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 Learning Vector Quantization for Semi-Supervised Classification

Distributed Semi-supervised Multi-label Classification with Quantized Communication

SemiCVT: Semi-Supervised Convolutional Vision Transformer for Semantic Segmentation

Semi-Supervised Dual Relation Learning for Multi-Label Classification

Vector-Valued Multi-View Semi-Supervsed Learning for Multi-Label Image Classification.

An Online Incremental Learning Vector Quantization

Learning From Semi-Supervised Weak-Label Data

Quantum Statistic Based Semi-Supervised Learning Approach for Industrial Soft Sensor Development

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

Multi-Modal Curriculum Learning for Semi-Supervised Image Classification

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

Deep dual incomplete multi-view multi-label classification via label semantic-guided contrastive learning

Efficient multi-class unlabeled constrained semi-supervised SVM.

Parallel and Sequential Support Vector Machines for Multi-label Classification

Semi-supervised Question Classification with Jointly Learning Question and Answer Representations

Multi-Modal Sentiment Classification With Independent and Interactive Knowledge via Semi-Supervised Learning

Query-Based Knowledge Sharing for Open-Vocabulary Multi-Label Classification

Semi-supervised support vector learning for face recognition

Graph-Based Semi-Supervised Learning with Multi-Label

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

Semi-Supervised Learning for Video Captioning.