AI Paper Atlas

Research map / Graph learning / GNN architectures

Revisiting 2D Convolutional Neural Networks for Graph-Based Applications

Yecheng Lyu, Xinming Huang, Ziming Zhang

TPAMI 2021 · Graph learning › GNN architectures · approach: Deep graph convolution and simplification · 12 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 vertex separation penalty (lambda>0) (ablation)GPGL · MUTAG (21x aug)accuracy (%): 80.51 -> 86.31Improved
use alpha=1.25 instead of 1.50 (ablation)GPGL · MUTAG (21x aug)accuracy (%): 83.04 -> 86.31Improved
use alpha=1.25 instead of 1.00 (ablation)GPGL · MUTAG (21x aug)accuracy (%): 85.14 -> 86.31Improved
use lambda=1000 instead of 200 or 5000 (ablation)GPGL · MUTAG (21x aug)accuracy (%): 85.26 -> 86.31Improved
add 2D max-pooling to PointNet on grid layoutPointNet · MUTAG (101x aug)accuracy (%): 66.55 -> 86.23Improved
use MSM-CNN instead of ResNet50 on GPGLGPGL · MUTAG (101x aug)accuracy (%): 91.02 -> 94.18Improved
add 2D max-pooling and upsampling to PointNet on H-GPGLPointNet · ShapeNetmcIoU (%): 80.1 -> 82.0Improved
use Delaunay triangulation instead of 5-NN graphH-GPGL MSM-CNN · ShapeNetmcIoU (%): 82.5 -> 83.8Improved
use circular initialization instead of spectralGPGL · MUTAGvertex loss ratio (%): 10.09 -> 1.06Improved
use hierarchical GPGL with 16 or 32 cuts instead of GPGL aloneH-GPGL · 2048-node graphrunning time: 90s -> 5.5sImproved
increase data augmentation of grid layouts from 1x to 101xGPGL MSM-CNN · MUTAGaccuracyImproved
overlapped points in grid cellsH-GPGL MSM-CNN · ShapeNetmcIoU (%): -0.22 mcIoU, -0.21 miIoUHurt

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