One-Class Transfer Learning with Uncertain Data

Bo Liu, Philip S. Yu, Yanshan Xiao, Zhifeng Hao

Abstract: One-class learning aims at constructing a distinctive classifier based on the labeled one class data. However, it is a challenge for the existing one-class learning methods to transfer knowledge from a source task to a target task for uncertain data. To address this challenge, this paper proposes a novel approach, called uncertain one-class transfer learning with SVM (UOCT-SVM), which first formulates the uncertain data and transfer learning into one-class SVM as an optimization problem and then proposes an iterative framework to build an accurate classifier for the target task. Our proposed method explicitly addresses the problem of one-class transfer learning with uncertain data. Extensive experiments has found our proposed method can mitigate the effect of uncertain data on the decision boundary and transfer knowledge to help build an accurate classifier for the target task, compared with state-of-the-art one-class learning methods.