Data scarcity is a regular problem in research, and in medicine it’s especially difficult to find datasets publicly available. The main reason is its rarity, by definition images of anomalies are scattered and/or not common, and also there are a lot of legal issues that prevent sharing personal information about patients. In this talk, I’m going to give an introduction to ensembles, and more specifically to GAN ensembles, applied to this problem of data scarcity, generating MRI images of demented brains that could be use in Alzheimer research.

The core idea of this talk is that a GAN ensemble might capture many axis of variations that one individual model wouldn’t, so as stated before, we will be exploring the usage of GAN ensembles to improve generation of demented brains MRIs, showing an implementation of the paper “Ensembles of Generative Adversarial Networks” (https://arxiv.org/abs/1612.00991). I will give a brief introduction to GANs, specifically DCGANs (https://arxiv.org/abs/1511.06434 ), and then explain the implementation of DCGAN ensembles.