This repository contains the code and dummy data for privacy-aware population synthesis using Generative Adversarial Networks (GAN). In addition to GAN, it also uses LSTM to generate both tabular and sequence data. Privacy is maintained using differential privacy, which can be turned on/off using the private flag in the WGAN class. We suggest using a very large epoch size (~10,000) to ensure stable convergence in training.

References

• Badu-Marfo, G., Farooq, B., & Patterson, Z. (2022). Composite travel generative adversarial networks for tabular and sequential population synthesis. IEEE Transactions on Intelligent Transportation Systems, 23(10), 17976-17985.
• Badu-Marfo, G., Farooq, B., & Patterson, Z. (2020). A differentially private multi-output deep generative networks approach for activity diary synthesis. arXiv preprint arXiv:2012.14574.