AI Paper Atlas

Research map / Point clouds / Point cloud classification

VBA: Vector Bundle Attention for Intrinsically Geometric Representation Learning

Shenglei Fang, Xianfang Sun, You Zhou

ICML 2026 · Point clouds › Point cloud classification · approach: Equivariant point cloud networks

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
replace standard attention with vector bundle attention (ablation)VBA-T · ModelNet40overall accuracy: 82.6% -> 92.9%Improved
replace standard attention with vector bundle attention (ablation)VBA-T · PBMCoverall accuracy: 66.76% -> 78.71%Improved
transport-then-attend geometry instead of additive geometric bias (GBT) (ablation)VBA-T · PBMCoverall accuracy: 70.32% -> 78.71%Improved
transport-then-attend geometry instead of additive geometric bias (GBT) (ablation)VBA-T · ModelNet40overall accuracy: 85.8% -> 92.9%Improved
use VBA instead of efficient MQA or GQA attention variants (ablation)VBA-T · PancreasF1: 0.67 (MQA-T), 0.66 (GQA-T) -> 0.86Improved
add self-supervised pretraining on HCL (ablation)VBA-SC · PBMCoverall accuracy: 78.71% -> 80.68%Improved
add self-supervised pretraining on HCL (ablation)VBA-SC · Pancreasoverall accuracy: 93.89% -> 97.64%Improved
pretrain scGPT and scBERT on same HCL subset as VBA-SC (ablation)scGPT, scBERT · PBMCoverall accuracy: scGPT 75.47% -> 74.42%; scBERT 75.52% -> 74.16%Hurt
restrict attention to top-k neighbors on learned manifold (ablation)VBA-SC · PBMCaccuracy: 78.71 -> 77.65 accuracy; 876 -> 560 s/epochMixed
use low-rank transport generators (ablation)VBA-SC · PBMCaccuracy: 78.71 -> 76.90 accuracy; 876 -> 529 s/epochMixed
use VBA-P on point clouds vs specialized modelsVBA-P · ModelNet40overall accuracy: 92.9% vs 93.8% (PointGA)No clear effect
VBA-ST vs graph-based spatial clustering methodsVBA-ST · DLPFC 12 slicesaverage ARI: 0.45 (SpaGCN), 0.43 (GraphST) -> 0.50Improved

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