zzf495/ktl-dds

Implementation of Knowledge Transfer Learning via Dual Density Sampling for Resource-Limited Domain Adaptation (KTL-DDS)

Knowledge Transfer Learning via Dual Density Sampling for Resource-Limited Domain Adaptation (KTL-DDS)

This repository contains the codes of the KTL-DDS.

Code files (matlab implementation)

├─+func: The functions used for DDSA. For the other algorithms, please refer to our repository.
├─data
│ ├─Adaptiope20: The .mat files of Adaptiope20 (ResNet50) gained by using a ResNet50 extractor proposed in [1].
│ │ Only the first 20 classes are used in the paper.
│ └─Office31: The .mat files of Office31 (ResNet50) gained by using a ResNet50 extractor proposed in [1].
└─util: The folder that stores KTL-DDS, DDSA and some auxiliary functions.
└─liblinear-2.30: A libsvm tool.

[1] Wang, Qian, and Toby Breckon. "Unsupervised domain adaptation via structured prediction based selective pseudo-labeling." Proc. AAAI Conf. Artif. Intell.. Vol. 34. No. 04. 2020.

Citation

@ARTICLE{10302459,
author={Zheng, Zefeng and Teng, Luyao and Zhang, Wei and Wu, Naiqi and Teng, Shaohua},
journal={IEEE/CAA Journal of Automatica Sinica},
title={Knowledge Transfer Learning via Dual Density Sampling for Resource-Limited Domain Adaptation},
year={2023},
volume={10},
number={12},
pages={2269-2291},
doi={10.1109/JAS.2023.123342}
}