README.md

An Effective and Secure Federated Multi-View Clustering Method with Information-Theoretic Perspective

Code for the paper "An Effective and Secure Federated Multi-View Clustering Method with Information-Theoretic Perspective". (ICML 2025)

Requirements

The code requires:

• Python 3.6 or higher

• Pytorch 1.9 or higher

Example execution

To train a new model, run:

python main.py 

Further settings for the dataset, number of clients, overlapping rate, and other parameters can be configured in main.py.

You can also apply it in incomplete scenarios by changing the missing_rate. For example, in the BDGP dataset with a missing rate of 0.5, run:

python main.py --dataset='BDGP' --num_users=2 --missing_rate=0.5

The contents of the mask folder are automatically generated by main.py, which applies to incomplete scenarios.

Citation

If you find our code useful, please cite:

@InProceedings{chen2025,
    author    = {Xinyue Chen, Jinfeng Peng, Yuhao Li, Xiaorong Pu, Yang Yang, Yazhou Ren},
    title     = {An Effective and Secure Federated Multi-View Clustering Method with Information-Theoretic Perspective},
    booktitle = {ICML},
    year      = {2025},
    pages     = {1-19}
}

If you have any problems, please contact me by martinachen2580@gmail.com.