Privacy-Preserving Dataset Publishing using Autoencoders

To improve the accuracy of learning models, it is essential to train them on larger datasets. Unfortunately, accessing such data is often restricted, as data providers are hesitant to share their data due to privacy concerns. Therefore, it is crucial to develop methods that ensure the desired privacy for data providers. In this paper, we present an approach where data providers utilize a neural network based on the autoencoder architecture to safeguard the sensitive components of their data while preserving the utility of the remaining parts. This method demonstrates superior performance in terms of the trade-off between utility and privacy compared to similar approaches, all the while maintaining a simpler structure.