└── README.md /README.md: -------------------------------------------------------------------------------- 1 | # Graph-based Neural Networks 2 | This page is to summarize important materials about *graph-based neural networks* and *relational networks*. If I miss some recent works or anyone wants to recommend other references, please let me know. 3 | 4 | ## Background 5 | (You can find many materials for deep neural networks in other places. Here, I mainly cover materials about graphs.) 6 | - [Basic Graph Theory](http://data-science-training-xb.com/) by Xavier Bresson, See Lecture 3 and 16 7 | - [Spectral Graph Theory](http://www.math.ucsd.edu/~fan/research/revised.html) by Fan Chung 8 | - [Graph Signal Processing GSP](https://arxiv.org/abs/1712.00468) by Ortega et al. 9 | - This paper provide an overview of core ideas in GSP and their connection to conventional digital signal processing. 10 | - Signal processing is required to understand the convolution in the spectral domain. 11 | - Keywords : graph theory, spectral graph theory, discrete Fourier transform (DFT) 12 | 13 | ## List of Related Works 14 | - **Early works using graph structure** 15 | - [A new model for learning in graph domains](http://ieeexplore.ieee.org/document/1555942/) 16 | - M. Gori, G. Monfardini, F. Scarselli, IJCNN 2005 17 | - **First attempts to generalize neural networks to graphs** 18 | - [The graph neural network model](http://ieeexplore.ieee.org/document/4700287/) 19 | - F. Scarselli, M. Gori, A. C. Tsoi, M. Hagenbuchner, and G. Monfardini, IEEE Trans. Neural Networks 2009 20 | - These works optimized over the parameterized steady state of some diffusion process (or random walk) on the graph. 21 | - **Review paper** (*highly recommend*) 22 | - [Geometric deep learning: going beyond Euclidean data](https://arxiv.org/abs/1611.08097) 23 | - Michael M. Bronstein, Joan Bruna, Yann LeCun, Arthur Szlam, Pierre Vandergheynst, IEEE Signal Processing Magazine 2017 24 | - **First review paper of geometric deep learning** 25 | 26 | - **Graph Convolutional Networks (GCNs)** 27 | - [Spectral Networks and Locally Connected Networks on Graphs](https://arxiv.org/abs/1312.6203) 28 | - Joan Bruna, Wojciech Zaremba, Arthur Szlam, Yann LeCun, ICLR 2014 29 | - **First formulation of CNNs on graphs in the spectral domain** 30 | - [Deep Convolutional Networks on Graph-Structured Data](https://arxiv.org/abs/1506.05163) 31 | - Mikael Henaff, Joan Bruna, Yann LeCun, 2015 32 | - **Spatial localization of smooth filters in the frequency domain** 33 | - [Convolutional Networks on Graphs for Learning Molecular Fingerprints](http://papers.nips.cc/paper/5954-convolutional-networks-on-graphs-for-learning-molecular-fingerprints) 34 | - David Duvenaud, Dougal Maclaurin, Jorge Iparraguirre, Rafael Bombarell, Timothy Hirzel, Alan Aspuru-Guzik, Ryan P. Adams, NIPS 2015 35 | - [Gated Graph Sequence Neural Networks](https://arxiv.org/abs/1511.05493) 36 | - Yujia Li, Daniel Tarlow, Marc Brockschmidt, Richard Zemel, ICLR 2016 37 | - **Sliding a filter on the vertices as conventional CNNs, not spectral filtering** 38 | - [Learning Convolutional Neural Networks for Graphs](https://arxiv.org/abs/1605.05273) 39 | - Mathias Niepert, Mohamed Ahmed, Konstantin Kutzkov, ICML 2016 40 | - [Generalizing the Convolution Operator to extend CNNs to Irregular Domains](https://arxiv.org/abs/1606.01166) 41 | - Jean-Charles