DYMAG: Rethinking Message Passing Using Dynamical-systems-based Waveforms

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Bhaskar, Dhananjay, Sun, Xingzhi, Zhang, Yanlei, Xu, Charles, Afrasiyabi, Arman, Viswanath, Siddharth, Fasina, Oluwadamilola, Nickel, Maximilian, Wolf, Guy, Perlmutter, Michael, Krishnaswamy, Smita
Natura: Preprint
Pubblicazione: 2023
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910969212436480
author Bhaskar, Dhananjay
Sun, Xingzhi
Zhang, Yanlei
Xu, Charles
Afrasiyabi, Arman
Viswanath, Siddharth
Fasina, Oluwadamilola
Nickel, Maximilian
Wolf, Guy
Perlmutter, Michael
Krishnaswamy, Smita
author_facet Bhaskar, Dhananjay
Sun, Xingzhi
Zhang, Yanlei
Xu, Charles
Afrasiyabi, Arman
Viswanath, Siddharth
Fasina, Oluwadamilola
Nickel, Maximilian
Wolf, Guy
Perlmutter, Michael
Krishnaswamy, Smita
contents We present DYMAG, a graph neural network based on a novel form of message aggregation. Standard message-passing neural networks, which often aggregate local neighbors via mean-aggregation, can be regarded as convolving with a simple rectangular waveform which is non-zero only on 1-hop neighbors of every vertex. Here, we go beyond such local averaging. We will convolve the node features with more sophisticated waveforms generated using dynamics such as the heat equation, wave equation, and the Sprott model (an example of chaotic dynamics). Furthermore, we use snapshots of these dynamics at different time points to create waveforms at many effective scales. Theoretically, we show that these dynamic waveforms can capture salient information about the graph including connected components, connectivity, and cycle structures even with no features. Empirically, we test DYMAG on both real and synthetic benchmarks to establish that DYMAG outperforms baseline models on recovery of graph persistence, generating parameters of random graphs, as well as property prediction for proteins, molecules and materials. Our code is available at https://github.com/KrishnaswamyLab/DYMAG.
format Preprint
id arxiv_https___arxiv_org_abs_2309_09924
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle DYMAG: Rethinking Message Passing Using Dynamical-systems-based Waveforms
Bhaskar, Dhananjay
Sun, Xingzhi
Zhang, Yanlei
Xu, Charles
Afrasiyabi, Arman
Viswanath, Siddharth
Fasina, Oluwadamilola
Nickel, Maximilian
Wolf, Guy
Perlmutter, Michael
Krishnaswamy, Smita
Machine Learning
Signal Processing
We present DYMAG, a graph neural network based on a novel form of message aggregation. Standard message-passing neural networks, which often aggregate local neighbors via mean-aggregation, can be regarded as convolving with a simple rectangular waveform which is non-zero only on 1-hop neighbors of every vertex. Here, we go beyond such local averaging. We will convolve the node features with more sophisticated waveforms generated using dynamics such as the heat equation, wave equation, and the Sprott model (an example of chaotic dynamics). Furthermore, we use snapshots of these dynamics at different time points to create waveforms at many effective scales. Theoretically, we show that these dynamic waveforms can capture salient information about the graph including connected components, connectivity, and cycle structures even with no features. Empirically, we test DYMAG on both real and synthetic benchmarks to establish that DYMAG outperforms baseline models on recovery of graph persistence, generating parameters of random graphs, as well as property prediction for proteins, molecules and materials. Our code is available at https://github.com/KrishnaswamyLab/DYMAG.
title DYMAG: Rethinking Message Passing Using Dynamical-systems-based Waveforms
topic Machine Learning
Signal Processing
url https://arxiv.org/abs/2309.09924