Dynamic Neural Network Structures Through Stochastic Rewiring

[@evankim4096]

Very interesting talk, I am curious about the third role, the role of the neurons that don't have a role in the current learning task. They sit there and don't seem to contribute towards learning seems to be the observation. Has any more research been done into understanding why that happens?

[@samlaf92]

It's a bit weird to compare the brain to a turing machine. These are on different Marrian levels... (the brain might be implementing a turing machine for all we know). Comparing to transistor structures would make more sense. And isn't there a way to argue that transistors are also not completely fixed? At least we can argue for external memory, if it gets corrupted or something, we have error correcting codes. Maybe that's what the brain is doing and all this dynamic connectome analysis is futile?

[@evankim4096]

I don't think that was a significant aspect of the entire talk