I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
11
,
No
.
4
,
A
u
g
u
s
t
2021
,
p
p
.
3
3
1
9
~
3
3
2
8
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
1
1
i
4
.
pp
3
3
1
9
-
3
3
2
8
3319
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Co
m
pa
ra
tive a
na
ly
sis
of sho
rt
-
t
er
m
de
m
a
nd p
redic
ting
m
o
dels
using
ARIM
A an
d deep
learning
H
a
li
m
a
B
o
us
qa
o
ui
1
,
I
l
ha
m
S
li
m
a
ni
2
,
Sa
id Ac
hcha
b
3
1,
3
Na
ti
o
n
a
l
Hig
h
e
r
S
c
h
o
o
l
f
o
r
C
o
m
p
u
ter
S
c
ien
c
e
a
n
d
S
y
ste
m
A
n
a
l
y
sis
,
M
o
h
a
m
m
e
d
V
Un
iv
e
rsity
,
M
o
ro
c
c
o
2
S
m
a
r
c
tICT
L
a
b
o
ra
to
ry
,
ENS
A
O,
M
o
h
a
m
m
e
d
F
irst
Un
iv
e
rsity
,
Ou
jd
a
,
M
o
ro
c
c
o
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Sep
9
,
2
0
20
R
ev
i
s
ed
Dec
11
,
2
0
2
0
A
cc
ep
ted
J
an
13,
20
2
1
T
h
e
f
o
re
c
a
stin
g
c
o
n
sists
o
f
t
a
k
in
g
h
isto
rica
l
d
a
ta
a
s
in
p
u
ts
th
e
n
us
i
n
g
th
e
m
to
p
re
d
ict
f
u
tu
re
o
b
se
rv
a
ti
o
n
s,
t
h
u
s
d
e
term
in
in
g
f
u
tu
re
tren
d
s
.
De
m
a
n
d
p
re
d
ictio
n
is
a
c
ru
c
ial
c
o
m
p
o
n
e
n
t
in
t
h
e
su
p
p
ly
c
h
a
in
’s
p
ro
c
e
ss
th
a
t
a
ll
o
w
s
e
a
c
h
m
e
m
b
e
r
to
e
n
h
a
n
c
e
it
s
p
e
rf
o
r
m
a
n
c
e
a
n
d
it
s
p
ro
f
it
.
Ne
v
e
rth
e
les
s,
b
e
c
a
u
se
o
f
d
e
m
a
n
d
u
n
c
e
rtain
ty
su
p
p
ly
c
h
a
in
s
u
su
a
ll
y
su
ff
e
r
fro
m
m
a
n
y
p
ro
b
lem
s
su
c
h
a
s
th
e
b
u
ll
w
h
ip
e
ff
e
c
t
.
A
s
a
so
lu
ti
o
n
to
th
o
se
lo
g
isti
c
s
issu
e
s,
th
is
p
a
p
e
r
p
re
se
n
ts
a
c
o
m
p
a
ra
t
iv
e
a
n
a
l
y
sis
o
f
f
o
u
r
ti
m
e
se
rie
s
d
e
m
a
n
d
f
o
re
c
a
stin
g
m
o
d
e
ls;
n
a
m
e
l
y
,
th
e
a
u
to
re
g
re
ss
iv
e
in
teg
ra
ted
m
o
v
in
g
Av
e
ra
g
e
(A
RIM
A
)
a
sta
ti
stica
l
m
o
d
e
l,
th
e
m
u
lt
i
-
lay
e
r
p
e
rc
e
p
tro
n
(M
L
P
)
a
f
e
e
d
f
o
r
w
a
rd
n
e
u
ra
l
n
e
tw
o
rk
,
th
e
lo
n
g
sh
o
rt
-
term
m
e
m
o
r
y
m
o
d
e
l
(L
S
TM
)
a
re
c
u
rre
n
t
n
e
u
ra
l
n
e
tw
o
rk
a
n
d
th
e
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
(CNN
o
r
Co
n
v
Ne
t)
a
d
e
e
p
lea
rn
i
n
g
m
o
d
e
l.
T
h
e
e
x
p
e
ri
m
e
n
tatio
n
s
a
re
c
a
rried
o
u
t
u
si
n
g
a
re
a
l
-
li
fe
d
a
tas
e
t
p
ro
v
id
e
d
b
y
a
s
u
p
e
rm
a
rk
e
t
in
M
o
ro
c
c
o
.
T
h
e
re
su
lt
s
c
lea
rl
y
sh
o
w
th
a
t
th
e
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
r
a
l
n
e
tw
o
rk
g
iv
e
s
slig
h
tl
y
b
e
tt
e
r
fo
re
c
a
stin
g
re
su
lt
s
th
a
n
th
e
L
o
n
g
sh
o
rt
-
term
m
e
m
o
r
y
n
e
t
w
o
rk
.
K
ey
w
o
r
d
s
:
AR
I
M
A
A
r
ti
f
icial
n
e
u
r
al
n
et
w
o
r
k
C
o
n
v
o
lu
tio
n
al
n
e
u
r
al
n
e
t
w
o
r
k
L
ST
M
Ma
ch
i
n
e
l
ea
r
n
i
n
g
Su
p
p
l
y
c
h
ain
m
a
n
ag
e
m
e
n
t
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Hali
m
a
B
o
u
s
q
ao
u
i
Natio
n
al
Hi
g
h
er
Sc
h
o
o
l f
o
r
C
o
m
p
u
ter
Scie
n
ce
an
d
S
y
s
te
m
An
al
y
s
i
s
Mo
h
a
m
m
ed
V
Un
iv
er
s
it
y
B
en
A
b
d
alla
h
R
e
g
r
ag
u
i,
Ma
d
in
at
A
l I
r
f
a
n
e,
B
P
7
1
3
,
A
g
d
al
R
ab
at,
Mo
r
o
cc
o
E
m
ail:
h
ali
m
a.
b
o
u
s
q
ao
u
i
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
A
s
u
p
p
l
y
c
h
ai
n
is
an
en
tire
n
e
t
w
o
r
k
co
m
p
o
s
ed
o
f
d
is
tr
ib
u
te
d
o
r
g
an
izatio
n
s
i
n
i
n
ter
ac
tio
n
(
s
u
p
p
lier
s
,
r
etailer
s
,
d
is
tr
ib
u
to
r
s
,
an
d
clien
t
s
)
in
v
o
l
v
ed
in
th
e
p
r
o
ce
s
s
o
f
p
u
r
ch
as
in
g
,
in
v
e
n
to
r
y
m
a
n
a
g
e
m
e
n
t,
p
r
o
d
u
ctio
n
,
d
is
tr
ib
u
tio
n
an
d
d
eliv
er
y
o
f
a
p
r
o
d
u
ct
o
r
a
s
er
v
ice
f
r
o
m
r
a
w
m
at
er
ial
s
to
a
f
i
n
al
co
s
t
u
m
er
[
1
]
.
Su
p
p
l
y
c
h
ai
n
m
an
a
g
e
m
e
n
t
co
v
er
s
i
n
f
o
r
m
ati
o
n
f
lo
w
s
,
p
h
y
s
ical
d
is
tr
ib
u
tio
n
as
w
ell
as
f
i
n
a
n
cial
tr
an
s
ac
tio
n
s
.
I
n
o
t
h
er
w
o
r
d
s
,
all
th
e
s
u
p
p
l
y
c
h
ain
m
e
m
b
er
s
s
h
ar
e
a
co
m
m
o
n
o
b
j
ec
tiv
e
t
h
a
t
is
i
m
p
r
o
v
i
n
g
th
e
ir
b
u
s
i
n
e
s
s
e
s
an
d
p
r
o
f
it
s
w
h
ile
cr
ea
tin
g
a
v
al
u
e
r
ep
r
esen
ted
b
y
th
e
p
r
o
d
u
ct
o
r
s
er
v
ice
d
eliv
e
r
ed
to
th
e
co
s
tu
m
er
[
2
]
.
T
h
e
p
r
o
b
lem
co
n
s
i
s
ts
o
f
th
e
d
em
a
n
d
f
o
r
ec
asti
n
g
w
h
ic
h
i
s
a
cr
u
c
ial
co
m
p
o
n
en
t
i
n
s
u
p
p
ly
c
h
ain
’
s
p
r
o
ce
s
s
.
I
n
d
ee
d
,
p
r
ed
ictio
n
co
n
s
is
ts
o
f
ta
k
i
n
g
m
o
d
els
f
r
o
m
h
is
to
r
ical
d
ata
i
n
o
r
d
er
to
u
s
e
th
e
m
to
p
r
ed
ict
f
u
tu
r
e
o
b
s
er
v
atio
n
s
.
Ne
v
er
th
e
less
,
t
h
e
u
n
ce
r
tain
ch
ar
ac
ter
o
f
t
h
e
co
s
t
u
m
er
’
s
d
e
m
a
n
d
ca
u
s
e
s
d
if
f
ic
u
ltie
s
f
o
r
d
ec
is
io
n
m
ak
er
s
w
h
o
m
u
s
t
r
ea
ct
in
an
e
f
f
icie
n
t
an
d
f
a
s
t
m
a
n
n
er
.
B
esid
es,
w
ith
th
e
ad
v
an
ce
m
en
t
o
f
tech
n
o
lo
g
y
,
s
u
p
p
l
y
c
h
ai
n
s
b
ec
o
m
e
m
o
r
e
co
m
p
le
x
.
S
o
,
in
o
r
d
er
to
k
ee
p
u
p
w
it
h
th
is
v
ital
e
v
o
lu
tio
n
a
n
d
f
o
r
an
ef
f
ec
tiv
e
s
u
p
p
l
y
c
h
ain
,
co
m
p
an
ies
m
u
s
t
b
e
ab
le
to
p
lan
n
o
t
o
n
l
y
t
h
e
p
r
ese
n
t
b
u
t
also
th
e
f
u
tu
r
e
s
ales
o
r
d
em
a
n
d
.
C
er
ta
in
l
y
,
s
ales
f
o
r
e
ca
s
tin
g
i
s
e
f
f
icie
n
t
f
o
r
s
u
p
p
l
y
ch
ai
n
s
’
p
la
n
n
i
n
g
ac
ti
v
itie
s
i
n
clu
d
in
g
p
u
r
ch
as
in
g
,
in
v
e
n
to
r
y
,
p
r
o
d
u
ctio
n
a
n
d
d
is
t
r
ib
u
tio
n
.
I
t h
elp
s
m
a
n
ag
er
s
to
m
ak
e
o
p
ti
m
al
d
ec
is
io
n
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
4
,
A
u
g
u
s
t 2
0
2
1
:
3
3
1
9
-
3328
3320
As b
ac
k
g
r
o
u
n
d
i
n
th
e
liter
at
u
r
e,
au
to
r
eg
r
es
s
i
v
e
i
n
te
g
r
ated
m
o
v
in
g
a
v
er
ag
e
(
A
R
I
M
A
)
,
is
a
lin
ea
r
t
i
m
e
s
er
ies
f
o
r
ec
ast
i
n
g
m
eth
o
d
th
a
t
h
as
b
ee
n
w
id
el
y
u
s
ed
in
m
a
n
y
ar
ea
s
to
m
o
d
el
d
if
f
er
en
t
p
r
o
b
le
m
s
.
So
m
e
o
f
t
h
e
ap
p
licatio
n
s
o
f
AR
I
M
A
ar
e
a
s
:
Fa
n
o
o
d
i
et
a
l.
[
3
]
u
s
es
AR
I
M
A
to
r
ed
u
ce
d
e
m
a
n
d
u
n
c
er
tain
t
y
in
a
b
lo
o
d
p
latelet
s
u
p
p
l
y
ch
a
in
.
J
i
et
a
l.
[
4
]
u
s
es
a
h
y
b
r
id
AR
I
M
A
a
n
d
d
ee
p
n
eu
r
al
n
et
w
o
r
k
m
o
d
el
to
f
o
r
ec
ast
f
u
tu
r
e
p
r
ices
o
f
ca
r
b
o
n
.
Ma
t
y
j
a
s
ze
k
et
a
l.
[
5
]
i
m
p
le
m
en
ts
a
n
AR
I
MA
m
o
d
el
i
n
o
r
d
er
to
f
o
r
ec
as
t
co
k
i
n
g
co
al
p
r
ices
in
t
h
e
m
in
in
g
i
n
d
u
s
tr
y
.
O
h
y
v
e
r
an
d
P
u
d
j
ih
astu
ti
[
6
]
u
s
es
th
e
AR
I
M
A
m
o
d
el
to
f
o
r
ec
ast
r
i
ce
p
r
ice.
T
h
ese
ar
e
on
l
y
a
f
e
w
ex
a
m
p
les
o
f
t
h
e
ap
p
licatio
n
o
f
AR
I
M
A
t
h
at
r
e
m
ai
n
s
w
id
el
y
e
m
p
lo
y
ed
i
n
m
u
ltip
le
d
o
m
ai
n
s
to
m
ak
e
ac
cu
r
ate
f
o
r
ec
ast
s
.
AR
I
MA
i
s
al
s
o
u
s
ed
f
o
r
d
e
m
an
d
f
o
r
ec
asti
n
g
.
Fo
r
in
s
tan
ce
,
Oliv
eir
a
et
a
l.
[
7
]
p
r
o
p
o
s
es
a
d
o
u
b
le
s
ea
s
o
n
al
AR
I
M
A
m
o
d
el
to
m
ak
e
ac
c
u
r
ate
s
h
o
r
t
-
ter
m
w
ater
d
e
m
an
d
f
o
r
ec
asts
.
