This repository is the official implementation of MvLNet.

Ubuntu 16.04 + cuda9.0

tensorflow-gpu==1.9.0
Keras==2.1.6
numpy==1.14.3
scikit-learn==0.19.1
munkres==1.0.12

Download pre-trained siamese networks, autoencoders and datasets here:

• MvLNet

To train and evaluate the model in the paper, run this command:

python run.py

📋 The default training scrip trains MvLNet on Noisy MNIST. Replace the config name in run.py for other datasets, i.e. Caltech101-20 or wiki.

Our model achieves the following performance :

If you find our work useful in your research, please consider citing:

@ARTICLE{huang2021deep,
  author={Huang, Zhenyu and Zhou, Joey Tianyi and Zhu, Hongyuan and Zhang, Changqing and Lv, Jiancheng and Peng, Xi},
  journal={IEEE Transactions on Image Processing}, 
  title={Deep Spectral Representation Learning From Multi-View Data}, 
  year={2021},
  volume={30},
  number={},
  pages={5352-5362},
  doi={10.1109/TIP.2021.3083072},
}