arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1301.2115stat.MLcs.LG
收藏

基于不变特征表示的领域泛化

Domain Generalization via Invariant Feature Representation

Max Planck Institute for Intelligent Systems(马克斯·普朗克智能系统研究所);ETH Zurich(苏黎世联邦理工学院)

Krikamol Muandet, David Balduzzi, Bernhard Schölkopf

摘要:

本文研究领域泛化:如何从任意数量的相关领域获得知识,并将其应用到以前看不见的领域?我们提出了域不变分量分析(DICA),一种基于内核的优化算法,通过最小化域之间的相异性来学习不变变换,同时保留输入和输出变量之间的函数关系。学习理论分析表明,减少相异度提高了分类器在新领域的预期泛化能力,激励所提出的算法。在合成数据集和真实数据集上的实验结果表明,DICA成功地学习了不变特征,提高了分类器的性能。

英文摘要:

This paper investigates domain generalization: How to take knowledge acquired from an arbitrary number of related domains and apply it to previously unseen domains? We propose Domain-Invariant Component Analysis (DICA), a kernel-based optimization algorithm that learns an invariant transformation by minimizing the dissimilarity across domains, whilst preserving the functional relationship between input and output variables. A learning-theoretic analysis shows that reducing dissimilarity improves the expected generalization ability of classifiers on new domains, motivating the proposed algorithm. Experimental results on synthetic and real-world datasets demonstrate that DICA successfully learns invariant features and improves classifier performance in practice.

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。

↑