GAN-supported data augmentation is a term used in artificial intelligence, big data, smart data and digital transformation. It describes a modern method in which so-called Generative Adversarial Networks (GANs) - a special form of artificial neural networks - are used to generate additional training data for computer programmes.
Imagine a company wants to develop an AI that can analyse X-ray images. However, there are not enough real X-ray images for training. This is where GAN-supported data augmentation comes into play: the AI automatically generates many artificial but realistic-looking X-ray images that help the system to learn better.
This allows researchers or companies to significantly increase the performance of their artificial intelligence without having to collect expensive or hard-to-obtain original data. The most important advantages are therefore better results, time and cost savings. Especially in areas where real data is rare or sensitive, GAN-supported data augmentation brings great progress - for example in medicine, autonomous driving or the detection of fraudulent activities.
Deutsch 
English (UK)