Partial domain adaptation based on shared class oriented adversarial network

Wenjie Qiu, Wendong Chen, Haifeng Hu

Abstract: Highlights•A deep-based model for solving partial domain adaptation.•Combine adversarial network and Maximum Mean Discrepancy (MMD) to bridge domain gap.•Propose a weighted class sampler module to circumvent negative transfer.•Improve the generalization ability of the network via multi-classifier module.•Experiments on benchmark datasets verify the effectiveness of the proposed model.