Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

16
Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering Defferrard, Michaël, Xavier Bresson, and Pierre Vandergheynst NIPS 2016

Transcript of Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Page 1: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Convolutional Neural Networks on Graphs

with Fast Localized Spectral FilteringDefferrard, Michaël, Xavier Bresson, and Pierre Van-

dergheynstNIPS 2016

Page 2: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Unstructured data as graphs• Majority of data is naturally unstructured, but can be

structured.• Irregular / non-Euclidean data can be structured with

graphs• Social networks: Facebook, Twitter.• Biological networks: genes, molecules, brain connectivity.• Infrastructure networks: energy, transportation, Internet, tele-

phony.

• Graphs can model heterogeneous pairwise relation-ships.• Graphs can encode complex geometric structures.

Page 3: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

CNN architecture

• Convolution filter translation or fast Fourier transform (FFT).• Down-sampling pick one pixel out of n.

Page 4: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Generalizing CNNs to graphs• Challenges• Formulate convolution and down-sam-

pling on graphs• How to define localized graph filters?• Make them efficient

Page 5: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Generalizing CNNs to graphs1. The design of localized convolutional filters on graphs2. Graph coarsening procedure (sub-sampling)3. Graph pooling operation

Page 6: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

• : undirected and con-nected graph

• Spectral graph theory• Graph Laplacians

• Normalized Laplacians

: set of vertices : set of edges : weighted adjacency matrix : diagonal degree matrix : identity matrix

Graph Fourier Transform

Page 7: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Graph Fourier Transform• Graph Fourier Transform• (Eigen value decomposition)• Graph Fourier basis • Graph frequencies = 1. Graph signal , 2. Transform

Page 8: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Spectral filtering of graph signals• Convolution on graphs

• filtered signal

• A non-parametric filter Non-localized in vertex domain Learning complexity in Computational complexity in

Page 9: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Polynomial parametrization for localized fil-ters•

order polynomials of the Laplacian -> -localized Learning complexity in Still, computational complexity in because of multiplication with Fourier

basis

• Filter localization on graph

Page 10: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Recursive formulation for fast filter-ing

• Chebyshev expansion • Filtered • multiplications by a sparse costs

Learning complexity in Computational complexity in

Page 11: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Graph coarsening and pooling

• Graph coarsening• To cluster similar vertices together, multilevel clustering algorithm is needed.• Pick an unmarked vertex and matching it with one of its unmarked neighbors that maxi-

mizes the local normalized cut • Pooling of graph signals

• Balanced binary tree structured coarsened graphs• ReLU activation with max pooling

• e.g.

level 0

level 1

level 2

Page 12: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Graph ConvNet (GCN) architecture

Page 13: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Experiments• MNIST

• CNNs on a Euclidean space• Comparable to classical CNN• Isotropic spectral filters

• edges in a general graph do not pos-sess an orientation

Page 14: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Experiments• 20NEWS

• structure documents with a feature graph

• 10,000 nodes, 132,834 edges

𝑂 (𝑛2)

𝑂 (𝑛)

Page 15: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Conclusion• Contributions• Spectral formulation of CNNs on graphs in GSP• Strictly localized spectral filters are proposed• Linear complexity of filters• Efficient pooling on graphs

• Limitation• Filters are not directly transferrable to a different graph

Page 16: Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

References• Deep Learning on Graphs, a lecture on A Network Tour of Data

Science (NTDS) 2016• Shuman, David I., et al. "The emerging field of signal processing

on graphs: Extending high-dimensional data analysis to networks and other irregular domains." IEEE Signal Processing Magazine 30.3 (2013): 83-98.• How powerful are Graph Convolutions? (

http://www.inference.vc/how-powerful-are-graph-convolutions-review-of-kipf-welling-2016-2/)• GRAPH CONVOLUTIONAL NETWORKS (

http://tkipf.github.io/graph-convolutional-networks/)