A
l
-
Mu
s
a
y
l
h
et
a
l.
[
8
]
ap
p
lies
AR
I
M
A
,
M
AR
S,
a
n
d
SV
R
i
n
s
h
o
r
t
ter
m
ele
ctr
icit
y
d
e
m
an
d
f
o
r
ec
asti
n
g
.
An
g
g
r
ae
n
i
et
a
l.
[
9
]
co
m
p
ar
es
b
et
w
ee
n
A
R
I
M
A
a
n
d
AR
I
M
A
X
r
es
u
lts
f
o
r
k
id
s
clo
th
es
d
e
m
a
n
d
f
o
r
ec
asti
n
g
.
Am
i
n
i
et
a
l.
[
1
0
]
ap
p
lie
s
A
R
I
M
A
f
o
r
a
n
elec
t
r
ic
v
e
h
icle
c
h
ar
g
i
n
g
d
e
m
an
d
ti
m
e
s
er
ie
s
f
o
r
ec
asti
n
g
.
An
d
W
an
g
et
a
l.
[
1
1
]
i
m
p
le
m
en
t
s
a
m
o
d
i
f
icatio
n
ap
p
r
o
ac
h
in
s
ea
s
o
n
a
l
A
R
I
M
A
f
o
r
elec
tr
icit
y
d
e
m
a
n
d
f
o
r
ec
asti
n
g
.
Neu
r
al
n
et
w
o
r
k
s
o
n
t
h
e
o
t
h
er
h
an
d
,
ar
e
u
s
ed
f
o
r
n
o
n
li
n
ea
r
t
i
m
e
s
er
ie
s
m
o
d
eli
n
g
.
So
m
e
ap
p
licatio
n
s
in
d
e
m
an
d
f
o
r
ec
asti
n
g
i
n
cl
u
d
e:
I
n
th
e
p
ap
er
[
1
2
]
,
n
eu
r
al
n
et
w
o
r
k
s
ar
e
ap
p
lied
f
o
r
d
e
m
a
n
d
f
o
r
ec
asti
n
g
o
f
a
n
en
g
i
n
e
o
il.
Sh
ar
m
a
et
a
l.
[
1
3
]
ap
p
lies
n
eu
r
al
n
et
w
o
r
k
s
to
p
r
ed
ict
en
er
g
y
d
e
m
a
n
d
,
ca
r
b
o
n
d
io
x
id
e
em
i
s
s
io
n
s
an
d
w
in
d
g
en
er
at
io
n
.
Ma
ti
n
o
et
a
l.
[
1
4
]
im
p
le
m
en
t
s
n
eu
r
al
n
et
w
o
r
k
-
b
ased
m
o
d
el
s
f
o
r
a
b
last
f
u
r
n
ac
e
g
a
s
p
r
o
d
u
ctio
n
an
d
d
em
a
n
d
p
r
ed
i
ctio
n
p
r
o
b
lem
.
A
n
d
Sil
v
a
et
a
l.
[
1
5
]
u
s
es
d
en
o
is
ed
n
eu
r
al
n
et
w
o
r
k
s
to
f
o
r
ec
ast
to
u
r
is
m
d
e
m
a
n
d
.
I
n
[
1
6
]
,
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
et
w
o
r
k
i
s
a
p
p
lied
to
tack
le
th
e
p
r
o
b
lem
o
f
p
r
ed
ictin
g
s
h
o
r
t
-
ter
m
s
u
p
p
l
y
-
d
e
m
a
n
d
g
ap
o
f
r
i
d
e
-
s
o
u
r
ci
n
g
s
er
v
ice
s
,
w
h
er
e
t
h
r
ee
h
ex
a
g
o
n
-
b
ased
co
n
v
o
l
u
tio
n
al
n
e
u
r
al
n
et
w
o
r
k
s
(H
-
C
N
N)
ar
e
p
r
o
p
o
s
ed
.
Ke
et
a
l.
[
1
7
]
,
p
r
o
p
o
s
e
a
h
y
b
r
id
ap
p
r
o
ac
h
o
r
th
e
f
u
s
io
n
co
n
v
o
lu
tio
n
al
lo
n
g
s
h
o
r
t
-
ter
m
m
e
m
o
r
y
n
et
w
o
r
k
,
w
h
er
e
m
u
l
tip
le
co
n
v
o
l
u
tio
n
a
l
lo
n
g
s
h
o
r
t
-
ter
m
m
e
m
o
r
y
(
L
ST
M)
lay
e
r
s
,
s
tan
d
ar
d
L
ST
M
la
y
er
s
,
an
d
co
n
v
o
l
u
tio
n
al
la
y
er
s
ar
e
f
u
s
ed
ai
m
i
n
g
to
f
o
r
ec
ast
p
ass
en
g
er
d
e
m
a
n
d
u
n
d
er
an
o
n
-
d
e
m
a
n
d
r
id
e
s
er
v
ice
p
lat
f
o
r
m
.
Am
ar
a
s
in
g
h
e
et
a
l.
[
1
8
]
co
m
p
ar
es
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
et
w
o
r
k
s
to
p
e
r
f
o
r
m
e
n
er
g
y
lo
ad
f
o
r
ec
asti
n
g
b
ased
o
n
h
is
to
r
ica
l
lo
ad
s
o
f
a
s
in
g
le
r
es
id
en
tial
cu
s
to
m
er
to
lo
n
g
s
h
o
r
t
-
ter
m
m
e
m
o
r
ie
s
(
L
ST
M)
s
eq
u
en
ce
-
to
-
s
eq
u
e
n
ce
(
L
ST
M
S2
S),
f
ac
to
r
ed
r
estricte
d
B
o
ltz
m
an
n
m
ac
h
i
n
es
(
F
C
R
B
M)
,
ar
tif
icial
n
e
u
r
al
n
et
w
o
r
k
s
(
A
NN)
an
d
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
i
n
es
(
SVM)
.
T
h
e
o
b
tain
ed
r
esu
lts
d
e
m
o
n
s
tr
ate
th
a
t
t
h
e
C
NN
o
u
tp
er
f
o
r
m
ed
SV
R
w
h
i
le
p
r
o
d
u
cin
g
co
m
p
ar
ab
le
r
esu
lts
to
th
e
n
eu
r
al
n
et
w
o
r
k
an
d
d
ee
p
lear
n
in
g
m
o
d
els.
Th
e
p
r
esen
t
p
ap
er
tack
les
th
e
p
r
o
p
o
s
ed
s
o
lu
tio
n
w
h
ich
is
u
s
in
g
ti
m
e
s
er
ies
f
o
r
ec
asti
n
g
m
o
d
el
s
a
n
d
d
ee
p
lear
n
in
g
m
o
d
els
f
o
r
d
em
an
d
p
r
ed
ictio
n
.
I
n
ad
d
itio
n
,
th
e
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
e
t
wo
r
k
h
as
b
ee
n
r
ar
el
y
u
s
ed
in
t
h
e
liter
at
u
r
e.
Usi
n
g
it
in
co
m
p
ar
is
o
n
w
it
h
o
th
er
m
o
d
els co
n
s
i
s
ts
o
f
th
e
m
ai
n
co
n
tr
i
b
u
tio
n
o
f
t
h
e
p
ap
er
.
Th
e
n
o
v
elt
y
p
r
ese
n
ted
in
t
h
is
p
ap
er
is
th
e
p
er
f
o
r
m
a
n
ce
ev
al
u
atio
n
a
n
d
co
m
p
ar
ativ
e
s
t
u
d
y
u
s
i
n
g
a
r
ea
l
d
ataset
p
r
o
v
id
ed
b
y
a
s
u
p
er
m
ar
k
et
in
Mo
r
o
cc
o
.
T
h
e
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
s
o
f
f
o
u
r
p
r
ed
ictio
n
m
eth
o
d
s
ar
e
p
r
esen
ted
.
Na
m
el
y
,
t
h
e
au
to
r
eg
r
ess
i
v
e
in
teg
r
ated
m
o
v
in
g
av
er
ag
e
(
A
R
I
M
A
)
as
s
tati
s
tic
al
m
o
d
el,
th
e
m
u
l
ti
-
la
y
er
p
er
ce
p
tr
o
n
(
ML
P
)
a
f
ee
d
f
o
r
w
ar
d
n
e
u
r
al
n
et
w
o
r
k
[
1
9
,
2
0
]
,
th
e
co
n
v
o
l
u
tio
n
a
l
n
e
u
r
al
n
et
w
o
r
k
(
C
NN
o
r
C
o
n
v
Net)
a
d
ee
p
lean
i
n
g
m
o
d
el
an
d
th
e
lo
n
g
s
h
o
r
t
ter
m
m
e
m
o
r
y
m
o
d
el
(
L
S
T
M)
a
r
ec
u
r
r
en
t
n
e
u
r
al
n
et
w
o
r
k
.
T
h
ese
m
o
d
els
w
er
e
ch
o
s
e
n
o
n
o
r
d
er
to
c
o
m
p
ar
e
b
et
w
ee
n
t
h
e
d
if
f
er
en
t
t
y
p
es
o
f
n
e
u
r
al
n
et
w
o
r
k
s
.
Ma
in
l
y
t
h
e
f
ee
d
f
o
r
w
ar
d
,
th
e
co
n
v
o
l
u
tio
n
a
l a
n
d
th
e
r
ec
u
r
r
en
t n
et
w
o
r
k
s
.
T
h
e
r
em
ai
n
d
er
o
f
th
is
p
ap
er
is
o
r
g
an
ized
as
f
o
llo
w
s
:
T
h
e
s
ec
o
n
d
p
ar
t
p
r
esen
ts
an
o
v
er
v
i
e
w
o
f
t
h
e
p
r
o
p
o
s
ed
p
r
e
d
ictio
n
m
et
h
o
d
s
b
ased
o
n
ti
m
e
s
er
ies
m
et
h
o
d
s
.
T
h
en
,
th
e
th
ir
d
s
ec
tio
n
is
d
ed
icate
d
to
ex
p
lain
th
e
ad
o
p
ted
m
et
h
o
d
o
lo
g
y
an
d
ill
u
s
tr
ate
t
h
e
o
b
tain
ed
f
o
r
ec
asti
n
g
r
es
u
lt
s
in
a
co
m
p
ar
ativ
e
an
al
y
s
is
b
ased
o
n
n
u
m
er
ical
e
x
p
er
i
m
e
n
tatio
n
s
u
s
in
g
a
r
ea
l
d
ataset
o
f
a
lo
ca
l
s
u
p
er
m
ar
k
et.
Fin
al
l
y
,
t
h
e
co
n
cl
u
d
in
g
r
e
m
ar
k
s
a
n
d
th
e
f
u
t
u
r
e
w
o
r
k
ar
e
p
r
esen
ted
.
2.
P
RO
P
O
SE
D
P
RE
D
I
C
T
I
O
N
M
E
T
H
O
DS
T
im
e
s
er
ies
ca
n
b
e
d
ef
in
ed
as
a
s
er
ies
o
f
d
ata
p
o
in
ts
r
ec
o
r
d
ed
an
d
an
al
y
ze
d
in
a
ti
m
e
o
r
d
er
,
it
is
a
s
eq
u
en
ce
ta
k
e
n
at
eq
u
all
y
s
p
a
ce
d
ti
m
e
p
er
io
d
s
[
2
1
]
.
T
h
e
o
n
l
y
in
d
ep
en
d
e
n
t
v
ar
iab
le
i
n
ti
m
e
s
er
ies
m
et
h
o
d
s
is
ti
m
e.
T
h
e
an
a
l
y
s
is
o
f
ti
m
e
s
er
ies
is
th
e
p
r
o
ce
s
s
o
f
ex
tr
ac
tin
g
m
ea
n
i
n
g
f
u
l
in
f
o
r
m
at
io
n
a
n
d
s
tatis
tics
f
r
o
m
t
h
e
d
ata.
T
h
e
f
o
r
ec
asti
n
g
o
f
ti
m
e
s
er
ies
is
t
h
e
p
r
o
ce
s
s
o
f
e
s
t
i
m
ati
n
g
t
h
e
f
u
t
u
r
e
v
a
lu
e
s
o
f
th
e
d
ata
b
ased
o
n
p
r
ev
io
u
s
l
y
o
b
s
er
v
ed
d
ata
p
lu
s
s
o
m
e
o
t
h
er
f
ea
t
u
r
es
[
2
2
]
.
C
o
n
tr
ar
y
to
s
u
p
er
v
i
s
ed
m
o
d
e
ls
f
o
r
ec
asti
n
g
w
h
er
e
d
ata
is
p
r
ed
icted
u
s
in
g
m
u
lt
ip
le
f
ea
t
u
r
es,
i
n
ti
m
e
s
er
ie
s
f
o
r
ec
asti
n
g
,
a
v
ar
iab
le
is
p
r
ed
icted
u
s
i
n
g
p
a
s
t
o
b
s
er
v
atio
n
s
o
f
th
e
s
a
m
e
v
ar
iab
le.
T
h
e
Fig
u
r
e
1
s
h
o
w
s
d
if
f
er
en
t
m
et
h
o
d
s
u
s
ed
i
n
ti
m
e
s
er
ies
f
o
r
ec
asti
n
g
:
Fo
r
ec
asti
n
g
tech
n
iq
u
es a
r
e
ca
t
eg
o
r
ized
in
to
t
w
o
f
a
m
ilie
s
:
P
ar
am
etr
ic
m
et
h
o
d
s
b
ased
o
n
m
at
h
e
m
atica
l
to
o
ls
a
n
d
s
tati
s
tical
a
n
al
y
s
i
s
u
s
in
g
h
i
s
to
r
ical
d
ata
s
u
c
h
a
s
lin
ea
r
an
d
n
o
n
lin
ea
r
r
eg
r
es
s
io
n
,
au
to
r
eg
r
es
s
i
v
e
in
te
g
r
ated
m
o
v
i
n
g
av
er
ag
e.
