Logo
0.9.4

Get Started

  • Installation
  • Structures in DHG
    • Introduction
    • Applications
    • Two Core Operations
      • The Spectral-Based Operations
      • The Spatial-Based Operations
    • What Can be Done with the Two Operations?
      • Add Early Self-loop and Late Self-Loop
        • Message Passing with Early Self-Loop
        • Message Passing with Late Self-Loop
      • Fuse Features Learned from the Spectral and Spatial Domain
      • Fuse Features Learned from different Structures
  • Learning on Low-Order Structures
    • Learning on Graph
      • Definition
      • Construction
      • Spectral-Based Learning
        • Smoothing with GCN’s Laplacian
        • Smoothing with Symmetrically Normalized Laplacian
        • Smoothing with Left (random-walk) Normalized Laplacian
      • Spatial-Based Learning
        • Message Propagation from Vertex to Vertex
        • Message Propagation from Vertex to Vertex with different Edge Weights
    • Learning on Directed Graph
      • Definition
      • Construction
      • Spectral-Based Learning
      • Spatial-Based Learning
        • Message Propagation from Source Vertex to Target Vertex
        • Message Propagation from Source Vertex to Target Vertex with different Edge Weights
        • Message Propagation from Target Vertex to Source Vertex
        • Message Propagation from Target Vertex to Source Vertex with different Edge Weights
    • Learning on Bipartite Graph
      • Definition
      • Construction
      • Spectral-Based Learning
        • Smoothing with GCN’s Laplacian
      • Spatial-Based Learning
        • Message Propagation from Vertices in Set \(U\) to Vertices in Set \(V\)
        • Message Propagation from Vertices in Set \(U\) to Vertices in Set \(V\) with different Edge Weights
        • Message Propagation from Vertices in Set \(V\) to Vertices in Set \(U\)
        • Message Propagation from Vertices in Set \(V\) to Vertices in Set \(U\) with different Edge Weights
  • Learning on High-Oder Structures
    • Learning on Hypergraph
      • Definition
      • Construction
      • Spectral-Based Learning
        • Smoothing with HGNN’s Laplacian
      • Spatial-Based Learning
        • Message Propagation from Vertex to Hyperedge
        • Message Propagation from Vertex to Hyperedge with different Edge Weights
        • Message Propagation from Hyperedge to Vertex
        • Message Propagation from Hyperedge to Vertex with different Edge Weights
        • Message Propagation from Vertex Set to Vertex Set
        • Message Propagation from Vertex Set to Vertex Set with different Edge Weights in Two Stages
  • Contribute to DHG
    • Coding Style
    • Testing
    • Building Documentation

Examples

  • Vertex Classification
    • On Graph
      • Models
      • Dataset
      • Results
      • GCN on Cora
        • Import Libraries
        • Define Functions
        • Main
        • Outputs
      • GAT on Cora
        • Import Libraries
        • Define Functions
        • Main
        • Outputs
      • HGNN on Cora
        • Import Libraries
        • Define Functions
        • Main
        • Outputs
      • HGNN+ on Cora
        • Import Libraries
        • Define Functions
        • Main
        • Outputs
    • On Hypergraph
      • Models
      • Dataset
      • Results
      • GCN on Cooking200
        • Import Libraries
        • Define Functions
        • Main
        • Outputs
      • HGNN on Cooking200
        • Import Libraries
        • Define Functions
        • Main
        • Outputs
      • HGNN+ on Cooking200
        • Import Libraries
        • Define Functions
        • Main
        • Outputs
  • User-Item Recommender
    • Models
    • Dataset
    • Results
    • NGCF on Gowalla
      • Import Libraries
      • Define Functions
      • Main
      • Outputs
    • LightGCN on Gowalla
      • Import Libraries
      • Define Functions
      • Main
      • Outputs
  • Auto Hyper-parameters Tuning
    • GCN on Cora
      • Configuration
      • Import Libraries
      • Define Functions
      • Main
      • Outputs
    • HGNN+ on Cooking200
      • Configuration
      • Import Libraries
      • Define Functions
      • Main
      • Outputs
    • LightGCN on Gowalla
      • Configuration
      • Import Libraries
      • Define Functions
      • Main
      • Outputs

