DecAug：基于分解特征表示和语义增强的分布外泛化
DecAug: Out-of-Distribution Generalization via Decomposed Feature Representation and Semantic Augmentation
摘要：
虽然深度学习展示了其处理独立同分布（IID）数据的强大能力，但它经常遭受分布外（OoD）泛化，其中测试数据来自另一个分布（w.r.t.训练1）。设计一个通用的面向对象的推广框架，以广泛的应用是具有挑战性的，主要是由于在现实世界中可能的相关性和多样性的转变。以往的方法大多只能解决一个特定的分布偏移，如跨域偏移或相关外推。为了解决这个问题，我们提出了DecAug，一种新的分解特征表示和语义增强方法，用于OoD泛化。DecAug解开了类别相关和上下文相关的特征。类别相关特征包含目标对象的因果信息，而上下文相关特征描述属性、风格、背景或场景，导致训练数据和测试数据之间的分布偏移。分解通过正交化两个梯度（w.r.t.中间特征）的损失，用于预测类别和上下文标签。此外，我们执行基于梯度的增强上下文相关的功能，以提高学习表示的鲁棒性。实验结果表明，DecAug在各种OoD数据集上的性能优于其他最先进的方法，这是为数不多的可以处理不同类型的OoD泛化挑战的方法之一。
英文摘要：
While deep learning demonstrates its strong ability to handle independent and identically distributed (IID) data, it often suffers from out-of-distribution (OoD) generalization, where the test data come from another distribution (w.r.t. the training one). Designing a general OoD generalization framework to a wide range of applications is challenging, mainly due to possible correlation shift and diversity shift in the real world. Most of the previous approaches can only solve one specific distribution shift, such as shift across domains or the extrapolation of correlation. To address that, we propose DecAug, a novel decomposed feature representation and semantic augmentation approach for OoD generalization. DecAug disentangles the category-related and context-related features. Category-related features contain causal information of the target object, while context-related features describe the attributes, styles, backgrounds, or scenes, causing distribution shifts between training and test data. The decomposition is achieved by orthogonalizing the two gradients (w.r.t. intermediate features) of losses for predicting category and context labels. Furthermore, we perform gradient-based augmentation on context-related features to improve the robustness of the learned representations. Experimental results show that DecAug outperforms other state-of-the-art methods on various OoD datasets, which is among the very few methods that can deal with different types of OoD generalization challenges.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
