发表机构
Harbin Institute of Technology; State Key Laboratory of Smart Farm Technologies and Systems（哈尔滨工业大学; 智慧农场技术与系统国家重点实验室）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
SAGE是两阶段生成增强框架，通过生成合成数据缓解虚假关联，在Waterbirds等三个数据集上的最差群体准确率优于无群体标签基线，提升最多7.7个百分点。
AI 中文摘要
虚假关联对现代机器学习的鲁棒性构成重大挑战。数据集分布的固有不平衡常导致传统经验风险最小化（ERM）模型依赖多数虚假属性进行分类，致使少数群体表现不佳；当虚假属性不可用时，该问题尤为棘手。现有无群体标签方法常对少数群体或被误分类的真实训练样本进行上采样，重复相同实例会降低有效多样性并加剧过拟合。为在缺乏先验知识时从数据中心视角缓解这些虚假关联，我们提出子群体感知生成增强（SAGE），这是一个两阶段生成增强框架。利用聚类得到的子标签和类别标签，我们微调条件生成模型与文本编码器，生成针对性合成数据以填充训练集的代表性不足区域，并构建平衡验证集用于最后一层重加权。实验表明，SAGE在Waterbirds、CelebA和MetaShift上的最差群体准确率分别达到89.5%、85.7%和79.1%，较最优无群体标签基线方法提升最多7.7个百分点。
英文摘要
Spurious correlations pose a significant challenge to the robustness of modern machine learning. The inherent imbalance in dataset distributions often leads traditional Empirical Risk Minimization (ERM) models to rely on majority spurious attributes for classification, resulting in poor performance on minority groups. This problem becomes particularly challenging when the spurious attributes are unavailable. Existing group-label-free methods often upsample minority groups or misclassified real training examples; repeating the same instances can reduce effective diversity and encourage overfitting. To mitigate these spurious correlations from a data-centric perspective in the absence of prior knowledge, we introduce Subpopulation-Aware Generative Enhancement (SAGE), a two-stage generative augmentation framework. Using cluster-derived sub-labels and class labels, we fine-tune a conditional generative model and text encoder, generating targeted synthetic data to fill underrepresented regions in the training set and construct a balanced validation set for last-layer reweighting. We experimentally show that SAGE achieves 89.5%, 85.7%, and 79.1% worst-group accuracy on Waterbirds, CelebA, and MetaShift, respectively, outperforming the best group-label-free baselines by up to 7.7 percentage points.
