Building a Deep Learning Network Based on the Visual Cortex

Jeffrey Sachs: Can we build a network whose architecture would be somewhat inspired by what we know of the visual cortex? He says there is a limit to what you can apply deep learning to today due to the fact that you need a lot of labeled data. If you want to train a system to interact with an environment, he says, you have to have an environment in which you can train yourself and it's not sufficient for all kinds of stuff. Goldman: We're missing something really essential in human and animal learning that is not reflected in the type of reinforcement running or supervised running that our machines can do.