1
$\begingroup$

Neural Net can be feed-forward or recurrent. Perceptron is only feed forward.

So, what is Hopfield Network then?

$\endgroup$
3
$\begingroup$

Hopfield stores some predefined patterns (lower energy states) and when an non seen pattern is fed to the Hopfield net, it tries to find the closest match among the stored patterns. Hence the output of a Hopfield is always one of the predefined patterns which matches closely to the unseen input pattern.

This is not the case with Feed Forward Neural Nets (where no such predefined patterns are stored) and every input generates a corresponding output. Here (in Feedforward) the output is generated by a predefined function (which is self-adjusted during training session) where as in Hopfield predefined patterns are stored and outputted (no such functions exist).

$\endgroup$
0
$\begingroup$

Hopfield Networks are recurrent. However, they are not as general as more modern Recurrent Neural Networks such as Long Short-Term Memory Networks as they cannot process sequential input. I've never worked with a Hopfield Network but I've been told that they are mostly of historical interest today due to their limitations.

$\endgroup$

Your Answer

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