Heterogeneous Domain Adaptation Using Manifold Alignment 论文

摘要

We propose a manifold alignment based approach for heterogeneous domain adaptation. A key aspect of this approach is to construct mappings to link different feature spaces in order to transfer knowl-edge across domains. The new approach can reuse labeled data from multiple source domains in a tar-get domain even in the case when the input domains do not share any common features or instances. As a pre-processing step, our approach can also be combined with existing domain adaptation ap-proaches to learn a common feature space for all input domains. This paper extends existing mani-fold alignment approaches by making use of labels rather than correspondences to align the manifolds. This extension signicantly broadens the applica-tion scope of manifold alignment, since the corre-spondence relationship required by existing align-ment approaches is hard to obtain in many applica-tions. 1