一种面向非线性回归的迁移学习模型  

Non-linear Transfer Learning Model

作　　者：杨沛[1] 谭琦[1,2] 丁月华[1] 

机构地区：[1]华南理工大学计算机学院,广州510640 [2]华南师范大学计算机学院,广州510631

出　　处：《计算机科学》2009年第8期212-214,242,共4页Computer Science

基　　金：国家自然科学基金项目(60574078)资助

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,HiRBF的预测性能大大优于其它两个算法。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. There is a comparison between them in the experiment section. The HiRBF algorithm is also compared with two transfer-unaware approaches. The experiments demonstrate that HiRBF significantly outperforms the others.

关 键 词：迁移学习 层次贝叶斯 回归 RBF神经网络 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]