arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2107.09562cs.LGcs.CV
收藏

深度度量学习中分布偏移外泛化的特征化

Characterizing Generalization under Out-Of-Distribution Shifts in Deep Metric Learning

LMU Munich(慕尼黑大学);Heidelberg University(海德堡大学);University of Toronto(多伦多大学);University of Washington(华盛顿大学);MIT(麻省理工学院)

Timo Milbich, Karsten Roth, Samarth Sinha, Ludwig Schmidt, Marzyeh Ghassemi, Björn Ommer

摘要:

深度度量学习(DML)旨在找到适合于zero-shot转移到先验未知测试分布的表示。然而,常见的评估协议仅测试单个固定的数据分割,其中训练和测试类被随机分配。更现实的评价应考虑具有潜在不同程度和难度的广泛的分布偏移。在这项工作中,我们系统地构建训练测试分裂的难度越来越大,并提出了ooDML基准,以表征泛化下的分布偏移在DML。ooDML旨在探索更具挑战性的多样化训练到测试分布偏移的泛化性能。基于我们的新基准,我们进行了彻底的实证分析国家的最先进的DML方法。我们发现,虽然泛化往往会不断降低难度,但随着分布偏移的增加,一些方法在保持性能方面会更好。最后,我们提出了Few-Shot DML作为一种有效的方法,以不断提高泛化响应未知的测试转移ooDML。代码可在这里:https://github.com/CompVis/Characterizing_Generalization_in_DML.

英文摘要:

Deep Metric Learning (DML) aims to find representations suitable for zero-shot transfer to a priori unknown test distributions. However, common evaluation protocols only test a single, fixed data split in which train and test classes are assigned randomly. More realistic evaluations should consider a broad spectrum of distribution shifts with potentially varying degree and difficulty. In this work, we systematically construct train-test splits of increasing difficulty and present the ooDML benchmark to characterize generalization under out-of-distribution shifts in DML. ooDML is designed to probe the generalization performance on much more challenging, diverse train-to-test distribution shifts. Based on our new benchmark, we conduct a thorough empirical analysis of state-of-the-art DML methods. We find that while generalization tends to consistently degrade with difficulty, some methods are better at retaining performance as the distribution shift increases. Finally, we propose few-shot DML as an efficient way to consistently improve generalization in response to unknown test shifts presented in ooDML. Code available here: https://github.com/CompVis/Characterizing_Generalization_in_DML.

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。

↑