Ne
v
er
th
ele
s
s
,
th
o
s
e
m
e
th
o
d
s
ar
e
co
m
p
licated
to
u
s
e.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
C
o
mp
a
r
a
tive
a
n
a
lysi
s
o
f sh
o
r
t
-
term d
ema
n
d
p
r
ed
ictin
g
mo
d
e
ls
u
s
in
g
A
R
I
MA
a
n
d
...
(
Ha
lima
B
o
u
s
q
a
o
u
i)
3321
No
n
-
p
ar
a
m
e
tr
ic
m
et
h
o
d
s
b
as
ed
o
n
m
ac
h
i
n
e
lear
n
i
n
g
w
it
h
th
e
ab
ilit
y
o
f
lear
n
in
g
a
n
d
a
p
p
r
o
ac
h
in
g
a
n
y
n
o
n
li
n
ea
r
f
u
n
ctio
n
.
T
h
o
s
e
te
ch
n
iq
u
es
ar
e
m
o
s
tl
y
b
ased
o
n
t
h
e
u
s
e
o
f
ar
ti
f
icial
i
n
tel
lig
e
n
ce
s
u
c
h
a
s
ar
tif
icial
n
eu
r
al
n
et
w
o
r
k
s
th
at
o
f
f
er
f
lex
ib
le
p
ar
a
m
eter
s
d
u
r
in
g
th
e
lear
n
i
n
g
a
n
d
i
m
p
le
m
en
ta
tio
n
p
h
ase
s
.
Fig
u
r
e
1
.
T
im
e
s
er
ies
f
o
r
ec
asti
n
g
m
et
h
o
d
s
[
2
2
]
2
.
1
.
Aut
o
re
g
re
s
s
iv
e
inte
g
ra
t
ed
m
o
v
ing
a
v
er
a
g
e
(
ARI
M
A)
T
h
e
A
R
I
M
A
m
eth
o
d
o
lo
g
y
was
o
r
ig
in
all
y
d
ev
elo
p
ed
b
y
B
o
x
an
d
J
en
k
i
n
s
i
n
th
e
1
9
7
0
s
.
A
m
o
d
el
i
s
d
en
o
ted
b
y
AR
I
M
A
(
p
,
d
,
q
)
an
d
is
co
m
p
o
s
ed
o
f
:
An
Au
to
r
eg
r
ess
iv
e
m
o
d
el
o
f
o
r
d
er
p
(
n
u
m
b
er
o
f
la
g
s
)
.
I
n
teg
r
ated
r
e
f
er
s
to
t
h
e
s
tep
o
f
r
e
m
o
v
i
n
g
n
o
n
s
ta
tio
n
ar
it
y
b
y
d
i
f
f
er
e
n
ci
n
g
th
e
d
ata.
d
i
s
t
h
e
d
eg
r
ee
o
f
d
if
f
er
e
n
ci
n
g
.
A
Mo
v
in
g
av
er
a
g
e
m
o
d
el
o
f
o
r
d
er
q
.
2
.
1
.
1
.
Aut
o
re
g
re
s
s
iv
e
m
o
del
T
h
e
au
to
r
eg
r
es
s
iv
e
m
o
d
el
A
R
(
p
)
o
f
o
r
d
er
p
s
tates
t
h
at
t
h
e
o
u
tp
u
t
v
ar
iab
le
d
ep
en
d
s
o
n
it
s
p
r
ev
io
u
s
v
alu
e
s
w
it
h
s
o
m
e
lag
p
l
u
s
a
r
a
n
d
o
m
ter
m
.
T
h
e
au
to
r
e
g
r
ess
i
v
e
m
o
d
el
AR
(
p
)
is
d
ef
in
ed
as:
=
+
1
−
1
+
2
−
2
+
⋯
+
−
2
.
1
.
2
.
Dif
f
er
encing
A
t
i
m
e
s
er
ies
i
s
ca
lled
n
o
n
s
t
atio
n
ar
y
w
h
e
n
th
e
v
al
u
es
o
f
t
h
e
d
ata
ar
e
d
ep
en
d
en
t
o
n
t
i
m
e.
A
n
o
n
-
s
tatio
n
ar
y
ti
m
e
s
er
ies
ca
n
b
e
ch
an
g
ed
in
to
s
tatio
n
ar
y
b
y
d
i
f
f
er
en
ci
n
g
t
h
e
ti
m
e
s
er
ies.
Di
f
f
er
en
ci
n
g
r
e
f
er
s
to
s
u
b
tr
ac
ti
n
g
s
o
m
e
p
a
s
t
v
a
lu
e
s
o
f
t
h
e
d
ata
a
n
u
m
b
er
o
f
ti
m
e
s
.
d
th
e
d
eg
r
ee
o
f
d
i
f
f
er
en
ci
n
g
r
ef
er
s
to
t
h
at
n
u
m
b
e
r
o
f
ti
m
e
s
.
Gen
er
all
y
,
a
s
er
ies
m
ig
h
t
n
ee
d
f
ir
s
t
-
d
i
f
f
er
en
ci
n
g
m
u
ltip
le
d
ti
m
es
to
attain
s
tatio
n
ar
it
y
[
2
3
]
.
A
d
i
f
f
er
e
n
ci
n
g
o
f
d
e
g
r
ee
1
is
as
:
′
=
−
−
1
2
.
1
.
3
.
M
o
v
ing
a
v
er
a
g
e
m
o
del
T
o
d
ea
l
w
it
h
n
o
n
-
s
tat
io
n
ar
it
y
,
w
e
c
h
ar
ac
ter
ize
t
h
e
ti
m
e
s
er
ies
as
t
h
e
s
u
m
o
f
a
n
o
n
-
co
n
s
tan
t
m
ea
n
v
alu
e
p
l
u
s
a
r
an
d
o
m
er
r
o
r
v
ar
i
ab
le
[
2
3
]
:
=
+
I
n
s
m
o
o
t
h
i
n
g
m
e
th
o
d
s
,
a
v
ar
iab
le
is
a
f
u
n
ctio
n
o
f
s
o
m
e
p
ast
o
b
s
er
v
atio
n
s
,
w
h
ic
h
m
ea
n
s
th
a
t
th
e
f
u
tu
r
e
v
a
lu
e
o
f
t
h
e
ti
m
e
s
er
i
es
is
th
e
w
ei
g
h
ted
av
er
ag
e
o
f
s
o
m
e
p
ast
o
b
s
er
v
at
io
n
s
.
A
s
s
u
ch
,
t
h
e
m
o
v
i
n
g
av
er
ag
e
M
A
(
q
)
o
f
o
r
d
er
q
ca
n
b
e
w
r
itte
n
as th
e
w
e
ig
h
ted
av
e
r
ag
e
o
f
th
e
p
ast q
er
r
o
r
s
.
=
+
+
1
−
1
+
2
−
2
+
⋯
+
−
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
4
,
A
u
g
u
s
t 2
0
2
1
:
3
3
1
9
-
3328
3322
2
.
1
.
4
.
Aut
o
re
g
re
s
s
iv
e
inte
g
ra
t
ed
m
o
v
ing
a
v
er
a
g
e
AR
I
M
A
is
a
m
i
x
ed
m
o
d
el
th
at
co
m
b
i
n
es
b
o
th
t
h
e
d
if
f
er
en
ce
d
au
to
r
eg
r
ess
i
v
e
an
d
m
o
v
in
g
av
er
ag
e
m
o
d
el
s
.
T
h
e
f
i
n
al
f
o
r
m
o
f
a
t
i
m
e
s
er
ie
s
m
o
d
el,
w
h
ic
h
d
ep
en
d
s
o
n
it
s
o
w
n
p
p
ast
v
al
u
e
s
an
d
o
n
t
h
e
q
p
ast
v
alu
e
s
o
f
w
h
ite
n
o
is
e
er
r
o
r
ter
m
s
[
2
1
]
,
is
as:
=
+
1
−
1
′
+
2
−
2
′
+
⋯
+
−
′
+
+
1
−
1
+
2
−
2
+
⋯
+
−
w
it
h
d
if
f
er
e
n
ce
d
to
a
d
eg
r
ee
d
.
T
h
is
r
ep
r
esen
ts
th
e
AR
I
M
A
(
p
,
d
,
q
)
m
o
d
el.
2
.
2
.
M
a
chine le
a
rning
ba
s
ed
m
et
ho
ds
Th
e
a
r
tif
icial
n
e
u
r
al
n
et
w
o
r
k
s
ar
e
p
o
p
u
lar
m
ac
h
i
n
e
lear
n
i
n
g
alg
o
r
ith
m
s
t
h
at
s
i
m
u
late
h
o
w
t
h
e
h
u
m
a
n
b
r
ain
b
eh
av
e
s
an
d
lear
n
s
.
Neu
r
al
r
ef
er
s
to
n
eu
r
o
n
s
,
th
e
y
ar
e
ce
lls
co
n
tai
n
ed
in
t
h
e
h
u
m
a
n
n
er
v
o
u
s
s
y
s
te
m
.
T
h
e
n
eu
r
o
n
s
ar
e
co
n
n
ec
ted
to
o
n
e
an
o
th
er
w
ith
t
h
e
u
s
e
o
f
d
en
d
r
ites
an
d
ax
o
n
s
,
w
h
ile
s
y
n
ap
s
es
ar
e
th
e
r
eg
io
n
s
co
n
n
ec
ti
n
g
b
et
w
ee
n
t
h
ese
a
x
o
n
s
[
2
4
]
.
A
r
ti
f
icial
n
e
u
r
al
n
e
t
w
o
r
k
s
(
ANN)
ar
e
an
ar
ti
f
ic
ial
i
n
telli
g
e
n
ce
m
et
h
o
d
in
s
p
ir
ed
f
r
o
m
t
h
e
f
u
n
ct
io
n
i
n
g
o
f
th
e
b
io
lo
g
ical
n
eu
r
al
n
et
w
o
r
k
s
c
h
ar
ac
ter
ized
b
y
a
cir
c
u
it
o
f
in
ter
co
n
n
ec
ted
n
e
u
r
o
n
s
o
r
g
an
ized
as
m
u
ltip
l
e
in
ter
co
n
n
ec
ted
la
y
er
s
: a
n
i
n
p
u
t la
y
er
,
o
n
e
o
r
m
o
r
e
h
id
d
en
la
y
er
s
a
n
d
an
o
u
tp
u
t la
y
er
.
A
N
N
s
ar
e
v
er
y
p
o
w
er
f
u
l
in
t
h
e
w
a
y
th
a
t
th
e
y
ca
n
id
en
t
if
y
co
m
p
le
x
li
n
ea
r
/n
o
n
-
l
in
ea
r
r
elatio
n
s
h
ip
s
b
et
w
ee
n
in
p
u
ts
a
n
d
o
u
tp
u
t
s
.
Neu
r
a
l
n
et
w
o
r
k
s
ar
e
u
s
ed
in
v
ar
io
u
s
f
ield
s
f
r
o
m
c
lass
if
ica
tio
n
,
to
f
o
r
ec
asti
n
g
,
to
ap
p
r
o
x
i
m
atio
n
,
to
d
iag
n
o
s
is
,
to
i
m
a
g
e
p
r
o
ce
s
s
in
g
o
r
ev
e
n
r
ec
o
g
n
i
tio
n
.
T
h
er
e
ar
e
tw
o
t
y
p
es o
f
n
e
u
r
al
n
et
w
o
r
k
s
:
Feed
f
o
r
w
ar
d
n
e
u
r
al
n
et
w
o
r
k
,
o
r
also
ca
lled
n
o
n
-
r
ec
u
r
r
en
t
n
e
u
r
al
n
et
w
o
r
k
w
h
er
e
in
f
o
r
m
a
tio
n
f
lo
w
s
o
n
l
y
i
n
o
n
e
d
ir
ec
tio
n
g
o
in
g
f
r
o
m
t
h
e
in
p
u
t
to
th
e
o
u
tp
u
t
la
y
er
all
th
e
w
a
y
t
h
r
o
u
g
h
th
e
h
id
d
en
lay
er
o
r
l
a
y
er
s
.
Ho
w
e
v
er
,
th
e
y
h
a
v
e
n
o
ab
ilit
y
to
m
ai
n
tai
n
h
is
to
r
ical
i
n
p
u
t
s
as th
e
y
o
n
l
y
co
n
s
id
er
th
e
cu
r
r
e
n
t in
p
u
ts
.
R
ec
u
r
r
en
t
n
e
u
r
al
n
et
w
o
r
k
,
w
h
er
e
in
f
o
r
m
at
io
n
c
y
c
le
s
i
n
a
lo
o
p
in
b
o
th
d
ir
ec
tio
n
s
,
f
r
o
m
t
h
e
in
p
u
t
la
y
er
to
th
e
o
u
tp
u
t
la
y
er
a
n
d
th
e
o
t
h
er
w
a
y
ar
o
u
n
d
.
So
m
e
n
et
w
o
r
k
s
a
llo
w
t
h
e
p
er
s
i
s
te
n
ce
o
f
i
n
f
o
r
m
atio
n
b
y
ta
k
i
n
g
in
to
ac
co
u
n
t th
e
c
u
r
r
en
t i
n
p
u
t
as
w
e
ll a
s
i
n
p
u
t
s
r
ec
eiv
ed
p
r
ev
io
u
s
l
y
.
2
.
2
.
1
.
