I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
pu
t
er
Science
Vo
l.
3
8
,
No
.
2
,
Ma
y
20
2
5
,
p
p
.
794
~
8
0
7
I
SS
N:
2
502
-
4
7
52
,
DOI
: 1
0
.
1
1
5
9
1
/ijee
cs
.v
3
8
.
i
2
.
pp
794
-
8
0
7
794
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs
.
ia
esco
r
e.
co
m
Spa
tial
-
tempo
ra
l
da
ta imputa
tion
f
o
r predic
tive mo
d
eling
in
intelligent
t
ra
nsp
o
rtation
sy
stems
Yo
ha
nes
P
ra
co
y
o
Widi
P
ra
s
et
y
o
1
,
L
ina
wa
t
i
2
,
Dew
a
M
a
de
Wiha
rt
a
2
,
Ny
o
m
a
n
P
utr
a
S
a
s
t
ra
2
1
F
a
c
u
l
t
y
o
f
En
g
i
n
e
e
r
i
n
g
,
U
n
i
v
e
r
s
i
t
a
s
1
7
A
g
u
s
t
u
s
1
9
4
5
,
B
a
n
y
u
w
a
n
g
i
,
I
n
d
o
n
e
si
a
2
F
a
c
u
l
t
y
o
f
En
g
i
n
e
e
r
i
n
g
,
U
d
a
y
a
n
a
U
n
i
v
e
r
si
t
y
,
B
a
l
i
,
I
n
d
o
n
e
si
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
May
21
,
2
0
2
4
R
ev
is
ed
Oct
24
,
2
0
2
4
Acc
ep
ted
No
v
10
,
2
0
2
4
Da
ta
imp
u
tatio
n
is
n
e
c
e
ss
a
ry
to
o
v
e
rc
o
m
e
d
a
ta
lo
ss
in
in
telli
g
e
n
t
tran
sp
o
rtati
o
n
sy
ste
m
s
(IT
S
)
d
u
e
to
t
h
e
m
a
n
y
se
n
so
rs
u
se
d
t
o
m
o
n
i
to
r
traffic
c
o
n
d
i
ti
o
n
s.
S
e
n
so
r
m
a
lfu
n
c
ti
o
n
,
h
a
rd
wa
re
li
m
it
a
ti
o
n
s,
a
n
d
tec
h
n
ic
a
l
g
li
tch
e
s
c
a
n
lea
d
t
o
in
c
o
m
p
lete
d
a
ta,
p
o
ten
ti
a
ll
y
lea
d
in
g
t
o
e
rro
rs
i
n
t
ra
ffic
d
a
ta
a
n
a
ly
sis.
T
h
is
a
n
a
l
y
sis
i
n
v
e
st
ig
a
ted
sp
a
t
ial
-
tem
p
o
ra
l
d
a
ta
i
m
p
u
tatio
n
a
p
p
ro
a
c
h
e
s
a
p
p
li
e
d
f
o
r
p
re
d
ict
iv
e
m
o
d
e
li
n
g
in
IT
S
.
Eac
h
a
p
p
r
o
a
c
h
'
s
stre
n
g
th
s,
we
a
k
n
e
ss
e
s,
a
n
d
a
p
p
li
c
a
b
il
it
y
i
n
th
e
c
o
n
tex
t
o
f
I
TS
a
re
e
v
a
lu
a
ted
.
We
a
n
a
ly
z
e
d
v
a
ri
o
u
s
imp
u
tatio
n
a
p
p
r
o
a
c
h
e
s
i
n
v
o
lv
in
g
sta
ti
stica
l,
m
a
c
h
in
e
lea
rn
in
g
,
a
n
d
c
o
m
b
i
n
e
d
m
e
th
o
d
s.
S
tatist
ica
l
m
e
th
o
d
s
a
re
m
o
re
stra
ig
h
tf
o
rwa
rd
b
u
t
c
o
u
ld
e
ffe
c
ti
v
e
ly
h
a
n
d
le
m
o
d
e
rn
traffic'
s
c
o
m
p
l
e
x
it
y
.
O
n
th
e
o
th
e
r
h
a
n
d
,
m
a
c
h
in
e
lea
rn
in
g
a
n
d
c
o
m
b
i
n
e
d
a
p
p
ro
a
c
h
e
s,
su
c
h
a
s
h
y
b
ri
d
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
(
CNN
)
-
lo
n
g
s
h
o
rt
-
term
m
e
m
o
ry
(
LS
TM
)
,
o
ffe
r
m
o
re
ro
b
u
st
c
a
p
a
b
i
li
ti
e
s
i
n
c
a
p
tu
ri
n
g
n
o
n
-
li
n
e
a
r
p
a
tt
e
rn
s
p
re
se
n
t
in
sp
a
ti
o
-
tem
p
o
ra
l
d
a
ta.
Th
is
re
se
a
rc
h
a
ims
to
in
v
e
stig
a
te
t
h
e
e
ffe
c
ti
v
e
n
e
ss
o
f
e
a
c
h
a
p
p
r
o
a
c
h
i
n
o
v
e
rc
o
m
in
g
d
a
ta
in
c
o
m
p
lete
n
e
ss
a
n
d
th
e
a
c
c
u
ra
c
y
o
f
p
re
d
ictin
g
f
u
t
u
re
traffic
c
o
n
d
it
io
n
s
wit
h
t
h
e
wid
e
s
p
re
a
d
a
d
o
p
t
io
n
o
f
I
o
T
,
e
lec
tri
c
v
e
h
icle
s,
a
n
d
a
u
t
o
n
o
m
o
u
s
v
e
h
icle
s.
T
h
e
re
su
lt
s
o
f
th
is
in
v
e
stig
a
ti
o
n
p
ro
v
id
e
a
n
u
n
d
e
rsta
n
d
in
g
o
f
t
h
e
m
o
st
su
i
tab
le
a
p
p
ro
a
c
h
e
s
t
o
a
d
d
re
ss
th
e
c
h
a
ll
e
n
g
e
s o
f
s
p
a
ti
o
-
tem
p
o
ra
l
d
a
t
a
imp
u
tatio
n
a
n
d
p
ro
v
id
e
p
ra
c
ti
c
a
l
g
u
id
a
n
c
e
fo
r
p
re
d
ictiv
e
m
o
d
e
li
n
g
i
n
ITS
.
K
ey
w
o
r
d
s
:
Data
i
m
p
u
tatio
n
I
n
tellig
en
t
t
r
an
s
p
o
r
tatio
n
s
y
s
tem
Miss
in
g
d
ata
Pre
d
ictiv
e
m
o
d
elin
g
Sp
atial
-
t
em
p
o
r
al
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
:
Y
o
h
an
es Pr
ac
o
y
o
W
id
i Pr
aset
y
o
Facu
lty
o
f
E
n
g
in
ee
r
in
g
,
Un
iv
e
r
s
itas
1
7
Ag
u
s
tu
s
1
9
4
5
B
an
y
u
wan
g
i
,
I
n
d
o
n
esia
E
m
ail:
wid
ip
r
asety
o
@
u
n
tag
-
b
an
y
u
wan
g
i.
ac
.
id
1.
I
NT
RO
D
UCT
I
O
N
L
o
s
t
d
ata
is
a
co
m
m
o
n
p
r
o
b
le
m
in
I
T
S
d
u
e
to
th
e
in
c
r
ea
s
in
g
n
u
m
b
er
o
f
v
eh
icles.
I
m
p
r
o
v
em
en
ts
to
th
e
I
T
S
f
r
a
m
ewo
r
k
ar
e
n
ee
d
e
d
as
s
en
s
o
r
f
ailu
r
es
an
d
h
ar
d
war
e
lim
itatio
n
s
r
esu
lt
in
in
c
o
m
p
lete
tr
af
f
ic
d
ata.
T
h
e
f
u
tu
r
e
d
ev
elo
p
m
en
t
o
f
au
to
n
o
m
o
u
s
v
eh
icles,
p
o
o
r
tr
af
f
i
c
m
an
ag
e
m
en
t
in
d
e
v
elo
p
in
g
co
u
n
tr
ies
r
esu
ltin
g
in
co
n
g
esti
o
n
,
an
d
th
e
d
r
i
v
e
t
o
war
d
s
s
m
ar
t,
g
r
ee
n
,
an
d
s
u
s
tain
ab
le
cities
m
ak
e
p
r
e
d
ictiv
e
m
o
d
elin
g
in
I
T
S
v
er
y
im
p
o
r
tan
t.
His
to
r
ically
m
is
s
in
g
d
ata
was
ad
d
r
ess
ed
b
y
h
is
to
r
ical
an
d
s
tatis
tical
m
e
th
o
d
s
[
1
]
.
H
o
wev
er
,
th
ese
m
eth
o
d
s
ar
e
less
ef
f
ec
tiv
e
d
u
e
to
t
h
e
lar
g
e
d
ata
s
ize
an
d
u
n
n
atu
r
al
p
atter
n
s
ca
u
s
e
d
b
y
d
elete
d
d
ata.
T
h
er
ef
o
r
e,
r
esear
ch
er
s
b
e
g
an
to
in
v
esti
g
ate
b
etter
im
p
u
tatio
n
m
eth
o
d
s
f
o
r
p
r
ed
ictiv
e
m
o
d
elin
g
in
I
T
S.
Fr
o
m
p
r
ev
io
u
s
r
esear
ch
,
th
e
d
ev
elo
p
m
en
t o
f
d
ata
im
p
u
tatio
n
alg
o
r
i
th
m
s
u
s
ed
in
ter
p
o
latio
n
,
e
x
tr
a
p
o
latio
n
,
o
r
m
o
d
el
-
b
ased
p
r
ed
ictio
n
tech
n
iq
u
es.
B
y
ad
d
r
ess
in
g
th
ese
is
s
u
es
,
th
e
q
u
ality
o
f
tr
af
f
ic
d
ata
an
d
th
e
s
y
s
tem
will
wo
r
k
m
o
r
e
ef
f
icien
tly
[
2
]
.
T
h
e
in
ter
p
o
latio
n
m
eth
o
d
ca
n
b
e
u
s
ed
ef
f
ec
ti
v
ely
[
3
]
.
Ho
we
v
er
,
th
ey
ca
n
n
o
t
ca
p
tu
r
e
c
o
m
p
lex
s
p
atial
o
r
tem
p
o
r
al
r
elatio
n
s
h
ip
s
in
th
e
d
ata,
h
en
ce
th
e
n
ee
d
to
c
o
m
b
in
e
th
em
with
o
th
er
m
et
h
o
d
s
.
Pro
p
o
s
in
g
an
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
S
p
a
tia
l
-
temp
o
r
a
l
d
a
ta
imp
u
ta
t
io
n
fo
r
p
r
ed
ictive
mo
d
elin
g
…
(
Yo
h
a
n
es P
r
a
co
y
o
Wid
i P
r
a
s
etyo
)
795
im
p
u
tatio
n
m
o
d
el
by
c
o
m
b
in
i
n
g
a
s
elf
-
atten
tio
n
m
ec
h
an
is
m
,
au
to
m
atic
en
co
d
er
,
a
n
d
g
e
n
er
ativ
e
ad
v
e
r
s
ar
ial
n
etwo
r
k
in
t
o
a
s
elf
-
atten
tio
n
g
en
er
ativ
e
a
d
v
er
s
ar
ial
im
p
u
tatio
n
n
et
(
SA
-
GAI
N)
[
4
]
.
Ho
w
ev
er
,
th
e
tr
ain
in
g
o
f
GAN
is
n
o
to
r
io
u
s
ly
co
m
p
licated
o
n
b
ala
n
ce
b
etwe
e
n
g
e
n
er
a
to
r
an
d
d
is
cr
im
in
at
o
r
,
t
h
is
tech
n
iq
u
e
ca
n
ca
p
tu
r
e
s
p
atio
-
tem
p
o
r
al
d
ep
e
n
d
en
cies
an
d
co
r
r
elatio
n
s
in
th
e
d
ata
,
wh
ich
is
v
er
y
u
s
ef
u
l
in
I
T
S.
Me
an
wh
ile,
[
5
]
p
r
o
p
o
s
ed
th
e
s
p
atio
-
tem
p
o
r
a
l
g
en
er
ativ
e
ad
v
e
r
s
ar
ial
n
etwo
r
k
(
STGA
N)
m
eth
o
d
wit
h
th
e
co
n
ce
p
t
o
f
m
in
im
izin
g
t
h
e
r
ec
o
n
s
tr
u
ctio
n
er
r
o
r
f
r
o
m
m
is
s
in
g
d
ata
en
tr
i
es
an
d
en
s
u
r
in
g
th
at
th
e
d
ata
en
tr
ies
f
it
th
e
lo
ca
l
s
p
atio
-
tem
p
o
r
al
d
is
tr
ib
u
tio
n
,
b
u
t
STGA
N
r
eq
u
i
r
es
lar
g
e
a
n
d
h
i
g
h
q
u
ality
d
ata
b
ec
a
u
s
e
p
o
o
r
d
ata
lead
s
to
in
ac
cu
r
ate
o
u
tp
u
t.
Hen
ce
,
it
n
ee
d
s
atten
tio
n
in
tr
ain
in
g
a
n
d
in
te
r
p
r
etatio
n
.
T
o
r
ec
o
n
s
tr
u
ct
s
p
atio
-
tem
p
o
r
al
s
tates
b
ased
o
n
GANs
,
[
6
]
p
r
o
p
o
s
ed
th
e
tr
af
f
ic
s
tate
r
ec
o
n
s
tr
u
ctio
n
GAN
(
T
SR
-
GAN)
m
o
d
el.
Ho
we
v
er
,
lik
e
o
th
er
GAN
m
o
d
els,
it
f
ac
es
th
e
p
r
o
b
lem
o
f
in
s
tab
ilit
y
d
u
r
in
g
tr
ain
in
g
,
s
o
it
n
ee
d
s
ca
r
ef
u
l
cu
s
to
m
izatio
n
.
Sp
ec
if
ically
,
th
e
tr
af
f
ic
lan
e
s
tates
ar
e
co
n
v
er
ted
i
n
to
tr
af
f
ic
s
tate
d
iag
r
am
s
(
T
SDs
)
,
wh
o
s
e
co
lo
r
s
r
ep
r
esen
t
th
e
v
alu
es
o
f
tr
af
f
ic
v
ar
ia
b
le
s
(
e.
g
.
,
s
p
ee
d
o
r
d
e
n
s
ity
)
.
T
o
r
ed
u
ce
in
s
tab
ilit
y
b
u
t
with
b
etter
ac
cu
r
ac
y
,
[
7
]
p
r
o
p
o
s
ed
s
p
atio
-
tem
p
o
r
al
lear
n
ab
le
b
id
ir
ec
tio
n
al
atten
tio
n
g
en
er
ativ
e
a
d
v
er
s
ar
ial
n
etwo
r
k
s
(
ST
-
L
B
AGAN
)
th
at
u
s
e
a
d
ee
p
lear
n
in
g
f
r
am
ewo
r
k
.
Alth
o
u
g
h
it
h
as
ex
ce
llen
t
p
o
ten
tial,
th
is
m
eth
o
d
r
is
k
s
o
v
er
f
itti
n
g
,
s
o
it
n
ee
d
s
a
p
r
o
p
e
r
ap
p
r
o
ac
h
.
So
m
etim
es,
d
ata
is
lo
s
t
o
n
a
la
r
g
e
s
ca
le
b
ec
au
s
e
s
o
m
e
r
o
ad
s
n
ee
d
s
en
s
o
r
s
.
T
o
o
v
er
co
m
e
th
is
p
r
o
b
lem
,
W
an
g
et
a
l.
[
8
]
p
r
o
p
o
s
ed
an
in
teg
r
ated
d
ee
p
lear
n
in
g
f
o
r
tr
af
f
ic
s
tate
r
ec
o
n
s
tr
u
ctio
n
(
I
DL
-
T
SR
)
f
r
am
ew
o
r
k
,
u
s
in
g
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
etwo
r
k
s
(
C
NNs)
to
ca
p
t
u
r
e
s
p
atial
f
ea
tu
r
es
an
d
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
L
STM
)
t
o
ca
p
tu
r
e
tem
p
o
r
al
f
ea
tu
r
es
t
o
r
ec
o
n
s
tr
u
ct
t
r
af
f
ic
s
tate
u
s
in
g
s
en
s
o
r
d
ata
f
r
o
m
lim
ited
lin
k
s
.
R
ea
l
-
tim
e
tr
af
f
ic
p
r
ed
ictio
n
m
o
d
elin
g
[
9
]
p
r
o
p
o
s
ed
t
h
e
d
y
n
am
ic
tem
p
o
r
al
ad
ja
ce
n
t
g
r
ap
h
co
n
v
o
l
u
tio
n
al
n
etwo
r
k
(
D
-
T
AGCN),
a
d
ee
p
lear
n
in
g
m
o
d
el
d
esig
n
ed
to
an
aly
ze
s
p
atio
-
te
m
p
o
r
al
d
ata
with
a
d
y
n
am
ic
g
r
ap
h
s
tr
u
ctu
r
e.
Alt
h
o
u
g
h
D
-
T
AGCN
is
r
o
b
u
s
t
i
n
ca
p
tu
r
in
g
d
y
n
am
ic
s
p
atio
-
te
m
p
o
r
al
p
atter
n
s
,
its
ap
p
licatio
n
r
eq
u
ir
es
s
p
ec
ial
atten
tio
n
to
m
o
d
el
d
esig
n
,
d
ata,
an
d
p
ar
am
ete
r
tu
n
in
g
.
P
r
o
p
o
s
ed
a
s
p
atio
tem
p
o
r
al
g
en
er
ativ
e
a
d
v
er
s
ar
ial
im
p
u
tat
io
n
n
et
(
ST
-
GAI
N)
m
o
d
el
t
h
at
r
elies
h
ea
v
ily
o
n
th
e
q
u
ality
o
f
av
ailab
le
d
ata,
as
d
ata
co
n
tain
in
g
m
an
y
o
u
tlier
s
ca
n
lead
to
p
o
o
r
im
p
u
tati
o
n
r
esu
lts
[
1
0
]
.
Sp
atio
-
tem
p
o
r
al
atten
tio
n
-
g
ated
r
ec
u
r
r
en
t
n
eu
r
al
n
etwo
r
k
(
ST
-
AGRNN)
is
a
d
ee
p
lear
n
in
g
m
o
d
el
d
esig
n
ed
to
p
r
o
ce
s
s
a
n
d
an
aly
ze
s
p
atio
-
tem
p
o
r
al
d
ata,
i.e
.
,
s
p
atial
an
d
tem
p
o
r
al
d
im
en
s
io
n
s
.
W
h
ile
th
ese
m
o
d
els
h
av
e
th
e
p
o
wer
to
h
an
d
le
co
m
p
lex
s
p
a
t
i
o
-
t
e
m
p
o
r
a
l
d
a
t
a
,
t
h
e
i
r
a
p
p
l
i
c
a
t
i
o
n
r
e
q
u
i
r
e
s
s
p
e
c
i
a
l
a
t
t
e
n
t
i
o
n
t
o
d
a
t
a
,
h
y
p
e
r
p
a
r
a
m
e
t
e
r
s
,
a
n
d
m
o
d
e
l
d
e
s
i
g
n
[
1
1
]
.
T
h
e
GANs f
am
ily
ca
n
f
ac
e
th
e
v
an
is
h
in
g
g
r
ad
ie
n
t p
r
o
b
lem
,
wh
er
e
th
e
d
is
cr
im
in
ato
r
b
ec
o
m
es to
o
g
o
o
d
,
s
o
th
e
g
en
er
ato
r
ca
n
n
o
t
lear
n
e
f
f
ec
ti
v
ely
.
T
h
e
ch
o
ice
o
f
m
eth
o
d
f
o
r
p
r
ed
ictiv
e
m
o
d
elin
g
i
n
I
T
S
d
ep
en
d
s
o
n
th
e
ch
ar
ac
ter
is
tics
o
f
th
e
d
ata
an
d
th
e
s
p
ec
if
ic
n
ee
d
s
[
1
2
]
.
Fo
r
s
p
atial
-
tem
p
o
r
al
d
ata,
h
y
b
r
i
d
m
eth
o
d
s
s
u
ch
as
C
NN
-
L
STM
ar
e
v
er
y
ef
f
ec
tiv
e
as
th
ey
co
m
b
in
e
th
e
s
tr
en
g
t
h
s
o
f
C
NN
an
d
L
STM
to
h
an
d
le
s
p
atial
-
tem
p
o
r
al
d
ata
in
p
r
e
d
ictiv
e
m
o
d
elin
g
.
E
n
s
em
b
le
an
d
d
ee
p
lear
n
in
g
c
an
p
r
o
v
id
e
a
co
m
p
etitiv
e
ad
v
an
tag
e
in
p
r
ed
ictin
g
co
m
p
lex
a
n
d
d
y
n
am
ic
c
o
n
d
iti
o
n
s
in
in
tellig
en
t tr
an
s
p
o
r
tatio
n
s
y
s
tem
s
[
1
3
]
.
T
h
is
ar
ticle
is
d
iv
i
d
ed
in
to
f
i
v
e
s
ec
tio
n
s
.
T
h
e
f
ir
s
t
s
ec
tio
n
in
tr
o
d
u
ce
s
t
h
e
s
tu
d
y
o
f
s
p
atial
-
tem
p
o
r
al
d
ata
im
p
u
tatio
n
f
o
r
p
r
ed
ictiv
e
m
o
d
elin
g
i
n
I
T
S.
T
h
e
n
e
x
t
s
ec
tio
n
p
r
esen
ts
m
eth
o
d
s
th
at
ar
e
o
f
ten
u
s
ed
in
d
ata
im
p
u
tatio
n
.
Sectio
n
3
d
is
cu
s
s
es e
ac
h
m
eth
o
d
'
s
s
tr
en
g
th
s
,
wea
k
n
ess
es,
ap
p
licab
ilit
y
,
an
d
ef
f
ec
tiv
en
ess
.
Sectio
n
4
p
r
esen
ts
th
e
r
esu
lts
o
f
th
e
in
v
esti
g
atio
n
an
d
d
is
cu
s
s
io
n
o
f
th
ese
m
eth
o
d
s
.
Fin
ally
,
th
e
p
ap
er
co
n
clu
d
es
with
co
n
clu
s
io
n
s
ab
o
u
t th
e
b
est m
et
h
o
d
f
o
r
ea
c
h
an
d
f
u
tu
r
e
r
esear
ch
d
ir
ec
tio
n
s
.
2.
T
H
E
CO
M
P
RE
H
E
NS
I
VE
T
H
E
O
RE
T
I
CA
L
B
ASI
S
Stu
d
ies
u
s
in
g
h
is
to
r
ical
tr
af
f
ic
d
ata
in
co
r
p
o
r
atin
g
3
D
co
n
v
o
l
u
tio
n
al
g
en
er
ativ
e
n
etwo
r
k
s
an
d
GANs
to
ac
co
u
n
t
f
o
r
m
is
s
in
g
tr
af
f
ic
d
ata
ar
e
[
1
4
]
.
I
n
co
n
tr
ast
to
m
o
s
t
s
tu
d
ies
th
at
ac
co
u
n
t
f
o
r
m
is
s
in
g
d
ata
at
th
e
g
r
an
u
lar
ity
o
f
r
o
ad
s
eg
m
en
t
s
an
d
c
o
m
b
i
n
ed
tim
e
in
ter
v
als
,
th
e
im
p
u
tatio
n
ap
p
r
o
ac
h
b
ased
o
n
g
ate
d
atten
tio
n
al
g
en
e
r
ativ
e
ad
v
er
s
a
r
ial
n
etwo
r
k
s
(
GaG
ANs)
is
h
ig
h
ly
r
esp
o
n
s
iv
e
to
d
y
n
am
ic
tr
af
f
ic
en
v
i
r
o
n
m
e
n
ts
on
s
ig
n
alize
d
r
o
ad
n
etwo
r
k
s
.
Ho
wev
er
,
it
r
e
q
u
ir
es
h
ig
h
co
m
p
u
tatio
n
al
r
eso
u
r
ce
s
an
d
lo
n
g
er
tr
ain
i
n
g
tim
es
[
1
5
]
.
Alth
o
u
g
h
m
an
y
s
p
atio
te
m
p
o
r
al
a
p
p
r
o
ac
h
es
h
a
v
e
b
ee
n
p
r
esen
ted
to
o
v
er
c
o
m
e
th
e
p
r
o
b
lem
o
f
m
is
s
in
g
s
p
atio
tem
p
o
r
al
d
ata,
W
an
g
et
a
l.
[
1
6
]
s
tated
th
at
th
er
e
ar
e
lim
itatio
n
s
to
ca
p
tu
r
in
g
s
p
atio
tem
p
o
r
al
d
ep
en
d
e
n
cies
in
s
p
atio
tem
p
o
r
al
g
r
ap
h
s
,
as
m
o
s
t
im
p
u
tatio
n
m
eth
o
d
s
d
o
n
o
t
co
n
s
id
er
t
h
e
d
y
n
am
ic
d
ata
h
id
d
en
in
g
r
a
p
h
n
o
d
es,
s
o
a
n
atten
tio
n
b
ased
m
ess
ag
e
p
ass
in
g
an
d
d
y
n
am
ic
g
r
a
p
h
c
o
n
v
o
lu
tio
n
n
etwo
r
k
is
p
r
o
p
o
s
ed
by
co
n
s
id
er
i
n
g
th
e
tr
af
f
ic
p
atte
r
n
s
o
f
n
ei
g
h
b
o
r
in
g
n
o
d
es a
n
d
t
em
p
o
r
al
ch
a
n
g
es in
th
e
d
ata
.
T
h
e
g
r
id
d
iv
is
io
n
m
et
h
o
d
[
1
7
]
is
an
ap
p
r
o
ac
h
to
m
is
s
in
g
d
at
a
im
p
u
tatio
n
i
n
tr
af
f
ic
p
ass
en
g
er
f
lo
w
b
y
d
iv
id
in
g
a
g
eo
g
r
ap
h
ical
ar
ea
i
n
to
g
r
id
s
o
r
b
o
x
es
an
d
an
aly
zin
g
th
e
tem
p
o
r
al
d
y
n
am
ics
in
ea
ch
g
r
id
.
Ho
wev
er
,
to
o
lar
g
e
g
r
id
s
d
o
n
o
t
ca
p
tu
r
e
lo
ca
l
v
ar
iatio
n
s
,
wh
ile
tin
y
g
r
id
s
ar
e
to
o
s
en
s
itiv
e
to
n
o
is
e.
A
d
ee
p
lear
n
in
g
m
o
d
el
d
esig
n
ed
to
h
an
d
le
task
s
in
v
o
lv
in
g
s
p
atio
tem
p
o
r
al
d
at
a,
s
u
ch
as
m
is
s
in
g
d
ata
im
p
u
tatio
n
o
r
p
r
ed
ictio
n
in
d
atasets
th
at
h
av
e
b
o
t
h
ti
m
e
an
d
s
p
ac
e
d
im
e
n
s
io
n
s
,
is
th
e
s
p
atio
tem
p
o
r
al
f
ea
t
u
r
e
-
e
n
h
an
ce
d
g
en
e
r
ativ
e
ad
v
er
s
ar
ial
n
etwo
r
k
(S
T
-
FVGAN
)
[
1
8
]
.
Ho
we
v
er
,
th
is
m
o
d
el
is
s
en
s
itiv
e
to
n
o
is
e
in
s
p
atial
an
d
tem
p
o
r
al
d
ata,
wh
ich
ca
u
s
es
less
ac
c
u
r
ate
p
r
ed
ictio
n
s
o
r
g
en
er
at
es
u
n
r
ea
lis
tic
s
y
n
th
etic
d
ata.
Du
e
to
co
m
p
lex
s
p
atio
tem
p
o
r
al
r
elatio
n
s
h
ip
s
[
1
1
]
,
[
1
9
]
p
r
o
p
o
s
ed
a
tr
af
f
ic
d
ata
co
m
p
letio
n
m
o
d
el
b
ased
o
n
a
g
r
ap
h
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
8
,
No
.
2
,
May
20
2
5
:
7
9
4
-
8
0
7
796
co
n
v
o
l
u
tio
n
al
n
etwo
r
k
m
o
d
el
to
ac
co
u
n
t
f
o
r
m
is
s
in
g
v
alu
e
s
f
r
o
m
a
d
e
ep
lear
n
in
g
p
e
r
s
p
ec
tiv
e.
T
h
is
m
o
d
el
u
s
es
g
r
ap
h
co
n
v
o
l
u
tio
n
to
m
o
d
el
lo
ca
l
s
p
atial
d
ep
en
d
en
cies,
co
m
b
in
in
g
s
elf
-
atten
tio
n
,
g
r
a
p
h
co
n
v
o
lu
tio
n
,
an
d
r
esid
u
al
n
etwo
r
k
m
ec
h
an
is
m
s
to
c
o
p
e
with
co
m
p
lex
g
r
ap
h
d
ata
.
Usi
n
g
m
u
lti
-
v
iew
p
ass
en
g
er
f
lo
w
(
MV
PF
)
,
th
e
OD
m
at
r
ix
-
b
ased
p
r
ed
icti
o
n
m
et
h
o
d
is
a
two
-
co
m
p
o
n
e
n
t
m
eth
o
d
t
h
at
u
s
es
m
u
lti
-
v
iew
d
ata
to
p
r
e
d
ict
co
n
g
esti
o
n
an
d
o
p
tim
ize
tr
a
n
s
p
o
r
tatio
n
r
o
u
tes
[
2
0
]
.
T
e
n
s
o
r
d
ec
o
m
p
o
s
itio
n
-
b
ased
m
et
h
o
d
s
ar
e
th
e
m
o
s
t
p
o
p
u
lar
f
o
r
d
ata
im
p
u
tatio
n
,
f
o
llo
wed
b
y
GANs
an
d
GNN
,
wh
ich
r
ely
o
n
e
x
ten
s
iv
e
tr
ain
i
n
g
d
ata
s
ets.
Usi
n
g
AI
an
d
d
ee
p
lear
n
in
g
m
o
d
els
f
o
r
d
ata
im
p
u
tatio
n
o
f
f
er
s
f
lex
i
b
ilit
y
an
d
th
e
a
b
ilit
y
to
ca
p
tu
r
e
co
m
p
lex
p
atter
n
s
b
y
co
m
b
in
in
g
m
u
ltip
le
d
ata
s
o
u
r
ce
s
[
2
1
]
.
R
esear
ch
m
eth
o
d
s
r
elate
d
to
f
illi
n
g
in
m
is
s
in
g
d
ata
an
d
co
m
p
ar
in
g
th
em
o
n
th
e
C
alif
o
r
n
ia
p
er
f
o
r
m
an
ce
m
ea
s
u
r
em
en
t
s
y
s
tem
(
PeMS)
n
ee
d
to
d
esig
n
r
esear
ch
th
at
in
clu
d
es
im
p
o
r
tan
t
asp
ec
ts
s
u
ch
as
r
ep
r
esen
tativ
e
m
eth
o
d
s
,
ass
u
m
p
tio
n
s
,
im
p
u
tatio
n
s
ty
les,
im
p
lem
en
tatio
n
c
o
n
d
i
t
i
o
n
s
,
l
i
m
i
t
a
t
i
o
n
s
,
a
n
d
t
h
e
u
s
e
o
f
p
u
b
l
i
c
d
a
t
a
s
e
t
s
[
2
2
]
.
H
o
w
e
v
e
r
,
t
h
e
r
e
i
s
a
l
i
m
i
t
a
t
i
o
n
i
n
t
h
a
t
t
h
e
i
m
p
u
t
a
t
i
o
n
e
f
f
e
c
t
i
v
e
n
e
s
s
i
s
o
n
l
y
b
a
s
e
d
o
n
t
h
e
P
e
M
S
d
a
t
a
s
e
t
a
n
d
m
a
y
n
o
t
r
e
f
l
e
c
t
t
h
e
c
o
m
p
l
e
x
i
t
y
o
f
t
h
e
a
c
t
u
a
l
d
a
t
a
.
T
h
e
iter
ativ
e
g
en
er
ativ
e
a
d
v
er
s
ar
ial
n
etwo
r
k
s
f
o
r
im
p
u
tatio
n
(
I
GANI
)
m
eth
o
d
p
r
o
p
o
s
ed
b
y
[
2
3
]
is
a
v
ar
ian
t
o
f
GANs
d
esig
n
e
d
to
p
er
f
o
r
m
m
is
s
in
g
d
ata
im
p
u
tatio
n
iter
ativ
ely
.
T
h
is
m
eth
o
d
u
ti
lizes
th
e
s
tr
en
g
th
o
f
GANs
in
g
en
er
atin
g
r
ea
lis
tic
d
ata
an
d
c
o
m
b
in
es
it
with
a
n
iter
ativ
e
ap
p
r
o
ac
h
to
i
m
p
r
o
v
e
th
e
q
u
ality
a
n
d
ac
cu
r
ac
y
o
f
m
is
s
in
g
d
ata
im
p
u
tatio
n
.
Ho
wev
er
,
th
e
ch
alle
n
g
e
is
th
at
th
is
ap
p
r
o
ac
h
b
ec
o
m
es
v
er
y
co
m
p
lex
an
d
r
eq
u
ir
es
s
ig
n
if
ica
n
t
co
m
p
u
tati
o
n
al
r
eso
u
r
ce
s
.
Dee
p
co
n
v
o
lu
t
io
n
al
g
e
n
er
ativ
e
a
d
v
er
s
ar
ial
n
e
two
r
k
s
(
DC
GANs)
ar
e
an
ex
citin
g
ap
p
r
o
ac
h
to
i
m
p
u
tin
g
m
is
s
in
g
d
ata,
esp
ec
ia
lly
tr
af
f
ic
tim
e
s
er
ies
d
ata
[
2
4
]
.
T
r
af
f
ic
d
ata
f
r
o
m
PeMS
is
co
n
v
er
ted
in
to
im
ag
es,
an
d
ea
ch
s
p
ec
if
ic
tim
e
win
d
o
w
(
e.
g
.
,
2
4
h
o
u
r
s
)
is
co
n
v
er
ted
in
to
an
im
ag
e
r
ep
r
esen
tatio
n
,
wh
er
e
tr
a
f
f
ic
v
alu
es
at
a
s
p
ec
if
ic
tim
e
ar
e
r
e
p
r
esen
ted
as
p
ix
els
in
th
e
im
ag
e
[
2
5
]
.
Ho
wev
er
,
th
is
tr
ain
in
g
s
till
r
eq
u
ir
es
s
ig
n
if
ican
t
co
m
p
u
tatio
n
.
Ad
d
in
g
a
m
u
ltimo
d
al
d
ee
p
le
ar
n
in
g
m
o
d
el
f
o
r
h
eter
o
g
en
e
o
u
s
tr
af
f
ic
d
ata
im
p
u
tatio
n
u
s
in
g
two
p
ar
allel
s
tack
ed
au
to
e
n
co
d
e
r
s
is
an
in
n
o
v
at
iv
e
s
tep
an
d
ca
n
co
n
s
id
er
s
p
atial
an
d
tem
p
o
r
al
d
ep
en
d
e
n
cies
s
im
u
ltan
eo
u
s
ly
,
it
ass
es
s
es
wh
eth
er
th
e
m
o
d
el
ca
n
b
e
g
en
er
alize
d
to
o
th
er
tr
af
f
ic
d
atasets
b
ey
o
n
d
PeMS
[
2
6
]
.
