Hierarchical Projection for Adaptive Knowledge Transfer

2026-06-07

AI Summary

This study addresses the challenge of negative transfer in multi-source heterogeneous data, where discrepancies in relevance or spurious signals hinder effective knowledge transfer. To enable reliable cross-domain learning, the authors propose ProjectionTL, a novel framework that integrates hierarchical Bayesian modeling with an adaptive projection mechanism. ProjectionTL performs selective knowledge transfer jointly at both the source and feature levels by synergistically optimizing global source alignment and local feature consistency. It further enhances robustness and interpretability through data-driven weighted aggregation and posterior projection. Extensive experiments on synthetic and real-world biomedical tasks demonstrate that ProjectionTL substantially outperforms existing methods, achieving significant improvements in prediction accuracy, stability, and model interpretability.

Abstract

Modern data-driven applications increasingly involve learning from multiple heterogeneous sources, where a target dataset is limited but related information is available across domains. Naively combining these sources can degrade performance when relevance varies or spurious signals are present, posing a fundamental challenge for trustworthy cross-domain learning. We propose Projection Transfer Learning (ProjectionTL), a unified framework that integrates hierarchical Bayesian modeling with adaptive projection for selective knowledge transfer. The key idea is to decouple transfer at two levels: first, we construct a source-guided hierarchical prior that aggregates information across sources using data-driven weights, capturing global alignment between each source and the target; second, we refine this borrowing through a posterior-projection step that operates at the feature level, selectively retaining coordinates that exhibit local agreement with the target signal. This two-stage design enables the method to simultaneously perform source selection and feature selection, thereby mitigating negative transfer while preserving interpretability. ProjectionTL provides a principled approach to integrating heterogeneous data across domains, bridging statistical modeling and modern machine learning paradigms for robust and interpretable transfer. Through simulations and real-world biomedical applications, we demonstrate improved accuracy, stability, and interpretability compared to existing methods. Our framework offers a scalable and generalizable strategy for trustworthy cross-domain learning in high-dimensional settings.

Problem

• cross-domain learning
• negative transfer
• heterogeneous data
• knowledge transfer
• trustworthy AI

Innovation

• Projection Transfer Learning
• hierarchical Bayesian modeling
• adaptive projection
• selective knowledge transfer
• negative transfer mitigation