Efficient Computation of Higher-Order Subgraph Attribution via Message Passing

Ping Xiong, Thomas Schnake, Grégoire Montavon, Klaus-Robert Müller, Shinichi Nakajima
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:24478-24495, 2022.

Abstract

Explaining graph neural networks (GNNs) has become more and more important recently. Higher-order interpretation schemes, such as GNN-LRP (layer-wise relevance propagation for GNN), emerged as powerful tools for unraveling how different features interact thereby contributing to explaining GNNs. GNN-LRP gives a relevance attribution of walks between nodes at each layer, and the subgraph attribution is expressed as a sum over exponentially many such walks. In this work, we demonstrate that such exponential complexity can be avoided. In particular, we propose novel algorithms that enable to attribute subgraphs with GNN-LRP in linear-time (w.r.t. the network depth). Our algorithms are derived via message passing techniques that make use of the distributive property, thereby directly computing quantities for higher-order explanations. We further adapt our efficient algorithms to compute a generalization of subgraph attributions that also takes into account the neighboring graph features. Experimental results show the significant acceleration of the proposed algorithms and demonstrate the high usefulness and scalability of our novel generalized subgraph attribution method.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-xiong22a, title = {Efficient Computation of Higher-Order Subgraph Attribution via Message Passing}, author = {Xiong, Ping and Schnake, Thomas and Montavon, Gr{\'e}goire and M{\"u}ller, Klaus-Robert and Nakajima, Shinichi}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {24478--24495}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/xiong22a/xiong22a.pdf}, url = {https://proceedings.mlr.press/v162/xiong22a.html}, abstract = {Explaining graph neural networks (GNNs) has become more and more important recently. Higher-order interpretation schemes, such as GNN-LRP (layer-wise relevance propagation for GNN), emerged as powerful tools for unraveling how different features interact thereby contributing to explaining GNNs. GNN-LRP gives a relevance attribution of walks between nodes at each layer, and the subgraph attribution is expressed as a sum over exponentially many such walks. In this work, we demonstrate that such exponential complexity can be avoided. In particular, we propose novel algorithms that enable to attribute subgraphs with GNN-LRP in linear-time (w.r.t. the network depth). Our algorithms are derived via message passing techniques that make use of the distributive property, thereby directly computing quantities for higher-order explanations. We further adapt our efficient algorithms to compute a generalization of subgraph attributions that also takes into account the neighboring graph features. Experimental results show the significant acceleration of the proposed algorithms and demonstrate the high usefulness and scalability of our novel generalized subgraph attribution method.} }
Endnote
%0 Conference Paper %T Efficient Computation of Higher-Order Subgraph Attribution via Message Passing %A Ping Xiong %A Thomas Schnake %A Grégoire Montavon %A Klaus-Robert Müller %A Shinichi Nakajima %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-xiong22a %I PMLR %P 24478--24495 %U https://proceedings.mlr.press/v162/xiong22a.html %V 162 %X Explaining graph neural networks (GNNs) has become more and more important recently. Higher-order interpretation schemes, such as GNN-LRP (layer-wise relevance propagation for GNN), emerged as powerful tools for unraveling how different features interact thereby contributing to explaining GNNs. GNN-LRP gives a relevance attribution of walks between nodes at each layer, and the subgraph attribution is expressed as a sum over exponentially many such walks. In this work, we demonstrate that such exponential complexity can be avoided. In particular, we propose novel algorithms that enable to attribute subgraphs with GNN-LRP in linear-time (w.r.t. the network depth). Our algorithms are derived via message passing techniques that make use of the distributive property, thereby directly computing quantities for higher-order explanations. We further adapt our efficient algorithms to compute a generalization of subgraph attributions that also takes into account the neighboring graph features. Experimental results show the significant acceleration of the proposed algorithms and demonstrate the high usefulness and scalability of our novel generalized subgraph attribution method.
APA
Xiong, P., Schnake, T., Montavon, G., Müller, K. & Nakajima, S.. (2022). Efficient Computation of Higher-Order Subgraph Attribution via Message Passing. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:24478-24495 Available from https://proceedings.mlr.press/v162/xiong22a.html.

Related Material