python train.py --dataset uci

To test the trained model, e.g., Digit, run:

python test.py --dataset uci

Due to the divergence based loss may fail into inferior local optima, there may be some fluctuations in the running results, which can be improved by more attempts.

Acknowledgements

We thank the PyTorch implementation on SiMVC (https://github.com/DanielTrosten/mvc).