Vialatte, Vincent Gripon, Grégoire Mercier, arXiv 2016 42 | - **Generalize CNNs to irregular domains using weight sharing and graph-based operators** 43 | - [Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering](https://arxiv.org/abs/1606.09375), [[PyTorch Code]](https://github.com/xbresson/graph_convnets_pytorch/blob/master/README.md) [[TF Code]](https://github.com/mdeff/cnn_graph) 44 | - Michaël Defferrard, Xavier Bresson, Pierre Vandergheynst, NIPS 2016 45 | - **Spectral CNN with Chebychev polynomial filters (ChebNet)** 46 | - [Learning Shape Correspondence with Anisotropic Convolutional Neural Networks](https://arxiv.org/abs/1605.06437) 47 | - Davide Boscaini, Jonathan Masci, Emanuele Rodolà, Michael M. Bronstein, NIPS 2016 48 | - **Anisotropic CNN framework** 49 | - [Semi-Supervised Classification with Graph Convolutional Networks](https://arxiv.org/abs/1609.02907), [[Code]](https://github.com/tkipf/gcn), [[Blog]](http://tkipf.github.io/graph-convolutional-networks/) 50 | - Thomas N. Kipf, Max Welling, ICLR 2017 51 | - **Graph Convolutional Networks (GCN) framework, a simplification of ChebNet** 52 | - [Geometric Deep Learning on Graphs and Manifolds using Mixture Model CNNs](https://arxiv.org/abs/1611.08402) 53 | - Federico Monti, Davide Boscaini, Jonathan Masci, Emanuele Rodolà, Jan Svoboda, Michael M. Bronstein, CVPR 2017 54 | - **MoNets** 55 | - [Geometric Matrix Completion with Recurrent Multi-Graph Neural Networks](https://arxiv.org/abs/1704.06803), [[Code]](https://github.com/fmonti/mgcnn) 56 | - Federico Monti, Michael M. Bronstein, Xavier Bresson, NIPS 2017 57 | - **Recommendation systems** 58 | - [CayleyNets: Graph Convolutional Neural Networks with Complex Rational Spectral Filters](https://arxiv.org/abs/1705.07664) 59 | - Ron Levie, Federico Monti, Xavier Bresson, Michael M. Bronstein, arXiv 2017 60 | - **Spectral CNN with complex rational filters (CayleyNet)** 61 | - [Residual Gated Graph ConvNets](https://arxiv.org/abs/1711.07553) 62 | - Xavier Bresson, Thomas Laurent, arXiv 2017 63 | 64 | - **Relational Networks (RNs), Relational Reasoning, Interactions** 65 | - [Interaction networks for learning about objects, relations and physics](https://arxiv.org/abs/1612.00222) 66 | - Peter W. Battaglia, Razvan Pascanu, Matthew Lai, Danilo Rezende, Koray Kavukcuoglu, NIPS 2016 67 | - [A simple neural network module for relational reasoning](https://arxiv.org/abs/1706.01427), [[Deepmind Article]](https://deepmind.com/blog/neural-approach-relational-reasoning/), [[Code]](https://github.com/kimhc6028/relational-networks) 68 | - Adam Santoro, David Raposo, David G.T. Barrett, Mateusz Malinowski, Razvan Pascanu, Peter Battaglia, Timothy Lillicrap, arXiv 2017 69 | - **Consider all possible pairs** 70 | - [Neural Message Passing for Quantum Chemistry](https://arxiv.org/abs/1704.01212) 71 | - Justin Gilmer, Samuel S. Schoenholz, Patrick F. Riley, Oriol Vinyals, George E. Dahl, ICML 2017 72 | - [Pointnet: Deep learning on point sets for 3d classification and segmentation](https://arxiv.org/abs/1612.00593) 73 | - Charles R. Qi, Hao Su, Kaichun Mo, Leonidas J. Guibas, CVPR2017 74 | - [SchNet: A continuous-filter convolutional neural network for modeling quantum interactions](https://arxiv.org/abs/1706.08566) 75 | - Kristof T. Schütt, Pieter-Jan Kindermans, Huziel E. Sauceda, Stefan Chmiela, Alexandre Tkatchenko, Klaus-Robert