OccamNets：通过偏好更简单假设缓解数据集偏差
OccamNets: Mitigating Dataset Bias by Favoring Simpler Hypotheses
AI总结：
针对数据集偏差损害深度网络泛化的问题，提出具备两种归纳偏置的OccamNets架构，通过偏好简单假设提升鲁棒性，性能优于或匹敌现有最优方法，且与现有去偏方法结合可进一步提升效果。
摘要：
数据集偏差和虚假相关性会显著损害深度神经网络的泛化能力。以往许多研究通过替代损失函数或聚焦稀有模式的采样策略来解决该问题。我们提出一个新方向：修改网络架构以引入归纳偏置，使网络对数据集偏差具备鲁棒性。具体而言，我们提出OccamNets（奥卡姆网络），其设计上偏向于更简单的解决方案。OccamNets具备两种归纳偏置：第一，它偏向于针对单个样本仅使用所需的最少网络深度；第二，它偏向于使用更少的图像位置进行预测。尽管OccamNets偏向更简单的假设，但在必要时也能学习更复杂的假设。实验中，OccamNets的表现优于或匹敌于在未融入这些归纳偏置的架构上运行的现有最优方法。此外，我们证明将现有最优去偏方法与OccamNets结合后，结果会进一步提升。
英文摘要：
Dataset bias and spurious correlations can significantly impair generalization in deep neural networks. Many prior efforts have addressed this problem using either alternative loss functions or sampling strategies that focus on rare patterns. We propose a new direction: modifying the network architecture to impose inductive biases that make the network robust to dataset bias. Specifically, we propose OccamNets, which are biased to favor simpler solutions by design. OccamNets have two inductive biases. First, they are biased to use as little network depth as needed for an individual example. Second, they are biased toward using fewer image locations for prediction. While OccamNets are biased toward simpler hypotheses, they can learn more complex hypotheses if necessary. In experiments, OccamNets outperform or rival state-of-the-art methods run on architectures that do not incorporate these inductive biases. Furthermore, we demonstrate that when the state-of-the-art debiasing methods are combined with OccamNets results further improve.
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。