P
r
o
p
o
s
ed
o
r
g
a
n
izin
g
lan
e
-
s
ca
le
tr
af
f
ic
d
ata
in
to
ten
s
o
r
p
atter
n
s
th
at
ca
n
s
im
u
ltan
e
o
u
s
ly
co
n
s
id
er
th
e
s
p
atio
-
tem
p
o
r
al
d
e
p
e
n
d
en
ce
o
f
tr
af
f
ic
f
lo
w
with
a
n
im
p
r
o
v
ed
t
u
ck
e
r
d
ec
o
m
p
o
s
itio
n
-
b
ased
im
p
u
tati
o
n
(
I
T
DI
)
m
eth
o
d
to
r
ec
o
v
er
th
e
m
is
s
in
g
v
alu
es
f
r
o
m
tr
af
f
ic
d
ata
b
y
ex
ten
d
in
g
th
e
T
u
ck
er
d
ec
o
m
p
o
s
itio
n
m
o
d
el
with
an
ad
ap
ti
v
e
r
an
k
c
alcu
latio
n
alg
o
r
ith
m
a
n
d
an
im
p
r
o
v
e
d
o
b
jectiv
e
f
u
n
ctio
n
[
2
6
]
.
P
r
o
p
o
s
ed
an
atten
tiv
e
g
r
ap
h
n
eu
r
al
p
r
o
ce
s
s
(
AGNP)
m
eth
o
d
f
o
r
s
h
o
r
t
-
ter
m
tr
af
f
ic
s
p
ee
d
p
r
ed
ictio
n
a
n
d
im
p
u
tatio
n
at
th
e
n
etwo
r
k
lev
el
wh
ile
c
o
n
s
id
er
in
g
r
eliab
ilit
y
f
ir
s
t,
a
g
au
s
s
ian
p
r
o
ce
s
s
(
GP)
is
u
s
ed
to
m
o
d
el
th
e
o
b
s
er
v
ed
tr
a
f
f
ic
s
p
ee
d
s
tate
[
2
7
]
.
As
a
to
o
l
to
ac
co
u
n
t
f
o
r
m
is
s
in
g
tr
af
f
ic
d
ata
[
2
8
]
,
d
esig
n
ed
a
n
o
v
el
d
ee
p
lear
n
in
g
ar
ch
itec
tu
r
e
ca
lled
d
y
n
am
ic
g
r
ap
h
co
n
v
o
l
u
tio
n
al
r
ec
u
r
r
en
t
im
p
u
tatio
n
n
etwo
r
k
(
DGCR
I
N)
.
DG
C
R
I
N
u
s
es
g
r
ap
h
g
en
e
r
ato
r
s
an
d
d
y
n
am
ic
g
r
ap
h
c
o
n
v
o
l
u
tio
n
al
g
ated
r
ec
u
r
r
en
t
u
n
its
(
DGCGR
U)
to
p
er
f
o
r
m
d
etailed
m
o
d
e
lin
g
o
f
th
e
d
y
n
a
m
ic
s
p
atio
tem
p
o
r
al
d
ep
en
d
en
cies
o
f
r
o
ad
n
etwo
r
k
s
.
An
in
n
o
v
ati
v
e
m
o
d
el
th
at
co
m
b
in
e
s
g
r
ap
h
atten
tio
n
n
etwo
r
k
s
(
GAT
s
)
an
d
r
ec
u
r
r
en
t
n
eu
r
al
n
etwo
r
k
s
(
R
NNs)
to
im
p
u
te
m
is
s
in
g
tr
af
f
ic
d
ata
b
y
co
n
s
id
er
in
g
s
p
atial
an
d
tem
p
o
r
al
d
e
p
en
d
e
n
cies
in
a
b
id
ir
ec
tio
n
al
m
an
n
er
b
y
[
2
9
]
ca
l
led
b
id
ir
ec
tio
n
al
g
r
ap
h
atten
ti
o
n
r
ec
u
r
r
e
n
t
n
e
u
r
al
n
etwo
r
k
(
GARNN)
wh
ich
n
e
ed
s
d
ev
elo
p
m
en
t
f
o
r
o
t
h
er
d
at
asets
.
I
t
ca
n
b
e
an
in
n
o
v
ativ
e
s
o
lu
tio
n
to
ad
d
r
ess
th
e
m
is
s
in
g
d
ata
p
r
o
b
lem
in
tr
af
f
ic
d
ata
f
o
r
t
h
e
im
p
u
tatio
n
o
f
m
u
ltis
tate
tim
e
s
er
ies
d
ata
[
3
0
]
.
Pro
p
o
s
in
g
m
u
ltis
tate
tim
e
s
er
ies
im
p
u
tat
io
n
u
s
in
g
a
g
en
er
ativ
e
ad
v
e
r
s
ar
ial
n
etwo
r
k
o
p
er
ates
b
y
u
s
in
g
a
g
e
n
er
ato
r
an
d
d
is
cr
im
in
ato
r
.
T
h
e
g
e
n
er
ato
r
aim
s
to
g
en
er
ate
an
im
p
u
tatio
n
o
f
m
is
s
in
g
v
alu
es
in
th
e
tim
e
s
er
ies
d
ata,
wh
ile
th
e
d
is
cr
im
in
ato
r
lea
r
n
s
to
d
is
cr
im
in
ate
b
etwe
en
th
o
s
e
g
en
er
ated
b
y
th
e
g
en
er
ato
r
.
T
h
e
in
ter
ac
tio
n
o
f
th
ese
two
co
m
p
o
n
en
ts
r
e
s
u
lts
in
s
tatis
tically
s
o
u
n
d
im
p
u
tatio
n
s
c
o
n
s
is
ten
t
with
th
e
u
n
d
e
r
ly
in
g
p
atter
n
o
f
th
e
tim
e
s
er
ies
.
Usi
n
g
a
laten
t
f
ac
to
r
m
o
d
el
-
b
ased
ap
p
r
o
ac
h
f
o
r
im
p
u
tin
g
tr
a
f
f
ic
d
ata
with
r
o
a
d
n
etwo
r
k
in
f
o
r
m
atio
n
ef
f
icien
tly
f
ills
d
ata
g
a
p
s
wh
i
le
co
n
s
id
er
in
g
th
e
r
o
ad
n
e
two
r
k
s
tr
u
ctu
r
e
[
3
1
]
.
T
h
e
m
o
d
el
in
co
r
p
o
r
ates
laten
t
f
a
c
t
o
r
s
t
o
c
a
p
t
u
r
e
c
o
m
p
l
e
x
p
a
t
t
e
r
n
s
i
n
t
r
a
f
f
i
c
d
a
t
a
a
n
d
r
o
a
d
n
e
t
w
o
r
k
i
n
f
o
r
m
a
t
i
o
n
t
o
i
m
p
r
o
v
e
i
m
p
u
t
a
t
i
o
n
a
c
c
u
r
a
c
y
.
I
n
m
o
d
el
f
u
s
io
n
,
th
e
h
y
b
r
id
C
NN
-
L
STM
en
s
em
b
le
m
eth
o
d
is
an
ap
p
r
o
ac
h
th
at
co
m
b
in
es
C
NNs
an
d
L
STM
s
in
an
en
s
em
b
le
m
o
d
el
.
T
h
is
m
eth
o
d
is
d
esig
n
ed
t
o
h
an
d
le
s
p
atial
-
tem
p
o
r
al
d
ata
e
f
f
ec
tiv
ely
,
m
ak
in
g
it
p
ar
ticu
lar
ly
s
u
itab
le
f
o
r
p
r
e
d
i
ctiv
e
m
o
d
elin
g
a
p
p
licatio
n
s
i
n
I
T
S
[
3
2
]
.
Utilizin
g
th
e
s
tr
en
g
th
s
o
f
C
NN
in
ca
p
tu
r
in
g
s
p
atial
in
f
o
r
m
atio
n
an
d
L
STM
in
ca
p
tu
r
i
n
g
tem
p
o
r
al
p
atter
n
s
,
it
is
well
s
u
ited
f
o
r
co
m
p
lex
s
p
atial
-
tem
p
o
r
al
d
ata.
Usi
n
g
an
en
s
e
m
b
le,
th
e
m
o
d
el
is
m
o
r
e
r
esil
ien
t
to
n
o
is
e
an
d
m
is
s
in
g
d
at
a
an
d
ca
n
p
r
o
d
u
ce
m
o
r
e
ac
cu
r
ate
p
r
ed
ictio
n
s
.
2
.
1
.
Da
t
a
c
o
llect
io
n
W
e
p
r
o
v
id
e
a
n
o
v
er
v
iew
o
f
i
n
v
esti
g
atio
n
s
in
to
tr
af
f
ic
d
ata
c
o
llectio
n
m
eth
o
d
s
,
d
ef
i
n
itio
n
s
o
f
m
is
s
in
g
d
ata
ty
p
es,
an
d
d
ata
p
r
ep
r
o
ce
s
s
in
g
m
eth
o
d
s
th
at
ca
n
b
e
u
s
ed
to
im
p
r
o
v
e
lim
ited
d
atasets
.
I
n
Fig
u
r
e
1
,
s
p
atial
-
tem
p
o
r
al
d
ata
co
llectio
n
t
h
r
o
u
g
h
d
atasets
in
clu
d
es
id
en
tify
in
g
I
o
T
s
en
s
o
r
d
ata
s
o
u
r
ce
s
,
h
is
to
r
ical
d
ata,
an
d
g
eo
s
p
atial
d
ata
[
3
3
]
.
Dete
r
m
in
in
g
r
an
d
o
m
m
is
s
in
g
d
ata,
m
is
s
in
g
d
ata
with
in
a
s
p
ec
if
ic
tim
e
r
an
g
e
,
an
d
m
is
s
in
g
d
ata
b
lo
ck
s
.
T
h
e
s
ec
o
n
d
p
ar
t
is
th
e
d
ata
im
p
u
tatio
n
m
eth
o
d
u
s
ed
,
in
clu
d
i
n
g
th
e
s
tatis
tical
m
eth
o
d
o
f
s
p
atial
in
ter
p
o
latio
n
b
y
esti
m
atin
g
m
is
s
in
g
v
alu
es
b
ased
o
n
a
s
tatis
t
ical
m
o
d
el
o
f
th
e
d
is
tan
ce
b
etwe
en
d
ata
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
S
p
a
tia
l
-
temp
o
r
a
l
d
a
ta
imp
u
ta
t
io
n
fo
r
p
r
ed
ictive
mo
d
elin
g
…
(
Yo
h
a
n
es P
r
a
co
y
o
Wid
i P
r
a
s
etyo
)
797
p
o
in
ts
,
m
ac
h
in
e
lear
n
in
g
b
ase
d
o
n
GANs
u
s
ed
to
g
en
er
ate
s
y
n
th
etic
d
ata
th
at
r
esem
b
les
th
e
o
r
ig
in
al
d
ata,
an
d
th
e
f
u
s
io
n
m
eth
o
d
r
ef
e
r
s
to
th
e
f
u
s
io
n
o
f
v
ar
io
u
s
d
ata
s
o
u
r
ce
s
,
tech
n
iq
u
es,
o
r
m
o
d
els to
ac
h
i
ev
e
m
o
r
e
ac
cu
r
ate
an
d
r
eliab
le
r
esu
lts
.
T
h
e
th
ir
d
p
ar
t
r
ef
er
s
to
th
e
p
r
o
ce
s
s
o
f
s
elec
tin
g
th
e
m
o
d
el
th
at
b
est
s
u
its
th
e
p
u
r
p
o
s
e
o
f
th
e
an
aly
s
is
an
d
th
e
ch
ar
ac
te
r
is
tics
o
f
t
h
e
d
ata.
I
n
v
o
l
v
in
g
th
e
s
elec
tio
n
o
f
f
ea
tu
r
es,
alg
o
r
ith
m
s
,
ev
alu
atio
n
m
eth
o
d
s
,
an
d
way
s
o
f
co
m
b
in
in
g
m
o
d
els
ac
r
o
s
s
r
o
ad
n
etwo
r
k
ty
p
es
an
d
ty
p
es
,
d
ata
lo
s
s
r
ef
er
s
to
s
itu
atio
n
s
wh
er
e
d
ata
th
at
s
h
o
u
ld
b
e
a
v
ailab
le
is
lo
s
t,
co
r
r
u
p
ted
,
o
r
in
ac
ce
s
s
ib
le.
Fu
zz
y
m
o
d
els
u
n
ce
r
tain
s
p
atial
-
tem
p
o
r
al
v
ar
ia
b
les,
s
u
ch
as
tr
av
el
tim
e
o
r
r
o
ad
co
n
d
itio
n
s
.
Fin
ally
,
we
r
ev
iew
f
u
tu
r
e
r
esear
ch
ch
allen
g
es
r
elate
d
to
s
p
atial
-
tem
p
o
r
al
d
a
ta
lim
itatio
n
s
,
h
is
to
r
ical
d
ata,
an
d
d
ata
q
u
ality
in
th
e
d
ata
s
et.
C
h
allen
g
es
in
s
y
s
tem
wo
r
k
lo
ad
a
s
well
as
t
h
e
ab
ilit
y
o
f
t
h
e
s
y
s
tem
to
r
ec
o
v
er
f
r
o
m
d
is
r
u
p
tio
n
s
with
clo
u
d
s
er
v
ices
f
o
r
elastic scala
b
ili
ty
.
Fig
u
r
e
1
.
Flo
wch
ar
t
o
f
r
esear
c
h
r
ev
iew
s
tep
s
2
.
1
.
1
.
Da
t
a
i
dentif
ica
t
io
n
T
r
af
f
ic
d
ata
id
en
tific
atio
n
r
e
f
er
s
to
co
llectin
g
an
d
an
aly
z
in
g
in
f
o
r
m
atio
n
r
elate
d
t
o
th
e
f
lo
w
o
f
v
eh
icles
o
r
r
o
ad
u
s
er
s
at
a
s
p
ec
if
ic
lo
ca
tio
n
.
Stan
d
ar
d
m
et
h
o
d
s
f
o
r
id
e
n
tify
in
g
tr
af
f
ic
d
ata
b
ased
o
n
o
n
lin
e
tech
n
o
lo
g
ies
s
u
c
h
as
GPS
lo
c
atio
n
,
m
o
d
elin
g
,
an
d
v
i
d
eo
a
n
aly
s
is
im
ag
e
p
r
o
ce
s
s
in
g
ar
e
d
i
s
cu
s
s
ed
,
as
well
as
th
e
co
m
p
lex
ity
o
f
th
e
d
ata
s
o
u
r
ce
[
3
4
]
.
Usi
n
g
ca
m
er
as
to
r
e
c
o
r
d
im
ag
es
an
d
i
d
en
tify
v
e
h
icles
is
s
u
g
g
ested
b
y
[
3
5
]
f
o
r
im
ag
e
p
r
o
ce
s
s
in
g
a
n
d
v
id
e
o
an
aly
s
is
in
d
etec
tin
g
m
o
v
em
e
n
ts
,
v
eh
icle
t
y
p
es,
a
n
d
tr
af
f
ic
p
atter
n
s
.
C
am
e
r
a
s
en
s
o
r
s
ar
e
also
f
r
eq
u
en
tly
u
s
ed
in
a
d
v
an
ce
d
tr
af
f
ic
m
o
n
ito
r
in
g
s
y
s
tem
s
[
3
6
]
.
Usi
n
g
Go
o
g
le
Ma
p
s
,
B
in
g
M
ap
s
,
W
az
e,
Nav
ig
atio
n
Pro
,
a
n
d
T
o
m
T
o
m
T
r
af
ic
p
latf
o
r
m
s
[
3
7
]
an
d
[
3
8
]
p
r
o
v
id
e
f
lex
i
b
ilit
y
in
o
b
tain
in
g
tr
af
f
ic
d
ata
u
s
in
g
r
ea
l
-
tim
e
m
eth
o
d
s
.
Ho
we
v
er
,
th
e
co
m
p
lex
ity
o
f
th
e
tr
af
f
ic
d
ata
r
eq
u
i
r
es
im
p
r
o
v
e
d
s
im
u
latio
n
m
o
d
els
th
at
ca
n
u
t
ilize
th
e
tech
n
o
lo
g
y
s
o
th
at
r
e
al
-
tim
e
tr
af
f
ic
d
ata
ca
n
b
e
u
s
ed
as
a
r
ef
er
en
ce
f
o
r
tr
af
f
ic
s
p
ee
d
s
er
v
ices
[
3
9
]
.
L
ik
e
PeMS
p
u
b
lic
d
ata,
th
e
tr
an
s
p
o
r
tatio
n
d
ep
ar
tm
en
t
ca
n
p
r
o
v
id
e
a
n
e
asil
y
ac
ce
s
s
ib
le,
in
ter
n
et
-
av
ai
lab
le
s
o
u
r
ce
o
f
r
ea
l
-
tim
e
h
is
to
r
ical
tr
af
f
ic
d
ata
co
n
tain
in
g
v
ar
io
u
s
a
n
aly
s
is
ca
p
ab
ilit
ies to
s
u
p
p
o
r
t v
a
r
io
u
s
u
s
er
s
[
4
0
]
.
2
.
1
.
2
.
M
is
s
ing
da
t
a
R
esear
ch
o
n
d
ata
im
p
u
tatio
n
h
as
d
if
f
er
en
t
class
if
icatio
n
s
f
o
r
th
e
ty
p
e
o
f
m
is
s
in
g
d
ata
[
4
1
]
d
escr
ib
in
g
r
an
d
o
m
,
u
n
iv
ar
iate,
an
d
m
u
lt
iv
ar
iate
m
is
s
in
g
d
ata.
I
n
o
th
e
r
p
ap
e
r
s
,
s
u
ch
as
[
4
2
]
an
d
[
4
3
]
,
u
n
iv
a
r
iate
an
d
m
u
ltiv
ar
iate
will
b
e
n
a
m
ed
f
ib
er
m
is
s
in
g
d
ata
an
d
b
lo
ck
o
r
p
an
el
m
is
s
in
g
d
ata;
o
th
er
p
a
p
er
s
m
ay
also
g
iv
e
d
if
f
er
en
t
n
am
es
to
s
im
ilar
ty
p
es
o
f
m
is
s
in
g
d
ata,
s
u
ch
as
c
o
n
tin
u
o
u
s
m
is
s
in
g
d
ata
t
o
r
ep
r
esen
t
f
ib
er
m
is
s
in
g
d
ata
[
4
4
]
.
