بسم الله الرحمن الرحیم

Graph Conv. Neural Networks

Mahmood Amintoosi

m.amintoosi @ hsu.ac.ir
پاییز ۱۴۰۱

These Slides are accessible from my github page

شبکه‌های عصبی پیچشی گراف

حوزه‌ی شبکه‌های عصبی پیچشی گراف از جمله موضوعاتی است که توجه بسیاری را در چند سال اخیر به خود معطوف نموده است. آشنایی با شبکه‌های عصبی، شبکه‌های عصبی پیچشی و گراف‌ها، لازمه ورود به این حوزه است.
چند منبع مرتبط:

Topics

  1. Deep Learning
  2. Graph Neural Networks
  3. Sample Code

Who is involved

Stanford Jure Leskovec
Germany Matthias Fey
Singapore Xavier Bresson
Google Thomas Kipf

موارد نظری مرتبط

  • Machine Learning and Data Mining
  • Calculus, Statistics
  • Graphs
    • Adjacency Matrix
    • Laplacian Matrix
    • Rayleigh Quotient
    • Spectral Clustering
    • Closeness, Eccentricity, ...
  • Vector spaces
    • Inner product spaces
    • Vector space of functions: The inner product of two vectors in this vector space is defined
      as the integral of the product of the two functions.
    • Fourier, Wavelet, DCT
    • Convolution
  • Regression
    • Linear Regression
    • Logistic Regression
    • LASSO, Least Square, Normal Equations, ...
  • Optimization
    • SGD, Adam, Conjugate gradient, Sparse Optimization, ...
  • Matrix Factorization
    • Gram-Schmit, QR-Decomposition, ...

موارد عملی مرتبط

  • Programming, OOP
  • Data mining
    • Web Mining
    • Text Mining
    • HTML, CSS, ...
  • Python
    • Pytorch or TF or JAX
    • NetworkX
    • SciPy
    • Pandas
    • Scrapy, Spacy, Beautiful-soup ...
  • SQL
  • Linux
  • GitHub
  • Google Colab
  • GNN Library
    • PyG
    • DGL
    • CogDL
  • Docker

Cora Dataset

Source: https://graphsandnetworks.com/the-cora-dataset/
GitHub: https://github.com/tomonori-masui/graph-neural-networks
The dataset consists of academic publications as the nodes and the citations between them as the links: if publication A cites publication B, then the graph has an edge from A to B. The nodes are classified into one of seven subjects, and the model will learn to predict this subject.

Run in Google Colab


Dataset description and code in TF