I think you may be on to something, because I was wondering how this might apply to deep learning as well; that is, such networks tend to have many layers and c

I think you may be on to something, because I was wondering how this might apply to deep learning as well; that is, such networks tend to have many layers and connections to generalize better, but at the expense of being slower (?). I'm not an expert either - but I think there may be something of interest there for study.