Broadly, the practice of creating
artifacts that mimic the
biophysical
properties of the
nervous system. Whereas most
neural networks
are, at best, 'inspired' by
biology, neuromorphic systems attempt to
accurately
replicate biology, right down to
ion channels.
NE could be seen as a form of
computational science, though it
is probably more strongly slanted toward engineering than most CS
(in part because of the emphasis on hardware).
The term was coined by Carver Mead (after "eclectronics" failed to
catch on), and originally referred to what is today known as
Neuromorphic Analog VLSI. This medium is still preferred, and
silicon neurons continue make up the bulk of neuromorphic
research. A side effect of this has been a small Renaissance in
analog computing, which was (in a roundabout way) the goal of
NE to begin with.