CoBA: Counterbias Text Augmentation for Mitigating Various Spurious Correlations via Semantic Triples

2025-08-26

AI Summary

Deep learning models often rely on spurious correlations in training data—such as gender or simplicity biases—leading to poor out-of-distribution (OOD) generalization. To address this, we propose CoBA, a unified counter-bias augmentation framework that, for the first time, jointly mitigates multiple types of spurious associations at the semantic triplet level. CoBA decomposes and reconstructs text into triplets and applies selective modification to generate adversarial augmented samples that break non-target correlations. It requires no additional annotations or model modifications and is plug-and-play. Extensive experiments across diverse natural language understanding tasks demonstrate that CoBA significantly improves OOD robustness, effectively alleviates multiple biases (e.g., gender, lexical, and syntactic), and yields stable, generalizable performance gains. These results empirically validate the efficacy of semantic-structured augmentation in disentangling spurious correlations.

Technology Category

• Natural Language Processing: Safety and Robustness
• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Deep learning models often learn and exploit spurious correlations in training data, using these non-target features to inform their predictions. Such reliance leads to performance degradation and poor generalization on unseen data. To address these limitations, we introduce a more general form of counterfactual data augmentation, termed counterbias data augmentation, which simultaneously tackles multiple biases (e.g., gender bias, simplicity bias) and enhances out-of-distribution robustness. We present CoBA: CounterBias Augmentation, a unified framework that operates at the semantic triple level: first decomposing text into subject-predicate-object triples, then selectively modifying these triples to disrupt spurious correlations. By reconstructing the text from these adjusted triples, CoBA generates counterbias data that mitigates spurious patterns. Through extensive experiments, we demonstrate that CoBA not only improves downstream task performance, but also effectively reduces biases and strengthens out-of-distribution resilience, offering a versatile and robust solution to the challenges posed by spurious correlations.

Problem

• Mitigates spurious correlations in training data
• Enhances out-of-distribution generalization robustness
• Simultaneously tackles multiple biases via semantic augmentation

Innovation

• Semantic triple-level text decomposition and modification
• Generating counterbias data via triple reconstruction
• Simultaneously mitigating multiple biases for robustness