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

2024-07-26

European Conference on Computer Vision

AI Summary

To address the performance degradation in semi-supervised multi-label learning caused by low-quality pseudo-labels, this paper proposes a dual-decoupled learning framework with a metric-adaptive thresholding mechanism. Methodologically: (1) a dual-view network is designed to decouple feature correlation from discriminability, enhancing prediction reliability; (2) a class-adaptive dynamic thresholding strategy, optimized via labeled data metrics, replaces fixed thresholds to better accommodate label imbalance and uncertainty; (3) consistency regularization, multi-label pseudo-label selection, and reweighting are jointly integrated to refine pseudo-label quality. Evaluated on multiple benchmark datasets, the method achieves state-of-the-art performance, improving average macro-F1 by 3.2–5.8 percentage points over existing approaches. It significantly enhances pseudo-label accuracy and model generalization under limited supervision.

Technology Category

• Machine Learning: Semi-Supervised Learning
• Search and Optimization: Learning to Search
• Computer Vision: Learning & Optimization for CV

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Semi-supervised multi-label learning (SSMLL) is a powerful framework for leveraging unlabeled data to reduce the expensive cost of collecting precise multi-label annotations. Unlike semi-supervised learning, one cannot select the most probable label as the pseudo-label in SSMLL due to multiple semantics contained in an instance. To solve this problem, the mainstream method developed an effective thresholding strategy to generate accurate pseudo-labels. Unfortunately, the method neglected the quality of model predictions and its potential impact on pseudo-labeling performance. In this paper, we propose a dual-perspective method to generate high-quality pseudo-labels. To improve the quality of model predictions, we perform dual-decoupling to boost the learning of correlative and discriminative features, while refining the generation and utilization of pseudo-labels. To obtain proper class-wise thresholds, we propose the metric-adaptive thresholding strategy to estimate the thresholds, which maximize the pseudo-label performance for a given metric on labeled data. Experiments on multiple benchmark datasets show the proposed method can achieve the state-of-the-art performance and outperform the comparative methods with a significant margin.

Problem

• Semi-supervised Multi-label Learning
• Pseudo Label Generation
• Prediction Quality Impact

Innovation

• Dual Perspective Approach
• Metric Adaptivity Strategy
• Semi-supervised Multi-label Learning