Müller, NIPS 2017 76 | - [VAIN: Attentional Multi-agent Predictive Modeling](http://papers.nips.cc/paper/6863-vain-attentional-multi-agent-predictive-modeling) 77 | - Yedid Hoshen, NIPS 2017 78 | - [Modeling Relational Data with Graph Convolutional Networks](https://arxiv.org/abs/1703.06103) 79 | - Michael Schlichtkrull, Thomas N. Kipf, Peter Bloem, Rianne van den Berg, Ivan Titov, Max Welling, arXiv 2017 80 | - [Graph Attention Networks](https://arxiv.org/abs/1710.10903), [[Code]](https://github.com/PetarV-/GAT) 81 | - Petar Veličković, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Liò, Yoshua Bengio, ICLR 2018 82 | 83 | - **Graph Auto-Encoder (GAE)** 84 | - [Variational Graph Auto-Encoders](https://arxiv.org/abs/1611.07308), [[Code]](https://github.com/tkipf/gae) 85 | - Thomas N. Kipf, Max Welling, NIPS Workshop on Bayesian Deep Learning 2016 86 | - Question: Why the adjacency matrix is reconstructed rather than the feature matrix? 87 | - [Modeling Relational Data with Graph Convolutional Networks](https://arxiv.org/abs/1703.06103) 88 | - Michael Schlichtkrull, Thomas N. Kipf, Peter Bloem, Rianne van den Berg, Ivan Titov, Max Welling, 2017 89 | - [Graph Convolutional Matrix Completion](https://arxiv.org/abs/1706.02263) 90 | - Rianne van den Berg, Thomas N. Kipf, Max Welling, 2017 91 | 92 | - **Other Applications using Graph-based Neural Networks** 93 | - [Diffusion Convolutional Recurrent Neural Network: Data-Driven Traffic Forecasting 94 | ](https://arxiv.org/abs/1707.01926), [[Code]](https://github.com/liyaguang/DCRNN) 95 | - Yaguang Li, Rose Yu, Cyrus Shahabi, Yan Liu, ICLR 2018 96 | - [Automatically Inferring Data Quality for Spatiotemporal Forecasting](https://openreview.net/forum?id=ByJIWUnpW) 97 | - Sungyong Seo, Arash Mohegh, George Ban-Weiss, Yan Liu, ICLR 2018 98 | 99 | ## Tutorials or Workshops 100 | - IPAM18 Workshop, [New Deep Learning Techniques](http://www.ipam.ucla.edu/programs/workshops/new-deep-learning-techniques/) 101 | - NIPS17 Tutorial, [Geometric Deep Learning on Graphs and Manifolds](https://nips.cc/Conferences/2017/Schedule?showEvent=8735) 102 | - CVPR17 Tutorial, [Geometric Deep Learning on Graphs](http://geometricdeeplearning.com/) 103 | 104 | ## Useful Resources 105 | - [Kipf's blog](http://tkipf.github.io/graph-convolutional-networks/) 106 | - [Geometric Deep Learning](http://geometricdeeplearning.com/) **highly recommended** 107 | - [CVPR17 tutorial, Geometric and Semantic 3D Reconstruction](https://www.dropbox.com/s/4l6m32tg9yecvow/CVPR%20GDL.pdf?dl=0), 240MB 108 | - [How do I generalize convolution of neural networks to graphs?](https://www.quora.com/How-do-I-generalize-convolution-of-neural-networks-to-graphs), Defferrard's answers in Quora 109 | - [PointNet](http://stanford.edu/~rqi/pointnet/) 110 | 111 | 112 | ## List of Researchers 113 | - [Thomas Kipf](http://tkipf.github.io/), University of Amsterdam 114 | - [Joan Bruna](http://cims.nyu.edu/~bruna/), NYU 115 | - [Michaël Defferrard](http://deff.ch/), EPFL 116 | - [Xavier Bresson](http://www.ntu.edu.sg/home/xbresson/index.html), NTU 117 | - [Federico Monti](https://www.ics.usi.ch/index.php/people-detail-page/268-federico-monti), Università della Svizzera Italiana 118 | - [Michael M. Bronstein](http://www.inf.usi.ch/bronstein/), Università della Svizzera Italiana 119 | --------------------------------------------------------------------------------