Readme for RWLTA

version Dec. 22, 2023

The algorithms for feature extraction and noise addition used in the section of the experiments can be found in the repository image_feature_intensity_LBP_Gabor

This repository includes the MATLAB code of the paper X. Pu, H. Che, B. Pan, M. -F. Leung and S. Wen, "Robust Weighted Low-Rank Tensor Approximation for Multiview Clustering With Mixed Noise," in IEEE Transactions on Computational Social Systems, doi: 10.1109/TCSS.2023.3331366

You will find an example of using this code in the repository (Demo_bbcsport.d, Demo_bbc4.m, Demo_msrc.m) and the corresponding datasets (bbcsport.mat, BBC.mat, msrc.mat)

MATLAB R2022a, Windows 10, 2.90-4.20 GHz AMD R7-4800H CPU, and 32 GB main memory.

• For $w$, we suggest it based on a priori knowledge, specifically, the range of $w_i$ can be set as $[0.1, 10]$. We suggest select $\lambda_1$, $\lambda_2$, and $\lambda_3$ from interval $[0.05, 10]$. For $\theta$, one can select from {0.2, 0.3, 0.5, 1, 3, 5}.
• For other parameters, you can keep the default values.

If you use this code please cite:

@ARTICLE{RWLTA,
  author={Pu, Xinyu and Che, Hangjun and Pan, Baicheng and Leung, Man-Fai and Wen, Shiping},
  journal={IEEE Transactions on Computational Social Systems}, 
  title={Robust Weighted Low-Rank Tensor Approximation for Multiview Clustering With Mixed Noise}, 
  year={2023},
  volume={},
  number={},
  pages={1-18},
  doi={10.1109/TCSS.2023.3331366}
}

This package is free for academic usage. You can run it at your own risk.

For other purposes, please contact Hangjun Che ([email protected])

This package was developed by Xinyu Pu.

For any problem concerning the code, please feel free to contact Xinyu Pu ([email protected])

Hangjun Che, SWU

Xinyu Pu, SWU