SPI-GCN: A Simple Permutation-Invariant Graph Convolutional Network

HAL archives-ouvertes 2019 Asma AtamnaNataliya SokolovskaJean-Claude Crivello

A wide range of machine learning problems involve handling graph-structured data. Existing machine learning approaches for graphs, however, often imply computing expensive graph similarity measures, preprocessing input graphs, or explicitly ordering graph nodes... (read more)

PDF Abstract

Code


No code implementations yet. Submit your code now

Evaluation Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK COMPARE
Graph Classification COIL-RAG SPI-GCN Accuracy 75.72 # 1
Graph Classification ENZYMES SPI-GCN Accuracy 50.17% # 14
Graph Classification HYDRIDES SPI-GCN Accuracy 82.25 # 1
Graph Classification IMDb-B SPI-GCN Accuracy 60.40% # 19
Graph Classification IMDb-M SPI-GCN Accuracy 44.13% # 16
Graph Classification MUTAG SPI-GCN Accuracy 84.40% # 28
Graph Classification NCI1 SPI-GCN Accuracy 64.11% # 31
Graph Classification PROTEINS SPI-GCN Accuracy 72.06% # 42
Graph Classification PTC SPI-GCN Accuracy 56.41 # 20
Graph Classification SYNTHIE SPI-GCN Accuracy 71.00 # 1