</> Code with Papers by Syntology

Papers › Twin Contrastive Learning for Online Clustering

Twin Contrastive Learning for Online Clustering

21 Oct 2022arXiv:2210.11680archive 2025-07-28

Yunfan Li, Mouxing Yang, Dezhong Peng, Taihao Li, Jiantao Huang, Xi Peng

This paper proposes to perform online clustering by conducting twin contrastive learning (TCL) at the instance and cluster level. Specifically, we find that when the data is projected into a feature space with a dimensionality of the target cluster number, the rows and columns of its feature matrix correspond to the instance and cluster representation, respectively. Based on the observation, for a given dataset, the proposed TCL first constructs positive and negative pairs through data augmentations. Thereafter, in the row and column space of the feature matrix, instance- and cluster-level contrastive learning are respectively conducted by pulling together positive pairs while pushing apart the negatives. To alleviate the influence of intrinsic false-negative pairs and rectify cluster assignments, we adopt a confidence-based criterion to select pseudo-labels for boosting both the instance- and cluster-level contrastive learning. As a result, the clustering performance is further improved. Besides the elegant idea of twin contrastive learning, another advantage of TCL is that it could independently predict the cluster assignment for each instance, thus effortlessly fitting online scenarios. Extensive experiments on six widely-used image and text benchmarks demonstrate the effectiveness of TCL. The code will be released on GitHub.

PaperPDFCode

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

Code

XLearning-SCU/2022-IJCV-TCL officialpytorch report
Yunfan-Li/Twin-Contrastive-Learning pytorch 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

ClusteringContrastive LearningDeep ClusteringImage ClusteringOnline ClusteringShort Text Clustering

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Clustering CIFAR-10 TCL ARI 0.780 #16 of 40 Archive leaderboard report
Image Clustering CIFAR-10 TCL Accuracy 0.887 #16 of 40 Archive leaderboard report
Image Clustering CIFAR-10 TCL Backbone ResNet-34 #16 of 40 Archive leaderboard report
Image Clustering CIFAR-10 TCL NMI 0.819 #16 of 40 Archive leaderboard report
Image Clustering CIFAR-10 TCL Train set Train #16 of 40 Archive leaderboard report
Image Clustering CIFAR-100 TCL ARI 0.357 #11 of 30 Archive leaderboard report
Image Clustering CIFAR-100 TCL Accuracy 0.531 #11 of 30 Archive leaderboard report
Image Clustering CIFAR-100 TCL NMI 0.529 #11 of 30 Archive leaderboard report
Image Clustering CIFAR-100 TCL Train Set Train #11 of 30 Archive leaderboard report
Image Clustering ImageNet-10 TCL ARI 0.837 #10 of 18 Archive leaderboard report
Image Clustering ImageNet-10 TCL Accuracy 0.895 #10 of 18 Archive leaderboard report
Image Clustering ImageNet-10 TCL NMI 0.875 #10 of 18 Archive leaderboard report
Image Clustering Imagenet-dog-15 TCL ARI 0.516 #9 of 20 Archive leaderboard report
Image Clustering Imagenet-dog-15 TCL Accuracy 0.644 #9 of 20 Archive leaderboard report
Image Clustering Imagenet-dog-15 TCL NMI 0.623 #9 of 20 Archive leaderboard report
Image Clustering STL-10 TCL ARI 0.757 #8 of 29 Archive leaderboard report
Image Clustering STL-10 TCL Accuracy 0.868 #8 of 29 Archive leaderboard report
Image Clustering STL-10 TCL Backbone ResNet-34 #8 of 29 Archive leaderboard report
Image Clustering STL-10 TCL NMI 0.799 #8 of 29 Archive leaderboard report
Image Clustering STL-10 TCL Train Split Train #8 of 29 Archive leaderboard report
Short Text Clustering Biomedical TCL Acc 49.8 #1 of 4 Archive leaderboard report
Short Text Clustering Biomedical TCL NMI 42.9 #1 of 4 Archive leaderboard report
Short Text Clustering Stackoverflow TCL Acc 88.2 #1 of 5 Archive leaderboard report
Short Text Clustering Stackoverflow TCL NMI 0.786 #1 of 5 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

Contrastive Learning

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