AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

DisCo: Remedying Self-Supervised Learning on Lightweight Models with Distilled Contrastive Learning

Yuting Gao, Jia-Xin Zhuang, Shaohui Lin, Hao Cheng, Xing Sun, Ke Li, Chunhua Shen

ECCV 2022 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Data-efficient and robust self-supervision · 56 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 embedding distillation loss to MoCo-V2 (ablation)MoCo-V2 student · ImageNet linear eval (teacher ResNet-50)top-1 accuracy: EfficientNet-B0 46.8 -> 65.6, MobileNet-v3 36.2 -> 63.7, ResNet-18 52.2 -> 57.1Improved
add contrastive loss to distillation loss (ablation)DisCo (distillation only) · ImageNet linear eval (teacher ResNet-50)top-1 accuracy: EfficientNet-B0 65.6 -> 65.6, MobileNet-v3 58.9 -> 63.7, ResNet-18 54.5 -> 57.1Mixed
enlarge MLP hidden dimension from 512 to 2048 (ablation)DisCo · ImageNet linear eval (teacher ResNet-50)top-1 accuracy: EfficientNet-B0 62.5 -> 66.5, ResNet-18 57.1 -> 60.6Improved
keep a hidden layer in the MLP instead of removing it (ablation)DisCo · ImageNet linear eval (teacher ResNet-50)top-1 accuracy: EfficientNet-B0 52.5 -> 66.5, ResNet-18 52.5 -> 60.6Improved
expand hidden layer to fix Distilling Bottleneck on ResNet-18/34 (ablation)DisCo* · ImageNet linear eval (teacher ResNet-50)top-1 accuracy: ResNet-18 57.1 -> 60.6, ResNet-34 58.9 -> 62.5Improved
use ResNet-152 teacher instead of ResNet-50 (ablation)DisCo ResNet-34 student · ImageNet linear evaltop-1 accuracy: 62.5 -> 68.1Improved
distill final embedding (DisCo) instead of SEED queue similarityMobileNet-v3-Large student, ResNet-50 teacher · ImageNet linear evaltop-1 accuracy: 55.2 -> 64.4Improved
combine AT with DisCoDisCo · ImageNet linear evaltop-1 accuracy: EfficientNet-B0 66.5 -> 66.7, ResNet-18 60.6 -> 60.0No clear effect
combine KD with DisCoDisCo · ImageNet linear evaltop-1 accuracy: EfficientNet-B0 66.5 -> 65.8, MobileNet-v3 64.4 -> 65.2Mixed
use SwAV R50*2 teacher with 200 epochs vs SEED 800 epochsDisCo EfficientNet-B1 student · ImageNet linear evaltop-1 accuracy: DisCo 64.0 vs SEED 68.0; 290 epochs gives 70.4Hurt
apply DisCo to DINO ViT-tiny with ViT-small teacherDINO ViT-tiny · ImageNet linear evaltop-1 accuracy: 63.2 -> 68.4Improved
apply DisCo to DINO XCiT-tiny with XCiT-small teacherDINO XCiT-tiny · ImageNet linear evaltop-1 accuracy: 67.0 -> 71.1Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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