AI Paper Atlas

Research map / Graph learning / Graph algorithms and clustering

Non-Graph Data Clustering via $\mathcal {O}(n)$ Bipartite Graph Convolution

Hongyuan Zhang, Jiankun Shi, Rui Zhang, Xuelong Li

TPAMI 2022 · Graph learning › Graph algorithms and clustering · approach: Neural and local graph clustering · 18 citations (OpenAlex)

Read the paper

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 adaptive update of anchors and transition probability (ablation)AnchorGAE-L2 · MNIST-fullACC: 0.468 -> 0.833Improved
add adaptive update of anchors and transition probability (ablation)AnchorGAE-L2 · MNIST-fullNMI: 0.635 -> 0.773Improved
add dynamically increasing sparsity k (ablation)AnchorGAE-L2 · MNIST-fullACC: 0.289 -> 0.833Improved
add dynamically increasing sparsity k (ablation)AnchorGAE-L2 · MNIST-testACC: 0.284 -> 0.823Improved
add dynamically increasing sparsity k (ablation)AnchorGAE-L2 · SEGMENTNMI: 0.125 -> 0.613Improved
use weighted graph instead of KNN graph (ablation)AnchorGAE-L2 · MNIST-fullACC: 0.793 -> 0.833Improved
use weighted graph instead of KNN graph (ablation)AnchorGAE-L2 · Fashion-MNISTACC: 0.533 -> 0.645Improved
use weighted graph instead of KNN graph (ablation)AnchorGAE-L2 · ISOLETACC: 0.654 -> 0.663No clear effect
fix anchors (no update)AnchorGAE-L2 · Fashion-MNISTACC: 0.645 -> 0.588Hurt
use 4-layer instead of 2-layer convolutionAnchorGAE · USPSACC: 0.853 -> 0.781Hurt
use too large initial sparsity k0AnchorGAE · SEGMENT and MNIST-testACCHurt

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