * University of Novi Sad(诺维萨德大学) ; Shanghai Key Lab of Intelligent Information Processing(上海智能信息处理重点实验室) ; College of Computer Science and Artificial Intelligence(计算机科学与人工智能学院) ; Fudan University(复旦大学) ; Institute of Automation(自动化研究所) ; Chinese Academy of Sciences(中国科学院)
AI总结 提出一种融合的PyTorch CUDA算子,通过单遍采样和聚合实现邻域均值估计,避免显式子图构建,在保持GraphSAGE-mean语义的同时,将端到端步延迟提升2.24-3.48倍,瞬态内存降低至1/160。
Comments 11 pages. Code and reproducibility scripts: https://github.com/SV25-22/FuseSampleAgg