W
e
illu
s
tr
ate
th
e
ca
teg
o
r
izatio
n
o
f
m
is
s
in
g
d
ata
in
Fig
u
r
e
2
.
Fig
u
r
e
2
(
a)
s
h
o
ws
th
at
r
a
n
d
o
m
m
is
s
in
g
d
ata
ca
n
o
cc
u
r
d
u
e
to
s
u
d
d
en
p
o
wer
d
is
co
n
n
ec
tio
n
o
f
s
en
s
o
r
s
,
f
ailed
d
ata
tr
a
n
s
m
is
s
i
o
n
d
u
e
to
n
etwo
r
k
in
ter
f
e
r
en
c
e,
an
d
r
an
d
o
m
e
r
r
o
r
s
in
s
u
r
v
e
y
d
ata
co
llectio
n
an
d
GPS
d
ev
ices.
Oth
er
v
ar
ian
ts
o
f
r
an
d
o
m
m
is
s
in
g
d
ata
ar
e
m
is
s
in
g
at
r
an
d
o
m
(
MA
R
)
an
d
m
is
s
in
g
n
o
t
at
r
an
d
o
m
(
MN
AR
)
b
u
t
[
4
5
]
,
[
4
6
]
s
tate
th
at
MN
AR
is
g
en
er
ally
n
o
t
co
n
s
id
er
e
d
.
T
h
er
ef
o
r
e,
MCAR
is
th
e
s
t
a
n
d
a
r
d
t
e
s
t
c
a
s
e
u
s
e
d
i
n
m
o
s
t
s
t
u
d
i
e
s
,
f
o
l
l
o
w
e
d
b
y
m
i
s
s
i
n
g
l
a
y
e
r
a
n
d
b
l
o
c
k
.
T
h
e
m
i
s
s
i
n
g
d
a
t
a
l
a
y
e
r
i
n
F
i
g
u
r
e
2
(
b
)
,
in
th
e
co
n
tex
t
o
f
s
p
atio
-
tem
p
o
r
al
d
ata
o
r
I
T
S
d
ata,
r
ef
er
s
to
a
s
p
ec
if
ic
tim
e
s
eg
m
en
t,
g
eo
g
r
ap
h
ic
lo
ca
tio
n
,
o
r
a
s
p
ec
if
ic
ca
teg
o
r
y
o
f
d
ata,
e.
g
.
,
v
eh
icle
ty
p
e
o
r
wea
th
er
co
n
d
it
io
n
s
.
T
h
e
m
is
s
in
g
d
ata
b
lo
ck
in
Fig
u
r
e
2
(
c)
o
f
ten
o
cc
u
r
s
in
th
e
co
n
te
x
t
o
f
tim
e
s
er
ies
d
ata
o
r
s
p
atio
-
tem
p
o
r
al
d
ata,
wh
er
e
an
en
tire
r
an
g
e
o
f
tim
e,
g
eo
g
r
a
p
h
ical
lo
ca
tio
n
,
o
r
o
th
er
v
ar
iab
les ar
e
s
u
cc
ess
iv
ely
m
is
s
in
g
.
2
.
2
.
Da
t
a
p
re
pro
ce
s
s
ing
Data
p
r
ep
r
o
ce
s
s
in
g
id
e
n
tifie
s
an
d
h
an
d
les
m
is
s
in
g
d
ata
,
v
alid
atin
g
th
e
m
is
s
in
g
v
a
lu
es
with
ap
p
r
o
p
r
iate
esti
m
ates.
P
r
o
p
o
s
ed
a
d
ata
d
en
o
is
in
g
an
d
co
m
p
r
ess
io
n
m
eth
o
d
b
ased
o
n
wa
v
elet
tr
an
s
f
o
r
m
an
d
d
ata
m
o
d
el
co
n
s
tr
u
ctio
n
[
4
7
]
.
S
aid
th
at
th
e
p
r
o
ce
s
s
o
f
r
em
o
v
in
g
u
n
r
ea
s
o
n
ab
le
o
u
tlier
s
s
h
o
u
ld
ad
ap
t
to
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
8
,
No
.
2
,
May
20
2
5
:
7
9
4
-
8
0
7
798
g
en
er
al
p
atter
n
o
f
th
e
d
ata
[
4
8
]
.
T
h
at
is
,
if
th
e
d
ataset
is
to
o
lar
g
e,
th
en
p
r
ep
r
o
ce
s
s
in
g
u
s
in
g
d
ata
s
am
p
les
to
m
ak
e
it
ea
s
ier
to
h
an
d
le,
th
en
ch
ec
k
in
g
th
e
q
u
ality
o
f
t
h
e
d
a
ta
an
d
en
s
u
r
in
g
t
h
at
th
e
d
ata
m
ee
ts
th
e
s
p
ec
if
ied
cr
iter
ia
b
ef
o
r
e
u
s
e
f
o
r
f
u
r
t
h
er
an
aly
s
is
.
(
a)
(
b
)
(
c)
Fig
u
r
e
2
.
I
ll
u
s
tr
atio
n
o
f
m
is
s
in
g
d
ata
wh
er
e
b
lack
ce
lls
ar
e
m
is
s
in
g
d
ata
:
(
a)
r
an
d
o
m
m
is
s
in
g
d
ata,
(
b
)
m
is
s
in
g
d
ata
lay
er
s
,
an
d
(
c)
m
is
s
in
g
d
a
ta
b
lo
ck
s
3.
I
NVE
ST
I
G
AT
I
O
N
M
E
T
H
O
D
L
iter
atu
r
e
r
ev
iews
o
n
t
h
e
ca
lc
u
latio
n
o
f
m
is
s
in
g
tr
af
f
ic
d
ata
o
f
ten
f
o
cu
s
o
n
t
h
e
r
esu
lt
o
f
th
e
m
eth
o
d
u
s
e
d
,
s
u
ch
as
[
7
]
-
[
1
0
]
,
b
u
t
m
ay
r
e
q
u
ir
e
f
u
r
th
er
ex
p
lo
r
atio
n
in
a
m
o
r
e
s
p
ec
if
ic
c
o
n
tex
t,
s
u
ch
as
th
e
r
o
a
d
n
etwo
r
k
o
r
t
h
e
ty
p
e
o
f
m
is
s
in
g
d
ata.
T
h
e
r
e
ar
e
th
r
ee
ca
te
g
o
r
ies
o
f
im
p
u
tatio
n
m
eth
o
d
s
f
o
r
m
is
s
in
g
tr
af
f
ic
d
ata
:
s
tatis
tical,
m
ac
h
in
e
lear
n
in
g
,
an
d
f
u
s
io
n
m
o
d
el
.
Statis
tical
im
p
u
tatio
n
to
esti
m
ate
m
is
s
in
g
v
alu
es
u
s
es
s
tatis
t
ical
m
ea
n
,
m
ed
ian
,
a
n
d
r
eg
r
ess
io
n
m
o
d
els.
Ma
ch
i
n
e
l
ea
r
n
in
g
m
et
h
o
d
s
in
v
o
lv
e
u
s
in
g
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
to
p
r
ed
ict
a
n
d
f
ill
in
m
is
s
in
g
v
alu
es
in
th
e
d
ata
s
et,
an
d
f
u
s
io
n
m
eth
o
d
s
ar
e
s
tatis
tical,
m
ac
h
in
e
lear
n
in
g
,
a
n
d
d
ee
p
lea
r
n
in
g
-
b
ased
ap
p
r
o
a
ch
es
.
T
h
e
h
y
b
r
id
C
NN
-
L
STM
en
s
em
b
le
m
eth
o
d
is
a
m
eth
o
d
th
at
co
m
b
in
es
th
e
ad
v
a
n
tag
es
o
f
C
NN
an
d
L
STM
with
in
f
o
r
m
ati
o
n
f
r
o
m
a
v
ailab
le
d
ata
to
p
e
r
f
o
r
m
im
p
u
tatio
n
a
n
d
p
r
ed
ictiv
e
m
o
d
elin
g
ac
c
u
r
atel
y
.
3
.
1
.
St
a
t
is
t
ica
l
m
et
ho
ds
Statis
t
ical
m
eth
o
d
s
an
aly
ze
e
x
is
tin
g
d
ata
to
d
e
v
elo
p
r
ep
r
es
en
tativ
e
m
o
d
els
an
d
ar
e
in
d
e
p
en
d
en
t
o
f
th
e
am
o
u
n
t
o
f
d
ata
[
4
9
]
.
As
th
e
m
o
s
t
p
o
p
u
lar
a
n
d
ea
s
y
-
to
-
p
r
o
ce
s
s
m
eth
o
d
,
it
r
ep
lace
s
m
is
s
in
g
v
al
u
es
with
t
h
e
m
ea
n
,
m
ed
ian
,
o
r
m
o
d
e
o
f
al
l
d
ata
in
th
e
co
lu
m
n
.
Pro
b
ab
ilit
y
p
r
in
cip
al
co
m
p
o
n
e
n
t
an
a
ly
s
is
(
P
PC
A)
is
a
s
tatis
t
ical
an
a
ly
s
is
tech
n
iq
u
e
u
s
ed
to
r
ed
u
ce
th
e
d
im
en
s
io
n
ali
ty
o
f
co
m
p
lex
d
ata
s
ets.
T
h
is
t
ec
h
n
iq
u
e
is
s
im
ilar
to
th
e
p
r
in
cip
al
co
m
p
o
n
en
t
an
aly
s
is
(
PC
A)
m
eth
o
d
b
u
t
u
s
es
a
p
r
o
b
a
b
ilis
tic
ap
p
r
o
ac
h
to
d
ete
r
m
in
e
th
e
p
r
in
cip
al
c
o
m
p
o
n
en
ts
o
f
th
e
d
ata.
PP
C
A
g
en
er
ally
w
o
r
k
s
b
y
f
in
d
i
n
g
a
p
r
o
b
a
b
ilis
tic
r
ep
r
es
en
tatio
n
o
f
t
h
e
d
ata
g
en
er
ated
b
y
a
lin
ea
r
co
m
b
in
atio
n
o
f
m
u
ltip
le
p
r
in
cip
al
co
m
p
o
n
en
ts
.
H
as
f
av
o
r
ab
ly
r
ev
iewe
d
a
s
p
atio
tem
p
o
r
al
PP
C
A
-
b
ased
d
ata
im
p
u
tatio
n
m
eth
o
d
f
o
r
tr
af
f
ic
f
lo
w
d
ata
in
u
r
b
an
n
etwo
r
k
s
[
5
0
]
.
T
o
o
v
er
co
m
e
th
e
s
h
o
r
tco
m
in
g
s
,
a
n
ew
m
eth
o
d
was
p
r
o
p
o
s
ed
to
im
p
r
o
v
e
th
e
im
p
u
tatio
n
p
er
f
o
r
m
an
ce
o
f
m
is
s
in
g
tr
af
f
ic
d
ata
b
y
f
u
lly
u
tili
zin
g
th
e
av
ailab
le
s
p
atial
-
tem
p
o
r
al
co
r
r
elatio
n
d
ata;
f
u
zz
y
m
ea
n
s
(
FC
M)
was
s
e
lecte
d
as th
e
b
asic a
lg
o
r
ith
m
[
5
1
]
.
3
.
1
.
1
.
Sp
a
t
ia
l
-
t
e
m
po
ra
l int
er
po
la
t
io
n
Sp
atial
-
tem
p
o
r
al
in
ter
p
o
latio
n
is
a
tech
n
iq
u
e
u
s
ed
in
g
eo
s
p
atial
d
ata
an
aly
s
is
to
esti
m
ate
v
alu
es
at
u
n
o
b
s
er
v
e
d
lo
ca
tio
n
s
an
d
ti
m
es
b
ased
o
n
o
b
s
er
v
atio
n
s
o
f
s
u
r
r
o
u
n
d
in
g
d
ata
[
5
2
]
.
As
p
ar
t
o
f
th
e
d
ata
in
ter
p
o
latio
n
tech
n
iq
u
e,
t
h
e
Kr
ig
in
g
m
et
h
o
d
f
u
lly
u
tili
ze
s
th
e
s
p
atio
tem
p
o
r
al
c
o
r
r
elatio
n
in
t
r
af
f
ic
d
ata
an
d
d
o
es
n
o
t
ass
u
m
e
th
at
th
e
d
ata
f
o
llo
ws
a
d
is
tr
ib
u
tio
n
[
5
3
]
.
T
h
e
p
er
f
o
r
m
an
ce
o
f
th
e
p
r
o
p
o
s
ed
m
eth
o
d
is
co
m
p
ar
ed
with
two
o
t
h
er
p
o
p
u
lar
m
eth
o
d
s
,
n
a
m
ely
h
is
to
r
ic
al
av
er
ag
in
g
a
n
d
KNN
.
T
h
e
r
esu
lts
s
h
o
w
th
at
th
e
p
r
o
p
o
s
ed
m
eth
o
d
h
as
t
h
e
h
ig
h
est
im
p
u
tatio
n
ac
c
u
r
ac
y
an
d
is
m
o
r
e
f
le
x
ib
le
th
an
o
th
e
r
m
eth
o
d
s
.
W
h
en
t
h
e
m
is
s
in
g
d
ata
r
ate
is
lo
wer
th
an
1
%,
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
h
is
to
r
ical
av
er
ag
e
m
eth
o
d
is
b
etter
th
an
th
e
p
r
o
p
o
s
ed
im
p
u
tatio
n
m
e
th
o
d
[
5
4
]
.
Alth
o
u
g
h
Kr
ig
in
g
is
a
p
o
wer
f
u
l
an
d
p
o
p
u
lar
in
ter
p
o
la
tio
n
m
eth
o
d
,
it
h
as
s
o
m
e
d
r
awb
ac
k
s
r
elate
d
to
t
h
e
s
tatio
n
ar
ity
ass
u
m
p
tio
n
,
d
ep
e
n
d
en
ce
o
n
v
ar
i
o
g
r
am
m
o
d
els,
s
u
f
f
icien
t
d
ata,
an
d
s
en
s
itiv
it
y
to
o
u
tlier
s
.
3
.
1
.
2
.
T
ens
o
r
deco
m
po
s
it
io
n a
nd
f
a
ct
o
riza
t
i
o
n m
et
ho
d
T
h
is
m
eth
o
d
b
elo
n
g
s
to
m
u
ltiv
ar
iate
s
tatis
tics
an
d
m
u
ltid
im
en
s
io
n
al
d
ata
a
n
aly
s
is
,
wh
ich
in
v
o
lv
es
u
s
in
g
ten
s
o
r
s
tr
u
ctu
r
es
to
f
ill
in
m
is
s
in
g
v
alu
es
in
m
u
ltid
i
m
en
s
io
n
al
d
ata.
T
h
e
tec
h
n
iq
u
e
in
v
o
lv
es
b
r
ea
k
in
g
d
o
wn
t
h
e
ten
s
o
r
s
tr
u
ctu
r
e
in
t
o
m
o
r
e
s
tr
u
ctu
r
ed
co
m
p
o
n
en
t
s
to
m
o
d
el
th
e
co
m
p
lex
p
atte
r
n
s
co
n
tain
e
d
in
th
e
d
ata
[
2
1
]
,
[
5
5
]
.
T
h
e
a
d
v
an
tag
e
o
f
d
ata
im
p
u
tatio
n
with
t
h
is
m
eth
o
d
is
th
e
ab
ilit
y
to
h
a
n
d
l
e
m
u
ltid
im
en
s
io
n
al
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
S
p
a
tia
l
-
temp
o
r
a
l
d
a
ta
imp
u
ta
t
io
n
fo
r
p
r
ed
ictive
mo
d
elin
g
…
(
Yo
h
a
n
es P
r
a
co
y
o
Wid
i P
r
a
s
etyo
)
799
d
ata
with
co
m
p
lex
p
atter
n
s
an
d
in
ter
ac
tio
n
s
b
etwe
en
d
im
en
s
io
n
s
th
r
o
u
g
h
a
B
ay
esian
ap
p
r
o
ac
h
co
m
b
in
e
d
with
im
p
u
tatio
n
tech
n
iq
u
es
to
p
r
e
d
ict
m
is
s
in
g
v
alu
es
in
th
e
d
ata
[
5
6
]
.
T
h
is
ap
p
r
o
ac
h
u
s
es
a
r
o
b
u
s
t
p
r
o
b
ab
ilis
tic
ap
p
r
o
ac
h
to
esti
m
ate
m
is
s
in
g
v
alu
es
b
y
co
n
s
id
er
in
g
th
e
u
n
ce
r
tain
ty
in
th
e
p
r
e
d
ictio
n
.
B
asic
ten
s
o
r
f
ac
to
r
izatio
n
m
et
h
o
d
s
h
a
v
e
s
h
o
wn
s
ig
n
if
ican
t
im
p
r
o
v
em
en
t
in
th
e
f
ield
o
f
m
is
s
in
g
tr
af
f
ic
d
ata
im
p
u
tatio
n
,
as
s
tated
b
y
[
3
]
,
[1
0]
,
[
1
3
]
,
[
1
8
]
,
[
2
0
]
,
[
2
5
]
,
[
5
6
]
-
[
5
8
]
I
t
ca
n
al
s
o
b
e
s
ee
n
t
h
at
m
o
s
t
o
f
th
ese
m
o
d
els
h
av
e
b
ee
n
test
ed
f
o
r
r
o
b
u
s
tn
ess
in
ac
co
u
n
tin
g
f
o
r
m
is
s
in
g
tr
af
f
ic
d
at
a
at
lev
els
r
an
g
in
g
f
r
o
m
1
%
to
9
0
%
wh
ile
s
till
h
av
in
g
a
h
ig
h
lev
el
o
f
ac
c
u
r
ac
y
[
5
7
]
,
[
5
9
]
.
T
h
is
m
eth
o
d
'
s
ad
v
a
n
tag
es
lie
i
n
its
s
im
p
licity
an
d
a
p
p
licab
ili
ty
,
ea
s
e
o
f
in
te
r
p
r
etatio
n
,
co
m
p
u
tatio
n
al
ef
f
icien
cy
,
an
d
s
u
itab
ilit
y
f
o
r
s
tr
u
ctu
r
ed
d
ata.