Tutorials

  • Overview
    • Preparations
    • Training and Evaluation
    • Others
  • Building Structure
    • Low-Order Structures
      • Building Graph
        • Common Methods
        • Reduced from High-Order Structures
      • Building Directed Graph
        • Common Methods
        • Reduced from High-Order Structures
      • Building Bipartite Graph
        • Common Methods
        • Reduced from High-Order Structures
    • High-Order Structures
      • Building Hypergraph
        • Common Methods
        • Prometed from Low-Order Structures
  • Building Dataset
    • Basic Usages
    • Architechture
    • Building Your Own Dataset
      • Example of Graph Dataset:
      • Example of Hypergraph Dataset
      • Example of User-Item Bipartite Dataset
  • Building Model
    • Building Spectral-based Model
    • Building Spatial-based Model
    • Building Hybrid Operation Model
    • Building Hybrid Structure Model
  • Building Evaluator
    • Initialization
    • Epoch Evaluation
    • Add Batches Then Do Epoch Evaluation
  • Training Model
    • Training without Batch Data
    • Training with Batch Data
  • Training with Auto ML
    • Builder Functions for Auto-ML
      • Defining the Structure Builder
      • Defining the Model Builder
      • Defining the Train Builder
    • Task Class for Auto-ML
    • Auto-ML for Vertex Classification Task
      • On Graph
      • On Hypergraph
    • Auto-ML for Item Recommender Task
  • Structure Generation
    • Random Graph Generation
    • Random Directed Graph Generation
    • Random Bipartite Graph Generation
    • Random Hypergraph Generation
  • Structure Visualization
    • Basic Usages
      • Visualization of Graph
      • Visualization of Directed Graph
      • Visualization of Bipartite Graph
      • Visualization of Hypergraph
    • Advanced Usages
      • Customize Labels
      • Customize Colors
      • Customize Sizes
      • Customize Layout
  • Feature Visualization
    • Basic Usages
      • Visualization of Features in Euclidean Space
      • Visualization of Features in Poincare Space
    • Make Animation
      • Rotating Visualization of Features in Euclidean Space
      • Rotating Visualization of Features in Poincare Space
    • Mathematical Principles of Hyperbolic Space

中文文档

  • DHG简介
  • 上手指南
    • 安装
    • DHG内的关联结构
      • 简介
      • 应用场景
      • 两个核心操作
        • 基于谱域的操作
        • 基于空域的操作
      • 基于两种操作可以实现什么?
        • 增加先自环以及后自环
        • 融合从谱域和空域中学习到的特征
        • 融合从不同关联结构中学习到的特征
    • 低阶关联结构上的表示学习
      • 图上的表示学习
        • 定义
        • 结构构建
        • 基于谱域的学习
        • 基于空域的学习
      • 有向图上的表示学习
        • 定义
        • 结构构建
        • 基于谱域的学习
        • 基于空域的学习
      • 二分图上的表示学习
        • 定义
        • 结构构建
        • 基于谱域的学习
        • 基于空域的学习
    • 高阶关联结构上的表示学习
      • 超图上的表示学习
        • 定义
        • 结构构建
        • 基于谱域的学习
        • 基于空域的学习
    • 如何加入DHG贡献团队
      • 编程风格
      • 代码测试
      • 构建文档
  • 代码样例
    • 节点分类
      • 图
        • 模型
        • 数据集
        • 结果汇总
        • Cora上使用GCN
        • Cora上使用GAT
        • Cora上使用HGNN
        • Cora上使用HGNN+
      • 超图
        • 模型
        • 数据集
        • 结果汇总
        • Cooking200上使用GCN
        • Cooking200上使用HGNN
        • Cooking200上使用HGNN+
    • <用户-物品>二分图上的推荐
      • 模型
      • 数据集
      • 结果汇总
      • 在Gowalla上使用NGCF
        • 导入依赖包
        • 定义函数
        • 主函数
        • 输出
      • 在Gowalla上使用LightGCN
        • 导入依赖包
        • 定义函数
        • 主函数
        • 输出
    • 自动化超参调优
      • 在Cora上使用GCN
        • 配置
        • 导入依赖包
        • 定义函数
        • 主函数
        • 输出
      • 在Cooking200上使用HGNN+
        • 配置
        • 导入依赖包
        • 定义函数
        • 主函数
        • 输出
      • 在Gowalla上使用LightGCN
        • 配置
        • 导入依赖包
        • 定义函数
        • 主函数
        • 输出
  • 相关教程
    • 总览
      • 准备工作
      • 训练和评测
      • 其他
    • 构建关联结构
      • 构建低阶关联结构
        • 构建图
        • 构建有向图
        • 构建二分图
      • 构建高阶关联结构
        • 构建超图
    • 构建输入数据
      • 使用方法
      • 模块架构设计
      • 建立自己的数据集
        • 图数据集示例
        • 超图数据集示例
        • <用户-物品>二分图示例
    • 构建模型
      • 构建基于谱域的模型
      • 构建基于空域的模型
      • 构建混合操作模型
      • 构建混合关联结构模型
    • 构建指标评测器
      • 初始化
      • 整轮评测
      • 添加批数据后整轮评测
    • 训练模型
      • 没有批数据的模型的训练
      • 有批数据的模型的训练
    • 自动化超参调优
      • 自动调优的构造函数
        • 定义结构调优构造函数
        • 定义模型调优构造函数
        • 定义训练调优构造函数
      • 自动化调优的任务类
      • 自动化节点分类任务
        • 自动化图节点分类任务
        • 自动化超图节点分类任务
      • 自动化物品推荐任务
    • 随机结构生成
      • 随机图生成
      • 随机有向图生成
      • 随机二分图生成
      • 随机超图生成
    • 关联结构可视化
      • 基本用法
        • 图的可视化
        • 有向图的可视化
        • 二分图的可视化
        • 超图的可视化
      • 高级用法
        • 自定义标签
        • 自定义颜色
        • 自定义大小
        • 自定义布局
    • 特征可视化
      • 基本用法
        • 在欧几里得空间中进行特征可视化
        • 在庞加莱空间中进行特征可视化
      • 制作动画
        • 欧几里得空间中特征的旋转可视化
        • 庞加莱空间中特征的旋转可视化
      • 双曲空间的数学原理

