CoBA: Counterbias Text Augmentation for Mitigating Various Spurious Correlations via Semantic Triples
This addresses the issue of poor model generalization due to spurious correlations for users of text-based AI systems, representing a novel method for a known bottleneck rather than an incremental improvement.
The paper tackles the problem of deep learning models exploiting spurious correlations in training data, which harms generalization, by introducing CoBA, a counterbias text augmentation method that improves downstream task performance and reduces biases while enhancing out-of-distribution resilience.
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.