M
ulti
-
la
y
er
perc
ept
ro
n (
M
L
P
)
T
h
is
s
p
ec
if
ic
ar
ch
itect
u
r
e
o
f
m
u
lti
-
la
y
er
n
eu
r
al
n
e
t
w
o
r
k
s
i
s
ca
lled
a
f
ee
d
-
f
o
r
w
ar
d
n
eu
r
a
l
n
et
w
o
r
k
,
b
ec
au
s
e
m
u
ltip
le
“
s
u
cc
es
s
i
v
e
l
a
y
er
s
f
ee
d
in
to
o
n
e
a
n
o
th
er
i
n
th
e
f
o
r
w
ar
d
d
ir
ec
tio
n
f
r
o
m
in
p
u
t
to
o
u
tp
u
t”
[
2
4
]
.
T
h
er
ef
o
r
e,
th
e
m
u
lti
-
la
y
er
p
er
ce
p
tr
o
n
n
et
w
o
r
k
g
en
er
ate
s
an
o
u
tp
u
t
f
r
o
m
s
o
m
e
g
iv
e
n
in
p
u
ts
.
T
h
ey
co
n
s
is
t
o
f
m
u
ltip
le
la
y
er
s
o
f
n
o
d
es
o
r
n
eu
r
o
n
s
(
at
leas
t
th
r
ee
)
.
On
e
i
n
p
u
t
la
y
er
,
s
o
m
e
h
id
d
en
la
y
e
r
s
,
an
d
o
n
e
o
u
tp
u
t
la
y
er
.
E
v
er
y
n
o
d
e
ex
ce
p
t
t
h
e
i
n
p
u
t
o
n
es
ar
e
ca
lled
n
eu
r
o
n
s
.
No
n
li
n
ea
r
ac
tiv
at
io
n
f
u
n
ctio
n
s
ar
e
u
s
ed
to
m
o
d
el
th
e
n
o
n
-
lin
ea
r
it
y
o
f
a
g
i
v
en
p
r
o
b
lem
.
T
h
e
co
m
p
u
ta
tio
n
h
ap
p
en
s
i
n
th
e
h
id
d
en
la
y
er
s
.
Firstl
y
,
th
e
f
ir
s
t
la
y
er
s
e
n
d
s
i
n
p
u
t
d
ata
to
th
e
h
id
d
en
n
o
d
es
i
n
t
h
e
h
id
d
en
la
y
er
.
T
h
ese
n
o
d
es
co
m
b
i
n
e
th
e
d
ata
w
it
h
a
s
et
o
f
co
ef
f
ic
ien
ts
o
r
w
ei
g
h
ts
t
h
at
eith
er
am
p
li
f
y
o
r
m
in
i
m
ize
th
e
in
p
u
t
,
th
en
th
e
r
es
u
lti
n
g
p
r
o
d
u
cts
ar
e
a
d
d
ed
.
Fin
all
y
,
a
n
ac
tiv
atio
n
f
u
n
c
tio
n
is
ap
p
lied
to
th
e
s
u
m
.
T
h
e
ac
tiv
atio
n
f
u
n
ct
io
n
d
eter
m
i
n
es
h
o
w
m
u
c
h
an
d
w
h
e
th
er
t
h
e
s
i
g
n
al
p
r
o
g
r
ess
es t
h
r
o
u
g
h
th
e
n
et
w
o
r
k
to
a
f
f
ec
t t
h
e
f
in
a
l r
es
u
l
t.
2
.
2
.
2
.
L
o
ng
s
ho
r
t
-
t
er
m
m
e
mo
ry
(
L
ST
M
)
T
h
e
L
ST
M
n
et
w
o
r
k
w
a
s
i
n
tr
o
d
u
ce
d
b
y
Ho
c
h
r
eiter
a
n
d
Sc
h
m
id
h
u
b
er
i
n
1
9
9
7
[
2
5
]
.
I
t
is
a
r
ec
u
r
r
en
t
n
eu
r
al
n
et
w
o
r
k
,
s
p
ec
if
icall
y
a
g
ated
ne
t
w
o
r
k
u
s
ed
i
n
d
ee
p
lear
n
i
n
g
.
Me
a
n
i
n
g
it
h
as
f
ee
d
b
ac
k
co
n
n
ec
tio
n
s
u
n
l
ik
e
th
e
f
ee
d
f
o
r
w
ar
d
n
eu
r
al
n
et
w
o
r
k
s
.
I
t
ca
n
b
e
u
s
ed
in
class
i
f
icatio
n
an
d
f
o
r
ec
asti
n
g
o
f
a
ti
m
e
s
er
ies
an
d
it
is
esp
ec
iall
y
e
f
f
ec
ti
v
e
i
n
s
o
lv
i
n
g
s
eq
u
e
n
tial p
r
o
b
le
m
s
li
k
e
s
p
ee
ch
an
d
h
a
n
d
w
r
itte
n
r
ec
o
g
n
i
t
io
n
.
L
ST
Ms
ar
e
ab
le
to
lear
n
lo
n
g
-
ter
m
d
ep
en
d
en
c
ies;
in
o
th
er
w
o
r
d
s
,
th
e
y
e
x
ce
l
at
r
e
m
e
m
b
er
i
n
g
in
f
o
r
m
atio
n
f
o
r
lo
n
g
p
er
io
d
s
o
f
ti
m
e
.
A
L
ST
M
u
n
it
is
co
m
p
o
s
ed
o
f
a
ce
ll
(
th
at
h
as
a
s
elf
-
l
o
o
p
)
,
an
in
p
u
t
g
ate,
an
o
u
tp
u
t
g
ate
an
d
a
f
o
r
g
et
g
ate.
T
h
e
ce
ll
r
em
e
m
b
er
s
v
al
u
es
o
v
er
in
ter
v
als
o
f
ti
m
e
wh
ile
th
e
t
h
r
ee
g
ates
co
n
tr
o
l th
e
f
lo
w
o
f
in
f
o
r
m
at
io
n
in
to
a
n
d
o
u
t o
f
it.
An
ex
a
m
p
le
o
f
a
ce
ll u
n
it is
s
h
o
w
n
in
t
h
e
Fig
u
r
e
2
.
=
ℎ
(
+
ℎ
−
1
+
)
.
(
1
)
=
(
+
ℎ
−
1
+
)
.
(
2
)
=
(
+
ℎ
−
1
+
)
.
(
3
)
=
−
1
⊗
+
⊗
(
4
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
C
o
mp
a
r
a
tive
a
n
a
lysi
s
o
f sh
o
r
t
-
term d
ema
n
d
p
r
ed
ictin
g
mo
d
e
ls
u
s
in
g
A
R
I
MA
a
n
d
...
(
Ha
lima
B
o
u
s
q
a
o
u
i)
3323
=
(
+
ℎ
−
1
+
)
.
(
5
)
ℎ
=
ta
n
h
(
)
⊗
(
6
)
T
h
e
ce
ll r
ec
eiv
es a
s
in
p
u
t t
h
e
cu
r
r
en
t i
n
p
u
t
an
d
th
e
p
r
ev
io
u
s
ce
ll o
u
tp
u
t
ℎ
−
1
.
I
t o
u
tp
u
ts
ℎ
.
Fig
u
r
e
2
.
T
h
e
s
tr
u
ctu
r
e
o
f
a
t
y
p
ical
m
e
m
o
r
y
ce
l
l o
f
L
ST
M
[
2
6
]
2
.
2
.
3
.
Co
nv
o
lutio
na
l
neura
l
net
w
o
rk
s
(
CNN
o
r
Co
nv
Net
)
A
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
et
wo
r
k
is
a
s
p
ec
if
ic
t
y
p
e
o
f
ar
ti
f
icial
n
eu
r
al
n
et
w
o
r
k
u
s
u
all
y
u
s
ed
f
o
r
co
g
n
iti
v
e
tas
k
s
s
u
c
h
as
i
m
ag
e
r
ec
o
g
n
itio
n
,
i
m
a
g
e
p
r
o
ce
s
s
in
g
an
d
n
atu
r
al
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
a
n
d
ti
m
e
s
er
ie
s
d
ata
an
d
m
o
r
e.
C
NN
i
s
t
h
e
m
o
s
t
p
o
p
u
lar
d
ee
p
lear
n
i
n
g
a
lg
o
r
ith
m
m
o
d
el
u
s
ed
f
o
r
i
m
a
g
e
p
r
o
ce
s
s
i
n
g
,
it
i
s
s
i
m
p
ler
to
tr
ai
n
a
n
d
m
o
r
e
e
f
f
e
ctiv
e
t
h
a
n
t
h
e
tr
ad
itio
n
a
l
n
e
u
r
al
n
et
w
o
r
k
s
s
in
ce
it
h
as
th
e
a
b
ilit
y
to
ca
p
t
u
r
e
t
h
e
te
m
p
o
r
al
an
d
s
p
atial
d
ep
en
d
e
n
cies
in
an
i
m
ag
e
b
y
ap
p
l
y
in
g
t
h
e
ap
p
r
o
p
r
iate
f
ilter
s
w
it
h
a
n
i
n
d
ep
en
d
en
ce
f
r
o
m
h
u
m
a
n
ef
f
o
r
t.
I
n
d
ee
d
,
th
e
la
y
e
r
s
in
C
NN
ar
ch
itec
tu
r
e
ar
e
ar
r
an
g
ed
i
n
o
r
d
er
to
co
v
er
th
e
en
tire
v
is
u
al
f
ield
th
a
t
h
elp
s
to
av
er
t t
h
e
p
iece
m
ea
l i
m
ag
e
-
p
r
o
ce
s
s
i
n
g
p
r
o
b
lem
i
n
tr
ad
itio
n
al
n
e
u
r
al
n
et
w
o
r
k
s
.
Si
m
i
lar
to
th
e
m
u
ltil
a
y
er
p
er
ce
p
tr
o
n
(
ML
P
)
s
tr
u
ctu
r
e
o
f
n
eu
r
al
n
et
w
o
r
k
s
,
th
e
la
y
er
s
in
C
NN
a
r
e
ca
teg
o
r
ized
in
to
t
h
r
ee
t
y
p
e
s
:
a
n
i
n
p
u
t
la
y
er
,
a
n
o
u
tp
u
t
la
y
er
an
d
a
h
id
d
en
la
y
er
.
T
h
er
e
ar
e
t
w
o
t
y
p
es
o
f
h
id
d
en
la
y
er
s
[
2
7
]
:
Featu
r
e
lear
n
in
g
la
y
er
s
,
p
er
f
o
r
m
in
g
t
h
r
ee
t
y
p
e
s
o
f
o
p
er
atio
n
s
:
co
n
v
o
l
u
tio
n
,
p
o
o
lin
g
,
a
n
d
r
ec
tif
ied
li
n
ea
r
u
n
i
t (
R
e
L
U)
o
n
th
e
i
n
p
u
t d
ata;
C
las
s
i
f
icatio
n
la
y
er
s
,
co
m
p
o
s
e
d
o
f
f
u
ll
y
co
n
n
ec
ted
la
y
er
s
an
d
n
o
r
m
a
lizatio
n
la
y
er
s
.
3.
E
XP
RIM
E
NT
AT
I
O
N
RE
S
UL
T
S AN
D
D
I
SCU
SS
I
O
N
3
.
1
.
Aut
o
re
g
re
s
s
iv
e
inte
g
ra
t
ed
m
o
v
ing
a
v
er
a
g
e
(
ARI
M
A)
3
.
1
.
1
.
T
he
pro
po
s
ed
a
pp
ro
a
ch
T
h
e
AR
I
M
A
(
p
,
d
,
q
)
m
o
d
el
is
d
eter
m
i
n
ed
b
y
ch
o
o
s
i
n
g
t
h
e
t
h
r
ee
p
ar
a
m
eter
s
p
,
d
,
an
d
q
.
T
h
e
p
lo
t
o
f
th
e
d
ata
w
ill
d
eter
m
in
e
t
h
e
o
r
d
er
o
f
d
if
f
er
en
c
in
g
d
,
th
e
AC
F
p
lo
t
w
il
l
d
eter
m
i
n
e
th
e
lag
q
,
an
d
th
e
P
AC
F
p
lo
t
th
e
la
g
p
.
B
esid
es,
th
e
m
o
d
el
is
i
m
p
r
o
v
ed
b
y
u
s
in
g
t
h
e
g
r
id
s
ea
r
ch
i
n
o
r
d
er
to
ch
o
o
s
e
th
e
b
est
p
,
d
,
q
p
ar
am
eter
s
.
T
h
e
m
o
d
els
ar
e
ev
al
u
ated
w
it
h
t
h
e
R
MSE
,
th
e
R
o
o
t
Me
an
Sq
u
ar
e
E
r
r
o
r
,
th
at
i
s
t
h
e
m
o
s
t
f
r
eq
u
en
tl
y
u
s
ed
m
etr
ic
s
o
f
p
r
ed
ictio
n
p
er
f
o
r
m
an
ce
an
d
ca
l
cu
lated
as
th
e
s
q
u
ar
e
r
o
o
t
o
f
th
e
Me
an
Sq
u
ar
e
E
r
r
o
r
,
w
h
er
e
th
e
M
SE
is
t
h
e
ar
ith
m
etic
m
ea
n
o
f
t
h
e
s
q
u
ar
es
o
f
t
h
e
d
i
f
f
er
e
n
ce
s
b
et
w
ee
n
f
o
r
ec
asts
a
n
d
o
b
s
er
v
atio
n
s
.
3
.
1
.
2
.
