Multi-Label Transfer Learning in Non-Stationary Data Streams

2025-09-09

AI Summary

To address the coexistence of independent and correlated label concept drifts in non-stationary multi-label data streams—and the lack of inter-label knowledge reuse in existing transfer learning approaches—this paper proposes two online multi-label transfer learning methods. BR-MARLENE enables efficient adaptation by sharing label-level knowledge across streams, while BRPW-MARLENE further models and transfers pairwise label dependencies to explicitly capture structured correlations in dynamic environments. Both methods are embedded within the Binary Relevance (BR) framework, supporting incremental model updates and real-time prediction. Extensive experiments on multiple real-world multi-label stream datasets demonstrate that our methods significantly outperform current state-of-the-art baselines. The results validate that leveraging label correlations for knowledge transfer enhances model robustness and predictive accuracy under concept drift.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Data Mining & Knowledge Management: Data Stream Mining
• Multiagent Systems: Multiagent Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Label concepts in multi-label data streams often experience drift in non-stationary environments, either independently or in relation to other labels. Transferring knowledge between related labels can accelerate adaptation, yet research on multi-label transfer learning for data streams remains limited. To address this, we propose two novel transfer learning methods: BR-MARLENE leverages knowledge from different labels in both source and target streams for multi-label classification; BRPW-MARLENE builds on this by explicitly modelling and transferring pairwise label dependencies to enhance learning performance. Comprehensive experiments show that both methods outperform state-of-the-art multi-label stream approaches in non-stationary environments, demonstrating the effectiveness of inter-label knowledge transfer for improved predictive performance.

Problem

• Addressing multi-label concept drift in non-stationary data streams
• Transferring knowledge between related labels for accelerated adaptation
• Modeling pairwise label dependencies to enhance classification performance

Innovation

• Leveraging knowledge from source and target streams
• Modelling pairwise label dependencies explicitly
• Transferring inter-label knowledge for classification