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.

Multi-label semi-supervised learning method learnt from Hilbert-Schmidt independence criterion

Dual-Classifier Collaborative Method Based on Semi-Supervised Active Learning

Semi-Supervised Dual Relation Learning for Multi-Label Classification

Semi-Supervised Multi-Label Dimensionality Reduction Learning by Instance and Label Correlations

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

Dual-Decoupling Learning and Metric-Adaptive Thresholding for Semi-Supervised Multi-Label Learning

Semi-supervised Discriminant Analysis Based on Dependence Estimation

Learning From Semi-Supervised Weak-Label Data

Research on Multi-Label Semi-Supervised Learning Algorithm Based on Dual Selection Criteria

Multi-Label Feature Selection Method Combining Unbiased Hilbert-Schmidt Independence Criterion With Controlled Genetic Algorithm

Semi-Supervised Partial Multi-Label Classification via Consistency Learning

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

Multi-Label Learning With Local Similarity Of Samples

Common and Discriminative Semantic Pursuit for Multi-Modal Multi-Label Learning

Multilabel Dimensionality Reduction Via Dependence Maximization.

Multi-label feature selection based on HSIC and sparrow search algorithm

Label Specific Multi-Semantics Metric Learning for Multi-Label Classification: Global Consideration Helps

XIV. Instruments for Use through Cylindrical Rectal Specula, with the Patient in the Knee-Chest Posture.

Joint multi-label multi-instance learning for image classification

Graph-Based Semi-Supervised Learning with Multi-Label

Multi-label learning by instance differentiation