liqunchen0606 / graph-optimal-transport

Code for ICML 2020 "Graph Optimal Transport for Cross-Domain Alignment"

Graph Optimal Transport framework for cross-domain alignment

Summary

In this work, both Gromov-Wasserstein and Wasserstein distance are applied to improve the performance of different cross-domain alignment tasks, such as VQA, Machine Translation, etc

Implemntation

Note that the VQA model is developed upon BAN model.

And the machine translation model is developed upon texar.

Citation

Please cite our ICML 2020 paper if you found the code useful.

@article{chen2020graph,
  title={Graph Optimal Transport for Cross-Domain Alignment},
  author={Chen, Liqun and Gan, Zhe and Cheng, Yu and Li, Linjie and Carin, Lawrence and Liu, Jingjing},
  journal={ICML},
  year={2020}
}

graph-optimal-transport's Issues

代码里没有实现算法3，共享矩阵版本

Can you provide the code of image-text retrieval task?

It is a good project~And I would ask if you plan to release the code of image-text retrieval task.
Thx~~

Please provide some annotations for helping us to understand.

First of all, thanks for your excellent work and code.

Could you please provide the detailed explanations for how you apply your GOT method to the BAN_vqa? We cannot find detailed annotations of where your work lies and there are too many functions in the OT_torch_.py.

Thanks for your kind help!