«A journey through multiple dimensions and transformations in SPACE» by Memo Akten

This is an expanded transcript of a guest lecture I gave at the School of Ma in Berlin, in July 2016, during Gene Kogan ’s “Machine Learning for Artists” course. As such this talk aims to give a very high-level, very particular perspective on ‘how Deep Learning works’, and is aimed at artists and creative folk who may or may not be well-versed in maths, computer science or machine learning (though it does assume the most basic knowledge of what an artificial neural network is). Many thanks to Mike Tyka and Jamie Ryan Kiros for taking the time to provide valuable feedback on this post. In addition to my general research in this field, some of the ideas below took shape whilst I was resident at Google’s Artists and Machine Intelligence program in Spring 2016. I’m finishing up two more (much shorter) posts regarding the specific work I made there, hopefully will be online shortly.