On reproduction of On the regularization of Wasserstein GANs
This work addresses reproducibility issues in machine learning research, specifically for GANs, but is incremental as it replicates prior findings.
The authors investigated the reproducibility of a 2018 paper on regularized Wasserstein GANs, focusing on aspects like learning speed and stability, and identified which contributions could be reproduced and at what resource cost.
This report has several purposes. First, our report is written to investigate the reproducibility of the submitted paper On the regularization of Wasserstein GANs (2018). Second, among the experiments performed in the submitted paper, five aspects were emphasized and reproduced: learning speed, stability, robustness against hyperparameter, estimating the Wasserstein distance, and various sampling method. Finally, we identify which parts of the contribution can be reproduced, and at what cost in terms of resources. All source code for reproduction is open to the public.