Papers › Dual Student: Breaking the Limits of the Teacher in Semi-supervised Learning

Dual Student: Breaking the Limits of the Teacher in Semi-supervised Learning

3 Sep 2019ICCV 2019 10arXiv:1909.01804archive 2025-07-28

Zhanghan Ke, Daoye Wang, Qiong Yan, Jimmy Ren, Rynson W. H. Lau

Recently, consistency-based methods have achieved state-of-the-art results in semi-supervised learning (SSL). These methods always involve two roles, an explicit or implicit teacher model and a student model, and penalize predictions under different perturbations by a consistency constraint. However, the weights of these two roles are tightly coupled since the teacher is essentially an exponential moving average (EMA) of the student. In this work, we show that the coupled EMA teacher causes a performance bottleneck. To address this problem, we introduce Dual Student, which replaces the teacher with another student. We also define a novel concept, stable sample, following which a stabilization constraint is designed for our structure to be trainable. Further, we discuss two variants of our method, which produce even higher performance. Extensive experiments show that our method improves the classification performance significantly on several main SSL benchmarks. Specifically, it reduces the error rate of the 13-layer CNN from 16.84% to 12.39% on CIFAR-10 with 1k labels and from 34.10% to 31.56% on CIFAR-100 with 10k labels. In addition, our method also achieves a clear improvement in domain adaptation.

PaperPDFConference PDFCode

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

Code

ZHKKKe/DualStudent officialmentioned on GitHubpytorch report
60972823l/SSL-DNLL mentioned 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

Semi-Supervised Image ClassificationUnsupervised Domain Adaptation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Semi-Supervised Image Classification CIFAR-10, 1000 Labels Dual Student (600) Accuracy 85.83 #5 of 9 Archive leaderboard report
Semi-Supervised Image Classification CIFAR-10, 2000 Labels Dual Student (600) Accuracy 89.28 #3 of 4 Archive leaderboard report
Semi-Supervised Image Classification CIFAR-10, 4000 Labels Dual Student (600) Percentage error 8.89 #39 of 49 Archive leaderboard report
Semi-Supervised Image Classification ImageNet - 10% labeled data Dual Student Top 1 Accuracy 63.52% #46 of 75 Archive leaderboard report
Semi-Supervised Image Classification ImageNet - 10% labeled data Dual Student Top 5 Accuracy 83.58% #46 of 75 Archive leaderboard report
Semi-Supervised Image Classification SVHN, 250 Labels Dual Student Accuracy 95.76 #10 of 15 Archive leaderboard report
Semi-Supervised Image Classification SVHN, 500 Labels Dual Student Accuracy 96.04 #4 of 6 Archive leaderboard report
Semi-Supervised Image Classification cifar-100, 10000 Labels Dual Student (480) Percentage error 32.77 #25 of 29 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