Combating Unknown Bias with Effective Bias-Conflicting Scoring and Gradient Alignment

Abstract

Models notoriously suffer from dataset biases which are detrimental to robustness and generalization. The identify-emphasize paradigm shows a promising effect in dealing with unknown biases. However, we find that it is still plagued by two challenges: A, the quality of the identified bias-conflicting samples is far from satisfactory; B, the emphasizing strategies just yield suboptimal performance. In this work, for challenge A, we propose an effective bias-conflicting scoring method to boost the identification accuracy with two practical strategies -- peer-picking and epoch-ensemble. For challenge B, we point out that the gradient contribution statistics can be a reliable indicator to inspect whether the optimization is dominated by bias-aligned samples. Then, we propose gradient alignment, which employs gradient statistics to balance the contributions of the mined bias-aligned and bias-conflicting samples dynamically throughout the learning process, forcing models to leverage intrinsic features to make fair decisions. Experiments are conducted on multiple datasets in various settings, demonstrating that the proposed solution can alleviate the impact of unknown biases and achieve state-of-the-art performance.

References in corpus (8)

• Towards Understanding Generalization of Deep Learning: Perspective of Loss Landscapes
• Learning Robust Representations by Projecting Superficial Statistics Out
• An Investigation of Why Overparameterization Exacerbates Spurious Correlations
• Just Train Twice: Improving Group Robustness without Training Group Information
• Learning Debiased Representation via Disentangled Feature Augmentation
• Model Patching: Closing the Subgroup Performance Gap with Data Augmentation
• No Subclass Left Behind: Fine-Grained Robustness in Coarse-Grained Classification Problems
• Gradient Starvation: A Learning Proclivity in Neural Networks