从损失景观视角引导泛化能力
Bootstrap Generalization Ability from Loss Landscape Perspective
AI总结：
本文将损失景观理论引入计算机视觉域泛化，从骨干网络、正则化、训练范式和学习率四方面提升模型泛化，并在多个数据集及 NICO 竞赛中验证有效性。
摘要：
域泛化旨在学习一个能够在未见过的测试数据集（即分布外数据，其分布与训练数据集不同）上良好泛化的模型。为解决计算机视觉中的域泛化问题，我们将损失景观理论引入该领域。具体而言，我们从损失景观视角出发，在骨干网络、正则化、训练范式和学习率四个方面引导深度学习模型的泛化能力。我们通过大量消融实验和可视化，在 NICO++、PACS 和 VLCS 数据集上验证了所提出的理论。此外，我们将该理论应用于 ECCV 2022 NICO Challenge1，在未使用任何域不变方法的情况下获得第三名。
英文摘要：
Domain generalization aims to learn a model that can generalize well on the unseen test dataset, i.e., out-of-distribution data, which has different distribution from the training dataset. To address domain generalization in computer vision, we introduce the loss landscape theory into this field. Specifically, we bootstrap the generalization ability of the deep learning model from the loss landscape perspective in four aspects, including backbone, regularization, training paradigm, and learning rate. We verify the proposed theory on the NICO++, PACS, and VLCS datasets by doing extensive ablation studies as well as visualizations. In addition, we apply this theory in the ECCV 2022 NICO Challenge1 and achieve the 3rd place without using any domain invariant methods.
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。
