Papers › Semi-supervised Vision Transformers at Scale

Semi-supervised Vision Transformers at Scale

11 Aug 2022arXiv:2208.05688archive 2025-07-28

Zhaowei Cai, Avinash Ravichandran, Paolo Favaro, Manchen Wang, Davide Modolo, Rahul Bhotika, Zhuowen Tu, Stefano Soatto

We study semi-supervised learning (SSL) for vision transformers (ViT), an under-explored topic despite the wide adoption of the ViT architectures to different tasks. To tackle this problem, we propose a new SSL pipeline, consisting of first un/self-supervised pre-training, followed by supervised fine-tuning, and finally semi-supervised fine-tuning. At the semi-supervised fine-tuning stage, we adopt an exponential moving average (EMA)-Teacher framework instead of the popular FixMatch, since the former is more stable and delivers higher accuracy for semi-supervised vision transformers. In addition, we propose a probabilistic pseudo mixup mechanism to interpolate unlabeled samples and their pseudo labels for improved regularization, which is important for training ViTs with weak inductive bias. Our proposed method, dubbed Semi-ViT, achieves comparable or better performance than the CNN counterparts in the semi-supervised classification setting. Semi-ViT also enjoys the scalability benefits of ViTs that can be readily scaled up to large-size models with increasing accuracies. For example, Semi-ViT-Huge achieves an impressive 80% top-1 accuracy on ImageNet using only 1% labels, which is comparable with Inception-v4 using 100% ImageNet labels.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

amazon-science/semi-vit mentioned on GitHubpytorchNOASSERTION 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

Inductive BiasSemi-Supervised Image Classification

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Semi-Supervised Image Classification ImageNet - 1% labeled data Semi-ViT (ViT-Huge) Top 1 Accuracy 80% #7 of 65 Archive leaderboard report
Semi-Supervised Image Classification ImageNet - 1% labeled data Semi-ViT (ViT-Huge) Top 5 Accuracy 93.1 #7 of 65 Archive leaderboard report
Semi-Supervised Image Classification ImageNet - 1% labeled data Semi-ViT (ViT-Large) Top 1 Accuracy 77.3% #8 of 65 Archive leaderboard report
Semi-Supervised Image Classification ImageNet - 1% labeled data Semi-ViT (ViT-Base) Top 1 Accuracy 71% #19 of 65 Archive leaderboard report
Semi-Supervised Image Classification ImageNet - 10% labeled data Semi-ViT (ViT-Huge) Top 1 Accuracy 84.3% #6 of 75 Archive leaderboard report
Semi-Supervised Image Classification ImageNet - 10% labeled data Semi-ViT (ViT-Huge) Top 5 Accuracy 96.6% #6 of 75 Archive leaderboard report
Semi-Supervised Image Classification ImageNet - 10% labeled data Semi-ViT (ViT-Large) Top 1 Accuracy 83.3% #7 of 75 Archive leaderboard report
Semi-Supervised Image Classification ImageNet - 10% labeled data Semi-ViT (ViT-Base) Top 1 Accuracy 79.7% #13 of 75 Archive leaderboard report
Semi-Supervised Image Classification ImageNet - 10% labeled data Semi-ViT (ViT-Small) Top 1 Accuracy 77.1% #21 of 75 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.

Methods

1x1 ConvolutionAverage PoolingConvolutionDropoutFixMatchInception-AInception-BInception-CInception-v4Max PoolingMixupReduction-AReduction-BSoftmax

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