</> Code with Papers by Syntology

Papers › TransBoost: Improving the Best ImageNet Performance using Deep Transduction

TransBoost: Improving the Best ImageNet Performance using Deep Transduction

26 May 2022arXiv:2205.13331archive 2025-07-28

Omer Belhasin, Guy Bar-Shalom, Ran El-Yaniv

This paper deals with deep transductive learning, and proposes TransBoost as a procedure for fine-tuning any deep neural model to improve its performance on any (unlabeled) test set provided at training time. TransBoost is inspired by a large margin principle and is efficient and simple to use. Our method significantly improves the ImageNet classification performance on a wide range of architectures, such as ResNets, MobileNetV3-L, EfficientNetB0, ViT-S, and ConvNext-T, leading to state-of-the-art transductive performance. Additionally we show that TransBoost is effective on a wide variety of image classification datasets. The implementation of TransBoost is provided at: https://github.com/omerb01/TransBoost .

PaperPDFCode

Code

omerb01/transboost officialmentioned in papermentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Image ClassificationTransductive Learningimage-classification

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Classification CIFAR-10 TransBoost-ResNet50 Percentage correct 97.61 #76 of 265 Archive leaderboard report
Image Classification DTD TransBoost-ResNet50 Accuracy 76.49 #9 of 11 Archive leaderboard report
Image Classification FGVC Aircraft TransBoost-ResNet50 Accuracy 83.80% #1 of 1 Archive leaderboard report
Image Classification Flowers-102 TransBoost-ResNet50 Accuracy 97.85% #34 of 52 Archive leaderboard report
Image Classification Food-101 TransBoost-ResNet50 Accuracy (%) 84.30 #7 of 11 Archive leaderboard report
Image Classification ImageNet TransBoost-ViT-S Number of params 22.05M #402 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ViT-S Top 1 Accuracy 83.67% #402 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ConvNext-T Number of params 28.59M #535 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ConvNext-T Top 1 Accuracy 82.46% #535 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-Swin-T Number of params 71.71M #570 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-Swin-T Top 1 Accuracy 82.16% #570 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ResNet50-StrikesBack Number of params 25.56M #657 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ResNet50-StrikesBack Top 1 Accuracy 81.15% #657 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ResNet152 Number of params 60.19M #690 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ResNet152 Top 1 Accuracy 80.64% #690 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ResNet101 Number of params 44.55M #732 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ResNet101 Top 1 Accuracy 79.86% #732 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ResNet50 Top 1 Accuracy 79.03% #788 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-EfficientNetB0 Number of params 5.29M #820 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-EfficientNetB0 Top 1 Accuracy 78.60% #820 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-MobileNetV3-L Number of params 5.48M #896 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-MobileNetV3-L Top 1 Accuracy 76.81% #896 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ResNet34 Number of params 21.8M #906 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ResNet34 Top 1 Accuracy 76.70% #906 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ResNet18 Number of params 11.69M #987 of 1060 Archive leaderboard report
Image Classification ImageNet TransBoost-ResNet18 Top 1 Accuracy 73.36% #987 of 1060 Archive leaderboard report
Image Classification SUN397 TransBoost-ResNet50 Accuracy 95.94% #1 of 1 Archive leaderboard report
Image Classification Stanford Cars TransBoost-ResNet50 Accuracy 90.80% #13 of 24 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections