<s>
A	O
large	B-Algorithm
memory	I-Algorithm
storage	I-Algorithm
and	I-Algorithm
retrieval	I-Algorithm
neural	I-Algorithm
network	I-Algorithm
(	O
LAMSTAR	O
)	O
is	O
a	O
fast	O
deep	B-Algorithm
learning	I-Algorithm
neural	B-Architecture
network	I-Architecture
of	O
many	O
layers	O
that	O
can	O
use	O
many	O
filters	O
simultaneously	O
.	O
</s>
<s>
These	O
filters	O
may	O
be	O
nonlinear	O
,	O
stochastic	O
,	O
logic	O
,	O
non-stationary	B-Algorithm
,	O
or	O
even	O
non-analytical	O
.	O
</s>
<s>
A	O
LAMSTAR	O
neural	B-Architecture
network	I-Architecture
may	O
serve	O
as	O
a	O
dynamic	O
neural	B-Architecture
network	I-Architecture
in	O
spatial	O
or	O
time	O
domains	O
or	O
both	O
.	O
</s>
<s>
Its	O
deep	B-Algorithm
learning	I-Algorithm
capability	O
is	O
further	O
enhanced	O
by	O
using	O
inhibition	O
,	O
correlation	O
and	O
by	O
its	O
ability	O
to	O
cope	O
with	O
incomplete	O
data	O
,	O
or	O
"	O
lost	O
"	O
neurons	O
or	O
layers	O
even	O
amidst	O
a	O
task	O
.	O
</s>
<s>
LAMSTAR	O
had	O
a	O
much	O
faster	O
learning	O
speed	O
and	O
somewhat	O
lower	O
error	O
rate	O
than	O
a	O
CNN	O
based	O
on	O
ReLU-function	O
filters	O
and	O
max	O
pooling	O
,	O
in	O
20	O
comparative	O
studies	O
.	O
</s>
