Treffer: Generating Deepfakes from MNIST Dataset in Python, Keras and Tensorflow

Deep learning is an Artificial Intelligent (AI) function that mimics the workings of the human brain in processing data such as speech recognition, visual object recognition, object detection, language translation, and making decisions. Deepfakes are images generated by artificial intelligence techniques where a person in an existing image or video is replaced by someone else’s likeness. A generative adversarial network (GAN) is a specific deep learning technique designed by Goodfellow et al. (2014) which generates new data from a given training set. This generates a new image which is referred to as a deepfake. In this work we developed deepfakes based on the public MNIST dataset using GAN. Deepfakes have become a societal challenge as they are difficult to impossible to distinguish from an authentic image. Future work will examine the accuracy of human subjects detecting deepfakes from authentic images.