Improving Multi-Label Contrastive Learning by Leveraging Label Distribution

2025-01-31

AI Summary

To address the challenges of complex positive/negative sample selection and unmodeled label importance in multi-label contrastive learning, this paper proposes a simplified sample selection method grounded in label distribution modeling. It is the first to incorporate label distributions into multi-label contrastive learning frameworks, jointly reconstructing soft label distributions via RBF kernel mapping and contrastive loss—thereby explicitly capturing inter-label semantic relationships and enabling importance-aware representation learning. Crucially, the method obviates computationally expensive hard-negative mining and instead adaptively identifies positive/negative sample intersections solely from label co-occurrence structures. Extensive experiments across nine mainstream multi-label datasets—spanning both image and vector modalities—demonstrate consistent superiority over state-of-the-art methods. The approach achieves significant improvements across six core metrics, including mAP and Hamming Loss, validating its effectiveness, robustness, and cross-modal generalizability.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Computer Vision: Multi-modal Vision
• Data Mining & Knowledge Management: Mining of Visual, Multimedia & Multimodal Data

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

In multi-label learning, leveraging contrastive learning to learn better representations faces a key challenge: selecting positive and negative samples and effectively utilizing label information. Previous studies selected positive and negative samples based on the overlap between labels and used them for label-wise loss balancing. However, these methods suffer from a complex selection process and fail to account for the varying importance of different labels. To address these problems, we propose a novel method that improves multi-label contrastive learning through label distribution. Specifically, when selecting positive and negative samples, we only need to consider whether there is an intersection between labels. To model the relationships between labels, we introduce two methods to recover label distributions from logical labels, based on Radial Basis Function (RBF) and contrastive loss, respectively. We evaluate our method on nine widely used multi-label datasets, including image and vector datasets. The results demonstrate that our method outperforms state-of-the-art methods in six evaluation metrics.

Problem

• Multi-label Learning
• Contrastive Learning
• Label Importance

Innovation

• Label Distribution
• Multi-label Contrastive Learning
• Radial Basis Function