Mitigating Negative Transfer via Reducing Environmental Disagreement

2025-10-27

AI Summary

This work addresses negative transfer in unsupervised domain adaptation (UDA), identifying discriminative “environmental divergence”—induced by cross-domain non-causal environmental features—as a key underlying cause. To mitigate it, we propose a novel causal disentanglement learning framework: a domain-specific adversarial environmental feature extractor explicitly separates causal (invariant) features from non-causal environmental features; environmental divergence is then minimized to enable robust knowledge transfer. Our approach offers the first causal interpretation of negative transfer mechanisms in UDA and uniquely formalizes environmental divergence as an optimizable barrier to transfer. Extensive experiments on multiple standard UDA benchmarks demonstrate significant suppression of negative transfer, achieving state-of-the-art performance.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Reasoning under Uncertainty: Causality
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data
• Economics, Online Markets and Human Computation: Fairness, privacy, and diversity in economic environments

Abstract

Unsupervised Domain Adaptation~(UDA) focuses on transferring knowledge from a labeled source domain to an unlabeled target domain, addressing the challenge of emph{domain shift}. Significant domain shifts hinder effective knowledge transfer, leading to emph{negative transfer} and deteriorating model performance. Therefore, mitigating negative transfer is essential. This study revisits negative transfer through the lens of causally disentangled learning, emphasizing cross-domain discriminative disagreement on non-causal environmental features as a critical factor. Our theoretical analysis reveals that overreliance on non-causal environmental features as the environment evolves can cause discriminative disagreements~(termed emph{environmental disagreement}), thereby resulting in negative transfer. To address this, we propose Reducing Environmental Disagreement~(RED), which disentangles each sample into domain-invariant causal features and domain-specific non-causal environmental features via adversarially training domain-specific environmental feature extractors in the opposite domains. Subsequently, RED estimates and reduces environmental disagreement based on domain-specific non-causal environmental features. Experimental results confirm that RED effectively mitigates negative transfer and achieves state-of-the-art performance.

Problem

• Mitigating negative transfer in unsupervised domain adaptation
• Reducing environmental disagreement between source and target domains
• Disentangling causal and environmental features to improve transfer

Innovation

• Disentangles causal and environmental features adversarially
• Reduces environmental disagreement via domain-specific features
• Mitigates negative transfer in unsupervised domain adaptation