AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Deep Companion Learning: Enhancing Generalization Through Historical Consistency

Ruizhao Zhu, Venkatesh Saligrama

ECCV 2024 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory · 0 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 DCL (companion consistency regularizer) to cross-entropy trainingResNet18 · CIFAR-100accuracy: 78.3 -> 82.4Improved
add DCL to cross-entropy trainingSwin-T · ImageNet-1Kaccuracy: 81.3 -> 81.7Improved
add DCL to cross-entropy trainingResNet50 · ImageNet-1Kaccuracy: 76.0 -> 77.1Improved
train from scratch with DCL instead of ImageNet pre-training then fine-tuningResNet18 · CIFAR-100accuracy: 80.5 -> 82.4Improved
add DCL on top of FixMatch with 40 labelsFixMatch WRN-28-2 · CIFAR-10 (40 labels)accuracy: 86.2 -> 88.7Improved
add DCL to MAE pre-training and fine-tuneMAE ViT-B · Stanford Caraccuracy: 82.5 -> 87.2Improved
add DCL to knowledge distillation (basic KD)ResNet8x4 student · CIFAR-100accuracy: 74.0 -> 75.2Improved
use MSE instead of KL-Divergence, InfoNCE or L1 as consistency distance (ablation)DCL · CIFAR-100accuracy: 81.5 -> 82.4Improved
set EMA alpha to 0.6 instead of extremes (0 or 0.99) (ablation)DCL companion update · CIFAR-100accuracy: 80.0 -> 82.4Improved
use fixed alpha instead of time-varying alpha_t=(t-1)/t (ablation)DCL · CIFAR-100accuracy: 81.5 -> 82.4Improved
use a smaller companion model (ResNet10-xs) instead of ResNet18 (ablation)DCL · CIFAR-100accuracy: 82.4 -> 80.2Hurt
train companion on 10% of data instead of 100% (ablation)DCL · CIFAR-100accuracy: 82.4 -> 80.7Hurt

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