Privacy-and-Performance-in-Power-Consumption-Curve-Generation-with-GANs

This repository contains several GAN models for artificial power consumption curve generation, along with the code to train them and evaluate their privacy using membership inference attacks. The DataSet folder contains numpy files with power consumption curves. Each numpy file corresponds to a different client and contains a different number of examples. Each power consumtion curve has 336 points, corresponding to a week of power consumption readings (readings every half-an-hour).