AI Paper Atlas

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

Self-Supervised Representation Learning From Multi-Domain Data

Zeyu Feng, Chang Xu, Dacheng Tao

ICCV 2019 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Data-efficient and robust self-supervision · 48 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 MI constraints to naive multi-domain combination (DeepAll)RotNet on AlexNet · PACS average, conv5 and fc7top-1 accuracy: 66.2 -> 67.3 (conv5), 59.3 -> 64.5 (fc7)Improved
add MI constraints to naive combination with AETAET on AlexNet · PACS average, conv5 and fc7top-1 accuracy: 66.1 -> 68.8 (conv5), 52.1 -> 59.3 (fc7)Improved
train on all four PACS domains naively instead of a single domainRotNet · PACS average, fc7top-1 accuracy: 59.3 vs 58.0 (Art), 61.5 (Cartoon)Mixed
add Places to ImageNet naively for pretrainingRotNet · ImageNet val, conv5top-1 accuracy: 31.9 -> 31.6Hurt
add MI constraints to ImageNet+Places combinationRotNet · ImageNet and Places average, fc7top-1 accuracy: 25.0 -> 28.9Improved
add MI constraints to ImageNet+Places pretrainingRotNet · PASCAL VOC 2007, fc7mAP: 73.8 -> 75.6Improved
add PASCAL training images to ImageNet pretrainingRotNet · PASCAL VOC 2007, conv5mAP: 74.3 -> 74.8Improved
add MI constraints to ImageNet+PASCAL pretrainingRotNet · PASCAL VOC 2007mAP: 74.8 -> 74.8 (conv5), 73.2 -> 75.3 (fc7)Mixed
add domain-invariance constraint only (ablation)DeepAll RotNet · PACS average, conv5 and fc7top-1 accuracy: 66.2 -> 67.0 (conv5), 59.3 -> 59.1 (fc7)Mixed
add domain-specific constraint only (ablation)DeepAll RotNet · PACS average, conv5 and fc7top-1 accuracy: 59.3 -> 62.8 (fc7), 66.2 -> 66.3 (conv5)Improved
combine both constraints (lambda_u=0.1, lambda_l=0.1) (ablation)DeepAll RotNet · PACS average, conv5 and fc7top-1 accuracy: 66.2 -> 67.2 (conv5), 59.3 -> 62.5 (fc7)Improved
raise lambda_l from 0.1 to 1 with lambda_u 0.1 (ablation)Full model · PACS average, fc7top-1 accuracy: 62.5 -> 64.5Improved

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