# mdeff/cnn_graph

Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Repository: https://github.com/mdeff/cnn_graph
Canonical: https://ross.abutalabs.com/products/cnn_graph
Homepage: https://arxiv.org/abs/1606.09375
Language: Jupyter Notebook
License: MIT
License Family: permissive
Topics: convolutional-neural-networks, deep-learning, graphs, graph-signal-processing, graph-neural-networks
Last push: 2020-06-13T21:47:49+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3641, "days_push": 2272, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1366, forks 386 (observed 2026-08-28T04:04:31.337752+00:00)

## What it is
Reference implementation of the NIPS 2016 paper 'Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering' (ChebNet), generalizing CNNs to arbitrary graphs via spectral graph theory. It ships TensorFlow code and Jupyter notebooks that reproduce the MNIST and 20NEWS experiments and demonstrate applying the graph ConvNet to custom data.

## Use cases
- run graph convolutional neural networks on graph-structured data
- reproduce the NIPS 2016 ChebNet paper experiments on MNIST and 20NEWS
- classify samples with an adjacency matrix encoding graph structure
- learn localized spectral filters on irregular domains like social networks or connectomes
- study implementations of spectral networks and deep convolutional nets on graphs

## When to choose
- you want the original Chebyshev graph convolution implementation from the NIPS 2016 paper
- you have data with an explicit graph/adjacency structure and want a simple TensorFlow 1.x model
- you are studying or teaching graph signal processing and graph neural networks

## When to avoid
- you need a maintained, production-grade GNN library (the repo targets TensorFlow 1.x and sees little activity)
- you want modern GNN architectures like GCN, GAT, or GraphSAGE with active support
- you work in PyTorch or need GPU-optimized large-scale training

## Facets
- artifact type: library
- maturity: maintenance
- function: deep-learning, machine-learning
- domain: deep-learning, machine-learning, graph-processing
- platform: python
- tags: graph-neural-networks, graph-convolutional-networks, spectral-filtering, tensorflow, jupyter-notebooks, research-code, nips-2016

## Member repositories
- mdeff/cnn_graph (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:31.337752+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:41:13.414385+00:00, confidence not recorded.
  - readme: https://github.com/mdeff/cnn_graph (fetched 2026-08-28T04:04:31.337752+00:00, sha a46ad5c43563)
  - homepage: https://arxiv.org/abs/1606.09375 (fetched 2026-08-29T11:58:31.434975+00:00, sha aaffbb452ea5)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T11:58:31.437748+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T11:58:31.441396+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T11:58:31.443364+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T11:58:31.439685+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.