We'll start with some quite recent history because GANs are among the newest ideas you'll find around AI and deep learning.
Everything started in 2014, when Ian Goodfellow and his colleagues (there is also Yoshua Bengio closing the list of contributors) at the Departement d'informatique et de recherche opérationnelle at Montreal University published a paper on Generative Adversarial Nets (GANs), a framework capable of generating new data based on a set of initial examples:
GOODFELLOW, Ian, et al. Generative Adversarial Nets. In: Advances in Neural Information Processing Systems. 2014. p. 2672-2680: https://arxiv.org/abs/1406.2661.
The initial images produced by such networks were astonishing, considering the previous attempts using Markov chains which were far from being credible. In the image, you can see some of the examples proposed in the...
