Pseudo-labels Traning with Uncertainty Weight for Semi-Supervised Domain Adaptation
Abstract
However, recent studies have revealed that noisy pseudo-labels in the target domain, caused by domain shift phenomena, can significantly
hinder performance. To mitigate the adverse effects of domain shift and noisy pseudo-labels on model performance, we introduce a novel approach termed Pseudo-labels Training with Uncertainty Weight for Semi-Supervised Domain Adaptation (UNW). This method achieves substantial improvements in model performance when partial labels of target samples are available. Unlike existing approaches that rely solely on
single-weight pseudo-label training, our method simultaneously addresses both domain discrepancy and the issue of noisy pseudo-labels. We
conduct extensive experiments on relevant datasets, and the results demonstrate the effectiveness of UNW in semi-supervised domain adaptation tasks.
Keywords
Full Text:
PDFReferences
[1] Qin C, Wang L, Ma Q, et al. Contradictory structure learning for semi-supervised domain adaptation[C]//Proceedings of the 2021 SIAM
International Conference on Data Mining (SDM). Society for Industrial and Applied Mathematics, 2021: 576-584.
[2] Saito K, Kim D, Sclaroff S, et al. Semi-supervised domain adaptation via minimax entropy[C]//Proceedings of the IEEE/CVF international conference on computer vision. 2019: 8050-8058.
[3] Li J, Li G, Shi Y, et al. Cross-domain adaptive clustering for semi-supervised domain adaptation[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2021: 2505-2514.
[4] Han K, Rebuffi S A, Ehrhardt S, et al. Automatically discovering and learning new visual categories with ranking statistics[J]. arXiv
preprint arXiv:2002.05714, 2020.
[5] Cubuk E D, Zoph B, Shlens J, et al. Randaugment: Practical automated data augmentation with a reduced search space[C]//Proceedings
of the IEEE/CVF conference on computer vision and pattern recognition workshops. 2020: 702-703.
[6] Ganin Y, Lempitsky V. Unsupervised domain adaptation by backpropagation[C]//International conference on machine learning. PMLR,
2015: 1180-1189.
[7] Peng X, Bai Q, Xia X, et al. Moment matching for multi-source domain adaptation[C]//Proceedings of the IEEE/CVF international conference on computer vision. 2019: 1406-1415.
[8] Venkateswara H, Eusebio J, Chakraborty S, et al. Deep hashing network for unsupervised domain adaptation[C]//Proceedings of the
IEEE conference on computer vision and pattern recognition. 2017: 5018-5027.
[9] Saenko K, Kulis B, Fritz M, et al. Adapting visual category models to new domains[C]//Computer visionECCV 2010: 11th European
conference on computer vision, Heraklion, Crete, Greece, September 5-11, 2010, proceedings, part iV 11. Springer Berlin Heidelberg,
2010: 213-226.
[10 ]Chen W Y, Liu Y C, Kira Z, et al. A closer look at few-shot classification[J]. arXiv preprint arXiv:1904.04232, 2019.
[11] Grandvalet Y, Bengio Y. Semi-supervised learning by entropy minimization[J]. Advances in neural information processing systems,
2004, 17.
DOI: http://dx.doi.org/10.70711/aitr.v2i8.6622
Refbacks
- There are currently no refbacks.