I
ts
wea
k
n
ess
es
ar
e
th
at
it
is
lim
ited
to
lin
ea
r
d
ata,
less
f
lex
ib
le,
an
d
less
ad
ap
tiv
e.
Sin
ce
it
wo
r
k
s
well
o
n
lin
ea
r
d
ata
p
atter
n
s
,
it
is
m
o
r
e
ef
f
ec
tiv
e
wh
en
wo
r
k
in
g
with
s
tr
u
ctu
r
ed
d
ata.
3
.
2
.
M
a
chine
l
ea
rning
Ma
ch
in
e
lear
n
in
g
alg
o
r
ith
m
s
ar
e
d
esig
n
ed
to
lea
r
n
f
r
o
m
ex
i
s
tin
g
d
ata,
id
en
tify
p
atter
n
s
,
a
n
d
ad
ap
t
t
o
en
v
ir
o
n
m
en
tal
ch
an
g
es
o
r
n
e
w
d
ata
[
6
0
]
.
M
o
d
els
ar
e
tr
ain
ed
u
s
in
g
d
atasets
co
n
tain
in
g
p
r
ed
ef
in
e
d
in
p
u
t
an
d
o
u
tp
u
t
p
air
s
in
s
u
p
er
v
is
ed
lear
n
in
g
.
I
n
co
n
tr
ast,
th
e
m
o
d
e
l
is
g
iv
en
d
ata
with
n
o
lab
els
o
r
ca
teg
o
r
ies
in
u
n
s
u
p
er
v
is
ed
lear
n
in
g
.
T
h
e
g
o
al
is
to
d
is
co
v
er
n
atu
r
al
p
atter
n
s
in
th
e
d
ata,
s
u
ch
as
clu
s
ter
s
o
r
h
id
d
en
s
tr
u
ctu
r
es.
N
eu
r
al
n
etwo
r
k
s
ar
e
th
e
m
o
d
el
m
o
s
t
s
y
n
o
n
y
m
o
u
s
with
m
ac
h
in
e
le
ar
n
in
g
,
alth
o
u
g
h
n
o
t
co
m
p
letely
[
6
1
]
,
it
is
a
p
o
wer
f
u
l
to
o
l
f
o
r
d
ata
im
p
u
tatio
n
,
in
clu
d
in
g
s
p
atial
-
tem
p
o
r
al
d
ata
im
p
u
tatio
n
.
I
n
th
is
co
n
tex
t,
n
eu
r
al
n
etwo
r
k
s
ca
n
p
r
ed
ict
m
is
s
in
g
v
alu
es
b
ased
o
n
p
atter
n
s
in
th
e
av
ailab
le
d
ata.
So
m
e
ap
p
r
o
ac
h
es
to
u
s
in
g
n
eu
r
al
n
etwo
r
k
s
in
d
ata
im
p
u
t
atio
n
ar
e
GANs
.
GAN
s
ar
e
a
ty
p
e
o
f
n
eu
r
al
n
etwo
r
k
a
r
ch
itectu
r
e
t
h
at
c
o
n
s
is
ts
o
f
two
n
eu
r
al
n
etwo
r
k
m
o
d
els:
g
en
er
ato
r
s
an
d
d
is
cr
im
in
ato
r
s
.
T
h
e
g
en
er
ato
r
is
r
esp
o
n
s
ib
le
f
o
r
cr
ea
tin
g
n
ew
d
ata,
f
o
r
ex
am
p
le,
im
ag
es,
s
o
u
n
d
s
,
o
r
tex
ts
,
s
im
ilar
to
th
e
tr
ain
in
g
d
ata
.
At
f
ir
s
t,
th
e
g
e
n
er
ato
r
g
en
er
ates
r
an
d
o
m
d
a
ta.
Du
r
in
g
tr
ai
n
in
g
,
th
e
g
en
e
r
ato
r
lear
n
s
to
g
e
n
er
a
te
d
ata
th
at
is
in
cr
ea
s
in
g
l
y
s
i
m
ilar
to
t
h
e
tr
ai
n
in
g
d
ata
t
h
r
o
u
g
h
f
ee
d
b
ac
k
f
r
o
m
th
e
d
is
cr
im
in
ato
r
[
6
2
]
.
T
h
e
d
i
s
cr
im
in
ato
r
is
r
esp
o
n
s
ib
le
f
o
r
d
is
tin
g
u
is
h
in
g
b
etwe
en
th
e
d
a
ta
g
en
er
ated
b
y
th
e
g
en
er
ato
r
an
d
th
e
o
r
ig
in
al
tr
ai
n
in
g
d
ata.
T
h
e
d
is
cr
im
in
ato
r
i
s
tr
ain
ed
to
d
is
tin
g
u
is
h
b
etwe
en
"r
ea
l"
d
ata
(
f
r
o
m
th
e
tr
ain
in
g
d
ataset)
an
d
"f
ak
e"
d
ata
(
g
en
e
r
ated
b
y
th
e
g
en
er
ato
r
)
.
T
h
e
d
is
cr
im
in
ato
r
u
p
d
ates
th
e
p
ar
am
eter
s
b
ased
o
n
th
e
er
r
o
r
in
class
if
y
in
g
th
e
f
ak
e
o
r
o
r
i
g
in
al
d
ata
[
2
2
]
,
[
2
4
]
.
T
h
e
GAN
tr
ain
in
g
p
r
o
ce
s
s
i
n
v
o
lv
es
iter
atio
n
s
wh
er
e
th
e
g
e
n
er
ato
r
an
d
d
is
cr
im
in
ato
r
p
lay
a
g
ain
s
t e
ac
h
o
th
er
.
A
n
eu
r
al
n
etwo
r
k
g
r
ap
h
is
a
v
is
u
al
r
ep
r
esen
tatio
n
o
f
t
h
e
ar
ch
itectu
r
e
an
d
s
tr
u
ctu
r
e
o
f
a
n
eu
r
a
l
n
etwo
r
k
u
s
ed
in
m
ac
h
i
n
e
lea
r
n
in
g
.
I
t
s
h
o
ws
h
o
w
n
eu
r
o
n
s
ar
e
o
r
g
a
n
ized
i
n
lay
er
s
an
d
co
n
n
ec
ted
th
r
o
u
g
h
weig
h
ted
co
n
n
ec
tio
n
s
.
R
ec
en
tly
,
[
6
3
]
h
as
co
n
d
u
cte
d
a
co
m
p
r
eh
en
s
iv
e
s
u
r
v
ey
o
n
GNNs
an
d
class
if
ied
v
ar
io
u
s
GNN
m
o
d
els
in
to
f
o
u
r
ca
teg
o
r
ies
-
r
ec
u
r
r
en
t
GNNs,
co
n
v
o
lu
tio
n
al
GNNs,
g
r
ap
h
au
to
-
en
co
d
er
s
,
an
d
s
p
atio
-
tem
p
o
r
al
GNN
s
.
Am
o
n
g
th
em
,
it
h
as
b
ee
n
f
o
u
n
d
th
at
co
n
v
o
lu
tio
n
al
GNN
h
as
r
ec
en
tly
b
ec
o
m
e
a
m
o
r
e
p
o
p
u
lar
ch
o
ice
in
tr
af
f
ic
r
ese
ar
ch
,
as
s
h
o
wn
b
y
[
5
]
,
[
1
5
]
,
[
3
6
]
.
G
r
ap
h
c
o
n
v
o
lu
tio
n
al
n
e
two
r
k
s
(
GC
Ns)
ar
e
n
eu
r
al
n
etwo
r
k
ar
c
h
itectu
r
es
d
esig
n
ed
to
p
er
f
o
r
m
lear
n
in
g
o
n
d
ata
s
tr
u
ctu
r
ed
as
g
r
ap
h
s
o
r
n
etwo
r
k
s
.
T
h
ey
ex
ten
d
t
h
e
c
o
n
v
o
lu
tio
n
co
n
ce
p
t
o
f
co
n
v
en
tio
n
al
n
eu
r
al
n
et
wo
r
k
s
to
th
e
g
r
ap
h
d
o
m
ain
,
en
ab
lin
g
th
e
u
s
e
o
f
to
p
o
lo
g
ical
in
f
o
r
m
atio
n
in
d
at
a
r
ep
r
esen
tatio
n
.
C
o
n
v
o
lu
tio
n
al
GNNs,
o
r
GC
Ns,
u
tili
ze
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
etwo
r
k
s
to
em
b
e
d
g
r
a
p
h
in
f
o
r
m
atio
n
in
to
ten
s
o
r
s
,
r
esu
ltin
g
in
a
u
n
if
o
r
m
f
r
a
m
ewo
r
k
f
r
o
m
ir
r
eg
u
lar
d
ata
s
ets
[
6
4
]
.
GNNs
ca
n
lear
n
co
m
p
le
x
r
ep
r
esen
tatio
n
s
o
f
s
p
atio
-
te
m
p
o
r
al
s
tr
u
ctu
r
es
in
g
r
ap
h
s
an
d
ex
tr
ac
t
p
atter
n
s
,
r
elatio
n
s
h
ip
s
,
an
d
d
ep
e
n
d
en
ci
es b
etwe
en
g
r
ap
h
en
titi
es
[
6
5
]
.
Ma
ch
in
e
lear
n
in
g
m
eth
o
d
s
ca
n
h
an
d
le
co
m
p
le
x
d
ata
s
u
ch
a
s
s
p
atial
-
tem
p
o
r
al,
ad
ap
tiv
e
to
ex
ten
s
iv
e
d
ata,
an
d
m
o
r
e
ac
cu
r
ate
p
r
e
d
ictio
n
r
esu
lts
.
Ma
ch
in
e
lear
n
in
g
m
o
d
els
s
u
ch
as
n
eu
r
al
n
e
two
r
k
s
o
r
r
an
d
o
m
f
o
r
ests
ca
n
ca
p
tu
r
e
co
m
p
lex
d
ata
p
atter
n
s
an
d
wo
r
k
with
v
ar
io
u
s
d
ata
ty
p
es,
in
clu
d
in
g
th
o
s
e
with
m
is
s
in
g
v
alu
es.
Alth
o
u
g
h
th
ey
h
av
e
d
r
awb
ac
k
s
,
s
u
ch
as
lar
g
e
d
ata
r
eq
u
ir
e
m
en
ts
an
d
t
h
e
r
is
k
o
f
o
v
e
r
f
itti
n
g
,
ap
p
r
o
ac
h
es
s
u
ch
as
r
eg
u
lar
izatio
n
an
d
ca
r
ef
u
l
f
ea
tu
r
e
s
elec
tio
n
ca
n
h
elp
o
v
er
co
m
e
th
em
.
I
n
ITS
,
th
is
m
eth
o
d
wo
r
k
s
well
wh
en
th
e
d
ata
is
d
y
n
am
ic
an
d
th
u
s
ca
n
p
r
ed
ict
t
r
af
f
ic,
d
etec
t
an
o
m
alies,
s
ig
n
al
o
p
tim
izatio
n
,
an
d
tr
av
el
tim
e
esti
m
atio
n
with
p
r
e
cisi
o
n
.
3
.
3
.
F
us
io
n
m
o
del
Mo
d
el
f
u
s
io
n
f
o
r
d
ata
im
p
u
ta
tio
n
r
ef
er
s
to
u
s
in
g
d
if
f
er
en
t
t
ec
h
n
iq
u
es
an
d
m
o
d
els
to
f
ill
in
m
is
s
in
g
v
alu
es
in
a
d
ata
s
et.
T
h
is
ap
p
r
o
ac
h
e
x
p
lo
its
th
e
s
tr
en
g
th
s
o
f
ea
ch
m
o
d
el
t
o
im
p
r
o
v
e
th
e
q
u
ality
o
f
im
p
u
tatio
n
an
d
r
ed
u
ce
th
e
wea
k
n
ess
es
o
f
a
s
in
g
le
im
p
u
tatio
n
m
eth
o
d
.
T
h
e
en
s
em
b
le
ap
p
r
o
ac
h
c
o
m
b
in
es
p
r
ed
ictio
n
s
f
r
o
m
s
ev
er
al
d
if
f
er
e
n
t
im
p
u
ta
tio
n
m
o
d
els
[
6
6
]
.
Fo
r
e
x
am
p
l
e,
b
ag
g
in
g
,
b
o
o
s
tin
g
,
o
r
s
tack
in
g
tech
n
iq
u
es
ca
n
in
teg
r
ate
th
e
r
esu
lts
f
r
o
m
m
u
l
tip
le
im
p
u
tatio
n
m
o
d
els
an
d
p
r
o
d
u
ce
m
o
r
e
ac
cu
r
ate
p
r
ed
ic
tio
n
s
.
C
o
m
b
in
atio
n
mo
d
els
co
m
b
in
e
im
p
u
tatio
n
r
e
s
u
lts
f
r
o
m
lin
ea
r
an
d
n
o
n
-
lin
e
ar
m
o
d
els
(
e.
g
.
,
r
an
d
o
m
f
o
r
est
o
r
n
eu
r
al
n
etwo
r
k
)
to
o
b
tain
b
etter
r
esu
lts
.
T
h
e
f
u
s
io
n
m
eth
o
d
with
h
y
b
r
id
C
NN
-
L
STM
en
s
em
b
le
in
th
e
co
n
te
x
t
o
f
im
p
u
tatio
n
o
f
s
p
atio
-
tem
p
o
r
al
d
ata
in
v
o
lv
es
co
m
b
in
in
g
two
t
y
p
es
o
f
m
o
d
els
t
o
u
tili
ze
th
e
s
tr
en
g
th
s
o
f
ea
ch
[
6
7
]
.
Fu
s
io
n
with
m
u
ltip
le
im
p
u
tatio
n
m
et
h
o
d
s
is
an
ap
p
r
o
ac
h
th
at
in
v
o
lv
es
u
s
in
g
s
ev
er
al
d
if
f
er
en
t
i
m
p
u
ta
tio
n
m
eth
o
d
s
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
8
,
No
.
2
,
May
20
2
5
:
7
9
4
-
8
0
7
800
s
u
ch
as
KNN,
r
eg
r
ess
io
n
,
an
d
in
ter
p
o
latio
n
,
a
n
d
th
en
co
m
b
in
in
g
th
e
im
p
u
tatio
n
r
esu
l
ts
f
r
o
m
th
e
v
ar
i
o
u
s
m
eth
o
d
s
to
p
r
o
d
u
ce
m
o
r
e
s
tab
le
p
r
ed
ictio
n
s
[
6
8
]
.
I
t
u
tili
ze
s
th
e
ad
v
a
n
tag
es
o
f
b
o
th
s
tatis
tical
an
d
m
ac
h
i
n
e
l
ea
r
n
in
g
m
eth
o
d
s
to
im
p
r
o
v
e
p
r
ed
ictiv
e
ac
cu
r
ac
y
an
d
h
as th
e
f
le
x
ib
ilit
y
an
d
a
d
ap
tiv
en
ess
to
o
v
e
r
co
m
e
o
v
er
f
itti
n
g
.
Alth
o
u
g
h
it h
a
s
th
e
d
is
ad
v
an
tag
e
o
f
h
ig
h
c
o
m
p
l
ex
ity
an
d
r
eq
u
i
r
es
ex
ten
s
iv
e
co
m
p
u
tatio
n
al
r
eso
u
r
ce
s
,
it
w
o
r
k
s
well
o
n
co
m
p
le
x
an
d
h
eter
o
g
en
e
o
u
s
d
ata
p
r
o
ce
s
s
in
g
,
m
ak
in
g
it
a
f
lex
ib
le
s
o
lu
ti
o
n
f
o
r
d
y
n
am
ic
d
ata
p
atter
n
s
.
E
f
f
ec
tiv
en
ess
f
o
r
h
eter
o
g
en
e
o
u
s
an
d
b
alan
ce
d
d
ata
o
n
v
ar
i
o
u
s
d
ata
v
a
r
iatio
n
m
o
d
els,
t
h
is
f
u
s
io
n
m
et
h
o
d
is
t
h
e
m
o
s
t e
f
f
ec
tiv
e.
3
.
4
.
O
v
er
v
iew
o
f
re
s
ea
rc
h met
ho
ds
T
o
p
r
o
v
i
d
e
m
o
r
e
s
p
ec
if
ic
in
f
o
r
m
atio
n
,
th
e
f
o
llo
win
g
is
a
d
etailed
d
escr
ip
tio
n
o
f
th
e
p
r
im
ar
y
m
eth
o
d
s
,
ex
am
p
le
r
e
f
er
en
ce
p
ap
er
s
,
r
o
ad
n
etwo
r
k
ty
p
es,
d
at
a
ac
q
u
is
itio
n
m
eth
o
d
s
,
an
d
ty
p
es
o
f
m
is
s
in
g
d
ata
test
ed
in
th
e
co
n
tex
t
o
f
s
p
atial
-
tem
p
o
r
al
d
ata
im
p
u
tatio
n
r
esear
ch
in
in
tellig
en
t tr
a
n
s
p
o
r
tati
o
n
in
T
a
b
le
1
.
Fro
m
T
ab
le
1
,
m
o
s
t
o
f
th
e
liter
atu
r
e
r
ev
iewe
d
was
co
n
d
u
ct
ed
o
n
u
r
b
a
n
n
etwo
r
k
s
.
T
h
is
i
s
b
ec
au
s
e
u
r
b
an
n
etwo
r
k
s
a
r
e
th
e
m
o
s
t
f
lu
ctu
atin
g
an
d
b
u
s
iest
an
d
th
u
s
r
eq
u
ir
e
th
e
s
u
p
p
o
r
t
o
f
in
tellig
en
t
tr
an
s
p
o
r
tatio
n
s
y
s
tem
s
.
Ho
wev
er
,
if
we
lo
o
k
d
ee
p
er
in
to
th
e
d
ata
s
ets
u
s
e
d
,
it
ca
n
b
e
s
ee
n
t
h
at
m
o
s
t
o
f
th
em
ar
e
th
e
s
am
e.