T
he
ACF
a
nd
P
ACF
p
lo
t
s
I
n
th
e
Fi
g
u
r
e
3
,
d
e
m
a
n
d
is
p
r
esen
ted
as
a
t
i
m
e
s
er
ies
d
at
a
h
av
in
g
t
h
e
d
a
y
i
n
t
h
e
x
-
a
x
is
an
d
t
h
e
d
em
a
n
d
q
u
an
tit
y
o
n
th
e
y
-
a
x
i
s
.
W
e
n
o
tice
th
at
th
e
ti
m
e
s
er
i
es
m
i
g
h
t
b
e
s
tatio
n
ar
y
a
n
d
s
o
m
i
g
h
t
n
o
t
r
eq
u
ir
e
d
if
f
er
e
n
ci
n
g
,
b
u
t
to
b
e
s
u
r
e,
at
least
a
d
i
f
f
er
e
n
ce
o
r
d
er
o
f
1
w
ill
b
e
ap
p
lied
.
Nex
t
let
u
s
lo
o
k
a
t
t
h
e
au
to
co
r
r
elatio
n
f
u
n
ctio
n
(
AC
F)
an
d
p
ar
tial
au
to
co
r
r
elatio
n
f
u
n
ctio
n
(
P
A
C
F)
p
lo
ts
o
f
th
e
ti
m
e
s
er
ies
i
n
Fig
u
r
e
4
.
T
h
e
A
C
F
p
lo
t
is
d
esig
n
ed
to
s
h
o
w
w
h
et
h
er
t
h
e
el
e
m
en
ts
o
f
a
t
i
m
e
s
er
ies
ar
e
p
o
s
iti
v
el
y
co
r
r
elate
d
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
4
,
A
u
g
u
s
t 2
0
2
1
:
3
3
1
9
-
3328
3324
n
eg
at
iv
el
y
co
r
r
elate
d
,
o
r
ze
r
o
co
r
r
elate
d
(
in
d
ep
en
d
en
t
o
f
ea
ch
o
th
er
)
.
I
n
o
th
er
w
o
r
d
s
,
it
d
escr
ib
es
h
o
w
w
ell
th
e
p
r
esen
t
v
a
lu
e
s
i
n
t
h
e
s
er
ies
is
r
el
ated
to
its
p
ast
v
alu
e
s
.
A
P
A
C
F
is
a
s
u
m
m
ar
y
o
f
th
e
r
e
latio
n
s
h
ip
b
et
w
ee
n
o
b
s
er
v
atio
n
s
i
n
a
ti
m
e
s
er
ies
w
it
h
o
b
s
er
v
atio
n
s
a
t
p
r
io
r
ti
m
e
s
tep
s
af
ter
ex
c
lu
d
i
n
g
t
h
e
co
n
tr
ib
u
tio
n
s
f
r
o
m
t
h
e
in
ter
m
ed
iate
lag
s
.
Fig
u
r
e
3
.
De
m
an
d
p
er
d
a
y
p
lo
t
Fig
u
r
e
4
.
AC
F
an
d
P
AC
F p
lo
t
s
f
o
r
d
e
m
an
d
q
u
a
n
tit
y
s
er
ies
W
e
in
itiall
y
ta
k
e
th
e
o
r
d
er
o
f
AR
ter
m
to
b
e
eq
u
al
to
a
s
m
a
n
y
la
g
s
th
a
t
cr
o
s
s
e
s
t
h
e
s
i
g
n
if
i
ca
n
ce
li
m
i
t
in
t
h
e
P
A
C
F
p
lo
t.
T
h
is
s
u
g
g
e
s
ts
t
h
at
a
g
o
o
d
s
tar
t
f
o
r
t
h
e
a
u
t
o
r
eg
r
ess
i
v
e
p
o
r
tio
n
o
f
t
h
e
m
o
d
el
is
AR
(
2
)
th
at
i
s
an
au
to
-
r
eg
r
es
s
io
n
m
o
d
el
w
it
h
t
w
o
lag
o
b
s
er
v
atio
n
u
s
ed
as
in
p
u
t.
In
th
e
s
a
m
e
w
a
y
t
h
a
t
w
e
lo
o
k
ed
at
th
e
P
A
C
F
p
lo
t
f
o
r
t
h
e
q
u
a
n
tit
y
o
f
th
e
AR
p
ar
t
o
f
t
h
e
m
o
d
el,
w
e
ca
n
lo
o
k
at
t
h
e
AC
F
p
lo
t
f
o
r
th
e
n
u
m
b
er
o
f
M
A
ter
m
s
.
O
n
l
y
o
n
e
lag
i
s
ab
o
v
e
t
h
e
s
i
g
n
if
ica
n
ce
li
n
e.
So
let
u
s
f
i
x
th
e
o
r
d
er
as M
A
(
1
)
.
3
.
1
.
3
.
F
it
t
ing
t
he
ARI
M
A(
2
,
1
,
1
)
m
o
del
No
w
th
at
w
e
h
a
v
e
d
eter
m
in
e
d
th
e
v
al
u
es
o
f
p
,
d
a
n
d
q
,
we
ca
n
f
i
t
t
h
e
AR
I
M
A
m
o
d
el.
Usi
n
g
th
e
AR
I
M
A
(
2
,
1
,
1
)
im
p
le
m
e
n
tatio
n
in
s
ta
ts
m
o
d
el
s
p
ac
k
ag
e
i
n
P
y
th
o
n
w
e
f
it
t
h
e
m
o
d
el
an
d
we
g
et
th
e
f
o
llo
w
i
n
g
r
esu
lt
s
.
T
h
e
Fig
u
r
e
s
5
an
d
6
s
h
o
w
t
h
at
t
h
e
r
esid
u
al
er
r
o
r
s
ar
e
n
ea
r
ze
r
o
m
ea
n
w
it
h
a
u
n
if
o
r
m
v
ar
ia
n
ce
.
Nex
t,
let
u
s
ca
lc
u
late
t
h
e
r
o
o
t
m
ea
n
s
q
u
ar
ed
er
r
o
r
(
R
MSE
)
an
d
p
lo
t
th
e
ac
t
u
al
v
s
p
r
ed
icted
v
al
u
es.
U
s
i
n
g
s
k
lear
n
an
d
s
tats
m
o
d
els p
ac
k
a
g
es o
f
p
y
th
o
n
w
e
f
it a
n
AR
I
M
A
(
2
,
1
,
1
)
m
o
d
el.
T
h
is
r
esu
lts
i
n
a
n
R
M
SE
o
f
8
2
3
.
9
4
5
.
Fig
u
r
e
5
.
AR
I
M
A
f
it r
es
id
u
al
er
r
o
r
lin
e
p
lo
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
C
o
mp
a
r
a
tive
a
n
a
lysi
s
o
f sh
o
r
t
-
term d
ema
n
d
p
r
ed
ictin
g
mo
d
e
ls
u
s
in
g
A
R
I
MA
a
n
d
...
(
Ha
lima
B
o
u
s
q
a
o
u
i)
3325
Fig
u
r
e
6
.
AR
I
M
A
f
it r
es
id
u
al
er
r
o
r
d
en
s
it
y
p
lo
t
A
li
n
e
p
lo
t
i
n
Fi
g
u
r
e
7
is
cr
ea
ted
s
h
o
w
i
n
g
t
h
e
e
x
p
ec
ted
v
alu
e
s
(
b
lu
e)
co
m
p
ar
ed
to
th
e
f
o
r
ec
ast
p
r
ed
ictio
n
s
(
r
ed
)
.
W
e
ca
n
co
n
clu
d
e
th
at
th
e
m
o
d
el
is
i
n
n
ee
d
o
f
f
u
r
th
er
tu
n
i
n
g
o
f
t
h
e
p
,
d
,
an
d
q
p
ar
am
eter
s
.
Fo
r
a
b
etter
f
it
o
f
th
e
m
o
d
el
to
th
e
d
ata,
o
th
er
p
ar
am
eter
s
p
,
d
,
an
d
q
ar
e
test
ed
u
s
in
g
th
e
g
r
id
s
ea
r
ch
i
n
g
p
ar
am
eter
s
ap
p
r
o
ac
h
.
Fig
u
r
e
7
.
P
r
ed
icte
d
v
s
ac
tu
al
d
e
m
an
d
q
u
a
n
tit
ies
2
.
1
.
4
.
G
rid sea
rc
hin
g
pa
ra
met
er
s
T
h
e
p
r
o
ce
s
s
o
f
tr
ain
in
g
an
d
ev
alu
ati
n
g
AR
I
M
A
m
o
d
els
o
n
d
if
f
er
en
t
co
m
b
i
n
atio
n
s
o
f
m
o
d
el
h
y
p
er
p
ar
a
m
eter
s
ca
n
b
e
au
to
m
ated
.
T
h
is
is
ca
lled
a
g
r
id
s
ea
r
ch
o
r
m
o
d
el
tu
n
in
g
in
m
ac
h
in
e
lear
n
in
g
.
W
e
s
p
lit
th
e
d
ata
s
et
in
t
w
o
:
6
6
%
f
o
r
t
h
e
tr
ain
i
n
g
s
et
a
n
d
3
4
%
f
o
r
t
h
e
test
s
et
.
W
e
c
h
o
s
e
t
h
e
p
v
al
u
e
s
to
r
an
g
e
f
r
o
m
1
to
1
0
,
th
e
d
v
alu
e
s
f
r
o
m
0
to
3
an
d
t
h
e
q
f
r
o
m
0
to
3
.
T
h
e
b
est
AR
I
M
A
m
o
d
el
t
h
at
y
ield
ed
th
e
b
est
R
MSE
is
AR
I
M
A
(
6
,
0
,
0
)
w
it
h
an
R
M
SE
o
f
4
8
5
.
6
9
0
.
T
h
is
is
s
h
o
w
n
in
th
e
F
ig
u
r
e
8
.
W
e
n
o
tice
t
h
at
th
e
lo
w
er
th
e
p
v
alu
e
t
h
e
h
ig
h
er
t
h
e
R
MSE
.
3
.
2
.
M
ulti
-
la
y
er
perc
ept
ro
n (
M
L
P
)
T
h
e
m
u
lti
-
la
y
er
p
er
ce
p
tr
o
n
n
eu
r
al
n
et
w
o
r
k
is
u
s
ed
to
f
o
r
ec
ast
th
e
d
e
m
an
d
q
u
a
n
tit
y
i
n
a
s
i
m
p
l
e
s
u
p
p
l
y
c
h
ai
n
co
m
p
o
s
ed
o
f
o
n
e
r
etailer
an
d
o
n
e
s
u
p
p
lier
.
T
h
e
d
ata
b
elo
n
g
s
to
a
Mo
r
o
cc
an
s
u
p
er
m
ar
k
et
d
ataset.
P
r
ed
ictin
g
t
h
e
d
e
m
a
n
d
o
f
a
g
i
v
en
d
a
y
d
i
n
v
o
lv
e
s
t
h
e
d
e
m
a
n
d
q
u
an
tit
ies o
f
t
h
e
t
h
r
ee
p
r
ev
io
u
s
w
ee
k
s
(
d
-
7
,
d
-
14
an
d
d
-
2
1
)
.
T
h
e
n
eu
r
al
n
et
w
o
r
k
i
s
co
m
p
o
s
ed
o
f
t
h
r
ee
h
id
d
en
la
y
er
s
ea
ch
co
n
ta
in
i
n
g
1
0
n
e
u
r
o
n
s
.
T
h
e
n
eu
r
a
l
n
et
w
o
r
k
is
tr
ain
ed
u
s
i
n
g
t
h
e
ad
am
v
er
s
io
n
o
f
t
h
e
g
r
ad
ien
t
d
escen
t,
th
e
m
s
e
lo
s
s
f
u
n
ctio
n
an
d
th
e
r
elu
f
u
n
c
tio
n
as
a
n
ac
ti
v
atio
n
f
u
n
ctio
n
.
T
h
i
s
r
es
u
lts
i
n
a
n
R
MSE
o
f
0
.
1
4
0
in
th
e
test
s
et
a
f
ter
n
o
r
m
ali
za
tio
n
a
n
d
4
6
4
.
2
6
1
b
ef
o
r
e
n
o
r
m
aliza
tio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
4
,
A
u
g
u
s
t 2
0
2
1
:
3
3
1
9
-
3328
3326
Fig
u
r
e
8
.
G
r
id
s
ea
r
ch
in
g
p
ar
am
eter
s
r
es
u
lt
s
3
.
3
.
L
o
ng
s
ho
rt
-
t
er
m
m
e
m
o
ry
(
L
ST
M
)
I
n
a
p
r
ev
io
u
s
s
tu
d
y
[
2
6
]
,
L
o
n
g
-
Sh
o
r
t
T
er
m
Me
m
o
r
y
i
s
u
s
e
d
to
f
o
r
ec
ast
d
em
a
n
d
.
I
n
ad
d
itio
n
to
th
e
in
p
u
t
s
p
r
o
p
o
s
ed
in
th
e
M
L
P
m
o
d
el,
th
e
L
ST
M
h
as
a
ti
m
e
d
i
m
e
n
s
io
n
.
I
t
h
a
s
a
th
r
e
e
-
d
i
m
e
n
s
io
n
al
s
h
ap
e
co
m
p
o
s
ed
o
f
t
h
e
n
u
m
b
er
o
f
e
x
a
m
p
les,
t
h
e
n
u
m
b
er
o
f
i
n
p
u
t
s
,
an
d
th
e
n
u
m
b
er
o
f
ti
m
e
s
tep
s
[
28]
.
