Generative modeling has been around for a few decades, but much of the field didn’t start to recognize itself without the discovery of the generative adversarial network (GAN). There is some debate on when GANs were discovered and by whom. One thing is for certain: Ian Goodfellow and his colleagues from the University of Montreal in 2014 deserve a good deal of credit for reinventing the technique of adversarial learning.
GANs are, after all, nothing more than autoencoders that have been split in the middle and flipped around. Goodfellow took this concept a step further by introducing ...
