Multi-view learning with Universum.pdfVIP

Knowledge-Based Systems 70 (2014) 376–391 Contents lists available at ScienceDirect Knowledge-Based Systems journal homepage: /locate/knosys Multi-view learning with Universum Zhe Wang ?, Yujin Zhu, Wenwen Liu, Zhihua Chen, Daqi Gao Department of Computer Science and Engineering, East China University of Science and Technology, Shanghai 200237, PR China article info Article history: Received 28 August 2013 Received in revised form 20 May 2014 Accepted 23 July 2014 Available online 30 July 2014 Keywords: Multi-view learning Universum learning Regularization learning Rademacher complexity Pattern classi?cation abstract The traditional Multi-view Learning (MVL) studies how to process patterns with multiple information sources. In practice, the MVL is proven to have a signi?cant advantage over the Single-view Learning (SVL). But in most real-world cases, there are only single-source patterns to be dealt with and the existing MVL is unable to be directly applied. In order to solve this problem, an alternative MVL technique was developed for the single-source patterns through reshaping the original vector representation of the single-source patterns into multiple matrix representations in our previous work. Doing so can effectively bring an improved classi?cation performance. This paper aims to generalize the previous MVL through taking advantage of the Universum examples which do not belong to either class of the classi?cation problem. The newly-proposed generalization can not only inherit the advantage of the previous MVL, but also get a prior domain knowledge of the whole data distribution. To our knowledge, it introduces the Universum technique into the MVL for the ?rst time. In the implementation, our previous MVL named MultiV-MHKS is selected as the learning paradigm and incorporate MultiV-MHKS with the Universum technique, which forms a more ?exible MVL with the Universum called UMultiV-MHKS for short. The subsequent experiments validate that the proposed UMultiV-M