Table 9 Experimental results with and without \(L_{MAGI}\) loss.

From: Simple yet effective heuristic community detection with graph convolution network

  

Min DBI

Max Q

Max NMI

Max ACC

Max F1

Max ARI

Cora

With contrastive loss

0.501831

0.746360

0.544831

0.666913

0.638700

0.488040

Without contrastive loss

0.458970

0.765320

0.561596

0.669129

0.663200

0.469243

Acm

With contrastive loss

0.916360

0.728609

0.516628

0.744463

0.553700

0.544837

Without contrastive loss

0.684481

0.745318

0.620495

0.844300

0.670900

0.692292

Amap

With contrastive loss

0.457712

0.643710

0.608214

0.615294

0.623500

0.466393

Without contrastive loss

0.377126

0.670459

0.646764

0.685882

0.684800

0.506579

Uat

With contrastive loss

0.206296

0.167113

0.227279

0.510924

0.491400

0.206899

Without contrastive loss

0.807088

0.280636

0.249595

0.546218

0.578200

0.242806

Back to article page