AI Paper Atlas

Research map / Graph learning / Graph self-supervised learning

Globally Correlation-Aware Hard Negative Generation

Wenjie Peng, Hongxiang Huang, Tianshui Chen, Qiuping Ke, Gang Dai, Shuangping Huang

IJCV 2024 · Graph learning › Graph self-supervised learning · approach: Contrastive learning with negatives and views · 10 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 GCA-HNG hard negative generation to N-pair+ lossResNet-50 R512 · CUB-200-2011Recall@1: 66.5 -> 70.3Improved
add GCA-HNG to Proxy Anchor lossResNet-50 R512 · Cars196Recall@1: 88.4 -> 91.1Improved
add GCA-HNG to Proxy Anchor lossDINO-S DN384 · SOPRecall@1: 85.2 -> 85.9Improved
add global sample correlations (node message propagation) (ablation)GCA-HNG NP+ · CUB-200-2011Recall@1: 69.1 -> 70.3Improved
add global sample correlations (node message propagation) (ablation)GCA-HNG PA · CUB-200-2011Recall@1: 69.3 -> 72.3Improved
add Hadamard product addition of edge info in node propagation (ablation)GCA-HNG PA · CUB-200-2011Recall@1: 71.7 -> 72.3Improved
use channel-wise random-weighted interpolation instead of single coefficient (ablation)GCA-HNG NP+ · CUB-200-2011Recall@1: 68.0 -> 70.3Improved
combine interpolated representations by random weighting (ablation)GCA-HNG NP+ · CUB-200-2011Recall@1: 69.9 -> 70.3Improved
add cross-entropy loss J_ce to generation loss (ablation)GCA-HNG NP+ · CUB-200-2011Recall@1: 67.7 -> 70.3Improved
add similarity loss J_sim and diversity loss J_div (ablation)GCA-HNG PA · CUB-200-2011Recall@1: 71.3 -> 72.3Improved
add GNN loss without hard negative generation (ablation)Baseline NP+ · CUB-200-2011Recall@1: 66.5 -> 66.8No clear effect
add GNN module (training-time only)GCA-HNG NP+ · Training timesec/iter: 0.22 sec/iter -> 0.29 sec/iterHurt

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