Transfer learning with one-class data

Jixu Chen, Xiaoming Liu

Abstract: Highlights • First work to identify the problem of transfer learning with one-class data. • A novel regression-based algorithm to address this problem. • A new approach to select the most transferable features for one-class data. • Experiment on two new application scenarios with one-class data. Abstract When training and testing data are drawn from different distributions, most statistical models need to be retrained using the newly collected data. Transfer learning is a family of algorithms that improves the classifier learning in a target domain of interest by transferring the knowledge from one or multiple source domains, where the data falls in a different distribution. In this paper, we consider a new scenario of transfer learning for two-class classification, where only data samples from one of the two classes (e.g., the negative class) are available in the target domain. We introduce a regression-based one-class transfer learning algorithm to tackle this new problem. In contrast to the traditional discriminative feature selection, which seeks the best classification performance in the training data, we propose a new framework to learn the most transferable discriminative features suitable for our transfer learning. The experiment demonstrates improved performance in the applications of facial expression recognition and facial landmark detection.