Inductive transfer learning for unlabeled target-domain via hybrid regularization

Inductive transfer learning for unlabeled target-domain via hybrid regularization
复制标题

通过混合正则化对未标记目标域进行归纳迁移学习

DOI:
10.1007/s11434-009-0171-x
复制
发表时间:
2009-07
期刊:
科学通报(英文版)
影响因子:
--
通讯作者:
Shi ZhongZhi
Shi ZhongZhi
中科院分区:
其他
文献类型:
--
作者:
Luo Ping;He Qing;Zhuang FuZhen;Shi ZhongZhi

文献摘要

参考文献

被引文献

相似文献

近年来,人们对迁移学习的兴趣越来越大。本文研究了目标域与源域分布不同且完全无标记的分类问题,旨在建立一个未知数据的归纳模型。首先,我们分析了在以前的工作中的类比漂移的问题的转导迁移学习,并提出使用归一化的方法来移动到所需的类比。此外,我们开发了一个混合正则化框架的归纳迁移学习。它考虑了三个因素,即流形正则化的目标域的分布几何、熵正则化的预测概率的熵值和期望正则化的类先验。该框架用于将从源域学习的归纳模型适应于目标域。最后,在真实文本数据上的实验表明了我们的迁移学习归纳方法的有效性。同时,它可以处理看不见的测试点。
Recent years have witnessed an increasing interest in transfer learning. This paper deals with the classification problem that thetarget-domainwith a different distribution from thesource-domain is totally unlabeled, and aims to build an inductive model for unseen data. Firstly, we analyze the problem of class ratio drift in the previous work of transductive transfer learning, and propose to use a normalization method to move towards the desired class ratio. Furthermore, we develop a hybrid regularization framework for inductive transfer learning. It considers three factors, including the distribution geometry of the target-domain bymanifold regularization, the entropy value of prediction probability byentropy regularization, and the class prior byexpectation regularization. This framework is used to adapt the inductive model learnt from the source-domain to the target-domain. Finally, the experiments on the real-world text data show the effectiveness of our inductive method of transfer learning. Meanwhile, it can handle unseen test points.
DOI: --
发表时间: 2009
期刊: --
影响因子: --
作者:
A. Ng;Rajat Raina
通讯作者: A. Ng;Rajat Raina
DOI: 10.1007/978-3-540-74976-9_31
发表时间: 2007-09
期刊: --
影响因子: --
作者:
Dikan Xing;Wenyuan Dai;Gui-Rong Xue;Yong Yu
通讯作者: Dikan Xing;Wenyuan Dai;Gui-Rong Xue;Yong Yu
DOI: --
发表时间: 1999-06
期刊: --
影响因子: --
作者:
T. Joachims
通讯作者: T. Joachims
DOI: --
发表时间: 2007-12
期刊: --
影响因子: --
作者:
M. H. Mahmud;S. Ray
通讯作者: M. H. Mahmud;S. Ray
DOI: 10.1145/1015330.1015436
发表时间: 2004-07
期刊: Proceedings of the twenty-first international conference on Machine learning
影响因子: --
作者:
Pengcheng Wu;Thomas G. Dietterich
通讯作者: Pengcheng Wu;Thomas G. Dietterich