CayleyNet

A CayleyNet is a Spectral Graph Convolutional Network that is based on Cayley polynomials.

• Context:
  • Source code available at:
    • https://github.com/amoliu/CayleyNet
  • It was first developed by Levie et al. (2018).
• Example(s):
  • …
• Counter-Example(s):
  • an Adaptive Graph Convolution Network,
  • a ChebNet,
  • a Spatial Graph Convolutional Network,
  • a Graph Attention Network,
  • a Graph Recurrent Network,
  • a Graph Wavelet Neural Network.
• See: Recurrent Neural Network, Feedforward Neural Network, Attention Mechanism.

---

---

References

2021

• (GitHub, 2021) ⇒ https://github.com/amoliu/CayleyNet Retrieved: 2021-08-29.
  • QUOTE: CayleyNet is a Graph CNN with spectral zoom properties able to effectively operate with signals defined over graphs. Thanks to its particular spectral properties, CayleyNet is well suited for dealing with a variety of different domains (e.g. citation networks, community graphs, user/item similarity graphs...). Variations of the architecture here implemented achieved state-of-the-art performance on vertex classification, community detection and matrix completion tasks.

2018

• (Levie et al., 2018) ⇒ Ron Levie, Federico Monti, Xavier Bresson, and Michael M. Bronstein. "CayleyNets: Graph Convolutional Neural Networks With Complex Rational Spectral Filters". In: IEEE Transactions on Signal Processing Volume: 67, Issue: 1.
  • QUOTE: In this paper, we introduce a new spectral domain convolutional architecture for deep learning on graphs. The core ingredient of our model is a new class of parametric rational complex functions (Cayley polynomials) allowing to efficiently compute spectral filters on graphs that specialize on frequency bands of interest. Our model generates rich spectral filters that are localized in space, scales linearly with the size of the input data for sparsely-connected graphs, and can handle different constructions of Laplacian operators.

---