Since its inception, at one night back in 2014, Generative Adversarial Network (GAN) became state-of-the-art in the rapidly advancing sub-field of machine learning and AI, the Unsupervised Learning.
Although AI researchers have made astonishing progress in devising new techniques under the field of Unsupervised Learning, and precisely in the Deep Learning branch, these techniques are limited to the recognition and differentiation between pairs/sets of images and rely heavily on supplementing enormous amounts of imagery data in order to produce their final recognition output.
Looking at these facts, and doubting its viability at first, Ian Goodfellow or so-called "Ganfather", a research scientist on the Google Brain team, devised a new Machine Learning technique based on a complex statistical analysis of the elements that make up a photograph to help machines come up with images by themselves, by implementing a rivalry between two artificial neural networks, that one would act as the image forger (generator) and the other as the art detective (discriminator) who repeatedly trying to outsmart one another until the generator draws the seamless realistic image possible and the discriminator can no longer differentiate between which image is real and which is bogus!
This technique fundamentally took the computer machines to a next level of mimicking the human consciousness by providing the machines, for the first time in history, with the ability to imagine and invent new images based on a limited amount of data. However, this technique is an incredible value generator, it comes at a heavy cost of cybersecurity and digital forensics. The discussion for its disadvantages is a debate for the next time!
That will mark a big leap forward in what’s known in AI as “unsupervised learning.