Skip to main content

Table 11 Performance of ML-GAT over the Koumbia multilayer networks with varying number of layers, and different types of input features

From: Graph convolutional and attention models for entity classification in multilayer networks

\(\ell\)

Koumbia-Fon

Koumbia-Foff

Koumbia-normal

Accuracy

MRR

Accuracy

MRR

Accuracy

MRR

2

\(\mathbf {0.9082 \pm 0.0113}\)

\(\mathbf {0.9541\pm 0.0057}\)

\(0.7426 \pm 0.0200\)

\(0.8713 \pm 0.0100\)

\(0.7556 \pm 0.0164\)

\(0.8778 \pm 0.0082\)

5

\(\mathbf {0.9373 \pm 0.0054}\)

\(\mathbf {0.9687 \pm 0.0027}\)

\(0.7656 \pm 0.0117\)

\(0.8828 \pm 0.0059\)

\(0.8355 \pm 0.01153\)

\(0.9193 \pm 0.0076\)

10

\(\mathbf {0.9385 \pm 0.0073}\)

\(\mathbf {0.9693 \pm 0.0036}\)

\(0.7588 \pm 0.0126\)

\(0.8794 \pm 0.0063\)

\(0.7874 \pm 0.0116\)

\(0.8937 \pm 0.0058\)

15

\(\mathbf {0.9355 \pm 0.0089}\)

\(\mathbf {0.9687 \pm 0.0044}\)

\(0.7522 \pm 0.0125\)

\(0.8761 \pm 0.0063\)

\(0.7723 \pm 0.0133\)

\(0.8826 \pm 0.0067\)

20

\(\mathbf {0.9404 \pm 0.0069}\)

\(\mathbf {0.9702 \pm 0.0035}\)

\(0.7488 \pm 0.0167\)

\(0.8744 \pm 0.0084\)

\(0.7706 \pm 0.0091\)

\(0.8853 \pm 0.0046\)

  1. Bold values correspond to best performances