基于神经网络的图布局技术综述  

A survey:neural network-based techniques for graph layout

作　　者：闫凯[1] 赵铁军[1] YAN Kai;ZHAO Tiejun(Machine Intelligence and Translation Laboratory,Harbin Institute of Technology,Harbin 150001,China)

机构地区：[1]哈尔滨工业大学语言技术研究中心,哈尔滨150001

出　　处：《智能计算机与应用》2023年第2期213-216,F0003,共5页Intelligent Computer and Applications

摘　　要：绘图技术研究中图布局问题面临着日益增长的性能需求,基于神经网络的方法由于其一次训练,随处使用的特点逐渐被图可视化领域的研究所注意,2017年后图神经网络成为研究热点,而基于图神经网络的图布局技术也不断出现。本文针对近5年中基于神经网络的图布局生成和验证技术进行梳理,总结当前领域的研究现状以及研究前景,同时帮助分析者合理选择自己需要的技术。There is an increasing performance demand on the graph layout problem in graph drawing.Because of its characteristics of once training and everywhere using,the neural network-based methods attract much attention from graph visualization researchers.Since 2017,graph neural network has become a research hotspot,and graph layout based on graph neural network has also been emerging.This paper summarizes the neural network-based graph layout generation and evaluation technologies in the past five years so that the readers can comprehensively understand the whole field and the analysts can choose proper technologies according to their requirement.

关 键 词：图可视化 图布局 绘图 神经网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]