In this paper, we propose a transductive transfer learning framework, referred to as Transfer Maximum Margin Criterion (T-MMC). This framework is suitable to transfer the knowledge acquired in one domain, the source domain, to another domain, the target domain, where no labeled examples are available in the target domain. We introduce an e ective feature weighting approach, which proceeds to reduce the domain di erence between the source and target domains. Moreover, we exploit maximum margin criterion to well discriminate various classes in the reduced domains. We simultaneously transfer knowledge from the source domain to target domain and also discriminate various classes in the reduced domains. Comprehensive experiments on the synthetic and real datasets demonstrate that T-MMC outperforms existing transfer learning methods.
Tahmoresnezhad, J., & Hashemi, S. (2016). Transductive transfer learning via maximum margin criterion. Scientia Iranica, 23(3), 1239-1250. doi: 10.24200/sci.2016.3892
MLA
J. Tahmoresnezhad; S. Hashemi. "Transductive transfer learning via maximum margin criterion". Scientia Iranica, 23, 3, 2016, 1239-1250. doi: 10.24200/sci.2016.3892
HARVARD
Tahmoresnezhad, J., Hashemi, S. (2016). 'Transductive transfer learning via maximum margin criterion', Scientia Iranica, 23(3), pp. 1239-1250. doi: 10.24200/sci.2016.3892
VANCOUVER
Tahmoresnezhad, J., Hashemi, S. Transductive transfer learning via maximum margin criterion. Scientia Iranica, 2016; 23(3): 1239-1250. doi: 10.24200/sci.2016.3892
)