API Reference

  • dhg
    • Load Structure
    • Low-Order Structures
      • Base Class
      • Graph
      • Directed Graph
      • Bipartite Graph
    • High-Order Structures
      • Base Class
      • Hypergraph
  • dhg.nn
    • Common Layers
      • MLP
      • MultiHeadWrapper
      • Discriminator
    • Layers on Graph
      • GCNConv
      • GraphSAGEConv
      • GATConv
      • GINConv
    • Layers on Hypergraph
      • HGNNConv
      • HGNNPConv
      • JHConv
      • HyperGCNConv
      • HNHNConv
      • UniGCNConv
      • UniGATConv
      • UniSAGEConv
      • UniGINConv
    • Loss Functions
      • BPRLoss
    • Regularizations
      • EmbeddingRegularization
  • dhg.models
    • Models on Graph
      • GCN
      • GraphSAGE
      • GAT
      • GIN
    • Models on Bipartite Graph
      • NGCF
      • LightGCN
      • BGNN_Adv
      • BGNN_MLP
    • Models on Hypergraph
      • HGNN
      • HGNNP
      • HyperGCN
      • DHCF
      • HNHN
      • UniGCN
      • UniGAT
      • UniSAGE
      • UniGIN
  • dhg.data
    • Base Class
    • Graph Datasets
      • Cora
      • Pubmed
      • Citeseer
      • BlogCatalog
      • Flickr
      • Github
      • Facebook
    • Bipartite Graph Datasets
      • MovieLens1M
      • AmazonBook
      • Yelp2018
      • Gowalla
      • TencentBiGraph
      • CoraBiGraph
      • PubmedBiGraph
      • CiteseerBiGraph
    • Hypergraph Datasets
      • Cooking200
      • CoauthorshipCora
      • CoauthorshipDBLP
      • CocitationCora
      • CocitationCiteseer
      • CocitationPubmed
      • YelpRestaurant
      • WalmartTrips
      • HouseCommittees
      • News20
      • DBLP4k
      • DBLP8k
      • IMDB4k
      • Recipe100k
      • Recipe200k
      • Yelp3k
      • Tencent2k
  • dhg.datapipe
    • Compose Datapipes
    • Transforms
    • Loaders
  • dhg.metrics
    • Basic Metrics
      • Classification
      • Recommender
      • Retrieval
    • Evaluators for Different Tasks
      • Base Class
      • Vertex Classification Task
        • On Graph
        • On Hypergraph
      • Recommender Task
        • On User-Item Bipartite Graph
      • Retrieval Task
  • dhg.experiments
    • Base Class
    • Vertex Classification Task
      • On Graph
      • On Hypergraph
    • Recommender Task
      • On User-Item Bipartite Graph
  • dhg.visualization
    • Structure Visualization
      • Graph
      • Directed Graph
      • Bipartite Graph
      • Hypergraph
    • Feature Visualization
      • Feature Visualization in Euclidean Space
      • Feature Visualization in Poincare Ball
      • Make Animations
  • dhg.random
    • Random Seed
    • Generating Features
    • Generating Graph
    • Generating Directed Graph
    • Generating Bipartite Graph
    • Generating Hypergraph
  • dhg.utils
    • Structure Helpers
    • Sparse Operations
    • Dataset Splitting
    • Dataset Wrapers
    • Log Helpers
    • Download Helpers
DHG
  • Overview
  • Edit on GitHub

Overview

Preparations

  1. Build the Correlation Structure for Your Task.

  2. Build the Input Data for Your Task.

  3. Build the Model for Your Task.

  4. Build the Evaluator for Your Task.

Training and Evaluation

  1. Build Your Training and Evaluation Code.

  2. Auto-Hyperparameter Tuning.

Others

  • Random Correlation Structure Generation.

  • Structure Visualization.

  • Feature Visualization.

Previous Next

© Copyright 2022, iMoonLab. Revision a09c7cfc.

Built with Sphinx using a theme provided by Read the Docs.