viethoang1512/mt-sgd

Stochastic Multiple Target Sampling Gradient Descent (NeurIPS 2022)

Stochastic Multiple Target Sampling Gradient Descent

This repository contains the Pytorch implementation of Stochastic Multiple Target Sampling Gradient Descent.

If you find our code useful in your research, please cite:

@article{phan2022stochastic,
  title={Stochastic Multiple Target Sampling Gradient Descent},
  author={Phan, Hoang and Tran, Ngoc and Le, Trung and Tran, Toan and Ho, Nhat and Phung, Dinh},
  journal={Advances in Neural Information Processing Systems},
  year={2022}
}

Our implementation consists of 4 subexperiments:

• CelebA
• Multi-MNIST/Fashion/Fashion and MNIST
• Multivariate Regression
• Toy example

Data:

• CelebA
• Multi-MNIST/Fashion/Fashion and MNIST
• Multivariate Regression

Reproducing:

Please refer to the bash script (*.sh) in each experiment to reproduce the reported results in the paper.

For the running time of our approach, you can take a look at this notebook in Google Colab as we wanted to utilize the cloud computing service for a fair comparison.

Contact

If you have any questions about the paper or the codebase, please feel free to contact [email protected].