TY - GEN
T1 - Classifying Me Softly
T2 - Joint IAPR International Workshop on Structural, Syntactic, and Statistical Pattern Recognition, S+SSPR 2022
AU - Bicciato, Alessandro
AU - Cosmo, Luca
AU - Minello, Giorgia
AU - Rossi, Luca
AU - Torsello, Andrea
N1 - Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2022/8
Y1 - 2022/8
N2 - Graph neural networks are increasingly becoming the framework of choice for graph-based machine learning. In this paper we propose a new graph neural network architecture based on the soft-alignment of the graph node features against sets of learned points. In each layer of the network the input node features are transformed by computing their similarity with respect to a set of learned features. The similarity information is then propagated to other nodes in the network, effectively creating a message passing-like mechanism where each node of the graph individually learns what is the optimal message to pass to its neighbours. We perform an ablation study to evaluate the performance of the network under different choices of its hyper-parameters. Finally, we test our model on standard graph-classification benchmarks and we find that it outperforms widely used alternative approaches, including both graph kernels and graph neural networks.
AB - Graph neural networks are increasingly becoming the framework of choice for graph-based machine learning. In this paper we propose a new graph neural network architecture based on the soft-alignment of the graph node features against sets of learned points. In each layer of the network the input node features are transformed by computing their similarity with respect to a set of learned features. The similarity information is then propagated to other nodes in the network, effectively creating a message passing-like mechanism where each node of the graph individually learns what is the optimal message to pass to its neighbours. We perform an ablation study to evaluate the performance of the network under different choices of its hyper-parameters. Finally, we test our model on standard graph-classification benchmarks and we find that it outperforms widely used alternative approaches, including both graph kernels and graph neural networks.
KW - Deep learning
KW - Graph neural network
UR - http://www.scopus.com/inward/record.url?scp=85147847628&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-23028-8_5
DO - 10.1007/978-3-031-23028-8_5
M3 - Conference article published in proceeding or book
AN - SCOPUS:85147847628
SN - 9783031230271
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 43
EP - 53
BT - Structural, Syntactic, and Statistical Pattern Recognition - Joint IAPR International Workshops, S+SSPR 2022, Proceedings
A2 - Krzyzak, Adam
A2 - Suen, Ching Y.
A2 - Nobile, Nicola
A2 - Torsello, Andrea
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 26 August 2022 through 27 August 2022
ER -