arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2202.02155cs.LGstat.AP
收藏

用于领域泛化的源数据选择

Source data selection for out-of-domain generalization

University of Wisconsin - Madison(威斯康星大学麦迪逊分校)

Xinran Miao, Kris Sankaran

摘要:

执行领域泛化的模型从异构源数据中借用知识,并将其应用于相关但不同的目标任务。迁移学习已被证明在许多应用中有效地实现了这种泛化。然而,源数据集的选择不当可能导致目标数据集的性能低下,这种现象称为负迁移。为了充分利用现有的源数据,这项工作研究了源数据选择相对于一个目标任务。我们提出了两种源选择方法,分别基于多强盗理论和随机搜索。我们对模拟和真实数据进行了全面的实证评估。我们的建议也可以被看作是诊断存在一个重新加权的源子样本,比随机选择的可用样本执行更好。

英文摘要:

Models that perform out-of-domain generalization borrow knowledge from heterogeneous source data and apply it to a related but distinct target task. Transfer learning has proven effective for accomplishing this generalization in many applications. However, poor selection of a source dataset can lead to poor performance on the target, a phenomenon called negative transfer. In order to take full advantage of available source data, this work studies source data selection with respect to a target task. We propose two source selection methods that are based on the multi-bandit theory and random search, respectively. We conduct a thorough empirical evaluation on both simulated and real data. Our proposals can be also viewed as diagnostics for the existence of a reweighted source subsamples that perform better than the random selection of available samples.

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。

↑