AI Paper Atlas

Research map / Graph learning / GNN architectures

Differentiable Graph Module (DGM) for Graph Convolutional Networks

Anees Kazi, Luca Cosmo, Seyed‐Ahmad Ahmadi, Nassir Navab, Michael M. Bronstein

TPAMI 2022 · Graph learning › GNN architectures · approach: Graph structure learning · 148 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add a DGM layer that learns the graph (ablation)GCN · Coraaccuracy: 80.90 -> 84.70Improved
add a DGM layer that learns the graph (ablation)GCN · CiteSeeraccuracy: 71.90 -> 74.70Improved
add a DGM layer that learns the graph (ablation)GCN · PubMedaccuracy: 86.40 -> 87.60Improved
stack a third DGM layer (ablation)dDGM · Coraaccuracy: 84.70 -> 82.40Hurt
use k=5 instead of k=20 neighbors (ablation)dDGM · Coraaccuracy: 70.60 -> 84.60Improved
use k=5 instead of k=1 neighbors (ablation)dDGM · CiteSeeraccuracy: 67.00 -> 74.80Improved
use Euclidean instead of hyperbolic latent space (ablation)dDGM · PubMedaccuracy: 86.60 -> 87.60Improved
use Euclidean instead of hyperbolic latent space (ablation)dDGM · Coraaccuracy: 84.40 -> 84.60No clear effect
use input graph with GCN embedding function instead of MLP (ablation)dDGM · Coraaccuracy: 63.30 -> 84.70Improved
use GCN diffusion instead of EdgeConv (ablation)dDGM · CiteSeeraccuracy: 52.20 -> 74.80Improved
use GCN diffusion instead of GAT (ablation)dDGM · Coraaccuracy: 81.30 -> 84.60Improved
add DGM before EdgeConv diffusion (ablation)EdgeConv · Coraaccuracy: 67.00 -> 71.60Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy