AI Paper Atlas

Research map / Graph learning / Graph algorithms and clustering

Learning Hierarchical Graph Neural Networks for Image Clustering

Yifan Xing, Tong He, Tianjun Xiao, Yongxin Wang, Yuanjun Xiong, Wei Xia, David Wipf, Zheng Zhang, Stefano Soatto

ICCV 2021 · Graph learning › Graph algorithms and clustering · approach: Neural and local graph clustering · 38 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 hierarchy (Hi-LANDER) to single-level LANDER (ablation)Hi-LANDER · HannahFb: 0.234 -> 0.700Improved
add hierarchy (Hi-LANDER) to single-level LANDER (ablation)Hi-LANDER · HannahNMI: 0.644 -> 0.810Improved
add hierarchy (Hi-LANDER) to single-level LANDER (ablation)Hi-LANDER · Hannahruntime (s): 44.9 -> 36.9Improved
use joint inference (single LANDER model) instead of two separate GNN models (ablation)LANDER · Hannahruntime (s): 256.2 -> 44.9Improved
use joint inference (single LANDER model) instead of two separate GNN models (ablation)LANDER · HannahFb: 0.224 -> 0.234Improved
early-stop agglomeration before full convergence (ablation)Hi-LANDER · clustering benchmarksFp/Fb/NMINo clear effect
train on a different cluster-size distribution than test (Hi-LANDER vs GCN-V+E)Hi-LANDER · HannahFb: 0.224 -> 0.700Improved
use Hi-LANDER pseudo labels instead of best baseline (GCN-V)face recognition · IJBC 1:1FNMR@FMR=1e-4: 0.181 -> 0.159Improved
use Hi-LANDER vs GCN-V on same-distribution iNat2018Hi-LANDER · iNat2018-Test (same distribution)Fb: 0.360 vs 0.350No clear effect
use Hi-LANDER instead of GCN-V on IMDB-Test-SameDistHi-LANDER · IMDB-Test-SameDistFp: 0.722 -> 0.793Improved

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