We recently had a visitor to the Gatsby Unit talk about his work in reinforcement learning, in particular the use of planning and forward models to speed up the learning of difficult tasks. The substance of his talk was good, but that’s not what I want to talk about: it was the motivation he gave ... [-->]
From my own experience, i also think, that learning on a small scale is solved. And i also believe, that structured data alone cannot enable semantic interpretation. So, building up models from huge amounts of data seems to be the only option today. But still - we as humans don't have to read the whole web to learn about a language... something is missing there.
