CONAN: Contrastive Fusion Networks for Multi-view Clustering

The official repos of "CONAN: Contrastive Fusion Networks for Multi-view Clustering".

• Submitted at: IEEE BigData 2021

• Status: Accepted (short paper).

Abstract

With the development of big data, deep learning has made remarkable progress on multi-view clustering. Multi-view fusion is a crucial technique for the model obtaining a common representation. However, existing literature adopts shallow fusion strategies, such as weighted-sum fusion and concatenating fusion, which fail to capture complex information from multiple views. In this paper, we propose a novel fusion technique, entitled contrastive fusion, which can extract consistent representations from multiple views and maintain the characteristic of view-specific representations. Specifically, we study multi-view alignment from an information bottleneck perspective and introduce an intermediate variable to align each view-specific representation. Furthermore, we leverage a single-view clustering method as a predictive task to ensure the contrastive fusion is working. We integrate all components into a unified framework called CONtrAstive fusion Network (CONAN). Experiment results on five multi-view datasets demonstrate that CONAN outperforms state-of-the-art methods.

Architecture

Environment

• Python 3.8
• PyTorch 1.8.0
• CUDA 10.2

Training

All our experiments are put in ./experiments, data files under data/processed.

Note: Before you run the program firstly, you should run datatool/load_dataset to generate dataset.

You could quickly run our experiments by: python train.py -c [config name].

For example: python train.py -mnist

Fusion Results

Citation

@inproceedings{CONAN,
  author    = {Guanzhou Ke and
               Zhiyong Hong and
               Zhiqiang Zeng and
               Zeyi Liu and
               Yangjie Sun and
               Yannan Xie},
  title     = {{CONAN:} Contrastive Fusion Networks for Multi-view Clustering},
  booktitle = {2021 {IEEE} International Conference on Big Data (Big Data), Orlando,
               FL, USA, December 15-18, 2021},
  pages     = {653--660},
  publisher = {{IEEE}},
  year      = {2021},
  url       = {https://doi.org/10.1109/BigData52589.2021.9671851},
  doi       = {10.1109/BigData52589.2021.9671851},
}