A Novel Doubly Reweighting Multisource Transfer Learning Framework

Abstract

In this paper, to effectively use the decision knowledge from multiple source domains to predict the labels of samples in the target domain, a novel doubly reweighting multisource transfer learning called DRMTL framework is proposed. DRMTL aims to simultaneously optimize the structural risk function, domain reweighting adaptation, pointwise reweighting adaptation and manifold consistency. The merits of DRMTL include the following: 1) The importance of every source domain can be evaluated using the proposed novel flexible weighting index; 2) The loss between an unknown label prediction and its prediction by some source decision function for a target sample can be reweighted using a novel domain separator; and 3) The manifold structure of the target domain is effectively used in this framework. Finally, a specific learning algorithm, i.e., a doubly reweighting multisource transfer learning using the regularized least-squares classifier called DRM-RLS, is proposed using the DRMTL framework and the classical regularized least-squares classifier, and its convergence is also proven. Our experimental results from several real-world datasets reveal that the proposed approach outperforms several state-of-the-art transfer learning algorithms.