ABSTRACT
We have surprisingly similar problems between machine learning and neuroscience – it is remarkably hard to know why both work. I will summarize a couple of recent approaches that we are using towards those two aims. However, as with many things in life, neither brains nor machine learning systems seem to overly care about the fact that they should not work. I will thus also discuss thus some recent collaborations with the laboratory of Michelle Johnson where we use pose tracking to ask scientific and medical questions.