A Transfer Learning Approach for Network Modeling.
A Transfer Learning Approach for Network Modeling.
复制标题
网络建模的转移学习方法。
DOI：
10.1080/0740817x.2011.649390
复制
发表时间：
2012-11-01
期刊：
影响因子：
--
通讯作者：
Yao L
中科院分区：
文献类型：
--
作者：
Huang S;Li J;Chen K;Wu T;Ye J;Wu X;Yao L
网络模型已广泛应用于许多领域来表征物理实体之间的交互关系。面临的一个典型问题是识别具有某些相似性的多个相关任务的网络。在这种情况下，非常需要一种能够利用在一项任务建模过程中获得的知识来帮助更好地建模另一项任务的迁移学习方法。在本文中，我们提出了一种迁移学习方法，该方法采用贝叶斯分层模型框架来表征任务相关性，并另外使用 L1 正则化来确保样本量有限的网络的鲁棒学习。进一步开发了一种基于期望最大化（EM）算法的方法来从数据中学习网络。进行了模拟研究，证明了所提出的迁移学习方法相对于单独学习每个任务的网络的单任务学习的优越性。所提出的方法还适用于从功能磁共振图像（fMRI）数据中识别阿尔茨海默病（AD）的大脑连接网络。研究结果与 AD 文献一致。
Networks models have been widely used in many domains to characterize the interacting relationship between physical entities. A typical problem faced is to identify the networks of multiple related tasks that share some similarities. In this case, a transfer learning approach that can leverage the knowledge gained during the modeling of one task to help better model another task is highly desirable. In this paper, we propose a transfer learning approach, which adopts a Bayesian hierarchical model framework to characterize task relatedness and additionally uses the L1-regularization to ensure robust learning of the networks with limited sample sizes. A method based on the Expectation-Maximization (EM) algorithm is further developed to learn the networks from data. Simulation studies are performed, which demonstrate the superiority of the proposed transfer learning approach over single task learning that learns the network of each task in isolation. The proposed approach is also applied to identification of brain connectivity networks of Alzheimer’s disease (AD) from functional magnetic resonance image (fMRI) data. The findings are consistent with the AD literature.