无需分组标签即可有效缓解模型偏见，提升最差群体表现。
Efficient Bias Mitigation Without Privileged Information
- 利用辅助模型训练历史识别虚假相关样本
- 生成平衡分组的训练集，显著提升最差群体准确率
- 无需超参数调优或分组信息，适合资源受限场景
通过经验风险最小化训练的深度神经网络常在不同群体间表现出显著性能差异，尤其当群体标签与任务标签存在虚假相关时（如“草地背景”与“牛”）。现有偏见缓解方法通常依赖分组标签进行训练或验证，或需大量超参数搜索，数据和计算成本高，限制了在大规模、无分组标注数据集上的实际应用。本文提出目标增强偏见缓解（TAB），一种简单且无需超参数调优的框架，利用辅助模型的完整训练历史识别虚假样本，并从中构建一个分组均衡的训练集，用于训练鲁棒模型。实验表明，TAB在不使用任何分组信息或模型选择的情况下，提升了最差群体性能，优于现有方法，同时保持整体准确率。
原文摘要 · Abstract (English)
Deep neural networks trained via empirical risk minimisation often exhibit significant performance disparities across groups, particularly when group and task labels are spuriously correlated (e.g., "grassy background" and "cows"). Existing bias mitigation methods that aim to address this issue often either rely on group labels for training or validation, or require an extensive hyperparameter search. Such data and computational requirements hinder the practical deployment of these methods, especially when datasets are too large to be group-annotated, computational resources are limited, and models are trained through already complex pipelines. In this paper, we propose Targeted Augmentations for Bias Mitigation (TAB), a simple hyperparameter-free framework that leverages the entire training history of a helper model to identify spurious samples, and generate a group-balanced training set from which a robust model can be trained. We show that TAB improves worst-group performance without any group information or model selection, outperforming existing methods while maintaining overall accuracy.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。