From a user's perspective, I'll describe what solutions I'd like to see regarding the learning of large scale graphical models. Also, at a recent vision workshop, I asked (one by one) numerous leading