I
n
th
e
p
r
o
p
o
s
ed
n
eu
r
al
n
et
w
o
r
k
,
t
h
e
d
e
m
a
n
d
o
f
a
g
iv
e
n
d
a
y
“
d
”
is
p
r
ed
icted
u
s
i
n
g
d
e
m
a
n
d
q
u
an
t
ities
o
f
th
e
d
a
y
s
:
d
-
6
,
d
-
1
2
,
an
d
d
-
1
8
.
T
h
er
ef
o
r
e,
w
e
m
o
d
el
th
e
p
r
o
b
lem
as
a
th
r
ee
ti
m
e
-
s
tep
s
p
r
ed
ictio
n
o
n
o
n
e
in
p
u
t
co
r
r
esp
o
n
d
in
g
to
t
h
e
q
u
an
tit
y
o
f
t
h
e
d
a
y
,
a
n
d
th
is
f
o
r
ea
ch
o
u
tp
u
t.
T
h
is
in
s
tead
o
f
c
o
n
s
id
er
in
g
t
h
at
t
h
e
d
ata
is
co
m
p
o
s
ed
o
f
th
r
ee
in
p
u
ts
as
w
e
d
id
in
t
h
e
ML
P
s
t
u
d
y
.
T
h
e
L
ST
M
n
et
w
o
r
k
is
co
m
p
o
s
ed
o
f
5
0
n
eu
r
o
n
s
in
th
e
h
id
d
en
la
y
er
an
d
1
n
eu
r
o
n
in
t
h
e
o
u
tp
u
t
la
y
er
.
T
h
e
r
o
o
t
m
ea
n
s
q
u
ar
ed
er
r
o
r
an
d
t
h
e
ef
f
icie
n
t
A
d
a
m
v
er
s
io
n
o
f
th
e
s
to
c
h
asti
c
g
r
ad
ien
t
is
u
s
ed
.
T
h
is
y
ie
ld
ed
an
R
M
SE
af
ter
n
o
r
m
aliza
tio
n
o
f
0
.
1
3
8
an
d
an
R
MSE
b
ef
o
r
e
n
o
r
m
aliza
t
io
n
o
f
4
5
7
.
9
5
8
.
3
.
4
.
Co
nv
o
lutio
na
l neura
l net
w
o
rk
(
CNN
o
r
Co
nv
Net
)
T
h
e
co
n
v
o
lu
t
io
n
al
n
e
u
r
al
n
et
w
o
r
k
(
C
NN
o
r
C
o
n
v
Net)
is
a
s
p
ec
ial
t
y
p
e
o
f
n
eu
r
al
n
et
wo
r
k
m
o
d
el
u
s
ed
f
o
r
w
o
r
k
i
n
g
w
it
h
t
w
o
-
d
im
en
s
io
n
a
l
an
d
th
r
ee
-
d
i
m
e
n
s
io
n
al
d
ata
lik
e
i
m
ag
e
s
,
it
ca
n
b
e
ap
p
lied
w
it
h
o
n
e
-
d
i
m
en
s
io
n
al
d
ata
li
k
e
ti
m
e
s
er
ies.
T
h
e
p
r
o
p
o
s
ed
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
e
t
w
o
r
k
i
s
co
m
p
o
s
ed
o
f
t
h
e
f
o
llo
w
in
g
la
y
er
s
:
A
o
n
e
-
d
i
m
e
n
s
io
n
al
co
n
v
o
lu
tio
n
al
la
y
er
w
it
h
6
4
f
ilter
s
,
a
k
er
n
el
s
ize
o
f
2
a
n
d
a
n
in
p
u
t
s
h
a
p
e
o
f
(
3
,
1
)
,
th
e
ac
tiv
atio
n
f
u
n
ctio
n
co
n
s
is
ts
o
f
th
e
r
elu
f
u
n
ctio
n
;
A
o
n
e
-
d
i
m
en
s
io
n
al
m
a
x
p
o
o
lin
g
la
y
er
w
it
h
a
f
il
ter
o
f
s
ize
2
;
A
f
latte
n
la
y
er
;
A
f
u
ll
y
co
n
n
ec
ted
la
y
er
co
m
p
o
s
ed
o
f
5
0
n
eu
r
o
n
s
w
it
h
r
elu
b
ein
g
t
h
e
ac
ti
v
atio
n
f
u
n
c
tio
n
.
W
e
tr
ain
th
e
m
o
d
el
u
s
in
g
t
h
e
ad
am
v
er
s
io
n
o
f
s
to
ch
a
s
tic
d
escen
t
a
n
d
th
e
MSE
lo
s
s
f
u
n
ctio
n
.
T
h
e
tr
ain
i
n
g
a
f
ter
2
0
0
ep
o
ch
s
r
esu
l
ts
in
a
R
MSE
o
f
0
.
1
3
8
af
ter
n
o
r
m
aliza
tio
n
,
an
d
4
5
7
.
0
7
9
b
ef
o
r
e
n
o
r
m
a
lizatio
n
.
3
.
5
.
Dis
cu
s
s
io
n
I
n
m
ac
h
i
n
e
lear
n
i
n
g
t
h
e
p
r
ed
ictio
n
ac
cu
r
ac
y
is
u
s
ed
f
o
r
d
is
cr
ete
p
r
o
b
lem
s
,
m
ea
n
i
n
g
cla
s
s
i
f
icatio
n
o
n
es.
Si
n
ce
w
e
ar
e
d
ea
lin
g
w
i
th
a
r
eg
r
ess
io
n
p
r
o
b
lem
,
w
e
c
h
o
s
e
to
ev
al
u
ate
th
e
m
o
d
els
u
s
in
g
t
h
e
r
o
o
t
m
ea
n
s
q
u
ar
ed
er
r
o
r
R
MSE
,
in
ad
d
iti
o
n
to
s
o
m
e
o
th
er
cr
iter
ia
.
T
h
e
R
MSE
m
ea
s
u
r
es
h
o
w
clo
s
e
th
e
r
ea
l
v
al
u
es
ar
e
to
th
e
p
r
ed
icted
o
n
es.
Hen
ce
its
u
s
e.
A
s
s
h
o
w
n
i
n
T
ab
le
1
,
th
e
w
o
r
s
t
p
r
ed
ictio
n
r
es
u
lt
s
ar
e
o
b
tain
ed
w
it
h
AR
I
M
A
m
o
d
e
l
w
h
ic
h
i
s
u
n
d
er
s
ta
n
d
ab
le
s
i
n
ce
it
i
s
a
l
in
ea
r
m
o
d
el.
T
h
e
n
e
u
r
al
n
e
t
w
o
r
k
s
p
r
o
p
o
s
ed
m
o
d
els,
o
n
t
h
e
o
th
er
h
an
d
,
ar
e
n
o
n
li
n
ea
r
m
o
d
els
t
h
at
ca
n
a
p
p
r
o
x
im
a
te
an
y
f
u
n
c
tio
n
,
h
e
n
ce
th
e
b
etter
r
esu
lt
s
.
C
NN
h
as
s
h
o
w
n
t
h
e
b
est
R
MSE
,
s
li
g
h
t
l
y
b
etter
t
h
an
th
e
L
ST
M
m
o
d
el.
T
h
is
m
i
g
h
t
b
e
th
a
n
k
s
to
its
ab
ilit
y
to
d
o
f
e
atu
r
e
s
e
lectio
n
o
n
i
ts
o
w
n
.
T
ab
le
1
.
R
esu
lts
o
f
th
e
f
o
u
r
ap
p
r
o
ac
h
es
ac
co
r
d
in
g
to
d
if
f
er
e
n
t c
r
iter
ias
A
R
I
M
A
M
L
P
L
S
T
M
C
N
N
R
M
S
E
b
e
f
o
r
e
n
o
r
mal
i
z
a
t
i
o
n
4
8
5
.
6
9
0
4
6
4
.
2
6
1
4
5
7
.
9
5
8
4
5
7
.
0
7
9
R
M
S
E
a
f
t
e
r
n
o
r
mal
i
z
a
t
i
o
n
-
0
.
1
4
0
0
.
1
3
8
0
.
1
3
8
M
o
d
e
l
t
r
a
i
n
i
n
g
t
i
me
(
s)
1
.
1
2
9
2
0
1
2
.
3
4
1
3
6
1
2
.
5
4
1
2
9
2
2
.
7
2
0
2
3
6
C
o
n
s
u
me
d
e
n
e
r
g
y
(
Jo
u
l
e
)
1
6
.
9
3
8
0
1
0
3
5
.
1
2
0
4
1
2
3
8
.
1
1
9
3
7
6
4
0
.
8
0
3
5
3
4
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
C
o
mp
a
r
a
tive
a
n
a
lysi
s
o
f sh
o
r
t
-
term d
ema
n
d
p
r
ed
ictin
g
mo
d
e
ls
u
s
in
g
A
R
I
MA
a
n
d
...
(
Ha
lima
B
o
u
s
q
a
o
u
i)
3327
Kn
o
w
i
n
g
th
a
t
th
e
c
o
n
s
u
m
ed
en
er
g
y
(
J
o
u
le)
=
m
o
d
el
tr
ain
i
n
g
ti
m
e
(
s
ec
o
n
d
s
)
X
GP
U/C
P
U
P
o
w
er
(
W
att)
.
W
e
ar
e
u
s
in
g
an
i
n
tel(
R
)
C
o
r
e
(
T
M)
i7
-
8
5
6
5
U
C
P
U
.
I
ts
t
h
er
m
al
d
esi
g
n
p
o
w
er
is
1
5
W
.
Mo
r
eo
v
er
,
th
e
o
p
er
atin
g
s
y
s
te
m
is
w
i
n
d
o
w
s
1
0
6
4
b
its
an
d
th
e
R
a
m
is
8
1
9
2
M
B
.
W
e
n
o
tice
th
at
co
n
t
r
ar
y
to
R
MSE
,
t
h
e
m
o
d
el
tr
ai
n
i
n
g
t
i
m
e
a
n
d
co
n
s
u
m
ed
en
er
g
y
ar
e
b
est f
o
r
A
R
I
MA
a
n
d
w
o
r
s
t
f
o
r
th
e
C
NN
n
e
t
w
o
r
k
s
.
4.
CO
NCLU
SI
O
N
T
h
e
m
ai
n
co
n
tr
ib
u
tio
n
o
f
t
h
is
s
tu
d
y
is
co
m
p
ar
in
g
t
h
e
p
er
f
o
r
m
an
ce
o
f
f
o
u
r
p
r
ed
ictio
n
m
o
d
els,
n
a
m
el
y
AR
I
M
A
a
s
tati
s
tical
m
o
d
el,
ML
P
a
f
ee
d
f
o
r
w
ar
d
n
e
u
r
al
n
e
t
w
o
r
k
an
d
L
ST
M
a
r
ec
u
r
r
en
t
n
eu
r
al
n
et
w
o
r
k
a
n
d
C
NN
a
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
et
w
o
r
k
.
T
h
e
an
al
y
s
i
s
o
f
t
h
e
ex
p
er
i
m
e
n
t
r
es
u
lts
s
h
o
w
s
t
h
at
th
e
co
n
v
o
lu
t
io
n
al
n
et
w
o
r
k
g
iv
e
s
b
etter
f
o
r
ec
asti
n
g
r
es
u
lt
s
th
a
n
th
e
L
ST
M,
ML
P
n
et
w
o
r
k
s
a
n
d
AR
I
M
A
m
o
d
el.
Ho
w
e
v
er
,
th
e
m
o
d
el
tr
ai
n
i
n
g
ti
m
e
a
n
d
co
n
s
u
m
ed
e
n
er
g
y
ar
e
i
n
th
e
o
p
p
o
s
ite
o
r
d
er
.
Me
an
in
g
,
th
e
y
ar
e
b
est
f
o
r
A
r
i
m
a.
T
h
is
m
ea
n
s
t
h
at
to
g
et
th
e
b
etter
p
r
ed
ictio
n
r
esu
lt
s
th
e
C
NN
o
r
L
ST
M
n
et
w
o
r
k
s
s
h
o
u
ld
b
e
u
s
ed
.
Ho
w
e
v
er
,
if
t
h
e
m
o
d
el
i
s
b
ig
,
A
r
i
m
a
m
i
g
h
t b
e
a
b
etter
s
o
lu
tio
n
s
i
n
ce
it
h
as le
s
s
tr
ain
in
g
ti
m
e
an
d
les
s
co
n
s
u
m
ed
en
er
g
y
.
T
h
e
r
esu
lts
o
f
t
h
is
p
ap
er
ca
n
b
e
f
u
r
t
h
er
ex
te
n
d
ed
in
s
e
v
er
al
d
ir
ec
tio
n
s
.
A
s
a
f
u
t
u
r
e
r
esear
ch
d
ir
ec
tio
n
,
it see
m
s
i
n
ter
esti
n
g
to
ex
p
lo
r
e
th
e
h
y
p
er
-
p
ar
a
m
eter
tu
n
i
n
g
m
eth
o
d
k
n
o
w
in
g
t
h
at
t
h
e
L
ST
M
is
m
o
s
tl
y
ef
f
ec
ti
v
e
in
s
eq
u
en
ce
p
r
o
b
le
m
s
a
n
d
ti
m
e
s
er
ies
a
llo
w
in
g
i
n
f
o
r
m
ati
o
n
to
p
er
s
is
t
th
at
m
a
y
lead
t
o
b
etter
f
o
r
ec
asti
n
g
r
esu
lt
s
.
I
t
is
al
s
o
p
o
s
s
ib
le
to
u
s
e
s
ta
n
d
ar
d
b
en
ch
m
ar
k
d
atase
t
s
to
v
alid
ate
t
h
e
p
r
o
p
o
s
ed
p
r
e
d
ictio
n
m
o
d
els.