Mo
s
t
s
tu
d
ies
co
n
s
id
er
in
g
tr
af
f
ic
f
lo
w
s
h
o
wn
in
T
ab
le
1
u
s
e
tax
i
GP
S
d
ata,
wh
ich
m
ay
n
o
t
h
av
e
ac
cu
r
at
e
tr
af
f
ic
s
p
ee
d
s
.
T
r
af
f
ic
v
o
lu
m
e
s
h
o
ws f
ewe
r
m
is
s
in
g
d
ata
im
p
u
tatio
n
s
tu
d
ies,
lik
ely
d
u
e
to
d
ata
av
ailab
ilit
y
an
d
th
e
in
ac
cu
r
ate
n
atu
r
e
o
f
tr
af
f
ic
v
o
lu
m
e.
Ho
wev
er
,
tr
af
f
ic
v
o
lu
m
e
also
p
r
o
v
id
es
a
g
o
o
d
p
ictu
r
e
o
f
tr
af
f
ic
c
o
n
d
i
t
i
o
n
s
,
t
r
a
v
e
l
t
i
m
e
,
a
n
d
c
o
n
g
e
s
t
i
o
n
l
e
v
e
l
s
;
t
h
e
s
e
a
r
e
p
a
r
a
m
e
t
e
r
s
t
o
c
o
n
s
i
d
e
r
.
I
m
p
u
t
a
t
i
o
n
m
e
t
h
o
d
s
s
h
o
u
l
d
c
o
n
s
i
d
e
r
b
o
t
h
s
p
a
t
i
a
l
a
n
d
t
e
m
p
o
r
a
l
a
s
p
e
c
t
s
t
o
p
r
o
d
u
c
e
a
c
c
u
r
a
t
e
e
s
t
i
m
a
t
e
s
,
a
s
t
r
a
n
s
p
o
r
t
a
t
i
o
n
d
a
t
a
a
r
e
o
f
t
e
n
d
e
r
i
v
e
d
f
r
o
m
s
e
n
s
o
r
s
a
n
d
d
i
r
e
c
t
o
b
s
e
r
v
a
t
i
o
n
s
t
o
o
v
e
r
c
o
m
e
u
n
c
e
r
t
a
i
n
t
i
e
s
a
n
d
d
i
s
t
u
r
b
a
n
c
e
s
t
h
a
t
m
a
y
o
c
c
u
r
i
n
o
b
s
e
r
v
a
t
i
o
n
a
l
d
a
t
a
.
T
a
b
l
e
2
l
i
s
t
s
t
h
e
c
h
a
r
a
c
t
e
r
i
s
t
i
c
a
d
v
a
n
t
a
g
e
s
a
n
d
g
a
p
s
o
f
t
h
e
p
o
p
u
l
a
r
m
e
t
h
o
d
s
m
e
n
t
i
o
n
e
d
i
n
T
a
b
l
e
1
.
T
ab
le
1
.
Su
m
m
a
r
y
o
f
liter
atu
r
e
s
tu
d
ies o
n
v
ar
iab
le
d
ata
im
p
u
tatio
n
M
e
t
h
o
d
s
A
r
t
i
c
l
e
R
o
a
d
n
e
t
w
o
r
k
D
a
t
a
l
o
ss
F
u
z
z
y
m
e
t
h
o
d
D
a
t
a
l
i
m
i
t
a
t
i
o
n
s
D
a
t
a
s
o
u
r
c
e
S
t
a
t
i
st
i
c
s
S
p
a
t
i
a
l
-
Te
m
p
o
r
a
l
I
n
t
e
r
p
o
l
a
t
i
o
n
[
4
9
]
,
[
5
0
]
-
[
5
4
]
[
5
2
]
-
[
5
4
]
2
2
3
1
0
0
0
0
0
0
Te
n
s
o
r
D
e
c
o
m
p
o
si
t
i
o
n
[
3
]
,
[
1
0
]
[
1
3
]
,
[
1
8
]
,
[
2
0
]
,
[
2
1
]
,
[
2
0
]
,
[
2
5
]
,
[
5
5
]
,
[
5
6
]
-
[
5
9
]
9
5
0
0
0
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
[
5
]
,
[
1
5
]
,
[
2
3
]
,
[
2
4
]
,
[
3
6
]
,
[
6
1
]
-
[
6
5
]
,
[
6
0
]
1
9
0
1
0
F
u
si
o
n
M
o
d
e
l
[
2
8
]
,
[
3
2
]
,
[
6
6
]
,
[
6
7
]
,
[
6
8
]
2
3
0
0
0
T
ab
le
2
.
C
h
ar
ac
ter
is
tics
o
f
p
o
p
u
lar
m
eth
o
d
s
M
e
t
h
o
d
s
S
t
a
t
e
o
f
t
h
e
a
r
t
R
e
se
a
r
c
h
g
a
p
s
S
t
a
t
i
st
i
c
s
−
A
b
i
l
i
t
y
t
o
h
a
n
d
l
e
t
i
me
a
n
d
sp
a
c
e
v
a
r
i
a
b
i
l
i
t
y
su
p
p
o
r
t
e
d
b
y
a
c
c
u
r
a
c
y
i
n
e
s
t
i
m
a
t
i
n
g
mi
ssi
n
g
v
a
l
u
e
s
a
n
d
f
l
e
x
i
b
i
l
i
t
y
i
n
m
o
d
e
l
s
a
n
d
a
p
p
r
o
a
c
h
e
s.
−
H
a
n
d
l
i
n
g
e
x
t
e
n
si
v
e
d
a
t
a
a
n
d
a
c
c
o
mm
o
d
a
t
i
n
g
u
n
c
e
r
t
a
i
n
t
y
i
n
e
s
t
i
m
a
t
i
o
n
a
r
e
l
i
m
i
t
a
t
i
o
n
s,
a
s
sp
a
t
i
a
l
-
t
e
mp
o
r
a
l
d
a
t
a
t
e
n
d
s
t
o
b
e
f
u
l
l
o
f
u
n
c
e
r
t
a
i
n
t
y
.
Te
n
s
o
r
d
e
c
o
m
p
o
si
t
i
o
n
−
I
t
h
a
n
d
l
e
s
c
o
mp
l
e
x
a
n
d
m
u
l
t
i
d
i
m
e
n
s
i
o
n
a
l
d
a
t
a
w
e
l
l
a
n
d
h
a
s
h
i
d
d
e
n
p
a
t
t
e
r
n
r
e
c
o
g
n
i
t
i
o
n
.
−
I
t
a
c
c
o
mm
o
d
a
t
e
s
u
n
c
e
r
t
a
i
n
t
y
i
n
d
a
t
a
a
n
d
e
st
i
mat
i
o
n
a
n
d
i
s
mo
r
e
c
o
mp
u
t
a
t
i
o
n
a
l
l
y
e
f
f
i
c
i
e
n
t
t
h
a
n
o
t
h
e
r
c
o
mp
u
t
a
t
i
o
n
a
l
l
y
i
n
t
e
n
s
i
v
e
a
p
p
r
o
a
c
h
e
s.
−
Li
mi
t
a
t
i
o
n
s
i
n
h
a
n
d
l
i
n
g
n
o
i
s
e
a
n
d
u
n
c
e
r
t
a
i
n
t
y
i
n
t
h
e
d
a
t
a
.
−
I
t
r
e
q
u
i
r
e
s
l
a
r
g
e
r
e
s
o
u
r
c
e
s
f
o
r
i
mp
l
e
men
t
a
t
i
o
n
i
n
r
e
a
l
-
t
i
me
o
r
b
i
g
d
a
t
a
e
n
v
i
r
o
n
m
e
n
t
s.
−
R
e
q
u
i
r
e
s
a
d
e
e
p
u
n
d
e
r
st
a
n
d
i
n
g
o
f
t
h
e
c
o
n
c
e
p
t
s
a
n
d
a
l
g
o
r
i
t
h
ms
i
n
v
o
l
v
e
d
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
−
A
b
l
e
t
o
h
a
n
d
l
e
l
a
r
g
e
v
o
l
u
mes
o
f
d
a
t
a
w
e
l
l
.
−
A
b
l
e
t
o
h
a
n
d
l
e
d
a
t
a
w
i
t
h
h
i
g
h
d
i
me
n
s
i
o
n
a
l
i
t
y
a
n
d
c
o
m
p
l
e
x
s
t
r
u
c
t
u
r
e
s,
s
u
c
h
a
s
sp
a
t
i
a
l
-
t
e
m
p
o
r
a
l
d
a
t
a
,
w
i
t
h
s
o
u
n
d
c
a
p
a
b
i
l
i
t
i
e
s
.
−
En
a
b
l
e
s
d
a
t
a
-
d
r
i
v
e
n
mo
d
e
l
t
r
a
i
n
i
n
g
t
h
a
t
c
a
n
i
mp
r
o
v
e
i
mp
u
t
a
t
i
o
n
p
e
r
f
o
r
ma
n
c
e
.
−
Th
e
p
e
r
f
o
r
ma
n
c
e
o
f
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
mo
d
e
l
s
i
s
h
i
g
h
l
y
d
e
p
e
n
d
e
n
t
o
n
t
h
e
d
a
t
a
q
u
a
l
i
t
y
u
se
d
f
o
r
t
r
a
i
n
i
n
g
.
−
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
mo
d
e
l
s
c
o
u
l
d
b
e
m
o
r
e
e
f
f
e
c
t
i
v
e
w
h
e
n
d
e
a
l
i
n
g
w
i
t
h
i
m
b
a
l
a
n
c
e
d
d
a
t
a
.
−
O
v
e
r
f
i
t
t
i
n
g
c
a
n
l
e
a
d
t
o
i
n
a
c
c
u
r
a
t
e
o
r
u
n
r
e
l
i
a
b
l
e
i
mp
u
t
a
t
i
o
n
r
e
su
l
t
s
.
F
u
si
o
n
m
o
d
e
l
s
−
W
e
a
r
e
u
t
i
l
i
z
i
n
g
e
a
c
h
m
o
d
e
l
's a
d
v
a
n
t
a
g
e
s
a
n
d
man
a
g
i
n
g
e
a
c
h
mo
d
e
l
's s
h
o
r
t
c
o
m
i
n
g
s.
−
I
t
h
a
s
i
m
p
r
o
v
e
d
p
e
r
f
o
r
m
a
n
c
e
a
n
d
f
l
e
x
i
b
i
l
i
t
y
i
n
r
e
so
u
r
c
e
u
t
i
l
i
z
a
t
i
o
n
.
−
R
e
si
l
i
e
n
t
t
o
c
h
a
n
g
e
s
i
n
d
a
t
a
a
n
d
e
n
v
i
r
o
n
m
e
n
t
a
n
d
i
n
c
r
e
a
s
e
d
r
o
b
u
st
n
e
ss
f
o
r
r
e
d
u
c
e
d
o
v
e
r
f
i
t
t
i
n
g
.
−
R
e
q
u
i
r
e
s m
o
r
e
e
x
c
e
l
l
e
n
t
c
o
mp
u
t
i
n
g
r
e
so
u
r
c
e
s.
−
Th
i
s
c
o
mp
l
e
x
i
t
y
c
a
n
c
o
m
p
l
i
c
a
t
e
t
h
e
i
n
t
e
r
p
r
e
t
a
t
i
o
n
o
f
r
e
s
u
l
t
s
a
n
d
i
n
c
r
e
a
s
e
t
h
e
c
o
m
p
u
t
a
t
i
o
n
a
l
a
n
d
man
a
g
e
me
n
t
c
o
st
s
o
f
t
h
e
mo
d
e
l
.
−
R
e
q
u
i
r
e
s
mo
r
e
c
o
mp
l
e
x
c
u
s
t
o
mi
z
a
t
i
o
n
a
n
d
mai
n
t
e
n
a
n
c
e
t
h
a
n
s
i
n
g
l
e
m
o
d
e
l
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
S
p
a
tia
l
-
temp
o
r
a
l
d
a
ta
imp
u
ta
t
io
n
fo
r
p
r
ed
ictive
mo
d
elin
g
…
(
Yo
h
a
n
es P
r
a
co
y
o
Wid
i P
r
a
s
etyo
)
801
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
is
p
ap
er
in
v
esti
g
ates
th
e
im
p
ac
t
o
f
s
p
atial
-
tem
p
o
r
al
d
at
a
lo
s
s
o
n
I
T
S.
Alth
o
u
g
h
p
r
e
v
io
u
s
p
ap
er
s
h
av
e
ex
p
lo
r
e
d
m
o
d
el
im
p
u
tat
io
n
an
d
p
r
ed
ictiv
e
m
o
d
elin
g
,
th
ey
d
id
n
o
t
ex
am
i
n
e
th
e
co
m
m
o
n
m
ec
h
an
is
m
s
u
s
ed
am
o
n
g
th
e
v
ar
i
o
u
s
m
o
d
els
r
ev
ie
wed
an
d
f
o
cu
s
ed
m
o
r
e
o
n
th
e
o
v
er
all
q
u
ality
o
f
ea
ch
m
o
d
el
.
Sp
atio
-
tem
p
o
r
al
atten
tio
n
n
e
two
r
k
s
(
STAN
)
in
co
r
p
o
r
ate
atten
tio
n
m
ec
h
an
is
m
s
to
ca
p
tu
r
e
th
e
co
m
p
lex
r
elatio
n
s
h
ip
s
b
etwe
en
s
p
atial
an
d
tem
p
o
r
al
d
ata
[
6
9
]
.
I
n
ad
d
itio
n
,
a
s
p
atial
-
tem
p
o
r
al
f
u
s
io
n
lay
er
co
m
b
in
es
s
p
atia
l
an
d
tem
p
o
r
al
f
ea
tu
r
e
r
ep
r
esen
tatio
n
s
an
d
a
n
e
n
co
d
e
r
-
d
ec
o
d
er
a
r
ch
itectu
r
e
th
at
p
r
o
d
u
ce
s
th
e
d
esire
d
o
u
tp
u
t.
STAN
ca
n
tak
e
in
to
ac
co
u
n
t
th
e
in
ter
ac
tio
n
s
b
etwe
en
s
en
s
o
r
lo
ca
tio
n
s
an
d
tim
e,
r
esu
ltin
g
in
f
ea
tu
r
e
r
ep
r
esen
tatio
n
s
f
o
r
m
o
r
e
ac
c
u
r
ate
p
r
e
d
ictio
n
s
[
7
0
]
.
T
h
e
d
is
cu
s
s
io
n
o
f
f
er
s
a
h
y
b
r
id
C
NN
-
L
STM
en
s
em
b
le
m
eth
o
d
th
at
c
o
m
b
in
es
t
h
e
s
tr
en
g
th
s
o
f
C
NN
an
d
L
STM
to
m
o
d
el
th
e
co
m
p
le
x
ity
o
f
s
p
a
tial
-
tem
p
o
r
al
d
ata.
I
n
itializatio
n
p
ar
am
ete
r
s
o
n
th
e
n
u
m
b
er
o
f
C
NN
lay
er
s
,
L
STM
u
n
its
,
e
n
s
em
b
le
s
ize,
an
d
m
eth
o
d
a
r
e
d
eter
m
in
ed
a
p
p
r
o
p
r
iately
,
a
n
d
th
en
d
ata
p
r
ep
r
o
ce
s
s
in
g
is
p
er
f
o
r
m
ed
t
o
h
an
d
le
m
is
s
in
g
v
alu
es
d
u
r
in
g
tr
ai
n
in
g
.
T
h
e
C
NN
ca
p
tu
r
es
th
e
s
p
atial
p
atter
n
s
,
wh
ile
th
e
L
STM
h
an
d
les
th
e
tem
p
o
r
al
d
ep
e
n
d
en
cie
s
.
T
h
en
,
th
e
C
NN
-
L
STM
m
o
d
el
is
co
m
b
in
ed
,
th
e
f
in
al
m
o
d
el
is
u
s
ed
to
im
p
u
te
th
e
m
is
s
in
g
v
alu
es,
an
d
th
e
p
er
f
o
r
m
a
n
ce
is
ev
alu
ated
u
s
in
g
s
tan
d
ar
d
m
etr
ics.
W
e
f
in
d
th
at
th
e
p
r
ed
ictio
n
ac
cu
r
ac
y
is
co
r
r
elate
d
with
th
e
co
m
p
le
x
ity
o
f
th
e
m
o
d
el
an
d
th
e
n
u
m
b
e
r
o
f
s
p
atio
-
tem
p
o
r
al
f
ea
t
u
r
es
u
s
ed
.
T
h
e
m
eth
o
d
p
r
o
p
o
s
ed
i
n
th
is
p
ap
er
h
as
a
m
u
ch
h
ig
h
er
p
r
o
p
o
r
tio
n
o
f
ac
c
u
r
a
cy
u
n
d
er
d
y
n
am
ic
tr
af
f
ic
c
o
n
d
itio
n
s
th
a
n
tr
a
d
itio
n
al
r
e
g
r
ess
io
n
-
b
ased
o
r
in
ter
p
o
latio
n
-
b
ased
im
p
u
tatio
n
m
eth
o
d
s
.
4
.
1
.
H
a
nd
lin
g
lo
s
t
da
t
a
Fu
s
io
n
m
o
d
els
u
tili
ze
th
e
ad
v
an
tag
es
o
f
ea
c
h
tech
n
i
q
u
e
t
o
o
v
er
c
o
m
e
th
e
lim
itatio
n
s
o
f
in
d
iv
id
u
al
m
eth
o
d
s
[
7
1
]
.
C
NNs
ca
p
tu
r
e
s
p
atial
p
atter
n
s
i
n
d
ata,
s
u
c
h
as
im
a
g
es
o
r
m
ap
s
,
with
a
two
-
d
im
e
n
s
io
n
al
s
tr
u
ctu
r
e.
T
o
g
en
e
r
ate
a
f
ea
tu
r
e
m
ap
,
C
NNs
u
s
e
co
n
v
o
lu
tio
n
an
d
p
o
o
lin
g
o
p
er
atio
n
s
to
h
ier
ar
ch
ically
ex
tr
ac
t
s
p
atial
f
ea
tu
r
es
b
y
ap
p
ly
in
g
f
ilter
s
o
r
s
m
all
k
er
n
els
to
th
e
in
p
u
t
d
ata
(
e.
g
.
