Hello

I'm translating a book called models of Mind by Grace Lindasy and I want to understand what is meant by noisy inputs - noisy outputs  - noisy neurons in the following:

- This means neurons do have the power to destroy noise – to take in noisy inputs while producing less noisy outputs.

- If noisy inputs led to outputs that were just as, or possibly more, noisy, this would be a perfectly self-consistent story; noisy neurons would beget noisy neurons.

Noisy inputs are what we were discussing earlier: input signals with too much noise in them.

Noisy outputs are neural output signals with noise in them, i.e. what you would get from the neuron with not enough averaging.

Apparently, noisy neurons create their own noise and this is necessary for learning:

https://news.stanford.edu/2015/03/18/decisions-noisy-neurons-031815/

Edited February 23, 20232 yr by Lorentz Jr

Thanks a million.