OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Clustergcn

Clustergcn is an open-source Python and PyTorch implementation of Cluster-GCN, an algorithm that trains graph convolutional networks by sampling clustered node blocks and restricting neighborhood searches to their subgraphs.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Clustergcn — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when you need a repository implementation of Cluster-GCN and can provide graph edges, sparse node features and node-class targets in the documented CSV formats.

Who it is for

  • Researchers exploring graph convolutional network training methods.
  • Practitioners preparing graph data for node-class prediction.

Documented uses

  • Train a graph convolutional network using edge, sparse-feature and node-target CSV files.
  • Adjust clustering and model settings, including cluster count, epochs, learning rate, dropout and layer sizes.
  • Explore a clustering-based approach to limiting graph neighborhood searches during training.
Before you choose

Requirements & limits

The public GitHub repository identifies the project as open source under GPL-3.0.

  • The documented training entry point is src/main.py and its inputs are CSV files for edges, features and targets.
  • The README lists older development versions, including Python 3.5.2, PyTorch 0.4.1 and PyTorch Geometric 0.3.1; it does not establish compatibility with newer versions.
  • METIS is the default clustering method, and the README documents an Ubuntu package prerequisite for it.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Prepare graph data for ClusterGCN

Format the graph, sparse features, and node targets as CSV files for the documented training inputs.

Before you start

  • Python 3.5.2; the README lists the package versions used during development.
  • On Ubuntu, install the METIS development library with sudo apt-get install libmetis-dev.
  • An edge-list CSV, a sparse-feature CSV, and a target CSV.

Steps

  1. Create an edge-list CSV with a header; put each edge on a row with its two node IDs separated by a comma, and index nodes starting at 0.

  2. Create a sparse-feature CSV with columns for node ID, feature ID, and value; index feature IDs consecutively from 0.

  3. Create a target CSV with headers for node ID and target; sort rows by node ID and index target classes from 0.

Expected outcome

CSV inputs are structured for the `--edge-path`, `--features-path`, and `--target-path` options documented for `src/main.py`.

Guide limitations

  • This recipe prepares input files only; it does not document running or configuring model training.
  • The README names default input paths (input/edges.csv, input/features.csv, and input/target.csv) but does not document how to execute training.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
6 November 2026
Original website title
benedekrozemberczki/ClusterGCN · Official homepage title · retained in its source language
How we review sourcesSuggest a correction