{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/semi-supervised-vision-transformers-at-scale","title":"Semi-supervised Vision Transformers at Scale","arxiv_id":"2208.05688","date":"2022-08-11","proceeding":null,"authors":["Zhaowei Cai","Avinash Ravichandran","Paolo Favaro","Manchen Wang","Davide Modolo","Rahul Bhotika","Zhuowen Tu","Stefano Soatto"],"abstract":"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.","url_abs":"https://arxiv.org/abs/2208.05688v1","url_pdf":"https://arxiv.org/pdf/2208.05688v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"semi-supervised-vision-transformers-at-scale","repo_url":"https://github.com/amazon-science/semi-vit","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"inductive-bias","task_name":"Inductive Bias"},{"task_slug":"semi-supervised-image-classification","task_name":"Semi-Supervised Image Classification"}],"methods":[{"method_slug":"1x1-convolution","method_name":"1x1 Convolution"},{"method_slug":"average-pooling","method_name":"Average Pooling"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"fixmatch","method_name":"FixMatch"},{"method_slug":"inception-a","method_name":"Inception-A"},{"method_slug":"inception-b","method_name":"Inception-B"},{"method_slug":"inception-c","method_name":"Inception-C"},{"method_slug":"inception-v4","method_name":"Inception-v4"},{"method_slug":"max-pooling","method_name":"Max Pooling"},{"method_slug":"mixup","method_name":"Mixup"},{"method_slug":"reduction-a","method_name":"Reduction-A"},{"method_slug":"reduction-b","method_name":"Reduction-B"},{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/semi-supervised-image-classification-on-1","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 1% labeled data","model":"Semi-ViT (ViT-Huge)","rank_in_archive_order":7,"of":65,"metrics":{"Top 1 Accuracy":"80%","Top 5 Accuracy":"93.1"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-1","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 1% labeled data","model":"Semi-ViT (ViT-Large)","rank_in_archive_order":8,"of":65,"metrics":{"Top 1 Accuracy":"77.3%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-1","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 1% labeled data","model":"Semi-ViT (ViT-Base)","rank_in_archive_order":19,"of":65,"metrics":{"Top 1 Accuracy":"71%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-2","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 10% labeled data","model":"Semi-ViT (ViT-Huge)","rank_in_archive_order":6,"of":75,"metrics":{"Top 1 Accuracy":"84.3%","Top 5 Accuracy":"96.6%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-2","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 10% labeled data","model":"Semi-ViT (ViT-Large)","rank_in_archive_order":7,"of":75,"metrics":{"Top 1 Accuracy":"83.3%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-2","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 10% labeled data","model":"Semi-ViT (ViT-Base)","rank_in_archive_order":13,"of":75,"metrics":{"Top 1 Accuracy":"79.7%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-image-classification-on-2","task":"Semi-Supervised Image Classification","dataset":"ImageNet - 10% labeled data","model":"Semi-ViT (ViT-Small)","rank_in_archive_order":21,"of":75,"metrics":{"Top 1 Accuracy":"77.1%"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2208.05688","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}