I
t
i
s
also
p
o
s
s
ib
le
to
u
s
e
o
th
er
m
o
d
els
in
th
e
co
m
p
ar
ativ
e
a
n
al
y
s
is
s
u
ch
as
t
h
e
B
o
ltz
m
an
n
m
ac
h
in
e
a
n
d
r
an
d
o
m
f
o
r
ests
a
m
o
n
g
o
th
er
s
.
RE
F
E
R
E
NC
E
S
[1
]
F
.
P
e
rss
o
n
a
n
d
J.
Ol
h
a
g
e
r,
“
P
e
rf
o
rm
a
n
c
e
si
m
u
latio
n
o
f
su
p
p
ly
c
h
a
in
d
e
sig
n
s,”
In
t.
J
.
Pro
d
.
Eco
n
.
,
v
o
l.
7
7
,
n
o
.
3
,
p
p
.
2
3
1
–
2
4
5
,
2
0
0
2
,
d
o
i:
1
0
.
1
0
1
6
/
S
0
9
2
5
-
5
2
7
3
(
0
0
)
0
0
0
8
8
-
8.
[2
]
P
.
Ha
u
g
u
e
l
a
n
d
E.
Via
rd
o
t,
“
De
la
su
p
p
ly
c
h
a
in
a
u
ré
se
a
u
in
d
u
str
iel,
”
L
’Exp
a
n
si
o
n
M
a
n
a
g
.
Rev
.
,
v
o
l.
1
0
1
,
n
o
.
1
,
p
p
.
9
4
–
1
0
0
,
2
0
0
1
.
[3
]
B.
F
a
n
o
o
d
i,
B
.
M
a
lm
ir,
a
n
d
F
.
F
.
Ja
h
a
n
ti
g
h
,
“
Re
d
u
c
i
n
g
d
e
m
a
n
d
u
n
c
e
rtain
ty
in
t
h
e
p
late
let
su
p
p
ly
c
h
a
in
th
r
o
u
g
h
a
rti
f
icia
l
n
e
u
ra
l
n
e
tw
o
rk
s
a
n
d
A
RIM
A
m
o
d
e
ls,”
Co
mp
u
t
.
Bi
o
l
.
M
e
d
.
,
v
o
l.
1
1
3
,
Ja
n
.
2
0
1
9
,
A
rt
.
n
o
.
1
0
3
4
1
5
,
d
o
i:
1
0
.
1
0
1
6
/j
.
c
o
m
p
b
io
m
e
d
.
2
0
1
9
.
1
0
3
4
1
5
.
[4
]
L
.
Ji,
Y.
Zo
u
,
K.
He
,
a
n
d
B.
Z
h
u
,
“
Ca
rb
o
n
f
u
tu
re
s
p
r
ice
f
o
re
c
a
stin
g
b
a
se
d
w
it
h
A
RIM
A
-
CNN
-
L
S
T
M
m
o
d
e
l,
”
Pro
c
e
d
ia
C
o
mp
u
t.
S
c
i
.
,
v
o
l.
1
6
2
,
p
p
.
3
3
–
3
8
,
2
0
1
9
,
d
o
i:
1
0
.
1
0
1
6
/j
.
p
ro
c
s.2
0
1
9
.
1
1
.
2
5
4
.
[5
]
M
.
M
a
ty
jas
z
e
k
,
P
.
R.
F
e
rn
á
n
d
e
z
,
A
.
Krz
e
m
ień
,
K.
W
o
d
a
rsk
i,
a
n
d
G
.
F
.
V
a
lv
e
rd
e
,
“
F
o
re
c
a
stin
g
c
o
k
in
g
c
o
a
l
p
rice
s
b
y
m
e
a
n
s
o
f
A
RIM
A
m
o
d
e
ls
a
n
d
n
e
u
ra
l
n
e
tw
o
rk
s,
c
o
n
sid
e
rin
g
th
e
tran
sg
e
n
ic
ti
m
e
se
ri
e
s
th
e
o
ry
,
”
Res
o
u
r.
Po
li
c
y
,
v
o
l.
6
1
,
p
p
.
2
8
3
–
2
9
2
,
F
e
b
.
2
0
1
9
,
d
o
i:
1
0
.
1
0
1
6
/j
.
re
so
u
rp
o
l.
2
0
1
9
.
0
2
.
0
1
7
.
[6
]
M
.
Oh
y
v
e
r
a
n
d
H.
P
u
d
ji
h
a
stu
ti
,
“
A
ri
m
a
M
o
d
e
l
f
o
r
F
o
re
c
a
stin
g
th
e
P
rice
o
f
M
e
d
iu
m
Qu
a
li
ty
Ric
e
to
A
n
ti
c
ip
a
te
P
rice
F
lu
c
tu
a
ti
o
n
s
,
”
Pro
c
e
d
i
a
Co
mp
u
t.
S
c
i
.
,
v
o
l.
1
3
5
,
p
p
.
7
0
7
–
7
1
1
,
2
0
1
8
,
d
o
i:
1
0
.
1
0
1
6
/j
.
p
ro
c
s.2
0
1
8
.
0
8
.
2
1
5
.
[7
]
P
.
J.
Oliv
e
ira,
J.
L
.
S
te
ff
e
n
,
a
n
d
P
.
Ch
e
u
n
g
,
“
P
a
ra
m
e
ter
Esti
m
a
ti
o
n
o
f
S
e
a
so
n
a
l
A
ri
m
a
M
o
d
e
ls
f
o
r
W
a
ter
De
m
a
n
d
F
o
re
c
a
stin
g
Us
in
g
th
e
Ha
rm
o
n
y
S
e
a
rc
h
A
l
g
o
rit
h
m
,
”
Pro
c
e
d
ia
En
g
.
,
v
o
l.
1
8
6
,
p
p
.
1
7
7
–
1
8
5
,
2
0
1
7
,
d
o
i:
1
0
.
1
0
1
6
/j
.
p
ro
e
n
g
.
2
0
1
7
.
0
3
.
2
2
5
.
[8
]
M
.
S
.
A
l
-
M
u
sa
y
lh
,
R.
C.
De
o
,
J.
F
.
A
d
a
m
o
w
sk
i,
a
n
d
Y.
L
i,
“
S
h
o
rt
-
term
e
lec
tri
c
it
y
d
e
m
a
n
d
f
o
re
c
a
sti
n
g
w
it
h
M
A
RS
,
S
V
R
a
n
d
A
RIM
A
m
o
d
e
ls
u
sin
g
a
g
g
re
g
a
ted
d
e
m
a
n
d
d
a
ta
in
Qu
e
e
n
sla
n
d
,
A
u
stra
li
a
,
”
Ad
v
.
En
g
.
I
n
fo
r
ma
ti
c
s
,
v
o
l.
3
5
,
p
p
.
1
–
1
6
,
2
0
1
8
,
d
o
i:
1
0
.
1
0
1
6
/
j.
a
e
i.
2
0
1
7
.
1
1
.
0
0
2
.
[9
]
W
.
A
n
g
g
ra
e
n
i,
R.
A
.
V
in
a
rti
,
a
n
d
Y
.
D.
Ku
r
n
iaw
a
ti
,
“
P
e
rf
o
rm
a
n
c
e
Co
m
p
a
riso
n
s
b
e
tw
e
e
n
A
ri
m
a
a
n
d
A
ri
m
a
x
M
e
th
o
d
in
M
o
sle
m
Kid
s
Clo
th
e
s
De
m
a
n
d
F
o
re
c
a
stin
g
:
Ca
se
S
tu
d
y
,
”
Pro
c
e
d
ia
Co
mp
u
t.
S
c
i
.
,
v
o
l.
7
2
,
p
p
.
6
3
0
–
6
3
7
,
2
0
1
5
,
d
o
i:
1
0
.
1
0
1
6
/
j.
p
r
o
c
s.2
0
1
5
.
1
2
.
1
7
2
.
[1
0
]
M
.
H.
Am
in
i,
A
.
Ka
r
g
a
rian
,
a
n
d
O.
Ka
ra
b
a
s
o
g
lu
,
“
A
RIM
A
-
b
a
s
e
d
d
e
c
o
u
p
led
ti
m
e
s
e
ries
f
o
re
c
a
stin
g
o
f
e
le
c
tri
c
v
e
h
icle
c
h
a
rg
in
g
d
e
m
a
n
d
f
o
r
sto
c
h
a
stic
p
o
w
e
r
s
y
ste
m
o
p
e
ra
ti
o
n
,
”
El
e
c
tr.
Po
we
r
S
y
st.
Res
.
,
v
o
l.
1
4
0
,
p
p
.
3
7
8
–
3
9
0
,
2
0
1
6
,
d
o
i:
1
0
.
1
0
1
6
/
j.
e
p
sr.
2
0
1
6
.
0
6
.
0
0
3
.
[1
1
]
Y.
W
a
n
g
,
J.
W
a
n
g
,
G
.
Zh
a
o
,
a
n
d
Y.
Do
n
g
,
“
A
p
p
li
c
a
ti
o
n
o
f
re
sid
u
a
l
m
o
d
if
ic
a
ti
o
n
a
p
p
r
o
a
c
h
in
se
a
so
n
a
l
A
RIM
A
f
o
r
e
lec
tri
c
it
y
d
e
m
a
n
d
f
o
re
c
a
stin
g
:
A
c
a
se
stu
d
y
o
f
Ch
in
a
,
”
En
e
rg
y
Po
li
c
y
,
v
o
l.
4
8
,
p
p
.
2
8
4
–
2
9
4
,
2
0
1
2
,
d
o
i:
1
0
.
1
0
1
6
/j
.
e
n
p
o
l.
2
0
1
2
.
0
5
.
0
2
6
.
[1
2
]
R.
S
h
a
rm
a
a
n
d
P
.
S
i
n
g
h
a
l,
“
De
m
a
n
d
f
o
re
c
a
stin
g
o
f
e
n
g
in
e
o
il
f
o
r
a
u
to
m
o
ti
v
e
a
n
d
in
d
u
str
ial
lu
b
r
ica
n
t
m
a
n
u
f
a
c
tu
rin
g
c
o
m
p
a
n
y
u
sin
g
n
e
u
ra
l
n
e
tw
o
rk
,
”
M
a
ter
.
T
o
d
a
y
Pro
c
.
,
v
o
l.
1
8
,
p
p
.
2
3
0
8
–
2
3
1
4
,
2
0
1
9
,
d
o
i:
1
0
.
1
0
1
6
/j
.
m
a
tp
r.
2
0
1
9
.
0
7
.
0
1
3
.
[1
3
]
K.
M
a
so
n
,
J
.
Du
g
g
a
n
,
a
n
d
E
.
Ho
w
le
y
,
“
F
o
re
c
a
stin
g
e
n
e
rg
y
d
e
m
a
n
d
,
w
in
d
g
e
n
e
ra
ti
o
n
a
n
d
c
a
rb
o
n
d
i
o
x
id
e
e
m
issio
n
s
in
Ire
lan
d
u
sin
g
e
v
o
lu
ti
o
n
a
ry
n
e
u
ra
l
n
e
tw
o
rk
s,”
E
n
e
rg
y
,
v
o
l.
1
5
5
,
p
p
.
7
0
5
–
7
2
0
,
2
0
1
8
,
d
o
i:
1
0
.
1
0
1
6
/j
.
e
n
e
rg
y
.
2
0
1
8
.
0
4
.
1
9
2
.
[1
4
]
I.
M
a
ti
n
o
,
S
.
De
tt
o
ri
,
V
.
Co
ll
a
,
V
.
W
e
b
e
r,
a
n
d
S
.
S
a
lam
e
,
“
F
o
re
c
a
stin
g
b
las
t
f
u
rn
a
c
e
g
a
s
p
ro
d
u
c
t
io
n
a
n
d
d
e
m
a
n
d
th
ro
u
g
h
e
c
h
o
sta
te
n
e
u
ra
l
n
e
tw
o
r
k
-
b
a
se
d
m
o
d
e
ls:
P
a
v
e
th
e
w
a
y
to
o
ff
-
g
a
s
o
p
ti
m
iz
e
d
m
a
n
a
g
e
m
e
n
t,
”
Ap
p
l.
En
e
rg
y
,
v
o
l.
2
5
3
,
p
.
1
1
3
5
7
8
,
Ju
n
.
2
0
1
9
,
d
o
i:
1
0
.
1
0
1
6
/
j.
a
p
e
n
e
rg
y
.
2
0
1
9
.
1
1
3
5
7
8
.
[1
5
]
E.
S
.
S
il
v
a
,
H.
Ha
s
sa
n
i,
S
.
He
ra
v
i,
a
n
d
X
.
Hu
a
n
g
,
“
F
o
re
c
a
stin
g
t
o
u
rism
d
e
m
a
n
d
w
it
h
d
e
n
o
ise
d
n
e
u
ra
l
n
e
tw
o
rk
s,”
An
n
.
T
o
u
r.
Res
.
,
v
o
l.
7
4
,
p
p
.
1
3
4
–
1
5
4
,
2
0
1
9
,
d
o
i:
1
0
.
1
0
1
6
/
j.
a
n
n
a
ls.2
0
1
8
.
1
1
.
0
0
6
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
4
,
A
u
g
u
s
t 2
0
2
1
:
3
3
1
9
-
3328
3328
[1
6
]
J.
Ke
e
t
a
l
.
,
“
He
x
a
g
o
n
-
Ba
se
d
C
o
n
v
o
l
u
ti
o
n
a
l
Ne
u
ra
l
Ne
tw
o
rk
f
o
r
S
u
p
p
ly
-
De
m
a
n
d
F
o
re
c
a
sti
n
g
o
f
Rid
e
-
S
o
u
rc
i
n
g
S
e
rv
ice
s,”
IEE
E
T
ra
n
s.
