Information Technology Reference
In-Depth Information
s
m
•
is a number of neutrons of a
m
-th layer of the neural network;
•
W
is a matrix of the coefficients of all inputs;
•
b
is neurons input bias;
F
m
•
is the transfer function of the
m
-th layer exit.
Now, we describe the IM-representation of the results of the work on the above
multilayered network.
Let
P
be an input vector in the form
p
1
...
p
R
P
=
s
R
.
p
0
s
1
...
Let the weight coefficients of the connections between the nodes of the input
vector and these from the first layer be given by the IM
a
1
,
1
...
a
1
,
s
1
p
1
W
1
,
1
W
1
,
s
1
...
W
1
=
,
.
.
.
. . .
p
R
W
R
,
1
...
W
R
,
s
1
while let the parameters of the moves of the neurons from the first layer be given by
the IM
a
1
,
1
...
a
1
,
s
1
B
1
=
b
1
,
s
1
.
p
0
b
1
,
1
...
Then,
a
1
is the IM with the values of the neurons in the first layer. It is obtained
by the formula
a
1
W
1
B
1
=
(
P
)
⊕
a
1
,
1
...
a
1
,
s
1
k
=
1
(
R
k
=
1
(
R
=
a
k
W
k
,
1
+
a
k
W
k
,
s
1
+
p
0
b
k
,
1
)...
b
k
,
s
1
)
a
1
,
1
...
a
1
,
s
1
=
a
s
1
.
p
0
a
1
...
. Let the IM of the weight
coefficients of the connections between the nodes of the
i
-th and
Let
i
be a natural number from the set
{
2
,
3
,...,
M
}
(
i
+
1
)
-st layers be
a
1
,
1
...
a
1
,
s
i
W
i
−
1
1
W
i
−
1
1
a
i
−
1
,
1
...
,
1
,
s
i
W
i
=
.
.
.
. . .
a
i
−
1
,
s
i
−
1
W
i
−
1
W
i
−
1
s
i
−
1
...
s
i
−
1
,
1
,
s
i
Search WWH ::
Custom Search