<s>
The	O
term	O
hybrid	B-Algorithm
neural	I-Algorithm
network	I-Algorithm
can	O
have	O
two	O
meanings	O
:	O
</s>
<s>
Artificial	B-Architecture
neural	I-Architecture
networks	I-Architecture
with	O
a	O
symbolic	O
part	O
(	O
or	O
,	O
conversely	O
,	O
symbolic	O
computations	O
with	O
a	O
connectionist	O
part	O
)	O
.	O
</s>
<s>
As	O
for	O
the	O
first	O
meaning	O
,	O
the	O
artificial	B-Algorithm
neurons	I-Algorithm
and	O
synapses	B-Application
in	O
hybrid	O
networks	O
can	O
be	O
digital	B-General_Concept
or	O
analog	O
.	O
</s>
<s>
For	O
the	O
digital	B-General_Concept
variant	O
voltage	O
clamps	O
are	O
used	O
to	O
monitor	O
the	O
membrane	O
potential	O
of	O
neurons	O
,	O
to	O
computationally	O
simulate	O
artificial	B-Algorithm
neurons	I-Algorithm
and	O
synapses	B-Application
and	O
to	O
stimulate	O
biological	O
neurons	O
by	O
inducing	O
synaptic	O
.	O
</s>
<s>
As	O
for	O
the	O
second	O
meaning	O
,	O
incorporating	O
elements	O
of	O
symbolic	O
computation	O
and	O
artificial	B-Architecture
neural	I-Architecture
networks	I-Architecture
into	O
one	O
model	O
was	O
an	O
attempt	O
to	O
combine	O
the	O
advantages	O
of	O
both	O
paradigms	O
while	O
avoiding	O
the	O
shortcomings	O
.	O
</s>
<s>
Representations	O
of	O
artificial	B-Architecture
neural	I-Architecture
networks	I-Architecture
,	O
on	O
the	O
other	O
hand	O
,	O
show	O
advantages	O
for	O
biological	O
plausibility	O
,	O
learning	O
,	O
robustness	B-Application
(	O
fault-tolerant	O
processing	O
and	O
graceful	O
decay	O
)	O
,	O
and	O
generalization	O
to	O
similar	O
input	O
.	O
</s>
