AI Paper Atlas

Research map / Label-efficient and robust learning / Semi-supervised learning

Probabilistic Contrastive Learning for Long-Tailed Visual Recognition

Chaoqun Du, Yulin Wang, Shiji Song, Gao Ming Huang

TPAMI 2024 · Label-efficient and robust learning › Semi-supervised learning · approach: Confidence-based pseudo-label selection · 127 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
model class features with vMF distribution instead of EMA prototype (ablation)ProCo · CIFAR-100-LT, imbalance 100 (200 epochs)top-1 accuracy: 51.6 -> 52.8Improved
model class features with vMF instead of normal distribution (ablation)ProCo · CIFAR-100-LT, imbalance 100 (200 epochs)top-1 accuracy: 52.1 -> 52.8Improved
use vMF instead of centroid prototype (ablation)ProCo · CIFAR-100-LT, imbalance 100 (200 epochs)top-1 accuracy: 52.0 -> 52.8Improved
add supervised contrastive loss to ProCo (ablation)ProCo · CIFAR-100-LT, imbalance 100 (200 epochs)top-1 accuracy: 52.8 -> 52.6Hurt
add logit adjustment to learnable class representation (ablation)ProCo · CIFAR-100-LT, imbalance 100 (200 epochs)top-1 accuracy: 50.5 -> 50.6No clear effect
use L_in instead of L_out (ablation)ProCo · CIFAR-100-LT, imbalance 100top-1 accuracy: 52.0 -> 52.8Improved
use L_in instead of L_out (ablation)ProCo · CIFAR-100-LT, imbalance 10top-1 accuracy: 65.1 -> 65.5Improved
add AutoAugment (ablation)ProCo · CIFAR-100-LT, imbalance 100 (200 epochs)top-1 accuracy: 50.5 -> 52.8Improved
add AutoAugment (ablation)BCL · CIFAR-100-LT, imbalance 100 (200 epochs)top-1 accuracy: 50.1 -> 51.9Improved
add ProCo to Faster R-CNNFaster R-CNN · LVIS v1AP_b: 22.1 -> 24.7Improved
add ProCo to Mask R-CNNMask R-CNN · LVIS v1AP_m: 21.7 -> 24.7Improved
add ProCo to NCL ensembleNCL · ImageNet-LT (180 epochs)top-1 accuracy: 59.2 -> 60.2Improved

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