,
im
ag
es
o
r
m
ap
s
)
.
Her
e
is
th
e
eq
u
atio
n
o
f
th
e
c
o
n
v
o
lu
tio
n
o
p
er
atio
n
.
,
=
∑
=
1
∑
=
1
+
−
1
,
+
−
1
.
,
+
(
1
)
,
:
o
u
tp
u
t
f
ea
tu
r
e
m
ap
;
:
in
p
u
t
d
ata;
:f
ilter
;
:b
ias;
an
d
ar
e
t
h
e
f
ilter
s
izes.
T
h
is
o
p
er
ati
o
n
allo
ws
C
NNs
to
r
ec
o
g
n
ize
b
asic
f
ea
tu
r
es
s
u
ch
as
ed
g
es,
co
r
n
er
s
,
o
r
tex
tu
r
es
ac
r
o
s
s
th
e
in
p
u
t
d
ata,
wh
ich
ar
e
th
e
n
co
m
b
in
ed
in
th
e
n
ex
t la
y
er
to
f
o
r
m
a
m
o
r
e
co
m
p
lex
r
e
p
r
esen
t
atio
n
.
Mu
ltip
le
C
NN
-
L
STM
m
o
d
els
ar
e
tr
ai
n
ed
i
n
d
ep
e
n
d
en
tly
with
d
if
f
er
en
t
p
ar
am
eter
v
ar
iatio
n
s
o
r
d
ata
s
u
b
s
ets,
an
d
all
m
o
d
els
'
p
r
ed
ictio
n
s
ar
e
co
m
b
in
ed
.
Fo
r
ex
am
p
le,
p
r
ed
ictio
n
s
ar
e
co
m
b
i
n
ed
b
y
tak
in
g
th
e
av
er
ag
e
o
f
all
p
r
ed
ictio
n
s
(
b
ag
g
in
g
)
o
r
weig
h
tin
g
th
e
p
r
ed
ictio
n
s
b
ased
o
n
m
o
d
el
p
er
f
o
r
m
a
n
ce
(
b
o
o
s
ti
n
g
)
.
Fo
r
ea
ch
m
is
s
in
g
v
alu
e,
th
e
p
r
ed
i
ctio
n
s
f
r
o
m
all
C
NN
-
L
STM
m
o
d
els
in
th
e
en
s
em
b
le
ar
e
c
o
m
b
in
ed
t
o
p
r
o
d
u
ce
th
e
f
in
al
esti
m
ate,
an
d
th
e
m
is
s
in
g
v
alu
e
is
r
ep
lace
d
with
t
h
e
p
r
ed
icted
r
esu
lt f
r
o
m
t
h
e
en
s
e
m
b
le.
4
.
2
.
F
e
a
t
ure
e
x
t
r
a
ct
io
n
Sp
atial
f
ea
tu
r
e
ex
tr
a
ctio
n
C
NNs
p
er
f
o
r
m
s
p
atial
d
ata
p
r
o
c
ess
in
g
,
s
u
ch
as
im
ag
es
o
r
m
ap
s
,
wh
er
e
C
NNs
au
to
m
atica
lly
lear
n
to
d
etec
t
ess
en
tial
p
atter
n
s
in
th
e
d
ata.
Sp
atial
d
ata
s
u
ch
as
im
ag
es,
m
ap
s
,
o
r
r
o
ad
n
etwo
r
k
g
r
id
s
ar
e
r
ep
r
esen
te
d
as
two
-
d
im
en
s
io
n
al
o
r
th
r
ee
-
d
im
en
s
io
n
al
m
a
tr
ices,
s
u
ch
as
ch
an
n
els
f
o
r
R
GB
im
ag
es.
E
ac
h
elem
en
t
in
th
is
m
atr
ix
r
ep
r
esen
ts
ce
r
tain
in
f
o
r
m
atio
n
,
s
u
ch
as
p
ix
el
in
te
n
s
ity
in
th
e
im
ag
e
o
r
a
s
p
ec
if
ic
v
alu
e
in
th
e
g
r
id
.
L
S
T
M
tem
p
o
r
al
f
ea
tu
r
e
ex
tr
ac
ti
o
n
is
u
s
ed
to
ca
p
tu
r
e
p
atter
n
s
in
s
eq
u
e
n
tial
d
ata,
s
u
ch
as
tim
e
s
eq
u
en
ce
s
,
wh
ich
ar
e
ess
en
tial
in
m
an
y
ap
p
li
ca
tio
n
s
,
s
u
ch
as
tim
e
s
er
ies
p
r
ed
ictio
n
,
wea
th
er
an
aly
s
is
,
an
d
in
tellig
en
t
tr
an
s
p
o
r
tatio
n
s
y
s
tem
s
.
T
h
e
in
p
u
t
o
f
L
STM
is
u
s
u
ally
a
s
eq
u
en
ce
o
f
d
ata
with
s
p
ec
if
ic
d
im
en
s
io
n
s
as in
(
2
)
:
=
{
1
,
2
,
3
,
…
,
}
(
2
)
W
h
er
e
is
th
e
len
g
th
o
f
th
e
tim
e
s
eq
u
en
ce
.
T
h
e
L
STM
m
ec
h
an
is
m
at
th
e
f
o
r
g
et
g
ate
r
eg
u
lates
h
o
w
m
u
ch
in
f
o
r
m
atio
n
f
r
o
m
th
e
p
r
ev
io
u
s
tim
e
s
tep
ℎ
−
1
will b
e
f
o
r
g
o
tten
b
y
th
e
m
em
o
r
y
ce
ll
.
=
(
.
[
ℎ
−
1
,
]
+
)
(
3
)
T
h
e
in
p
u
t
g
ate
co
n
tai
n
s
h
o
w
m
u
ch
n
ew
i
n
f
o
r
m
atio
n
t w
ill b
e
ad
d
ed
t
o
th
e
m
em
o
r
y
ce
ll.
=
(
.
[
ℎ
−
1
,
]
+
)
=
ℎ
(
.
[
ℎ
−
1
,
]
+
)
(
4
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
5
0
2
-
4
7
52
In
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
3
8
,
No
.
2
,
May
20
2
5
:
7
9
4
-
8
0
7
802
T
o
u
p
d
ate
th
e
ce
ll
s
tate,
th
e
m
em
o
r
y
ce
ll
is
u
p
d
ated
b
y
co
m
b
in
in
g
th
e
o
ld
in
f
o
r
m
atio
n
f
ilter
ed
b
y
th
e
f
o
r
g
e
t
g
ate
an
d
th
e
n
ew
in
f
o
r
m
atio
n
s
elec
ted
b
y
th
e
in
p
u
t g
ate.
=
.
−
1
+
.
(
5
)
T
h
e
o
u
tp
u
t g
ate
d
ete
r
m
in
es th
e
cu
r
r
en
t
o
u
tp
u
t b
ased
o
n
th
e
u
p
d
ated
m
em
o
r
y
ce
ll in
f
o
r
m
at
io
n
.
=
(
.
[
ℎ
−
1
,
]
+
ℎ
=
.
ta
n
h
(
)
(
6
)
W
h
er
e
is
th
e
in
p
u
t
at
tim
e
;
ℎ
−
1
is
th
e
h
id
d
en
s
tate
f
r
o
m
th
e
p
r
ev
io
u
s
tim
e;
is
th
e
ce
ll
s
tate
at
tim
e
;
,
,
,
ar
e
th
e
lear
n
e
d
weig
h
ts
;
,
,
,
is
th
e
b
ias;
is
th
e
s
ig
m
o
id
f
u
n
ctio
n
,
an
d
ta
n
h
is
th
e
ac
tiv
ity
f
u
n
ctio
n
ta
n
h
.
4
.
3
.
F
uzzy
m
et
ho
d
Fu
zz
y
th
eo
r
y
allo
ws
m
o
d
elin
g
u
n
ce
r
tain
ty
with
f
u
zz
y
r
u
le
s
th
at
ca
n
f
lex
ib
ly
r
ep
r
esen
t
k
n
o
wled
g
e,
s
u
ch
as
g
en
er
al
tr
af
f
ic
p
atter
n
s
,
r
o
ad
u
s
er
h
ab
its
,
o
r
v
eh
icle
m
o
v
em
en
t
p
atter
n
s
[
7
2
]
.
Fo
r
i
n
co
m
p
lete
d
ata
d
u
e
to
d
am
a
g
ed
s
en
s
o
r
s
,
o
u
tag
es,
o
r
s
y
s
tem
f
ailu
r
es,
f
u
zz
y
m
et
h
o
d
s
ca
n
p
r
o
v
i
d
e
a
m
o
r
e
ac
c
u
r
ate
ap
p
r
o
ac
h
th
a
n
d
eter
m
in
is
tic
o
n
es
[
7
3
]
u
s
es
g
eo
g
r
ap
h
ic
in
f
o
r
m
atio
n
f
o
r
s
p
atial
d
ata
th
r
o
u
g
h
m
o
v
e
m
e
n
t
p
atter
n
s
,
tr
a
f
f
ic
d
is
tr
ib
u
tio
n
,
an
d
tem
p
o
r
al
d
at
a
to
ca
p
tu
r
e
tr
en
d
s
an
d
tim
e
p
atter
n
s
af
f
ec
tin
g
t
r
af
f
ic
co
n
d
itio
n
s
.
T
h
e
f
u
zz
y
-
s
p
atial
-
tem
p
o
r
al
m
o
d
el
is
u
s
ed
to
d
ev
elo
p
a
h
y
b
r
id
m
o
d
el
t
h
at
co
m
b
i
n
es
f
u
zz
y
th
e
o
r
y
wi
th
s
p
atial
-
tem
p
o
r
al
an
aly
s
is
m
eth
o
d
s
f
o
r
d
ata
im
p
u
tatio
n
.
T
h
e
m
o
d
el
ca
n
d
y
n
am
ically
ad
ju
s
t
to
tr
af
f
ic
co
n
d
itio
n
s
an
d
o
th
e
r
ex
ter
n
al
f
ac
to
r
s
[
7
4
]
.
Desp
ite
its
m
in
im
al
u
s
ef
u
ln
ess
,
th
e
au
th
o
r
ar
g
u
es
th
at
th
is
m
eth
o
d
i
s
wo
r
th
m
en
tio
n
in
g
b
ec
au
s
e
tr
af
f
ic
d
ata
te
n
d
s
to
b
e
im
p
r
ec
is
e
d
u
e
to
m
an
y
ex
te
r
n
al
v
ar
iab
les,
a
n
d
f
u
zz
y
th
e
o
r
y
ca
n
im
p
r
o
v
e
th
e
p
er
f
o
r
m
an
ce
o
f
o
th
e
r
m
o
d
els in
a
h
y
b
r
id
s
ettin
g
,
as sh
o
wn
b
y
th
e
ab
o
v
e
r
esear
ch
.
4
.
4
.
Cha
lleng
e
I
n
Fig
u
r
e
1
,
r
o
ad
an
d
en
v
ir
o
n
m
en
tal
co
n
d
itio
n
s
r
elate
d
to
d
ata
av
ailab
ilit
y
an
d
i
n
f
r
astru
ct
u
r
e
ar
e
th
e
m
ain
o
b
jects
o
f
d
ata
av
ailab
il
ity
in
m
o
d
er
n
tr
an
s
p
o
r
tatio
n
.
T
h
e
ex
is
tin
g
tr
a
n
s
p
o
r
tatio
n
in
f
r
astru
ctu
r
e
m
ay
b
e
in
ad
eq
u
ate
to
h
an
d
le
th
e
g
r
o
win
g
v
o
lu
m
e
o
f
v
eh
icles,
lead
in
g
to
c
o
n
g
esti
o
n
an
d
ac
cid
en
ts
,
wh
ich
ar
e
s
till
s
ig
n
if
ican
t p
r
o
b
lem
s
in
m
o
d
er
n
tr
an
s
p
o
r
tatio
n
s
y
s
tem
s
.
4
.
4
.
1
.
Da
t
a
l
im
it
a
t
i
o
ns
Data
s
ets
co
llected
f
r
o
m
I
T
S
s
y
s
tem
s
ar
e
o
n
ly
s
o
m
etim
es
co
m
p
lete
d
u
e
to
s
en
s
o
r
lim
itatio
n
s
,
s
ig
n
al
in
ter
f
er
en
ce
,
o
r
tech
n
ical
er
r
o
r
s
th
a
t
af
f
ec
t
d
ata
q
u
ality
[
7
5
]
.
Data
th
at
is
n
o
t
r
ea
l
-
tim
e
h
as
im
p
licatio
n
s
f
o
r
d
elay
s
in
d
ata
p
r
o
ce
s
s
in
g
an
d
co
llectio
n
,
r
esu
ltin
g
in
s
lo
w
r
esp
o
n
s
e
to
r
ap
id
ly
ch
a
n
g
in
g
tr
af
f
ic
co
n
d
itio
n
s
.
T
h
e
Hy
b
r
i
d
C
NN
-
L
STM
E
n
s
em
b
le
m
eth
o
d
is
an
ap
p
r
o
ac
h
th
at
co
m
b
in
es
C
NN
an
d
L
ST
M
to
h
an
d
le
s
p
atial
-
tem
p
o
r
al
d
ata
ef
f
ec
tiv
ely
.
W
h
en
th
er
e
is
lim
ited
d
ata
av
ailab
le,
th
is
m
eth
o
d
ca
n
m
ax
im
ize
in
f
o
r
m
atio
n
u
tili
za
tio
n
b
y
lev
er
ag
i
n
g
t
h
e
s
tr
en
g
th
s
o
f
ea
c
h
m
o
d
el.
C
NNs
in
th
is
m
eth
o
d
ca
n
u
tili
ze
tr
an
s
f
er
lear
n
in
g
b
y
u
s
in
g
p
r
e
-
tr
ain
ed
m
o
d
els
tr
ai
n
ed
o
n
s
im
ilar
lar
g
e
d
atasets
to
s
tr
en
g
th
en
s
p
atial
f
ea
tu
r
es
.
At
th
e
s
am
e
tim
e
,
L
STM
s
ca
n
u
tili
ze
tr
an
s
f
er
le
ar
n
in
g
i
f
p
r
e
-
tr
ain
ed
m
o
d
els
ar
e
av
ailab
le
f
o
r
t
h
e
r
elev
a
n
t
t
em
p
o
r
al
d
ata
ty
p
e
.
E
n
s
em
b
le
lear
n
in
g
co
m
b
in
es
m
u
ltip
le
in
d
ep
en
d
en
tly
tr
ain
ed
C
NN
-
L
STM
m
o
d
els
with
d
if
f
er
en
t
in
itializatio
n
s
o
r
d
if
f
er
en
t
s
u
b
s
ets
o
f
d
ata.
T
h
is
ap
p
r
o
ac
h
im
p
r
o
v
es
th
e
m
o
d
el'
s
r
eliab
ili
t
y
an
d
ac
cu
r
ac
y
b
y
r
ed
u
cin
g
b
ias an
d
v
ar
ian
ce
,
es
p
ec
i
ally
u
n
d
e
r
d
ata
lim
itatio
n
s
.
4
.
4
.
2
.
Da
t
a
s
o
urce
I
T
S
r
elies
o
n
v
a
r
io
u
s
d
ata
s
o
u
r
ce
s
to
o
p
tim
ize
tr
a
n
s
p
o
r
tatio
n
s
y
s
tem
s
,
an
d
th
e
s
en
s
o
r
s
u
s
ed
ca
n
v
ar
y
in
ty
p
e
an
d
s
co
p
e
[
7
6
]
.
So
m
e
s
en
s
o
r
s
o
n
l
y
m
ea
s
u
r
e
tr
af
f
i
c
d
ata
o
n
r
o
a
d
s
,
wh
ile
o
th
er
s
in
clu
d
e
d
ata
f
r
o
m
p
u
b
lic
tr
an
s
p
o
r
tatio
n
o
r
o
th
er
m
o
d
es
o
f
tr
an
s
p
o
r
tatio
n
.
Data
q
u
ality
lim
itatio
n
s
,
s
u
ch
a
s
ir
r
eg
u
lar
ities
an
d
n
o
is
e,
ca
n
co
m
p
licate
th
e
im
p
u
tatio
n
p
r
o
ce
s
s
an
d
m
ak
e
th
e
im
p
u
tatio
n
r
esu
lts
less
a
cc
u
r
ate
[
7
7
]
,
[
7
8
]
.
Ur
b
an
ar
ea
s
m
ay
h
av
e
d
en
s
er
s
en
s
o
r
n
etwo
r
k
s
co
m
p
ar
ed
to
r
u
r
a
l
ar
ea
s
.
T
h
is
m
ay
lead
to
an
im
b
alan
ce
in
th
e
av
ailab
ilit
y
o
f
s
p
atial
-
tem
p
o
r
a
l d
ata.
Data
o
b
tain
ed
f
r
o
m
s
en
s
o
r
s
m
ay
h
av
e
p
r
iv
ac
y
an
d
o
w
n
er
s
h
ip
r
estrictio
n
s
th
at
af
f
ec
t
its
ac
ce
s
s
ib
ilit
y
an
d
u
s
e
f
o
r
im
p
u
tatio
n
p
u
r
p
o
s
es
[
7
9
]
.
T
h
e
lim
itatio
n
s
o
f
th
e
f
o
r
m
at
an
d
s
tr
u
ctu
r
e
o
f
d
ata
o
b
tain
ed
f
r
o
m
v
ar
io
u
s
s
en
s
o
r
s
ca
n
b
e
d
if
f
e
r
en
t
f
o
r
m
ats
an
d
s
tr
u
ct
u
r
es
[
8
0
]
.
L
ar
g
e
v
o
lu
m
es
o
f
s
p
atio
-
tem
p
o
r
al
d
ata
r
eq
u
i
r
e
lar
g
e
s
to
r
ag
e
a
n
d
p
r
o
ce
s
s
i
n
g
ca
p
ac
ities
,
an
d
th
ese
lim
itatio
n
s
ca
n
h
in
d
er
th
e
ab
ilit
y
to
s
to
r
e
an
d
p
r
o
ce
s
s
d
ata
ef
f
icien
tly
[
8
1
]
.