In
tell.
T
ra
n
sp
.
S
y
st.
,
v
o
l.
2
0
,
n
o
.
1
1
,
p
p
.
4
1
6
0
–
4
1
7
3
,
2
0
1
9
,
d
o
i:
1
0
.
1
1
0
9
/T
IT
S
.
2
0
1
8
.
2
8
8
2
8
6
1
.
[1
7
]
J.
Ke
,
H.
Zh
e
n
g
,
H.
Ya
n
g
,
a
n
d
X.
(M
ich
a
e
l)
Ch
e
n
,
“
S
h
o
rt
-
term
f
o
re
c
a
stin
g
o
f
p
a
ss
e
n
g
e
r
d
e
m
a
n
d
u
n
d
e
r
o
n
-
d
e
m
a
n
d
rid
e
se
rv
i
c
e
s:
A
sp
a
ti
o
-
tem
p
o
ra
l
d
e
e
p
lea
rn
in
g
a
p
p
ro
a
c
h
,
”
T
ra
n
sp
.
Res
.
Pa
rt
C
Eme
rg
.
T
e
c
h
n
o
l
.
,
v
o
l.
8
5
,
p
p
.
5
9
1
–
6
0
8
,
J
u
n
.
2
0
1
7
,
d
o
i
:
1
0
.
1
0
1
6
/j
.
trc.
2
0
1
7
.
1
0
.
0
1
6
.
[1
8
]
K.
Am
a
ra
sin
g
h
e
,
D.
L
.
M
a
rin
o
,
a
n
d
M
.
M
a
n
ic,
“
De
e
p
n
e
u
ra
l
n
e
tw
o
rk
s
f
o
r
e
n
e
rg
y
lo
a
d
f
o
re
c
a
sti
n
g
,
”
IEE
E
In
t.
S
y
mp
.
I
n
d
.
El
e
c
tro
n
.
,
p
p
.
1
4
8
3
–
1
4
8
8
,
2
0
1
7
,
d
o
i:
1
0
.
1
1
0
9
/I
S
IE.
2
0
1
7
.
8
0
0
1
4
6
5
.
[1
9
]
I.
S
li
m
a
n
i,
I.
El
F
a
rissi,
a
n
d
S
.
A
c
h
c
h
a
b
,
“
Co
n
f
ig
u
ra
ti
o
n
a
n
d
imp
lem
e
n
tatio
n
o
f
a
d
a
il
y
a
rti
f
icia
l
n
e
u
ra
l
n
e
tw
o
rk
-
b
a
se
d
f
o
re
c
a
stin
g
s
y
ste
m
u
sin
g
r
e
a
l
su
p
e
r
m
a
rk
e
t
d
a
ta,”
In
t.
J
.
L
o
g
ist.
S
y
st.
M
a
n
a
g
.
,
v
o
l.
2
8
,
n
o
.
2
,
p
p
.
1
4
4
–
1
6
3
,
2
0
1
7
,
d
o
i:
1
0
.
1
5
0
4
/IJL
S
M
.
2
0
1
7
.
0
8
6
3
4
5
.
[2
0
]
A
.
S
a
id
,
I.
S
li
m
a
n
i,
a
n
d
I.
El
F
a
rissi,
“
A
rti
f
icia
l
Ne
u
ra
l
Ne
t
w
o
rk
s/o
r
De
m
a
n
d
F
o
re
c
a
stin
g
:
A
p
p
li
c
a
ti
o
n
Us
in
g
M
o
ro
c
c
a
n
S
u
p
e
rm
a
rk
e
t
Da
ta,”
IE
EE
In
t
.
Co
n
f.
C
o
n
s
u
m.
El
e
c
tro
n
(
ICCE)
,
Be
rli
n
,
2
0
1
5
,
p
p
.
2
6
6
–
2
7
1
.
[2
1
]
V
.
A
.
P
ro
f
il
li
d
is an
d
G
.
N.
Bo
tzo
ris,
“
T
re
n
d
P
ro
jec
ti
o
n
a
n
d
T
im
e
S
e
ries
M
e
th
o
d
s
,”
M
o
d
e
li
n
g
o
f
T
ra
n
sp
o
rt
De
m
a
n
d
,
p
p
.
2
2
5
–
2
7
0
,
2
0
1
9
.
[2
2
]
V
.
K
o
tu
a
n
d
B.
De
sh
p
a
n
d
e
,
“
Da
ta S
c
ien
c
e
Co
n
c
e
p
ts a
n
d
P
ra
c
ti
c
e
,
”
M
o
rg
a
n
Ka
u
fma
n
n
,
v
o
l.
2
,
n
o
.
C.
2
0
1
9
.
[2
3
]
T
.
C.
M
il
ls,
“
A
RIM
A
M
o
d
e
ls
f
o
r
No
n
sta
ti
o
n
a
ry
T
i
m
e
S
e
rie
s,”
Ap
p
l.
T
ime
S
e
r.
An
a
l.
,
p
p
.
5
7
–
6
9
,
2
0
1
9
,
d
o
i:
1
0
.
1
0
1
6
/b
9
7
8
-
0
-
12
-
8
1
3
1
1
7
-
6
.
0
0
0
0
4
-
1.
[2
4
]
C.
C.
A
g
g
a
r
w
a
l,
“
Ne
u
ra
l
Ne
t
w
o
rk
s a
n
d
De
e
p
L
e
a
rn
in
g
,
”
S
p
rin
g
e
r
,
2
0
1
8
.
[2
5
]
S
.
Ho
c
h
re
it
e
r
a
n
d
J.
Urg
e
n
S
c
h
m
id
h
u
b
e
r,
“
L
o
n
g
S
h
o
rt
-
T
e
rm
M
e
m
o
r
y
,
”
N
e
u
ra
l
Co
mp
u
t.
,
v
o
l.
9
,
n
o
.
8
,
p
p
.
1
7
3
5
–
1
7
8
0
,
1
9
9
7
,
d
o
i:
1
0
.
1
1
6
2
/n
e
c
o
.
1
9
9
7
.
9
.
8
.
1
7
3
5
.
[2
6
]
H.
Bo
u
s
q
a
o
u
i,
S
.
A
c
h
c
h
a
b
,
a
n
d
K.
T
ik
it
o
,
“
M
a
c
h
i
n
e
lea
rn
i
n
g
a
p
p
li
c
a
ti
o
n
s
i
n
su
p
p
ly
c
h
a
in
s:
L
o
n
g
sh
o
rt
-
ter
m
m
e
m
o
r
y
f
o
r
d
e
m
a
n
d
f
o
re
c
a
stin
g
,
”
in
L
e
c
t
u
re
No
tes
in
Ne
two
rk
s a
n
d
S
y
ste
ms
,
v
o
l.
4
9
,
p
p
.
3
0
1
–
3
1
7
,
2
0
1
9
.
[2
7
]
R.
T
h
a
n
k
i
a
n
d
S
.
Bo
rra
,
“
A
p
p
li
c
a
ti
o
n
o
f
M
a
c
h
in
e
L
e
a
rn
in
g
A
l
g
o
rit
h
m
s
f
o
r
Clas
sif
ic
a
ti
o
n
a
n
d
S
e
c
u
r
it
y
o
f
Dia
g
n
o
stic Im
a
g
e
s
,”
El
se
v
ier In
c
.
,
2
0
1
9
.
[2
8
]
H.
Bo
u
sq
a
o
u
i,
I.
S
li
m
a
n
i,
a
n
d
S
.
A
c
h
c
h
a
b
,
“
I
m
p
ro
v
in
g
Co
o
rd
i
n
a
ti
o
n
in
S
u
p
p
ly
Ch
a
in
Us
in
g
A
rti
f
icia
l
Ne
u
ra
l
Ne
tw
o
rk
s
a
n
d
M
u
l
ti
-
a
g
e
n
t
A
p
p
ro
a
c
h
,
”
in
En
g
Op
t
2
0
1
8
Pro
c
e
e
d
in
g
s
o
f
t
h
e
6
th
I
n
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
En
g
i
n
e
e
rin
g
O
p
ti
miza
ti
o
n
,
S
p
ri
n
g
e
r
In
tern
a
ti
o
n
a
l
P
u
b
l
ish
i
n
g
,
2
0
1
9
,
p
p
.
1
3
5
3
–
1
3
5
9
.
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
H
a
li
m
a
B
o
u
s
q
a
o
u
i
o
b
tain
e
d
h
e
r
d
e
g
re
e
f
ro
m
Na
ti
o
n
a
l
Hig
h
e
r
S
c
h
o
o
l
f
o
r
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
S
y
st
e
m
A
n
a
l
y
sis
(ENS
I
A
S
).
A
n
d
is
c
u
rre
n
tl
y
p
u
rsu
i
n
g
a
P
h
D
d
e
g
r
e
e
th
e
re
.
He
r
w
o
rk
f
o
c
u
se
s
o
n
m
a
c
h
in
e
lea
rn
in
g
a
lg
o
rit
h
m
s
su
c
h
a
s
f
e
e
d
f
o
r
w
a
rd
a
n
d
re
c
u
rre
n
t
n
e
u
ra
l
n
e
tw
o
rk
s
in
th
e
su
p
p
ly
c
h
a
in
.
Ilh
a
m
S
li
m
a
n
i
In
g
.
&
P
h
D
f
ro
m
th
e
Na
ti
o
n
a
l
S
c
h
o
o
l
o
f
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
S
y
s
te
m
A
n
a
l
y
sis
(ENS
IA
S
).
S
h
e
is
c
u
rre
n
tl
y
P
ro
f
e
ss
o
r
a
t
th
e
De
p
a
rt
m
e
n
t
o
f
Co
m
p
u
ter
S
c
ien
c
e
a
t
M
o
h
a
m
m
e
d
F
irst
Un
iv
e
rsity
,
F
a
c
u
lt
y
o
f
S
c
i
e
n
c
e
Ou
jd
a
.
He
r
re
se
a
rc
h
a
n
d
p
u
b
li
c
a
ti
o
n
i
n
tere
sts
in
c
lu
d
e
m
a
c
h
in
e
lea
rn
in
g
,
su
p
p
ly
c
h
a
in
m
a
n
a
g
e
m
e
n
t,
d
e
m
a
n
d
f
o
re
c
a
stin
g
,
ro
a
d
traf
f
i
c
f
o
re
c
a
stin
g
,
in
tru
si
o
n
d
e
tec
ti
o
n
o
f
c
o
m
p
u
ter n
e
tw
o
rk
s.
S
a
id
Ac
h
c
h
a
b
,
h
o
ld
e
r
o
f
a
P
h
D
in
A
p
p
li
e
d
M
a
th
e
m
a
ti
c
s
f
ro
m
th
e
M
o
h
a
m
m
e
d
ia
S
c
h
o
o
l
o
f
En
g
in
e
e
rin
g
in
Ra
b
a
t
a
s
w
e
ll
a
s
a
u
n
iv
e
rsity
h
a
b
il
it
a
ti
o
n
i
n
Bu
si
n
e
ss
In
telli
g
e
n
c
e
f
ro
m
th
e
sa
m
e
in
stit
u
ti
o
n
,
h
e
a
lso
a
tt
e
n
d
e
d
t
h
e
h
ig
h
e
r
c
y
c
le
o
f
m
a
n
a
g
e
m
e
n
t
a
t
th
e
ENCG
o
f
S
e
tt
a
t
.
He
is
c
u
rre
n
tl
y
P
ro
f
e
ss
o
r
o
f
Qu
a
n
ti
tat
iv
e
F
in
a
n
c
e
,
A
rti
f
icia
l
In
telli
g
e
n
c
e
a
n
d
Risk
M
a
n
a
g
e
m
e
n
t
a
t
ENS
IA
S
,
Dire
c
to
r
o
f
th
e
Co
n
ti
n
u
in
g
Ed
u
c
a
ti
o
n
Ce
n
ter
o
f
th
e
sa
m
e
sc
h
o
o
l
b
e
tw
e
e
n
Ja
n
u
a
ry
2
0
1
3
a
n
d
M
a
y
2
0
1
7
a
n
d
He
a
d
o
f
th
e
De
p
a
rtme
n
t
"
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
De
c
isio
n
S
u
p
p
o
r
t"
sin
c
e
Ja
n
u
a
ry
2
0
1
8
.
He
is
c
o
o
rd
i
n
a
to
r
o
f
th
e
S
p
e
c
ialize
d
M
a
ste
r
"
En
g
in
e
e
rin
g
f
o
r
S
u
sta
in
a
b
le
F
in
a
n
c
e
a
n
d
Risk
M
a
n
a
g
e
m
e
n
t"
a
n
d
o
f
th
e
e
n
g
in
e
e
rin
g
p
ro
g
ra
m
"
Di
g
it
a
l
E
n
g
in
e
e
rin
g
f
o
r
F
in
a
n
c
e
"
.
H
e
is
th
e
f
o
u
n
d
in
g
P
re
si
d
e
n
t
o
f
th
e
A
f
r
ica
n
In
stit
u
t
e
o
f
F
i
n
tec
h
s.
F
o
r
m
o
re
th
a
n
1
5
y
e
a
rs,
h
e
h
a
s
b
e
e
n
c
a
rr
y
in
g
o
u
t
c
o
n
s
u
lt
i
n
g
m
issio
n
s fo
r
p
u
b
l
ic an
d
p
riv
a
te o
rg
a
n
iza
ti
o
n
s.
Evaluation Warning : The document was created with Spire.PDF for Python.