Efficient Face Generation and Clustering Using Generative Adversarial Networks

Abstract

Generative Adversarial Network (GAN) is an unsupervised learning technique in performing task such as prediction, classification and clustering. The GAN algorithm can learn the internal representation of data and can act as good features extractor. Training on a dataset of faces, we show convincing evidence that our deep convolutional adversarial pair learnt well and generated new images of fake human faces that look as realistic as possible. The unsupervised clustering model divides and groups faces based on their characteristics. In this paper, we present DCGAN (Deep Convolutional Generative Adversarial Network) in performing classification and clustering.