[EN] Over the past years, supervised models with end-to-end backpropagation have dominated the Deep Learning paradigm. However, great potential can be observed in out-of-the-box approaches like Greedy InfoMax, an ...