robustOT
robustOT copied to clipboard
Robust Optimal Transport code
Robust Optimal Transport with Applications in Generative Modeling and Domain Adaptation
This is the official codebase of our NeurIPS 2020 paper "Robust Optimal Transport with Applications in Generative Modeling and Domain Adaptation".
GAN Experiments
Go to GAN folder.
To train robust WGAN experiments on CIFAR-10 dataset corrupted with MNIST, run
python main.py --base_cfg_path configs/experiments/CIFAR10_MNIST.json --cfg_path configs/unconditional/WGAN.json
By changing the configs and experiment in base configs, different models and datasets can be run.
Domain adaptation experiments
To train domain adaptation models, go to DA folder and run
python main.py --cfg-path configs/robust_adversarial.json