Abstract:In a real-world scenario, an object could contain multiple tags instead of a single categorical label. To this end, multi-label learning (MLL) emerged. In MLL, the feature distributions are long-tailed and the complex semantic label relation and the long-tailed training samples are the main challenges. Semi-supervised learning is a potential solution. While, existing methods are mainly designed for single class scenario while ignoring the latent label relations. In addition, they cannot well handle the distribution shift commonly existing across source and target domains. To this end, a Semi-supervised Dual Relation Learning (SDRL) framework for multi-label classification is proposed. SDRL utilizes a few labeled samples as well as large scale unlabeled samples in the training stage. It jointly explores the inter-instance feature-level relation and the intra-instance label-level relation even from the unlabeled samples. In our model, a dual-classifier structure is deployed to obtain domain invariant representations. The prediction results from the classifiers are further compared and the most confident predictions are extracted as pseudo labels. A trainable label relation tensor is designed to explicitly explore the pairwise latent label relations and refine the predicted labels. SDRL is able to effectively and efficiently explore the feature-label relation as well as the label-label relation knowledge without any extra semantic knowledge. We evaluated SDRL in general and zero-shot multi-label classification tasks and we concluded that SDRL is superior to other SOTA baselines. Furthermore, extensive ablation studies have been done which reveal the effectiveness of each component in our framework.

Semi-Supervised Partial Multi-Label Classification via Consistency Learning

Semi-Supervised Dual Relation Learning for Multi-Label Classification

Partial multi-label learning via semi-supervised subspace collaboration

Dlsa: Semi-supervised Partial Label Learning Via Dependence-Maximized Label Set Assignment

Partial Multi-label Learning with Label and Feature Collaboration

Learning From Semi-Supervised Weak-Label Data

Partial-Label Regression

Multi-label classification with weak labels by learning label correlation and label regularization

Partial Label Learning with Unlabeled Data.

Conditional Consistency Regularization for Semi-Supervised Multi-label Image Classification

Learning from Multi-Dimensional Partial Labels.

MixLab: An Informative Semi-supervised Method for Multi-label Classification

Controller-Guided Partial Label Consistency Regularization with Unlabeled Data

PLMCL: Partial-Label Momentum Curriculum Learning for Multi-Label Image Classification

Combining Supervised Learning and Reinforcement Learning for Multi-Label Classification Tasks with Partial Labels

Exploiting Unlabeled Data via Partial Label Assignment for Multi-Class Semi-Supervised Learning

Semi-supervised Contrastive Learning Using Partial Label Information

On the consistency of multi-label learning

Semi-supervised imbalanced multi-label classification with label propagation

A survey of multi-label classification based on supervised and semi-supervised learning

Semisupervised deep learning using consistency regularization and pseudolabels for hyperspectral image classification