Is there a Hebb neural network? What kind of functions can it implement? Or, are there multiple "Hebb networks", that is, neural networks that learn in a Hebbian fashion?

  • $\begingroup$ Can you please tell us where you heard of the expression "Hebb network"? Aren't you actually interested in Hebbian learning, as opposed to a particular neural network that learns in an Hebbian fashion? $\endgroup$ – nbro Apr 26 '19 at 0:54
  • $\begingroup$ @nbro I am confused on the same thing if Hebbian learning and Hebb Network are same thing. $\endgroup$ – Lexi Apr 26 '19 at 2:53
  • 1
    $\begingroup$ There is no such thing as a Hebb Network. Perhaps you mean a Hopfield Network? Or its probabilistic version the Boltzmann Machine? There are the models which are trained by Hebbian Learning. $\endgroup$ – samlaf Apr 27 '19 at 6:24

In machine learning, the idea behind Hebbian learning is to strengthen (or weaken) the connection (the weight) between the neurons that have similar (or, respectively, dissimilar) outputs, where "similar" can be defined in different ways (e.g. it could be based on the sign of the output of the neurons).

Hebbian learning is a more biologically plausible way of learning than back-propagation, because it is a "local learning strategy" (you locally update the connections and not all the connections of the model at the same time), as opposed to back-propagation, which is a "global learning strategy" (where all connections are usually updated at once, given a "global error" of the network).

There are several neural networks (or models) that can learn in a Hebbian fashion: for example, the Hopfield network or Numenta's temporal memory.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.