Investigating Local Differential Privacy and Generative Adversarial Network in Collecting Data

For recent decades there has been a great demand for data collection to build machine learning models. But collecting data has to do with the privacy of its owners. Therefore, it is important to protect data privacy. There are many methods for protecting the privacy of user data. The differential privacy approach is often applied to machine learning models that protect data privacy from queries to the model. Furthermore, local differential privacy is recently used to protect privacy in data collection. Besides that, the machine learning model that wants to have high accuracy, and requires a lot of data to train. One of the ways to generate data is to use a generative model, specifically generative adversarial network. In this paper, we combine the generative adversarial network model with local differential privacy to protect privacy during data collection. The experimental results show that the generative adversarial networks method can be used to perturb the data in local setting for data collection, and experiment on MNIST dataset with ε parameter in local differential privacy.