Acc
o
r
d
in
g
t
o
[
8
2
]
,
[
8
3
]
,
d
ata
s
ec
u
r
ity
lim
itatio
n
s
an
d
tr
an
s
p
o
r
tatio
n
d
ata
ar
e
s
en
s
itiv
e
to
p
r
iv
ac
y
s
ec
u
r
ity
,
s
o
th
ey
m
u
s
t
b
e
p
r
o
tecte
d
f
r
o
m
u
n
au
t
h
o
r
ized
ac
ce
s
s
.
T
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esian
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
S
p
a
tia
l
-
temp
o
r
a
l
d
a
ta
imp
u
ta
t
io
n
fo
r
p
r
ed
ictive
mo
d
elin
g
…
(
Yo
h
a
n
es P
r
a
co
y
o
Wid
i P
r
a
s
etyo
)
803
h
y
b
r
id
C
NN
-
L
STM
en
s
em
b
le
m
eth
o
d
ca
n
b
e
ad
ap
ted
t
o
h
a
n
d
le
d
if
f
er
e
n
t
d
ata
in
f
o
r
m
,
ty
p
e,
an
d
s
o
u
r
ce
.
T
h
is
ap
p
r
o
ac
h
in
v
o
lv
es
th
e
c
o
m
b
i
n
atio
n
o
f
C
NN
an
d
L
STM
i
n
an
e
n
s
em
b
le
f
r
am
ew
o
r
k
t
o
m
ax
im
ize
m
o
d
el
p
er
f
o
r
m
an
ce
wh
en
wo
r
k
in
g
w
ith
d
if
f
er
en
t
ty
p
es
o
f
d
ata,
s
u
c
h
as
s
p
atial,
tem
p
o
r
al,
o
r
a
co
m
b
in
atio
n
o
f
b
o
th
.
I
n
th
is
h
y
b
r
id
a
r
ch
itectu
r
e
,
C
NN
ex
tr
ac
ts
s
p
atial
f
ea
tu
r
es
f
r
o
m
d
if
f
er
e
n
t
d
ata
.
Af
ter
s
p
atia
l
f
ea
tu
r
e
ex
tr
ac
tio
n
,
L
STM
is
u
s
ed
to
ex
tr
ac
t
te
m
p
o
r
al
f
ea
tu
r
es
f
r
o
m
th
e
d
at
a
th
at
C
NN
h
as
p
r
o
ce
s
s
ed
.
L
STM
ca
p
tu
r
es
t
h
e
r
elatio
n
s
h
ip
b
etwe
en
tim
e
an
d
em
er
g
in
g
s
p
atial
p
atter
n
s
.
E
n
s
em
b
le
lear
n
in
g
co
m
b
in
es
p
r
ed
ictio
n
s
f
r
o
m
d
if
f
er
en
t
C
NN
-
L
STM
m
o
d
els
th
at
m
ay
b
e
tr
ain
e
d
i
n
d
ep
en
d
en
tly
o
n
d
if
f
e
r
en
t
d
ata
ty
p
es.
T
ec
h
n
iq
u
es
s
u
ch
as
v
o
tin
g
,
a
v
er
ag
in
g
,
o
r
s
tack
in
g
ca
n
b
e
ap
p
lied
to
p
r
o
d
u
ce
m
o
r
e
ac
cu
r
ate
f
in
al
p
r
ed
ictio
n
s
.
4
.
4
.
3
.
Sca
la
bil
it
y
a
nd
r
esil
ience
Scalab
ilit
y
r
ef
er
s
to
th
e
s
y
s
tem
'
s
ab
ili
ty
to
h
an
d
le
in
cr
ea
s
i
n
g
v
o
l
u
m
es
o
f
d
ata
with
o
u
t
ex
p
er
ien
cin
g
p
er
f
o
r
m
an
ce
d
eg
r
a
d
atio
n
,
g
iv
en
th
e
g
r
o
wth
in
th
e
n
u
m
b
e
r
o
f
u
s
er
s
,
d
ata
tr
af
f
ic,
an
d
s
er
v
ic
e
r
eq
u
ests
[
8
4
]
.
T
o
ac
h
iev
e
s
ca
lab
ilit
y
,
th
e
I
T
S sy
s
tem
ar
ch
itectu
r
e
s
h
o
u
ld
b
e
d
e
s
ig
n
ed
co
n
s
id
er
in
g
d
is
tr
ib
u
ted
ar
ch
itectu
r
e,
clo
u
d
co
m
p
u
tin
g
tech
n
o
lo
g
y
,
an
d
s
u
f
f
icien
t
n
etwo
r
k
ca
p
ac
ity
[
8
5
]
.
E
ar
l
y
d
etec
tio
n
tech
n
o
lo
g
i
es
an
d
m
an
a
g
em
en
t
s
y
s
tem
s
ar
e
im
p
o
r
tan
t
to
h
elp
r
esp
o
n
d
q
u
ick
l
y
to
d
is
r
u
p
tio
n
s
o
r
in
cid
en
ts
in
th
e
tr
an
s
p
o
r
tatio
n
s
y
s
tem
[
8
6
]
.
I
n
teg
r
atin
g
n
ew
tech
n
o
lo
g
ies,
s
u
ch
as
I
o
T
,
b
ig
d
ata
a
n
aly
tics
,
an
d
AI
,
ca
n
h
elp
im
p
r
o
v
e
th
e
s
ca
lab
ilit
y
an
d
r
esil
ien
ce
o
f
I
T
S
s
y
s
tem
s
b
y
en
ab
lin
g
r
ea
l
-
tim
e
m
o
n
ito
r
in
g
,
p
r
e
d
ictiv
e
an
aly
tics
,
a
n
d
m
o
r
e
ef
f
icie
n
t
ce
n
tr
alize
d
m
an
ag
e
m
en
t
[
8
7
]
,
[
8
8
]
.
Hy
b
r
id
C
NN
-
L
STM
ca
n
b
e
im
p
lem
en
ted
u
s
in
g
p
ar
alle
l
p
r
o
ce
s
s
in
g
.
C
NN
an
d
L
STM
ca
n
r
u
n
in
d
ep
en
d
en
tly
o
n
d
if
f
er
e
n
t
d
ata
b
ef
o
r
e
co
m
b
i
n
in
g
th
eir
o
u
t
p
u
ts
in
t
h
e
en
s
em
b
le
s
tag
e.
T
h
is
m
eth
o
d
ca
n
ef
f
icien
tly
p
r
o
ce
s
s
lar
g
e
v
o
lu
m
es
o
f
d
ata
b
y
u
t
ilizin
g
p
ar
allel
co
m
p
u
tin
g
,
th
u
s
im
p
r
o
v
in
g
s
ca
lab
ilit
y
.
I
n
th
is
s
ce
n
ar
io
,
e
x
ten
s
iv
e
d
ata
ca
n
b
e
d
iv
id
e
d
in
to
s
ev
er
al
p
a
r
ts
an
d
p
r
o
ce
s
s
ed
an
d
d
is
tr
ib
u
ted
ac
r
o
s
s
m
u
ltip
le
n
o
d
es,
s
ig
n
if
ic
an
tly
im
p
r
o
v
in
g
t
h
e
m
o
d
el'
s
a
b
ilit
y
to
h
an
d
le
lar
g
e
-
s
ca
le
d
at
a
.
R
o
b
u
s
tn
ess
i
s
im
p
r
o
v
ed
th
r
o
u
g
h
en
s
em
b
le
lear
n
in
g
,
wh
er
e
m
u
ltip
le
C
NN
-
L
STM
m
o
d
els
ar
e
tr
ain
ed
in
d
ep
en
d
en
tly
,
an
d
th
e
r
esu
lts
ar
e
co
m
b
in
ed
.
B
y
co
m
b
in
i
n
g
p
r
ed
ictio
n
s
f
r
o
m
m
u
ltip
le
m
o
d
els,
e
n
s
em
b
le
lear
n
in
g
r
ed
u
ce
s
t
h
e
r
is
k
o
f
m
o
d
el
f
ailu
r
e
ca
u
s
ed
b
y
d
ata
o
u
tl
ier
s
o
r
n
o
is
e.
T
ec
h
n
iq
u
e
s
s
u
ch
as
iter
ativ
e
im
p
u
tatio
n
to
f
ill
in
m
is
s
in
g
d
ata
o
r
s
p
atial/tem
p
o
r
al
f
ilter
s
t
o
r
em
o
v
e
n
o
is
e
ca
n
b
e
ap
p
lie
d
,
en
s
u
r
in
g
th
at
th
e
m
o
d
el
r
em
ain
s
ac
cu
r
ate
d
esp
ite
p
r
o
b
lem
s
with
th
e
d
ata.
Usi
n
g
in
cr
em
en
tal
lear
n
in
g
o
r
f
in
e
-
tu
n
in
g
ap
p
r
o
ac
h
e
s
to
ad
ju
s
t
th
e
m
o
d
el
b
ased
o
n
th
e
latest
d
ata
en
s
u
r
es
th
a
t
p
r
ed
ictio
n
s
r
em
ain
ac
cu
r
ate
d
esp
ite
ch
an
g
in
g
en
v
ir
o
n
m
en
tal
co
n
d
itio
n
s
.
T
ab
le
3
s
h
o
ws th
e
r
esu
lts
o
f
th
e
d
is
cu
s
s
io
n
o
f
th
e
v
ar
i
o
u
s
m
eth
o
d
s
u
s
ed
.
T
ab
le
3
.
R
esu
lts
o
f
d
is
cu
s
s
io
n
o
n
m
eth
o
d
ch
a
r
ac
ter
is
tics
M
e
t
h
o
d
s
S
t
r
e
n
g
t
h
W
e
a
k
n
e
ss
A
p
p
l
i
c
a
t
i
o
n
Ef
f
e
c
t
i
v
e
n
e
ss
S
t
a
t
i
st
i
c
s
−
S
i
mp
l
e
a
n
d
e
a
s
y
t
o
i
mp
l
e
m
e
n
t
,
e
f
f
i
c
i
e
n
t
o
n
c
o
m
p
u
t
i
n
g
r
e
s
o
u
r
c
e
s
,
i
n
t
e
r
p
r
e
t
a
b
i
l
i
t
y
,
a
n
d
st
r
u
c
t
u
r
e
d
d
a
t
a
.
−
Li
mi
t
e
d
t
o
l
i
n
e
a
r
r
e
l
a
t
i
o
n
s
h
i
p
s
,
l
e
ss
f
l
e
x
i
b
l
e
,
l
e
ss
a
d
a
p
t
i
v
e
.
−
W
o
r
k
s
w
e
l
l
w
h
e
n
t
h
e
d
a
t
a
p
a
t
t
e
r
n
i
s
si
mp
l
e
a
n
d
l
i
n
e
a
r
w
h
i
l
e
p
r
e
d
i
c
t
i
n
g
i
n
t
h
e
sh
o
r
t
-
t
e
r
m
.
−
Ef
f
e
c
t
i
v
e
o
n
l
y
o
n
s
i
mp
l
e
v
a
r
i
a
b
l
e
s
a
n
d
st
r
u
c
t
u
r
e
d
l
i
n
e
a
r
d
a
t
a
.
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
−
A
b
l
e
t
o
h
a
n
d
l
e
c
o
mp
l
e
x
a
n
d
d
y
n
a
m
i
c
d
a
t
a
,
m
o
r
e
a
d
a
p
t
i
v
e
t
o
c
h
a
n
g
e
s
i
n
d
a
t
a
p
a
t
t
e
r
n
s
,
e
f
f
e
c
t
i
v
e
w
i
t
h
l
a
r
g
e
-
sc
a
l
e
d
a
t
a
,
a
n
d
m
o
r
e
a
c
c
u
r
a
t
e
p
r
e
d
i
c
t
i
o
n
r
e
s
u
l
t
s.
−
I
t
r
e
q
u
i
r
e
s
e
x
t
e
n
si
v
e
a
n
d
h
i
g
h
-
q
u
a
l
i
t
y
d
a
t
a
,
i
s
c
h
a
l
l
e
n
g
i
n
g
t
o
i
n
t
e
r
p
r
e
t
,
r
e
q
u
i
r
e
s
h
i
g
h
c
o
m
p
u
t
a
t
i
o
n
a
l
r
e
so
u
r
c
e
s
,
a
n
d
i
s
p
r
o
n
e
t
o
o
v
e
r
f
i
t
t
i
n
g
.
−
W
o
r
k
s
w
e
l
l
o
n
d
y
n
a
m
i
c
,
l
a
r
g
e
,
a
n
d
c
o
m
p
l
e
x
d
a
t
a
a
n
d
f
o
r
l
o
n
g
-
t
e
r
m
p
r
e
d
i
c
t
i
o
n
s.
−
P
r
e
d
i
c
t
i
o
n
a
c
c
u
r
a
c
y
i
n
I
TS
i
s st
r
o
n
g
e
r
−
Ef
f
e
c
t
i
v
e
l
y
h
a
n
d
l
e
s
c
o
m
p
l
e
x
a
n
d
d
y
n
a
m
i
c
d
a
t
a
.
−
S
u
i
t
a
b
l
e
f
o
r
p
r
o
b
l
e
ms
t
h
a
t
r
e
q
u
i
r
e
a
c
c
u
r
a
t
e
p
r
e
d
i
c
t
i
o
n
s
i
n
d
y
n
a
m
i
c
a
n
d
n
o
n
-
l
i
n
e
a
r
d
a
t
a
F
u
si
o
n
m
o
d
e
l
s
−
U
t
i
l
i
z
e
st
a
t
i
st
i
c
a
l
a
n
d
mac
h
i
n
e
l
e
a
r
n
i
n
g
met
h
o
d
s,
i
mp
r
o
v
e
a
c
c
u
r
a
c
y
,
b
e
f
l
e
x
i
b
l
e
a
n
d
a
d
a
p
t
i
v
e
,
a
n
d
a
v
o
i
d
o
v
e
r
f
i
t
t
i
n
g
.
−
H
i
g
h
c
o
m
p
l
e
x
i
t
y
,
r
e
q
u
i
r
e
s
l
a
r
g
e
c
o
m
p
u
t
i
n
g
r
e
s
o
u
r
c
e
s,
a
n
d
d
i
f
f
i
c
u
l
t
y
i
n
t
u
n
i
n
g
.
−
W
o
r
k
s
w
e
l
l
o
n
c
o
mp
l
e
x
a
n
d
h
e
t
e
r
o
g
e
n
e
o
u
s
d
a
t
a
.
−
P
r
e
d
i
c
t
i
o
n
a
c
c
u
r
a
c
y
a
n
d
mo
d
e
l
r
o
b
u
s
t
n
e
ss
a
c
r
o
ss
d
i
f
f
e
r
e
n
t
t
y
p
e
s
o
f
d
y
n
a
mi
c
d
a
t
a
.
−
M
o
s
t
e
f
f
e
c
t
i
v
e
f
o
r
h
e
t
e
r
o
g
e
n
e
o
u
s
d
a
t
a
.
−
I
t
h
a
s
a
b
a
l
a
n
c
e
o
n
v
a
r
i
o
u
s
d
a
t
a
mo
d
e
l
s
a
n
d
i
s
h
i
g
h
l
y
a
c
c
u
r
a
t
e
i
n
p
r
e
d
i
c
t
i
o
n
.
Ou
r
s
tu
d
y
s
h
o
ws
th
at
h
ig
h
e
r
m
o
d
el
co
m
p
lex
ity
is
n
o
t
ass
o
ciate
d
wi
th
p
o
o
r
p
e
r
f
o
r
m
a
n
ce
.
W
h
en
ap
p
lied
to
lar
g
e
tr
a
f
f
ic
d
atasets
,
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
ca
n
b
en
ef
it
f
r
o
m
in
c
r
ea
s
ed
f
ea
tu
r
es
an
d
s
p
atio
tem
p
o
r
al
v
ar
iab
les
with
o
u
t
ad
v
er
s
ely
af
f
ec
tin
g
co
m
p
u
tatio
n
al
e
f
f
icien
cy
o
r
p
r
ed
ictio
n
ac
cu
r
ac
y
.
T
h
is
s
tu
d
y
ex
p
lo
r
es
a
co
m
p
r
eh
e
n
s
i
v
e
s
p
atio
-
tem
p
o
r
al
d
ata
im
p
u
tatio
n
ap
p
r
o
ac
h
u
s
in
g
v
ar
io
u
s
m
eth
o
d
s
,
in
clu
d
i
n
g
m
ac
h
in
e
lea
r
n
in
g
an
d
co
m
b
in
ed
tech
n
iq
u
es.
Ho
wev
er
,
f
u
r
th
er
in
-
d
e
p
th
s
tu
d
ies
m
ay
b
e
n
ee
d
ed
to
co
n
f
ir
m
th
is
m
eth
o
d
'
s
r
o
b
u
s
tn
ess
an
d
g
en
er
aliza
b
ilit
y
,
esp
ec
ially
r
eg
ar
d
in
g
th
e
in
f
l
u
e
n
ce
o
f
tr
af
f
ic
en
v
ir
o
n
m
en
t v
ar
iatio
n
s
an
d
lo
wer
s
en
s
o
r
d
ata
q
u
ality
o
n
p
r
ed
icti
o
n
p
er
f
o
r
m
a
n
ce
.
Ou
r
r
esear
ch
s
h
o
ws
th
at
C
N
N
-
L
STM
-
b
ased
im
p
u
tatio
n
m
eth
o
d
s
ar
e
m
o
r
e
r
o
b
u
s
t
th
an
tr
ad
itio
n
al
in
ter
p
o
latio
n
m
eth
o
d
s
i
n
th
e
f
ac
e
o
f
d
y
n
am
ic
s
p
atio
tem
p
o
r
al
d
ata
in
co
m
p
leten
ess
.
Fu
tu
r
e
r
esear
ch
ca
n
Evaluation Warning : The document was created with Spire.PDF for Python.