一种面向非线性回归的迁移学习模型

计算机科学 ›› 2009, Vol. 36 ›› Issue (8): 212-214.

一种面向非线性回归的迁移学习模型

杨沛,谭琦,丁月华

(华南理工大学计算机学院广州 510640);(华南师范大学计算机学院广州510631)

出版日期:2018-11-16 发布日期:2018-11-16
基金资助:
本文受国家自然科学基金项目(60574078)资助。

Non-linear Transfer Learning Model

YANG Pei,TAN Qi, DING Yue-hua

Online:2018-11-16 Published:2018-11-16

摘要/Abstract

摘要： 迁移学习能够有效地在相似任务之间进行信息的共享和迁移。之前针对多任务回归的迁移学习研究大多集中在线性系统上。针对非线性回归问题，提出了一种新的多任务回归模型—HiRBF。HiRBF基于层次贝叶斯模型，采用RBF神经网络进行回归学习，假设各个任务的输出层参数服从某种共同的先验分布。根据各个任务是否共享隐藏层，在构造HiRBF模型时有两种可选方案。在实验部分，将两种方案进行了对比，也将HiRBF与两种非迁移学习算法进行了对比，实验结果表明，HiRI3F的预测性能大大优于其它两个算法。

关键词: 迁移学习，层次贝叶斯，回归，RBF神经网络

Abstract: Multi-task learning utilizes labeled data from other "similar" tasks and can achieve efficient knowledge-sharing between tasks. Previous research mainly focused on multi task learning for linear regression. A novel Bayesian multi-task learning model for non-linear regression, i. e. HiRBF, was proposed. HiRBF is constructed under a hierarchical Bayesian framework. According to whether the input to-hidden is shared by all tasks or not, we have two options to build the HiRBF model. Inhere is a comparison between them in the experiment section. The HiRBF algorithm is also compared with two transfer-unaware approaches. The experiments demonstrate that HiRI3F significantly outperforms the others.

Key words: Transfer learning,Bayesian hierarchical model,Regression,RBF network

杨沛,谭琦,丁月华. 一种面向非线性回归的迁移学习模型[J]. 计算机科学, 2009, 36(8): 212-214. https://doi.org/

YANG Pei,TAN Qi, DING Yue-hua. Non-linear Transfer Learning Model[J]. Computer Science, 2009, 36(8): 212-214. https://doi.org/

参考文献

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed