I
n
t
ern
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
E
l
ect
ri
ca
l
a
n
d
C
o
m
p
u
t
er E
n
g
i
n
eeri
n
g
(
I
J
E
C
E
)
V
o
l.
8
,
No
.
5
,
O
c
t
obe
r
20
1
8
, p
p
.
32
32~
3
237
I
S
S
N
:
2088
-
8708
,
D
O
I
:
10.
11
591/
i
j
ece
.
v8
i
5
.
pp
323
2
-
3237
3232
Jou
r
n
al
h
om
e
p
age
:
h
ttp
:
//ia
e
s
c
o
r
e
.
c
o
m/
j
our
nal
s
/
i
nde
x
.
php/
I
J
E
C
E
O
pt
i
m
i
z
a
t
io
n o
f
B
a
ck
pro
pa
g
a
t
io
n f
o
r E
a
rly
Det
ec
t
io
n o
f
Dia
bet
es
M
elli
t
us
R
os
i
t
a S
of
i
a
n
a
1
,
Sut
i
kno
2
1
D
ep
ar
t
m
en
t
o
f
C
o
m
p
u
t
er
S
ci
en
ce
/
I
n
f
o
r
m
at
i
cs
,
F
acu
l
t
y
o
f
S
ci
en
ce an
d
M
at
h
em
at
i
cs
,
D
i
pone
g
or
o U
n
i
v
e
r
s
i
t
y
,
I
ndo
ne
s
i
a
2
C
o
m
p
u
t
er
S
ci
en
ce f
r
o
m
t
h
e G
ad
j
ah
M
ad
a U
n
i
v
er
s
i
t
y
,
I
ndo
ne
s
i
a
A
rt
i
cl
e I
n
f
o
AB
S
T
RAC
A
r
tic
le
h
is
to
r
y
:
R
ecei
v
ed
J
a
n 4
,
201
7
Re
v
i
se
d
J
a
n 9
,
201
8
A
ccep
t
ed
S
e
p 21
,
2
01
8
D
ia
b
e
te
s
m
e
llitu
s
is
o
n
e
o
f
th
e
u
r
g
e
n
t h
e
a
lth
p
r
o
b
le
m
s
in
th
e
w
o
r
ld
.
D
ia
b
e
te
s
is
a
c
o
n
d
iti
o
n
p
r
im
a
r
ily
d
e
f
in
e
d
b
y
th
e
le
v
e
l o
f
h
y
p
e
r
g
l
y
c
e
m
i
a
g
i
v
in
g
r
is
e
to
r
i
s
k
o
f
m
i
cr
o
v
as
cu
l
ar
d
a
m
ag
e.
T
h
o
s
e w
h
o
s
u
f
f
er
f
r
o
m
t
h
i
s
d
i
s
eas
e
g
en
er
al
l
y
do n
ot
r
e
a
l
i
z
e
a
nd t
e
nd t
o ov
e
r
l
o
ok
t
he
e
a
r
l
y
s
y
m
pt
o
m
s
.
L
a
t
e
r
e
c
og
ni
t
i
o
n of
th
e
s
e ear
l
y
s
y
m
p
t
o
m
s
m
a
y
d
r
i
v
e
t
h
e d
i
s
eas
e t
o
a m
o
r
e co
n
cer
n
i
n
g
l
ev
el
.
O
n
e
s
ol
ut
i
on t
o s
ol
v
e
t
hi
s
pr
obl
e
m
i
s
t
o c
r
e
a
t
e
a
n a
ppl
i
c
a
t
i
on t
ha
t
m
a
y
pe
r
f
or
m
e
a
r
l
y
d
e
te
c
tio
n
o
f
d
ia
b
e
te
s
m
e
llitu
s
s
o
th
a
t
it
d
o
e
s
n
o
t
g
r
o
w
la
r
g
e
r
.
I
n
th
is
ar
t
i
cl
e,
a n
e
w
m
et
h
o
d
i
n
p
er
f
o
r
m
in
g
e
a
r
l
y
d
e
te
c
tio
n
o
f
d
ia
b
e
te
s
m
e
llitu
s
is
s
ug
g
e
s
t
e
d.
T
hi
s
m
e
t
hod i
s
ba
c
k
pr
opa
g
a
t
i
on w
i
t
h t
hr
e
e
opt
i
m
i
z
a
t
i
on na
m
e
l
y
e
a
r
l
y
in
itia
liz
a
tio
n
w
ith
N
g
u
y
e
n
-
W
id
r
o
w
a
lg
o
r
ith
m
,
le
a
r
n
in
g
r
a
te
a
d
a
p
tiv
e
de
t
e
r
m
i
na
t
i
on,
a
n
d
de
t
e
r
m
i
na
t
i
on of
w
e
i
g
ht
c
ha
ng
e
by
a
ppl
y
i
ng
m
o
me
n
t
u
m
c
oe
f
f
i
c
i
e
nt
.
T
he
obs
e
r
v
a
t
i
on i
s
c
onduc
t
e
d by
c
ol
l
e
c
t
i
ng
150 da
t
a
c
ons
i
s
t
i
ng
of
79
di
a
be
t
e
s
m
e
l
l
i
t
us
pa
t
i
e
nt
a
n
d
71
no
n di
a
be
t
e
s
m
e
l
l
i
t
us
pa
t
i
e
nt
da
t
a
.
T
he
r
e
s
ul
t
of
t
hi
s
s
t
udy
i
s
t
he
s
ugg
e
s
t
e
d a
l
g
or
i
t
hm
s
uc
c
e
e
ds
i
n de
t
e
c
t
i
ng
di
a
be
t
e
s
m
e
llitu
s
w
ith
accu
r
ac
y
r
at
e o
f
9
9
.
3
3
%
.
O
p
t
i
m
i
zed
b
ack
p
r
o
p
ag
at
i
o
n
al
g
o
r
i
t
h
m
m
a
y
al
l
o
w
t
h
e t
r
ai
n
i
n
g
p
r
o
ces
s
g
o
es
1
2
.
4
t
i
m
es
f
as
t
er
t
h
an
s
t
an
d
ar
d
ba
c
k
pr
opa
g
a
t
i
on.
Ke
y
wo
rd
s
:
A
d
ap
t
i
v
e l
ear
n
i
n
g
r
at
e
B
ack
p
r
o
p
ag
at
i
o
n
M
o
m
en
t
u
m
co
ef
f
i
ci
e
n
t
N
guy
e
n
W
i
dr
o
w
O
p
ti
m
iz
a
tio
n
C
opy
r
i
g
ht
©
201
8
I
ns
t
i
t
ut
e
o
f
A
d
v
anc
e
d E
ngi
ne
e
r
i
ng
an
d
Sc
i
e
nc
e
.
A
l
l
ri
g
h
t
s re
se
rv
e
d
.
Co
rre
sp
o
n
d
i
n
g
Au
t
h
o
r
:
S
ut
i
k
no
,
D
ep
ar
t
me
n
t
o
f
C
o
m
p
u
t
er
S
ci
e
n
ce/
I
n
f
o
r
m
at
i
cs
,
D
i
p
o
ne
go
r
o
U
ni
ve
r
s
i
t
y,
P
r
o
f
.
S
o
ed
ar
t
o
S
t
r
eet
,
T
em
b
al
an
g
,
S
e
m
ar
a
n
g
,
I
n
d
o
n
e
s
i
a.
E
m
a
il:
t
i
k@
un
di
p.
a
c
.
i
d
1.
I
NT
RO
D
UCT
I
O
N
D
i
a
be
t
e
s
m
e
l
l
i
t
u
s
i
s
o
n
e
of
t
he
w
or
l
d
'
s
u
r
g
e
n
t
h
e
a
l
t
h
pr
obl
e
m
s
[
1]
.
A
c
c
or
di
ng
t
o t
h
e
da
t
a
obt
a
i
n
e
d
f
r
o
m
W
or
d H
e
a
l
t
h
O
r
g
a
n
i
z
a
t
i
on
(
W
H
O
)
,
t
h
e
r
e
w
e
r
e
1.
5
m
i
l
l
i
on
pe
opl
e
di
e
d f
r
om
di
a
be
t
e
s
m
e
l
l
i
t
us
i
n
2012
[
1
]
.
I
t is
p
r
e
d
ic
te
d
t
h
a
t
t
h
e
n
um
be
r
of
pe
opl
e
s
u
f
f
e
r
i
n
g
f
r
om
di
a
be
t
e
s
m
e
l
l
i
t
us
w
i
l
l
r
e
a
c
h 366
m
i
l
l
i
on
i
n
2030
[
2
]
.
D
i
a
b
e
t
e
s
i
s
a
c
o
nd
i
t
i
o
n
p
r
i
m
a
r
i
l
y d
e
f
i
ne
d
b
y t
he
l
e
ve
l
o
f
h
yp
e
r
gl
yc
e
m
i
a
gi
vi
ng
r
i
s
e
t
o
r
i
s
k o
f
m
i
c
r
o
v
as
c
u
l
ar
d
a
m
ag
e
[
2
]
.
T
h
e
ear
l
y
s
y
m
p
t
o
m
s
o
f
t
h
i
s
d
i
s
ea
s
e
i
n
cl
u
de
pol
y
di
ps
i
a
(
e
x
c
e
s
s
i
v
e
t
h
i
r
s
t
)
,
pol
y
p
h
a
gi
a
(
ex
ces
s
i
v
e h
u
n
g
er
)
an
d
p
o
l
y
u
r
i
a (
ex
ce
s
s
i
v
e u
r
i
n
at
i
o
n
v
o
l
u
m
e)
.
D
i
ab
et
es
m
el
l
i
t
u
s
m
a
y
cau
s
e o
r
g
an
d
ys
f
u
nc
t
i
o
n a
nd
f
a
i
l
ur
e
[
3
]
.
T
h
e
p
at
i
en
t
s
h
ar
d
l
y
r
eco
g
n
i
ze
an
d
t
e
n
d
t
o
o
v
er
l
o
o
k
t
h
es
e
e
ar
l
y
s
y
m
p
t
o
m
s
i
n
d
i
cat
i
n
g
t
h
e
r
i
sk
s
t
h
e
d
i
s
eas
e b
r
i
n
g
s
.
M
ed
i
cal
a
n
d
l
ab
o
r
at
o
r
y
ex
a
m
i
n
at
i
o
n
ar
e o
n
l
y
en
f
o
r
ced
w
h
e
n
t
h
e p
at
i
en
t
s
s
u
f
f
er
f
r
o
m
s
ev
er
e
p
ai
n
.
L
at
e r
eco
g
n
i
t
i
o
n
o
f
t
h
e
ear
l
y
s
y
m
p
t
o
m
s
r
es
u
l
t
s
i
n
t
h
e
i
n
cr
eas
e o
f
t
h
e
n
u
m
b
er
o
f
d
i
a
b
et
es
m
e
l
l
i
t
u
s
cas
e.
A
n
o
t
h
er
a
f
t
er
ef
f
ect
o
f
l
a
t
e
ex
a
m
i
n
at
i
o
n
an
d
t
r
eat
m
en
t
i
s
t
h
at
t
h
e
y
m
a
y
d
e
v
el
o
p
t
h
e d
i
s
eas
e t
o
b
e
m
o
r
e
d
an
g
er
o
u
s
.
O
n
e s
o
l
u
t
i
o
n
t
o
s
o
l
v
e t
h
i
s
p
r
o
b
l
e
m
i
s
t
o
cr
eat
e an
ap
p
l
i
cat
i
o
n
t
h
at
m
a
y
p
er
f
o
r
m
ear
l
y
d
et
ect
i
o
n
o
f
d
ia
b
e
te
s
m
e
llit
u
s
s
o
th
a
t it d
o
e
s
n
o
t g
r
o
w
la
r
g
e
r
.
S
tu
d
ie
s
c
o
n
c
e
r
n
in
g
th
i
s
m
at
t
er
h
a
v
e b
een
car
r
i
ed
o
u
t
i
n
s
e
v
er
al
m
et
h
o
d
s
,
s
u
ch
as
c
o
m
b
i
n
i
n
g
R
eg
r
es
s
i
o
n
T
r
ee an
d
R
an
d
o
m
F
o
r
es
t
(
R
F
)
[
4
]
,
F
u
zz
y
H
i
er
ar
ch
i
cal
M
o
d
el
[
5
]
,
G
en
et
i
c P
r
o
g
r
am
m
i
n
g
[
6
]
,
S
u
p
p
o
r
t
V
ect
o
r
Mach
i
n
e
s
(
S
V
M
)
,
N
aï
v
e B
a
y
es
[
7
]
an
d
ar
t
i
f
i
ci
a
l
n
e
u
r
al
n
et
w
o
r
k
[
8
]
,
[
9
]
.
I
n
p
u
t d
a
ta
to
b
e
in
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t
J
E
l
e
c
&
C
o
m
p
E
n
g
I
S
S
N
:
2088
-
8708
O
pt
i
m
i
z
at
i
on of
B
ac
k
pr
o
p
agat
i
on f
or
E
a
r
ly
D
e
te
c
tio
n
o
f.
.
.
(
R
os
i
t
a Sof
i
ana
)
3233
t
h
o
u
g
h
o
n
t
h
e
s
e
m
et
h
o
d
s
ca
n
b
e a d
i
g
i
t
al
i
m
a
g
e,
v
o
i
ce,
el
ec
t
r
o
car
d
i
o
g
r
am
(
E
C
G
)
s
i
g
n
al
a
n
d
n
u
m
er
i
c.
I
m
ag
e
da
t
a
i
n
pu
t
i
n
c
l
u
de
s
hum
a
n
body
pa
r
t
s
.
S
o
m
e
e
x
a
m
pl
e
s
a
r
e
i
r
i
s
i
m
a
g
e
[
10]
,
[
11]
,
f
a
c
e
a
r
e
a
[
12]
,
[
13]
a
n
d
m
ag
n
et
i
c
r
es
o
n
a
n
ce
i
m
ag
i
n
g
o
f
t
h
e
br
a
i
n
[
14]
.
V
oi
c
e
da
t
a
m
a
y
a
l
s
o
be
i
n
c
l
u
de
d
a
s
a
d
a
t
a
i
n
put
ba
s
e
d
on
s
e
v
e
r
a
l p
a
r
a
m
e
te
r
s
t
h
a
t c
o
n
s
is
t o
f
a
b
s
o
l
u
te
j
itte
r
,
s
h
i
m
m
e
r
,
a
m
p
lit
u
d
e
p
e
r
tu
r
b
a
tio
n
q
u
o
tie
n
t,
n
o
is
e
-
to
-
h
a
r
m
o
n
ic
r
a
tio
,
s
m
o
o
th
e
d
a
m
p
litu
d
e
p
e
r
tu
r
b
a
tio
n
q
u
o
tie
n
t a
n
d
r
e
la
tiv
e
a
v
e
r
a
g
e
pe
r
t
u
r
ba
t
i
on [
15]
.
A
r
t
i
f
i
ci
al
n
eu
r
al
n
et
w
o
r
k
i
s
a
n
ex
cel
l
e
n
t
m
et
h
o
d
t
o
d
i
ag
n
o
s
e
d
i
s
eas
e
[
8
]
,
[
9
]
,
[
1
6
-
20]
.
J
a
y
a
l
a
ks
m
i
a
n
d S
a
n
s
t
h
a
kum
a
r
a
n poi
n
t
out
t
h
a
t
a
r
t
i
f
i
c
i
a
l
n
e
u
r
a
l
n
e
t
w
or
k
m
a
y
be
i
m
pl
a
n
t
e
d i
n
di
a
g
n
os
i
n
g di
a
be
t
e
s
m
e
l
l
i
t
u
s
an
d
cl
as
s
i
f
y
i
n
g
t
h
e ear
l
y
d
et
e
ct
i
o
n
o
f
g
es
t
at
i
o
n
al
d
i
ab
et
es
m
el
l
i
t
u
s
[
8
]
.
T
h
e act
i
v
e p
ar
a
m
et
er
s
i
n
v
o
l
v
e t
h
e
n
um
be
r
of
pr
e
gn
a
nt
t
i
m
e
s
,
pl
a
s
m
a
g
l
u
c
os
e
c
on
c
e
n
t
r
a
t
i
on
,
bl
ood p
r
e
s
s
u
r
e
,
t
r
i
c
e
ps
s
k
i
n
f
ol
d
t
h
i
c
k
n
e
s
s
,
i
n
s
u
l
i
n
s
er
u
m
,
b
o
d
y
m
a
s
s
i
n
d
ex
,
d
i
ab
et
i
c p
ed
i
g
r
ee f
u
n
ct
i
o
n
an
d
ag
e.
I
n
an
o
t
h
er
s
t
u
d
y
,
b
ack
p
r
o
p
ag
at
i
o
n
w
a
s
e
m
p
l
o
y
ed
t
o
cl
as
s
i
f
y
t
h
e ear
l
y
d
et
ect
i
o
n
o
f
g
es
t
at
i
o
n
al
d
i
ab
et
es
m
el
l
i
t
u
s
[
9
]
.
T
h
i
s
s
t
u
d
y
o
b
s
er
v
ed
1
1
0
d
at
a an
d
pr
om
ot
e
d 10 pa
r
a
m
e
t
e
r
i
n
put
s
:
f
a
m
i
l
y
h
i
s
t
or
y
o
f
di
a
be
t
e
s
,
pr
e
-
p
r
e
g
na
nc
y b
o
d
y
m
a
s
s
i
nd
e
x,
hi
s
t
o
r
y o
f
g
e
s
ta
t
io
n
a
l d
i
ab
et
es
,
d
el
i
v
er
y
o
f
a l
ar
g
e
i
n
f
a
n
t
,
h
i
s
t
o
r
y
o
f
m
i
s
car
r
i
ag
e,
ab
n
o
r
m
al
b
ab
y
i
n
p
r
ev
i
o
u
s
p
r
eg
n
a
n
c
y
,
h
is
to
r
y
o
f
s
til
lb
ir
th
,
in
f
e
c
tio
n
s
,
a
n
d
h
is
to
r
y
o
f
p
o
ly
c
y
s
ti
c
o
v
a
r
y
s
y
n
d
r
o
m
e
.
T
h
e
w
e
a
k
n
e
s
s
o
f
a
p
p
l
y
i
n
g
b
a
c
kp
r
o
p
a
ga
t
i
o
n ne
ur
a
l
ne
t
w
o
r
k i
s
i
t
ha
s
s
l
o
w
e
r
c
o
n
ve
r
ge
n
c
e
a
n
d
l
o
n
g
e
r t
ra
i
n
i
n
g
t
i
m
e
s
[2
1
],
[2
2
].
T
h
er
e ar
e
s
ev
er
al
act
i
o
n
s
t
a
k
en
t
o
r
eco
n
d
i
t
i
o
n
t
h
i
s
w
ea
k
n
es
s
,
s
u
c
h
as
s
el
ec
t
i
n
g
o
r
ad
j
u
s
t
i
n
g
t
h
e
a
c
t
i
va
t
i
o
n f
unc
t
i
o
n
us
e
d
[
2
2
]
,
[
2
3
]
,
p
r
e
p
a
r
i
ng t
he
d
a
t
a
b
e
f
o
r
e
t
he
t
r
a
i
ni
n
g s
t
a
r
t
s
[
2
4
]
,
r
e
f
i
ni
n
g t
he
w
e
i
g
ht
ch
an
g
e o
f
t
he
ne
t
w
o
r
k
w
i
t
h
m
o
m
e
nt
u
m
c
o
e
f
f
i
c
i
e
nt
[
2
1
]
,
[
2
5
]
,
m
e
nd
i
n
g t
he
i
ni
t
i
a
l
i
z
a
t
i
o
n o
f
e
a
r
l
y
w
e
i
ght
s
[
2
6
]
,
r
ect
i
f
y
i
n
g
t
h
e l
ear
n
i
n
g
r
at
e [
2
1
]
,
[
2
7
]
,
an
d
r
ev
i
v
i
n
g
t
h
e
i
n
i
t
i
al
i
zat
i
o
n
o
f
t
h
e n
et
w
o
r
k
'
s
ear
l
y
w
ei
g
h
t
s
[
2
8
]
.
I
n t
hi
s
a
r
t
i
c
l
e
,
a
ne
w
m
e
t
ho
d
i
n p
e
r
f
o
r
m
i
n
g
e
a
r
l
y
de
t
e
c
t
i
o
n
of
di
a
be
t
e
s
m
e
l
l
i
t
u
s
i
s
pr
opos
e
d.
T
h
i
s
m
e
t
h
od i
s
b
a
c
k
p
r
o
p
a
g
a
tio
n
w
it
h
t
h
r
e
e
o
p
tim
iz
a
tio
n
n
a
m
e
l
y
i
n
itia
liz
a
tio
n
o
f
e
a
r
l
y
w
e
ig
h
t
s
w
ith
N
g
u
y
e
n
-
W
i
d
r
o
w
al
g
o
r
i
t
h
m
,
l
ear
n
i
n
g
r
at
e
ad
ap
t
i
v
e,
an
d
d
et
er
m
i
n
at
i
o
n
o
f
w
e
i
g
h
t
c
h
a
n
g
e
b
y
ap
p
l
y
i
n
g
m
o
m
en
t
u
m
co
e
f
f
ic
ie
n
t.
T
he
r
e
s
ul
t
o
f
t
hi
s
s
t
ud
y
m
a
y
gi
ve
c
o
nt
r
i
b
ut
i
o
ns
i
n t
he
ne
w
a
l
go
r
i
t
h
m
,
t
he
o
p
t
i
m
i
z
e
d
b
a
c
kp
r
op
a
ga
t
i
o
n
a
l
g
or
i
t
hm
.
I
n
a
ddi
t
i
on
,
t
h
e
pr
opos
e
d a
l
g
or
i
t
hm
c
a
n be
us
e
d
f
or
e
a
r
l
y
de
t
e
c
t
i
on of
di
a
be
t
e
s
m
e
l
l
i
t
us
di
s
e
a
s
e
,
s
o
t
h
e
num
be
r
of
de
a
t
h
s
c
a
u
s
e
d by
t
h
i
s
d
i
s
eas
e ca
n
b
e r
ed
u
ced
.
2.
R
ES
EA
R
C
H
M
ETH
O
D
2.
1.
B
ac
k
p
r
op
agat
i
on
O
p
t
i
m
i
z
at
i
on
M
e
th
o
d
i
m
p
le
m
e
n
te
d
i
n
th
i
s
s
tu
d
y
o
f
e
a
r
l
y
d
e
te
c
tio
n
o
f
d
ia
b
e
te
s
m
e
llit
u
s
i
s
th
e
o
p
tim
iz
e
d
b
ack
p
r
o
p
ag
at
i
o
n
al
g
o
r
i
t
h
m
.
O
p
t
i
m
i
zat
i
o
n
i
s
p
er
f
o
r
m
ed
i
n
t
h
r
ee ap
p
r
o
ach
es
n
a
m
el
y
i
n
itia
liz
a
tio
n
o
f
e
a
r
l
y
w
e
i
g
ht
s
w
i
t
h N
g
u
ye
n
-
W
i
d
r
o
w
al
g
o
r
i
t
h
m
,
l
ear
n
i
n
g
r
at
e a
d
ap
t
i
v
e,
an
d
d
et
er
m
i
n
at
i
o
n
o
f
w
ei
g
h
t
ch
a
n
g
e b
y
a
ppl
y
i
ng
m
o
m
e
n
t
um
c
oe
f
f
i
c
i
e
n
t
.
T
h
e co
m
p
l
et
e
s
u
g
g
es
t
ed
al
g
o
r
i
t
h
m
i
s
d
ep
i
ct
ed
i
n
F
i
g
u
r
e
1
.
T
h
e ex
p
l
an
at
i
o
n
s
f
o
r
each
p
r
o
ces
s
ar
e as
f
o
llo
w
:
a.
T
r
a
i
ni
ng
D
a
ta
T
h
e d
at
a
w
h
i
ch
u
s
ed
i
n
t
h
i
s
r
es
ear
ch
o
r
i
g
i
n
at
ed
f
r
o
m
t
h
e
m
e
d
i
cal
r
eco
r
d
s
o
f
D
r
.
H
.
S
u
w
o
n
d
o
K
en
d
al
H
os
pi
t
a
l
'
s
pa
t
i
e
n
t
s
i
n
2016.
T
h
e
r
e
a
r
e
150 da
t
a
w
hi
c
h
c
ov
e
r
s
79 da
t
a
of
m
e
l
l
i
t
u
s
di
a
be
t
e
s
pa
t
i
e
n
t
s
a
n
d
71 da
t
a
o
f n
o
n
-
m
e
l
lit
u
s
d
ia
b
e
te
s
p
a
tie
n
ts
.
T
h
e
s
e
le
c
tio
n
m
e
th
o
d
f
o
r
th
e
tr
a
i
n
i
n
g
a
n
d
te
s
ti
n
g
is
h
o
l
d
-
ou
t
m
e
t
h
od w
hi
c
h
d
i
vi
d
e
s
t
he
d
a
t
a
r
a
nd
o
m
l
y i
n
t
o
t
w
o
s
e
ts
th
a
t a
r
e
tr
a
i
n
in
g
d
a
t
a
a
n
d
te
s
tin
g
d
a
ta
.
T
h
e
d
a
ta
c
o
m
p
o
s
i
tio
n
is
2
/3
o
f
t
r
a
i
n
i
n
g
da
t
a
a
n
d 1/
3 of
t
e
s
t
i
ng da
t
a
.
b.
I
n
itia
liz
a
tio
n
o
f
n
e
t
w
o
r
k
w
e
i
g
h
t
O
n
t
h
e
s
t
a
n
da
r
d ba
c
k
pr
opa
g
a
t
i
on
a
l
g
or
i
t
hm
i
n
i
t
i
a
l
i
z
a
t
i
o
n
of
n
e
t
w
or
k
w
e
i
gh
t
i
s
don
e
b
y
g
e
n
e
r
a
t
i
ng
ra
n
d
o
m
s
m
a
ll n
u
m
b
e
r
,
m
e
a
n
w
h
ile
i
n
th
i
s
a
r
tic
le
,
N
g
u
y
e
n
-
W
id
r
o
w
te
c
h
n
iq
u
e
w
ill b
e
u
s
e
d
.
T
h
is
te
c
h
n
iq
u
e
is
i
n
t
r
odu
c
e
d b
y
N
guy
e
n
a
n
d W
i
dr
ow
on
t
w
o l
a
y
e
r
s
n
e
u
t
r
a
l
n
e
t
w
or
k
[
28]
.
c.
S
to
p
p
in
g
c
o
n
d
itio
n
T
h
e
pr
oc
e
s
s
of
n
e
t
w
or
k t
r
a
i
n
i
n
g
w
i
l
l
be
s
t
oppe
d i
f
t
h
e
c
on
di
t
i
on
h
a
d
al
r
ead
y
b
ee
n
f
u
l
f
i
l
l
ed
.
T
h
er
e ar
e
t
w
o r
e
qui
r
e
m
e
n
t
s
of
s
t
oppi
n
g c
on
di
t
i
on
t
h
a
t
a
r
e
:
i
f
t
h
e
v
a
l
u
e
of
M
e
a
n S
qu
a
r
e
d E
r
r
or
(
M
S
E
)
r
e
s
u
l
t
e
d b
y
t
h
e
n
et
w
o
r
k
i
s
s
m
al
l
er
t
h
an
t
h
e s
p
eci
f
i
ed
er
r
o
r
v
al
u
e o
r
t
h
e ep
o
ch
o
f
t
r
ai
n
i
n
g
p
r
o
ces
s
i
s
eq
u
al
t
o
t
h
e ep
o
ch
t
h
at
ha
s
b
een
s
p
eci
f
i
e
d.
d.
F
eed
f
o
r
w
ar
d
F
e
e
d f
or
w
a
r
d pr
oc
e
s
s
i
s
us
e
d t
o c
ou
n
t
a
l
l
t
h
e
o
u
t
pu
t
v
a
l
u
e
on h
i
dde
n
l
a
y
e
r
a
n
d ou
t
pu
t
l
a
y
e
r
.
e.
B
ack
p
r
o
p
ag
at
i
o
n
B
ack
p
r
o
p
ag
at
i
o
n
p
r
o
ces
s
i
s
u
s
ed
t
o
cal
cu
l
a
t
e t
h
e r
at
e o
f
w
ei
g
h
t
ch
a
n
g
es
o
n
a
l
l
n
e
t
w
o
r
k
s
.
T
h
i
s
cal
cu
l
at
i
o
n
i
n
cl
u
d
es
t
h
e
va
l
ue
o
f
l
e
a
r
ni
ng r
a
t
e
a
nd
va
l
ue
o
f
ne
t
w
o
r
k i
np
ut
a
nd
a
l
s
o
o
ut
p
ut
o
n e
ve
r
y
hi
d
d
e
n
l
a
ye
r
.
f.
N
e
t
w
o
r
k
w
e
i
g
ht
c
ha
n
ge
s
T
he
ne
t
w
o
r
k
w
e
i
ght
c
ha
nge
s
o
n
s
t
a
nd
a
r
d
b
a
c
kp
r
o
p
a
ga
t
i
o
n
a
l
go
r
i
t
h
m
a
r
e
c
a
l
c
ul
a
t
e
d
b
y
a
d
d
i
ng
t
he
p
r
e
vi
o
us
ne
t
w
o
r
k
w
e
i
g
ht
a
nd
w
e
i
g
ht
c
ha
n
ge
s
r
a
t
e
.
I
n
t
h
i
s
pr
opos
e
d a
l
g
or
i
t
hm
i
n
a
ddi
t
i
on
t
o a
ddi
n
g
t
h
e
w
e
i
ght
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
20
88
-
8708
In
t
J
E
l
e
c
&
C
o
m
p
E
n
g
,
V
o
l.
8
, N
o
.
5
,
O
c
t
obe
r
20
18
:
323
2
-
3237
3234
ch
an
g
e r
at
e al
s
o
ad
d
s
t
h
e
m
o
m
en
t
u
m
co
e
f
f
i
c
i
en
t
.
T
h
i
s
t
ec
h
n
i
q
u
e
h
ad
b
een
ap
p
l
i
ed
b
y
Y
u
an
d
L
i
t
o
o
p
t
i
m
i
ze
ba
c
k
pr
opa
g
a
t
i
on
a
l
g
or
i
t
hm
[
2
1]
.
g.
C
al
cu
l
at
i
n
g
t
h
e M
ean
S
q
u
ar
ed
E
r
r
o
r
(
M
S
E
)
M
S
E
cal
cu
l
at
i
o
n
p
r
o
ces
s
i
s
p
r
act
i
ced
t
o
co
u
n
t
t
h
e av
er
a
g
e v
al
u
e i
n
ev
er
y
ep
o
ch
o
n
t
h
e t
r
ai
n
i
n
g
p
r
o
ces
s
.
O
n
t
h
e
n
ex
t
s
t
ep
,
MS
E
i
s
co
m
p
ar
ed
t
o
t
h
e s
p
eci
f
i
ed
er
r
o
r
v
al
u
e t
o
en
d
t
h
e t
r
ai
n
i
n
g
p
r
o
ces
s
.
h.
C
a
lc
u
la
t
i
ng t
he
l
e
a
r
ni
n
g r
a
t
e
v
a
l
ue
I
n s
t
a
n
da
r
d ba
c
k
pr
opa
g
a
t
i
on
a
l
g
or
i
t
hm
,
le
a
r
ni
n
g r
a
t
e
i
s
n
o
t
c
ha
n
ge
d
i
n i
t
s
e
ve
r
y e
p
o
c
h.
O
n t
h
i
s
a
l
go
r
i
t
h
m
,
l
e
a
r
ni
n
g r
a
t
e
va
l
ue
i
s
c
ha
nge
d
d
ur
i
n
g t
he
t
r
a
i
ni
ng
p
r
o
c
e
s
s
t
o
m
a
i
nt
a
i
n t
he
s
t
a
b
i
l
i
t
y o
f
t
he
a
l
go
r
i
t
h
m
.
T
h
i
s
t
e
c
h
ni
qu
e
i
s
s
ugg
e
s
t
e
d b
y
Y
u
a
n
d
L
i
[
21]
.
i.
S
a
vi
n
g
t
he
ne
t
w
o
r
k w
e
i
g
ht
T
h
e
la
s
t s
te
p
o
f
t
he
ne
t
w
o
r
k t
r
a
i
ni
n
g p
r
o
c
e
s
s
i
s
p
r
o
d
uc
i
ng
t
he
ne
t
w
o
r
k
w
e
i
gh
t
.
T
he
w
e
i
g
ht
i
s
s
a
ve
d
an
d
u
s
ed
o
n
t
h
e t
es
t
i
n
g
p
r
o
ces
s
.
F
i
gu
r
e
1.
T
h
e
pr
opos
e
d m
e
t
h
o
d i
n
t
h
i
s
r
es
ear
ch
2.
2.
A
rt
i
f
i
ci
a
l
N
eu
t
ra
l
N
et
w
o
rk
A
rch
i
t
ect
u
re
A
r
t
i
f
i
ci
al
n
eu
r
al
n
et
w
o
r
k
ar
ch
i
t
ect
u
r
e u
s
ed
i
n
t
hi
s
r
e
s
e
a
r
c
h
c
o
ns
i
s
t
s
o
f
1
2
i
np
ut
u
ni
t
b
a
s
e
d
o
n t
he
a
m
oun
t
o
f
i
n
di
c
a
t
or
v
a
r
i
a
bl
e
of
di
a
be
t
e
s
m
e
l
l
i
t
us
,
on
e
hi
dde
n
l
a
y
e
r
,
a
n
d on
e
o
u
t
pu
t
un
i
t
s
h
o
w
n i
n
F
i
gu
r
e
2.
T
h
e i
n
p
u
t
u
s
ed
i
n
t
h
i
s
ar
c
h
i
t
e
ct
u
r
e ar
e 1
2
p
ar
am
et
er
s
:
a
g
e (
x
1
)
,
h
er
ed
i
t
y
(
x
2
)
,
P
o
l
y
u
r
i
a (
x
3
)
,
P
ol
y
di
ps
i
a
(
x
4)
,
P
ol
y
ph
a
g
i
a
(
x
5)
,
bl
ood s
u
g
a
r
(
x
6)
,
I
n
f
e
c
t
i
on
(
x
7)
,
w
e
i
gh
t
l
os
s
(
x
8)
,
t
i
ng
l
i
ng
(
x
9)
,
f
a
t
i
gu
e
a
n
d dr
o
w
s
i
n
e
s
s
(
x1
0
)
,
ne
a
r
s
i
g
ht
e
d
ne
s
s
(
x1
1
)
,
a
nd
w
o
und
(
x1
2
)
.
T
he
u
n
it o
f
o
u
tp
u
t is
t
h
e
s
ta
tu
s
o
f
d
ia
b
e
te
s
m
illit
u
s
(
y
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t
J
E
l
e
c
&
C
o
m
p
E
n
g
I
S
S
N
:
2088
-
8708
O
pt
i
m
i
z
at
i
on of
B
ac
k
pr
o
p
agat
i
on f
or
E
a
r
ly
D
e
te
c
tio
n
o
f.
.
.
(
R
os
i
t
a Sof
i
ana
)
3235
F
ig
u
r
e
2
.
A
r
ti
f
ic
ia
l
n
eu
t
r
al
n
et
w
o
r
k
ar
c
h
i
t
ect
u
r
e u
s
ed
i
n
t
h
e r
es
ear
ch
3.
R
ES
U
LT
S
A
ND AN
AL
Y
S
I
S
T
he
e
xp
e
r
i
m
e
nt
o
n t
hi
s
r
es
ear
ch
is
c
la
s
s
i
f
ie
d
in
to
t
w
o
e
x
p
e
r
i
m
e
n
ts
b
e
lo
w
:
3.
1.
E
x
pe
r
i
m
e
nt
1
T
h
e p
u
r
p
o
s
e o
f
t
h
i
s
ex
p
er
i
m
e
n
t
i
s
t
o
d
et
er
m
i
n
e t
h
e ef
f
ect
o
f
m
o
m
e
n
t
u
m
v
al
u
es
o
n
t
h
e ac
cu
r
ac
y
o
f
th
e
o
p
ti
m
iz
e
d
b
a
c
k
p
r
o
p
a
g
a
tio
n
a
l
g
o
r
ith
m
to
id
e
n
ti
f
y
d
ia
b
e
t
e
s
m
e
llit
u
s
.
T
h
e
g
iv
e
n
m
o
m
e
n
tu
m
v
a
lu
e
v
a
r
ia
tio
n
s
t
ar
t
s
f
r
o
m
t
h
e
r
an
g
e
o
f
0
.
1
u
p
t
o
0
.
9
.
Mean
w
h
i
l
e,
o
t
h
er
v
ar
i
ab
l
es
ar
e
l
ear
n
i
n
g
r
at
e
(
α
)
o
f
0
.
6
,
m
o
m
en
t
u
m
pa
r
a
m
e
t
e
r
(
µ
)
of
0.
9,
m
a
x
i
m
u
m
e
poc
h
of
10
0
0
,
t
ar
g
et
er
r
o
r
o
f
0
.
0
0
0
1
,
r
at
i
o
o
f
l
ear
n
i
n
g
r
a
t
e i
n
cr
eas
e (
l
r
_
i
n
c)
o
f
1
.
0
5
,
r
at
i
o
o
f
l
ear
n
i
n
g
r
at
e
d
ecr
eas
e (
l
r
_
d
ec
)
o
f
0
.
6
an
d
m
ax
i
m
a
l
i
n
cr
eas
e o
f
p
er
f
o
r
m
an
ce (
m
a
x
_
p
er
f
)
o
f
0
.
6
.
R
e
s
u
lt o
f
t
h
is
te
s
t is
i
n
T
a
b
le
1
.
F
r
o
m
t
ab
l
e 1
,
i
t
ca
n
b
e s
ee
n
t
h
at
t
h
e
h
i
gh
e
s
t
e
poc
h num
be
r
i
s
64
w
h
e
n
t
h
e
m
o
m
e
n
t
um
i
s
0.
1 a
n
d t
he
l
o
w
es
t
ep
o
ch
n
u
m
b
er
i
s
3
0
w
h
en
t
h
e
m
o
m
e
n
t
u
m
v
al
u
e 0
.
9
i
s
g
i
v
e
n
.
B
as
ed
o
n
t
h
e
s
e r
es
u
l
t
s
can
b
e co
n
cl
u
d
ed
t
h
at
t
h
e p
r
o
v
i
s
i
o
n
o
f
m
o
m
e
n
t
u
m
can
accel
er
at
e t
h
e l
ear
n
i
n
g
p
r
o
ces
s
.
T
h
e cau
s
e i
s
t
h
e ad
d
i
t
i
o
n
o
f
m
o
me
n
t
u
m
v
al
u
e
w
i
l
l
r
ed
u
ce t
h
e er
r
o
r
s
o
t
h
at
t
h
e l
ear
n
i
n
g
p
r
o
ces
s
t
i
m
e i
s
s
m
al
l
er
.
F
r
o
m
t
ab
l
e 1
,
i
t
ca
n
al
s
o
b
e s
e
en
t
h
at
i
f
t
h
e
m
o
m
en
t
u
m
v
al
u
e i
s
r
ai
s
ed
,
t
h
e
n
t
h
e acc
u
r
ac
y
l
ev
el
t
e
n
d
s
t
o i
n
c
r
e
a
s
e
.
T
h
i
s
i
s
pr
ov
e
n
w
h
e
n
gi
v
i
ng
t
h
e
m
o
m
e
n
t
um
v
a
l
ue
0.
1 t
o 0.
6
y
i
el
d
s
t
h
e s
a
m
e l
e
v
el
o
f
acc
u
r
ac
y
an
d
t
h
e
m
o
m
en
t
u
m
v
a
l
u
e 0
.
7
t
o
0
.
9
i
n
cr
eas
es
f
r
o
m
t
h
e p
r
ev
i
o
u
s
v
al
u
e.
T
h
e g
r
eat
es
t
accu
r
acy
i
s
9
9
.
3
3
%
o
n
t
h
e
m
o
m
e
n
t
um
c
oe
f
f
i
c
i
e
n
t
of
0.
9.
T
h
i
s
r
e
s
u
l
t
i
s
be
t
t
e
r
t
h
a
n
t
h
e
s
t
a
n
da
r
d
ba
c
k
pr
opa
g
a
t
i
on
m
e
t
h
od
pr
op
os
e
d
by
D
u
r
ai
r
aj
an
d
K
al
ai
s
el
v
i
w
i
t
h
an
accu
r
ac
y
o
f
9
1
%
[
1
6
]
.
T
ab
le
1
.
M
o
m
en
t
u
m
V
al
u
e E
f
f
ect
t
o
w
ar
d
s
R
es
u
l
t
ed
A
cc
u
r
acy
L
e
v
el
M
o
m
e
n
t
u
m
E
p
o
ch
A
ccu
r
acy
(
%
)
M
SE
0
.1
6
4
9
7
.
33
0
.
0
00
0
9
98
08
0
.2
6
2
9
7
.
33
0
.
0
00
0
9
76
47
0
.3
5
9
9
7
.
33
0
.
0
00
0
9
97
04
0
.4
5
7
9
7
.
33
0
.
0
00
0
9
94
56
0
.5
6
3
9
7
.
33
0
.
0
00
0
9
92
24
0
.6
5
4
9
7
.
33
0
.
0
00
0
9
82
52
0
.7
5
3
9
8
.
67
0
.
0
00
0
9
75
97
0
.8
5
8
9
8
.
67
0
.
0
00
0
9
71
22
0
.9
3
0
9
9
.
33
0
.
0
00
0
9
49
20
3.
2.
E
x
pe
r
i
m
e
nt
2
T
h
e p
u
r
p
o
s
e o
f
t
h
i
s
s
eco
n
d
ex
p
er
i
m
en
t
i
s
t
o
co
m
p
ar
e t
h
e p
er
f
o
r
m
a
n
ce b
et
w
ee
n
s
t
an
d
ar
d
ba
c
k
pr
opa
g
a
t
i
on
a
l
g
or
i
t
hm
a
nd opt
i
m
i
z
e
d ba
c
k
pr
opa
g
a
t
i
on
a
l
g
or
i
t
hm
f
or
e
a
r
l
y
de
t
e
c
t
i
o
n
of
di
a
be
t
e
s
m
e
l
l
i
t
u
s
d
i
s
eas
e.
T
es
t
s
u
s
ed
s
e
v
er
al
v
a
r
i
at
i
o
n
s
o
f
α
v
al
u
e
o
f
0
.
1
t
o
0
.
9
,
m
a
x
i
m
um
e
poc
h
of
1000 a
n
d e
r
r
or
t
a
r
g
e
t
of
0.
000
1.
T
h
e
pa
r
a
m
e
t
e
r
s
v
a
l
u
e
of
opt
i
m
i
z
e
d ba
c
k
r
opa
g
a
t
i
on
a
r
e
m
o
m
e
n
t
um
c
oe
f
f
i
c
i
e
n
t
o
f
0.
9,
r
a
t
i
o of
l
e
a
r
n
i
ng
r
at
e i
n
cr
eas
e
(
l
r
_
i
n
c)
o
f
1
.
0
5
,
r
at
i
o
o
f
l
ear
n
i
n
g
r
at
e d
ec
r
eas
e (
l
r
_
d
ec)
o
f
0
.
6
an
d
m
ax
i
m
al
i
n
cr
eas
e
o
f
p
er
f
o
r
m
a
n
ce (
m
a
x
_
p
er
f
)
o
f
1
.
0
6
.
T
h
e r
es
u
l
t
s
o
f
t
h
i
s
ex
p
er
i
m
en
t
ar
e s
h
o
w
n
i
n
T
ab
l
e
2.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
20
88
-
8708
In
t
J
E
l
e
c
&
C
o
m
p
E
n
g
,
V
o
l.
8
, N
o
.
5
,
O
c
t
obe
r
20
18
:
323
2
-
3237
3236
B
as
ed
o
n
t
ab
l
e 2
,
i
t
ca
n
b
e s
e
en
t
h
at
t
h
e i
n
cr
eas
e o
f
l
ear
n
i
n
g
r
at
e
v
al
u
e
i
n
b
ack
p
r
o
p
ag
at
i
o
n
s
t
a
n
d
ar
d
g
i
v
e
s
de
c
r
e
a
s
e
of
e
poc
h
n
um
be
r
i
n
r
a
n
g
e
354 u
p t
o 1000.
T
h
e
de
c
r
e
a
s
e
o
f
e
poc
h num
b
e
r
a
l
s
o
h
a
ppe
n
e
d a
t
opt
i
m
i
z
e
d ba
c
k
pr
opa
g
a
t
i
on
.
I
f
t
h
e
v
a
l
u
e
of
l
e
a
r
n
i
ng
r
a
t
e
i
s
0.
1 t
o 0.
7 i
n
opt
i
m
i
z
e
d ba
c
k
pr
opa
g
a
t
i
on
t
h
e
n
t
he
e
p
o
c
h nu
m
b
e
r
w
i
l
l
t
e
nd
t
o
f
a
l
l
i
n t
he
r
a
n
ge
3
8
t
o 6
2 a
nd
i
f
t
h
e
va
l
ue
o
f
l
e
a
r
ni
n
g r
a
t
e
i
s
gi
ve
n 0
.
8
a
nd
0.
9
t
he
n
th
e ep
o
ch
n
u
m
b
er
i
n
cr
eas
ed
t
o
4
9
.
T
h
e cau
s
e i
s
t
h
e
v
al
u
e
o
f
l
ear
n
i
n
g
r
at
e
i
s
t
o
o
s
m
al
l
r
es
u
l
t
ed
t
h
e
l
ear
n
i
n
g
p
r
o
ces
s
t
ak
es
a l
o
n
g
t
i
m
e t
o
r
each
co
n
v
er
g
e
n
t
,
b
u
t
i
f
t
h
e v
a
l
u
e o
f
l
ear
n
i
n
g
r
at
e i
s
t
o
o
l
ar
g
e p
r
o
ces
s
,
t
h
en
t
h
e
l
ear
n
i
n
g
w
i
l
l
b
eco
m
e u
n
s
t
ab
l
e.
F
r
o
m t
h
e t
e
s
t
r
es
u
l
t
s
i
n
T
ab
l
e 2
,
i
t
i
s
s
een
t
h
at
o
p
t
i
m
i
zed
b
ack
p
r
o
p
ag
at
i
o
n
can
d
ecr
eas
e
t
h
e ep
o
ch
n
um
be
r
s
i
gn
i
f
i
c
a
n
t
l
y
b
y
12.
4 t
i
m
e
s
c
o
m
pa
r
e
d t
o
s
t
a
n
da
r
d ba
c
k
pr
opa
g
a
t
i
on
.
I
t
c
a
n
be
c
on
c
l
u
de
d t
h
a
t
o
p
t
i
m
i
zed
b
ack
p
r
o
p
ag
at
i
o
n
r
eq
u
i
r
es
a
m
u
ch
f
as
t
er
l
ear
n
i
n
g
p
r
o
ces
s
t
ha
n t
he
s
t
a
nd
a
r
d
b
a
c
kp
r
op
a
ga
t
i
o
n
pr
op
os
e
d by
[
16]
a
n
d [
17
]
i
n
t
h
e
s
a
m
e
c
a
s
e
.
T
h
e
v
a
l
u
e
of
l
e
a
r
n
i
ng
r
a
t
e
0.
7 r
e
s
u
l
t
e
d i
n
t
he
n
um
be
r
of
e
poc
h
471 on
t
h
e
ba
c
k
pr
opa
g
a
t
i
on
s
t
a
n
da
r
d a
n
d t
h
e
e
poc
h
num
be
r
of
38 i
n
t
h
e
opt
i
m
i
z
e
d ba
c
k
pr
opa
g
a
t
i
i
on
.
T
ab
l
e 2
.
P
er
f
o
r
m
a
n
ce
Co
m
p
a
r
i
so
n
b
e
t
we
en
S
t
an
d
ar
d
B
ack
p
r
o
p
ag
at
i
o
n
a
n
d O
pt
i
m
i
z
e
d B
a
c
k
pr
opa
g
a
t
i
on
L
e
ar
n
i
n
g
R
at
e
(
α
)
S
t
a
n
da
r
d
B
a
c
kp
r
o
pa
ga
t
i
o
n
O
p
t
i
m
i
zed
B
a
ck
p
r
o
p
a
g
at
i
o
n
E
p
o
ch
M
SE
E
p
o
ch
M
SE
0
.1
1
0
0
0
0
.
0
00
5
0
42
65
6
2
0
.
0
00
0
9
92
77
4
0
.2
1
0
0
0
0
.
0
00
1
9
72
36
6
1
0
.
0
00
0
9
71
70
0
0
.3
1
0
0
0
0
.
0
00
1
1
80
91
5
1
0
.
0
00
0
9
81
28
9
0
.4
8
5
8
0
.
0
00
0
9
99
02
4
5
0
.
0
00
0
9
75
99
3
0
.5
6
8
1
0
.
0
00
0
9
98
90
4
4
0
.
0
00
0
9
74
00
9
0
.6
5
7
9
0
.
0
00
0
9
98
59
3
8
0
.
0
00
0
9
67
79
1
0
.7
4
7
1
0
.
0
00
0
9
98
68
3
8
0
.
0
00
0
9
77
89
6
0
.8
4
1
2
0
.
0
00
0
9
98
34
4
9
0
.
0
00
0
9
84
44
5
0
.9
3
5
4
0
.
0
00
0
9
98
29
4
9
0
.
0
00
0
9
90
08
0
4.
CO
NCL
U
S
I
O
N
O
p
ti
m
iz
e
d
b
a
c
k
p
r
o
p
a
g
a
tio
n
a
lg
o
r
ith
m
w
i
th
i
n
itia
liz
a
t
io
n
o
f
e
a
r
l
y
w
e
i
g
h
ts
w
i
th
N
g
u
y
e
n
-
W
i
d
r
o
w
al
g
o
r
i
t
h
m
,
l
ear
n
i
n
g
r
at
e ad
ap
t
i
v
e,
an
d
d
et
er
m
i
n
at
i
o
n
o
f
w
ei
g
h
t
ch
a
n
g
e b
y
ap
p
l
y
i
n
g
m
o
m
e
n
t
u
m
co
e
f
f
i
c
i
en
t
f
o
r
ear
l
y
d
et
ect
i
o
n
o
f
d
i
ab
et
es
m
el
l
i
t
u
s
t
h
at
p
r
o
d
u
ced
t
h
e b
es
t
accu
r
ac
y
l
ev
e
l
o
f
9
9
.
3
3
%
.
O
p
t
i
m
i
zed
b
ack
p
r
o
p
ag
at
i
o
n
i
s
ab
l
e t
o
accel
er
at
e t
h
e t
r
ai
n
i
n
g
p
r
o
ces
s
b
y
1
2
.
4
t
i
m
es
co
m
p
ar
e
t
o
s
t
a
n
d
ar
d
ba
c
k
pr
opa
g
a
t
i
on
.
R
EF
ER
EN
C
ES
[
1]
W
o
r
ld
H
e
a
lth
O
r
g
a
n
iz
a
tio
n
,
“G
l
o
b
al
R
ep
o
r
t
o
n
D
i
ab
et
es
,
”
W
or
l
d H
e
al
t
h O
r
g
an
i
z
at
i
on
(
W
HO)
,
20
16.
[
2]
W
o
r
ld
H
e
a
lth
O
r
g
a
n
iz
a
tio
n
,
“D
e
f
i
n
i
t
i
o
n
an
d
D
i
ag
n
o
s
i
s
o
f
D
i
ab
et
es
M
el
l
i
t
u
s
an
d
I
n
t
er
m
ed
i
at
e
H
y
p
er
g
l
y
ce
m
i
a,
”
W
or
l
d H
e
al
t
h O
r
ga
ni
z
a
t
i
o
n (
W
H
O
)
,
2
00
6.
[
3]
V
.
J
a
k
h
m
o
la
a
n
d
P
.
T
a
n
g
r
i,
“
D
ia
b
e
te
s
M
e
llitu
s
a
S
i
le
n
t K
ille
r
: R
o
le
o
f
D
P
P
4
I
n
h
ib
i
to
r
s
i
n
T
r
e
a
t
m
e
n
t,
”
J
our
nal
of
P
h
a
r
m
a
ceu
t
i
ca
l
S
ci
en
ce A
n
d
B
i
o
s
ci
en
t
i
f
i
c R
es
ea
r
ch
(
JP
S
B
R
)
,
v
o
l
. 2
,
p
p
. 4
9
–
53
,
2
01
2.
[
4]
M
. T
. M
. K
. S
a
b
a
r
i
a
h
,
e
t a
l.
,
“E
ar
l
y
D
et
ect
i
o
n
o
f
T
y
p
e
I
I
D
ia
b
e
te
s
M
e
llitu
s
w
ith
R
a
n
d
o
m
F
o
r
e
s
t a
n
d
C
la
s
s
if
ic
a
tio
n
an
d
R
eg
r
es
s
i
o
n
T
r
ee (
C
A
R
T
)
,
”
A
dv
anc
e
d
I
nf
or
m
at
i
c
s
:
C
onc
e
pt
,
T
he
or
y
an
d
A
p
pl
i
c
at
i
on
(
I
C
A
I
C
T
A
)
,
I
E
E
E
, p
p
.
238
-
2
42,
2
01
4.
[
5]
R
.
B
.
L
u
m
an
t
o
an
d
E
.
I
r
w
an
s
y
a
h
,
“T
h
e E
ar
l
y
D
et
ect
i
o
n
o
f
D
i
ab
et
es
M
el
l
i
t
u
s
(
D
M
)
U
s
i
n
g
F
u
z
z
y
H
i
er
ar
ch
i
ca
l
M
o
d
el
,
”
P
r
oc
e
di
a
C
om
p
ut
e
r
Sc
i
e
nc
e
,
v
ol
.
59,
p
p.
31
2
–
31
9,
20
15.
[
6]
M
. A
. P
r
a
d
h
a
n
,
e
t a
l.
,
“
D
e
s
ig
n
o
f
C
la
s
s
i
f
ie
r
f
o
r
D
e
te
c
tio
n
o
f
D
ia
b
e
te
s
M
e
llitu
s
U
s
in
g
G
e
n
e
tic
P
r
o
g
r
a
m
m
in
g
,
”
I
nt
e
r
nat
i
o
nal
C
o
nf
e
r
e
nc
e
on C
o
m
put
e
r
Sc
i
e
nc
e
and I
nf
or
m
a
t
i
o
n
T
e
c
hnol
ogy
(
I
C
C
SI
T
’
2
01
1)
,
I
E
E
E
,
pp.
12
5
–
13
0,
201
1.
[
7]
B
.
A
.
T
a
m
a an
d
F
.
S
.
R
o
d
i
y
at
u
l
,
“A
n
E
ar
l
y
D
et
ect
i
o
n
M
et
h
o
d
o
f
T
y
p
e
-
2
D
ia
b
e
te
s
M
e
llitu
s
in
P
u
b
lic
H
o
s
p
ita
l,
”
T
el
ko
m
n
i
ka
,
v
ol
.
9,
pp
.
2
87
–
2
94
,
20
11.
[
8]
T
.
J
a
y
al
a
k
s
h
m
i
an
d
A
.
S
an
t
h
ak
u
m
ar
an
,
“
A
N
o
v
el
C
l
as
s
i
f
i
cat
io
n
M
e
th
o
d
f
o
r
D
ia
g
n
o
s
is
o
f
D
ia
b
e
te
s
M
e
llitu
s
U
s
in
g
A
r
t
i
f
i
ci
al
N
eu
r
al
N
et
w
o
r
k
s
,
”
2010 I
nt
e
r
n
at
i
o
na
l
C
o
nf
e
r
e
nc
e
o
n
D
at
a
St
or
age
an
d D
at
a
E
n
gi
ne
e
r
i
ng
, I
E
E
E
, p
p
.
159
-
1
53,
20
10
.
[
9]
P
.
S
.
M
u
lle
r
,
e
t a
l.
,
“
A
ppl
i
c
a
t
i
on
of
C
om
put
a
t
i
o
na
l
T
e
c
hni
q
ue
i
n D
e
s
i
g
n
of
C
l
a
s
s
i
f
i
er
f
o
r
E
ar
l
y
D
et
ect
i
o
n
o
f
G
e
s
ta
tio
n
a
l D
ia
b
e
te
s
M
e
llit
u
s
,
”
A
ppl
i
e
d
M
at
he
m
a
t
i
c
al
Sc
i
e
nc
e
s
,
vol
.
9
,
pp
.
3
32
7
–
3
33
6,
20
15.
[
1
0]
I
.
P
. D
. L
e
s
m
a
n
a
,
e
t a
l.
,
“A
b
n
o
r
m
al
C
o
n
d
i
t
i
o
n
D
et
ect
i
o
n
o
f
P
an
cr
eat
i
c B
et
a
-
C
el
l
s
as
T
h
e C
a
u
s
e o
f
D
i
ab
et
es
M
el
l
i
t
u
s
b
as
ed
o
n
I
r
i
s
I
m
ag
e,
”
I
ns
t
r
um
e
nt
at
i
on
,
C
om
m
u
ni
c
at
i
ons
,
I
nf
or
m
at
i
on T
e
c
h
nol
ogy
,
a
nd B
i
om
e
di
c
a
l
E
ngi
ne
e
r
i
ng (
I
C
I
C
I
-
B
M
E
)
,
2
01
1
2nd
I
nt
e
r
n
at
i
on
al
C
onf
e
r
e
nc
e
, I
E
E
E
,
2
0
1
1
.
[
1
1]
A
.
D
.
W
i
ba
w
a
a
nd M
.
H
.
P
ur
no
m
o,
“
E
a
r
l
y
de
t
e
c
t
i
on on t
he
c
on
di
t
i
o
n of
P
a
nc
r
e
a
s
or
g
a
n a
s
t
he
c
a
us
e
of
di
a
be
t
e
s
m
e
llitu
s
b
y
r
e
a
l ti
m
e
ir
is
i
m
a
g
e
p
r
o
c
e
s
s
in
g
,
”
C
i
r
c
ui
t
s
a
nd
Sy
s
t
e
m
s
,
200
6.
A
P
C
C
A
S 2
00
6.
I
E
E
E
A
s
i
a P
ac
i
f
i
c
C
o
n
f
er
en
ce o
n
, I
E
E
E
,
p
p
.
1
0
0
8
-
10
1
0,
20
06
.
[
1
2]
B
.
Z
h
an
g
,
et
al
.
,
“N
o
n
i
n
v
as
i
v
e
d
i
ab
et
es
m
el
l
i
t
u
s
d
et
ect
i
o
n
u
s
i
n
g
f
aci
al
b
l
o
ck
co
l
o
r
w
i
t
h
a
s
p
ar
s
e r
ep
r
es
en
t
at
i
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t
J
E
l
e
c
&
C
o
m
p
E
n
g
I
S
S
N
:
2088
-
8708
O
pt
i
m
i
z
at
i
on of
B
ac
k
pr
o
p
agat
i
on f
or
E
a
r
ly
D
e
te
c
tio
n
o
f.
.
.
(
R
os
i
t
a Sof
i
ana
)
3237
cl
as
s
i
f
i
er
,
”
I
E
E
E
T
r
ans
ac
t
i
o
ns
o
n
B
i
om
e
di
c
a
l
E
ngi
ne
e
r
i
ng
,
vo
l
.
61,
p
p.
10
27
–
1
03
3,
20
14
.
[
1
3]
S
.
T
i
n
g
an
d
B
.
Z
h
an
g
,
“D
i
ab
et
es
M
el
l
i
t
u
s
D
et
ect
i
o
n
B
as
ed
o
n
F
aci
al
B
l
o
ck
T
ex
t
u
r
e F
eat
u
r
es
U
s
i
n
g
t
h
e G
ab
o
r
F
ilte
r
,
”
C
om
p
ut
at
i
o
nal
Sc
i
e
nc
e
and E
ngi
ne
e
r
i
ng (
C
SE
)
,
2
01
4 I
E
E
E
17t
h I
nt
e
r
n
at
i
on
al
C
on
f
e
r
e
nc
e
on
, I
E
E
E
, p
p
. 1
-
6,
2
01
4.
[
1
4]
G
. K
h
a
n
,
et
al
.
,
“D
et
ect
i
o
n
o
f
C
er
eb
r
al
A
t
r
o
p
h
y
i
n
T
y
p
e
-
I
I
D
i
ab
et
es
M
el
l
i
t
u
s
b
y
M
ag
n
et
i
c R
es
o
n
an
ce I
m
a
g
i
n
g
o
f
B
r
ai
n
,
”
J
o
ur
nal
of
A
y
u
b M
e
di
c
al
C
ol
l
e
ge
A
bb
ot
t
a
ba
d
,
v
o
l
.
22,
p
p.
67
–
7
0,
20
10
.
[
1
5]
D
.
C
h
i
t
k
ar
a an
d
R
.
K
.
S
h
ar
m
a,
“V
o
i
ce B
as
ed
D
et
ect
i
o
n
o
f
T
y
p
e 2
D
i
ab
et
es
M
el
l
it
u
s
,
”
A
d
va
n
ces
i
n
E
l
ect
r
i
ca
l
,
E
l
e
c
t
r
oni
c
s
,
I
nf
or
m
a
t
i
o
n,
C
om
m
u
ni
c
at
i
on
a
nd
B
i
o
-
I
nf
or
m
at
i
c
s
(
A
E
E
I
C
B
)
,
20
16
2
nd
I
nt
e
r
n
at
i
on
al
C
onf
e
r
e
nc
e
o
n
,
I
E
E
E
, 2
0
1
6
.
[
1
6]
M
.
D
u
r
ai
r
aj
a
n
d
G
,
K
al
ai
s
ev
i
,
“
P
r
e
d
ic
tio
n
O
f
D
ia
b
e
te
s
U
s
in
g
B
a
c
k
P
r
o
p
a
g
a
tio
n
A
lg
o
r
ith
m
,
”
I
nt
e
r
nat
i
o
nal
J
o
ur
na
l
of
E
m
e
r
gi
ng
T
e
c
hn
ol
o
gy
a
nd I
nn
ov
at
i
v
e
E
n
gi
ne
e
r
i
ng
,
vol
.
1
,
p
p
.
21
-
2
5,
20
15.
[
1
7]
S
.
J
os
hi
a
nd M
.
B
or
s
e
,
“
D
et
ect
i
o
n
an
d
P
r
ed
i
ct
i
o
n
o
f
D
i
ab
et
es
M
el
l
i
t
u
s
U
s
i
n
g
B
ack
P
r
o
p
ag
at
i
o
n
N
eu
r
al
N
et
w
o
r
k
,
”
Mi
c
r
o
-
E
l
e
c
t
r
oni
c
s
and T
e
l
e
c
om
m
uni
c
at
i
o
n E
ng
i
ne
e
r
i
ng (
I
C
M
E
T
E
)
,
20
16 I
nt
e
r
nat
i
o
nal
C
o
nf
e
r
e
nc
e
on
, I
E
E
E
,
pp.
11
0
-
1
13,
2
01
6.
[
1
8]
W
ih
a
r
to
,
et
al
.
,
“H
y
b
r
i
d
S
y
s
t
e
m
o
f
T
i
er
ed
M
u
l
t
i
v
ar
i
at
e A
n
al
y
s
i
s
a
n
d
A
r
t
i
f
i
ci
al
N
eu
r
al
N
et
w
o
r
k
f
o
r
C
o
r
o
n
ar
y
H
ear
t
Di
se
a
se
Di
a
g
n
o
si
s,
”
I
nt
e
r
nat
i
o
na
l
J
our
nal
of
E
l
e
c
t
r
i
c
al
an
d C
om
p
ut
e
r
E
ngi
ne
e
r
i
ng (
I
J
E
C
E
)
,
v
ol
.
7,
pp.
1
02
3
-
10
31
,
201
7.
[
1
9]
B
.
S
ab
i
r
,
et
al
.
,
“I
m
p
r
o
v
ed
A
l
g
o
r
ith
m
f
o
r
P
a
th
o
lo
g
ic
a
l a
n
d
N
o
r
m
a
l V
o
ic
e
s
I
d
e
n
tif
ic
a
tio
n
,
”
I
nt
e
r
n
at
i
ona
l
J
o
ur
na
l
o
f
E
l
e
c
t
r
i
c
al
a
nd C
om
p
ut
e
r
E
n
gi
ne
e
r
i
ng (
I
J
E
C
E
)
,
v
o
l
. 7
, p
p
.
2
3
8
-
24
3,
20
17.
[
2
0]
M
.
A
b
d
ar
,
et
al
.
,
“C
o
m
p
ar
i
n
g
P
er
f
o
r
m
an
ce o
f
D
at
a M
i
n
i
n
g
A
l
g
o
r
i
t
h
m
s
i
n
P
r
ed
i
ct
i
o
n
H
ear
t
D
i
s
eas
es
,
”
In
t
e
r
nat
i
on
al
J
our
n
al
o
f
E
l
e
c
t
r
i
c
al
an
d C
om
pu
t
e
r
E
ngi
ne
e
r
i
ng (
I
J
E
C
E
)
,
vo
l
.
5
,
pp
.
15
69
-
157
6,
20
15
.
[
2
1]
C.
-
C
.
Y
u a
n
d B
.
-
D
.
L
i
u,
“
A
ba
c
k
pr
opa
g
a
t
i
on
a
l
g
or
i
t
hm
w
i
t
h a
da
pt
i
v
e
l
e
a
r
ni
ng
r
a
t
e
a
n
d m
om
e
nt
um
c
oe
f
f
i
c
i
e
nt
,
”
N
e
ur
al
N
e
t
w
or
k
s
,
20
02.
I
J
C
N
N
.
P
r
oc
e
e
di
ngs
o
f
t
he
2
00
2 I
nt
e
r
n
at
i
ona
l
J
oi
n
t
C
on
f
e
r
e
nc
e
on
,
I
E
E
E
, p
p
. 1
2
1
8
–
12
23
,
200
2.
[
2
2]
K
.
E
om
,
e
t
a
l
.
,
“
P
e
r
f
or
m
a
nc
e
i
m
pr
ov
e
m
e
nt
of
ba
c
k
pr
opa
g
a
t
i
on
a
l
g
or
i
t
hm
b
y
a
ut
om
a
t
i
c
a
c
t
i
v
a
t
i
on f
unc
t
i
on g
a
i
n
t
uni
ng
us
i
ng
f
uz
z
y
l
og
i
c
,
”
N
e
ur
oc
om
put
i
n
g
,
vo
l
.
50
,
p
p.
43
9
–
46
0,
20
0
3.
[
2
3]
K.
V.
N.
B
a
b
u
a
n
d
D
.
R
.
E
d
la
,
“
N
e
w
A
lg
e
b
r
a
ic
A
c
tiv
a
tio
n
F
u
n
c
tio
n
f
o
r
M
u
lti
-
L
a
y
er
ed
F
eed
F
o
r
w
ar
d
N
eu
r
al
N
et
w
o
r
k
s
,
”
I
E
T
E
J
our
nal
of
R
e
s
e
ar
c
h
,
v
ol
.
2
06
3,
pp.
7
1
–
79,
2
01
6.
[
2
4]
R
.
A
s
ad
i
,
et
al
.
,
“N
e
w
S
u
p
er
v
i
s
ed
M
u
l
t
i
L
a
y
er
F
eed
F
o
r
w
ar
d
N
e
u
r
al
N
et
w
o
r
k
M
o
d
el
t
o
A
ccel
er
at
e C
l
as
s
i
f
i
cat
i
o
n
w
i
t
h
H
i
g
h
A
c
cu
r
ac
y
,
”
E
ur
ope
an
J
our
n
al
o
f
Sc
i
e
n
t
i
f
i
c
R
e
s
e
ar
c
h
,
vol
.
33
,
p
p
.
16
3
–
1
78,
2
00
9.
[
2
5]
G
.
P
.
D
r
a
g
o,
M
.
M
or
a
n
do
,
a
n
d S
.
R
i
de
l
l
a
,
“
A
n A
da
pt
i
v
e
M
om
en
t
u
m
B
ack
P
r
o
p
ag
at
i
o
n
(
A
M
B
P
)
,
”
N
e
ur
al
C
om
put
i
n
g &
A
ppl
i
c
at
i
o
n
,
v
ol
.
3
,
pp.
2
13
–
2
21,
1
99
5.
[
2
6]
R
.
A
s
ad
i
an
d
S
.
A
b
d
u
l
,
“R
ev
i
ew
o
f
F
eed
F
o
r
w
ar
d
N
eu
r
al
N
et
w
o
r
k
cl
as
s
i
f
i
cat
i
o
n
p
r
e
p
r
o
ces
s
i
n
g
t
ech
n
i
q
u
es
,
”
P
r
oc
e
e
di
ngs
of
t
he
3r
d I
n
t
e
r
nat
i
o
nal
C
on
f
e
r
e
nc
e
on M
at
he
m
at
i
c
a
l
S
ci
en
ces
,
AI
P
,
pp.
5
67
–
5
73,
2
01
4.
[
2
7]
S
.
I
r
an
m
an
es
h
,
“A
D
i
f
f
er
en
t
i
al
A
d
ap
t
i
v
e L
ear
n
i
n
g
R
at
e M
et
h
o
d
f
o
r
B
ack
-
P
r
o
p
ag
at
i
o
n
N
eu
r
al
N
et
w
o
r
k
s
,
”
P
r
oc
e
e
di
ngs
of
t
he
1
0t
h W
SE
A
S I
nt
e
r
nat
i
on
al
C
on
f
e
r
e
nc
e
on N
e
ur
al
N
e
t
w
or
k
s
, A
C
M
, p
p
.
3
0
–
34
,
20
1
0.
[
2
8]
D
.
N
g
u
y
e
n a
nd B
.
W
i
dr
ow
,
“
I
m
pr
ov
i
ng
t
he
L
e
a
r
ni
ng
S
pe
e
d of
2
-
L
a
y
e
r
N
e
ur
a
l
N
e
t
w
or
k
s
b
y
C
hoos
i
ng
I
ni
t
i
a
l
V
al
u
es
o
f
t
h
e A
d
ap
t
i
v
e W
ei
g
h
t
s
,
”
N
eu
r
a
l
N
et
w
o
r
ks
,
1
99
0 I
J
C
N
N
I
nt
e
r
nat
i
o
nal
Jo
i
n
t
C
o
n
f
er
en
ce o
n
, I
E
E
E
, p
p
.I
I
I
-
21
-
III
-
26,
1
99
0.
Evaluation Warning : The document was created with Spire.PDF for Python.