Conference paper Open Access

Neighborhood Contrastive Learning for Novel Class Discovery

Zhun Zhong; Enrico Fini; Subhankar Roy; Zhiming Luo; Elisa Ricci; Nicu Sebe

In this paper, we address Novel Class Discovery (NCD), the task of unveiling new classes in a set of unlabeled samples given a labeled dataset with known classes. We exploit the peculiarities of NCD to build a new framework, named Neighborhood Contrastive Learning (NCL), to learn discriminative representations that are important to clustering performance. Our contribution is twofold. First, we find that a feature extractor trained on the labeled set generates representations in which a generic query sample and
its neighbors are likely to share the same class. We exploit this observation to retrieve and aggregate pseudo-positive pairs with contrastive learning, thus encouraging the model to learn more discriminative representations. Second, we notice that most of the instances are easily discriminated by the network, contributing less to the contrastive loss. To overcome this issue, we propose to generate hard negatives by mixing labeled and unlabeled samples in the feature space. We experimentally demonstrate that these two ingredients significantly contribute to clustering performance and lead our model to outperform state-of-the-art methods by a large margin (e.g., clustering accuracy +13% on CIFAR-100 and +8% on ImageNet).

IEEE/CVF International Conference on Computer Vision and Pattern Recognition (CVPR'21)
Files (2.6 MB)
Name Size
Zhong_Neighborhood_Contrastive_Learning_for_Novel_Class_Discovery_CVPR_2021_paper.pdf
md5:b3d0f396ca90899d1a104b2a6c603540
2.6 MB Download
14
9
views
downloads
All versions This version
Views 1414
Downloads 99
Data volume 23.7 MB23.7 MB
Unique views 1111
Unique downloads 77

Share

Cite as