I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
p
u
t
er
Science
Vo
l.
41
,
No
.
2
,
Feb
r
u
ar
y
20
2
6
,
p
p
.
700
~
709
I
SS
N:
2
502
-
4
7
52
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ee
cs
.v
41
.
i
2
.
pp
700
-
7
0
9
700
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ee
cs
.
ia
esco
r
e.
co
m
Predic
ting no
n
-
p
e
rfor
m
ing
loa
ns in
Viet
na
m
’s
f
ina
nc
ia
l sector:
a
deep
Q
-
lea
rni
ng
appro
a
ch
L
uy
en
An
h Do
,
H
uo
ng
T
hi
Viet
P
ha
m
,
T
hin
h Duc
L
e,
O
a
nh
T
hi
T
ra
n
I
n
t
e
r
n
a
t
i
o
n
a
l
S
c
h
o
o
l
,
V
i
e
t
n
a
m
N
a
t
i
o
n
a
l
U
n
i
v
e
r
si
t
y
,
H
a
n
o
i
,
V
i
e
t
n
a
m
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Sep
4
,
2025
R
ev
i
s
ed
J
an
3
,
2
0
2
6
A
cc
ep
ted
J
an
11
,
2
0
2
6
No
n
-
p
e
rf
o
rm
in
g
lo
a
n
s
(NP
L
s)
p
re
d
ictio
n
is
a
v
e
r
y
i
m
p
o
rtan
t
tas
k
in
risk
m
a
n
a
g
e
m
e
n
t
o
f
f
in
a
n
c
ial
in
stit
u
ti
o
n
s.
N
P
L
s
o
f
ten
lea
d
t
o
s
u
b
sta
n
t
ial
lo
ss
e
s
w
h
e
n
lo
a
n
s
a
re
n
o
t
p
a
i
d
b
a
c
k
o
n
ti
m
e
.
W
h
il
e
trad
it
io
n
a
l
m
a
c
h
in
e
lea
rn
in
g
(M
L
)
m
o
d
e
ls
h
a
v
e
b
e
e
n
c
o
n
v
e
n
ti
o
n
a
ll
y
e
x
p
lo
it
e
d
f
o
r
c
re
d
it
risk
a
ss
e
ss
m
e
n
t,
th
e
y
f
re
q
u
e
n
tl
y
fa
c
e
c
h
a
ll
e
n
g
e
s
with
h
a
n
d
li
n
g
im
b
a
lan
c
e
d
d
a
ta.
T
o
d
e
a
l
w
it
h
th
is
p
ro
b
lem
,
th
is
p
a
p
e
r
in
tr
o
d
u
c
e
s
a
n
o
v
e
l
a
p
p
ro
a
c
h
u
s
in
g
d
e
e
p
re
in
f
o
rc
e
m
e
n
t
lea
rn
in
g
(DRL
),
sp
e
c
if
ica
ll
y
d
e
e
p
Q
-
l
e
a
rn
in
g
,
to
e
n
h
a
n
c
e
t
h
e
p
re
d
ictio
n
o
f
N
P
L
s.
T
o
v
e
rify
th
e
e
ff
e
c
ti
v
e
n
e
ss
o
f
th
e
m
e
th
o
d
,
w
e
in
tro
d
u
c
e
a
n
e
w
d
a
tas
e
t
c
o
m
p
risin
g
8
3
,
7
3
2
c
u
s
to
m
e
r
r
e
c
o
rd
s
(e
a
c
h
d
e
sc
rib
e
d
w
it
h
2
2
k
e
y
f
e
a
tu
re
s)
f
ro
m
o
n
e
o
f
V
iet
n
a
m
'
s
larg
e
st
f
in
a
n
c
ial
e
n
ti
ti
e
s.
Ou
r
m
e
th
o
d
is
c
o
m
p
a
re
d
w
it
h
sta
n
d
a
rd
ML
tec
h
n
iq
u
e
s
su
c
h
a
s
ra
n
d
o
m
f
o
re
st,
d
e
c
isio
n
tree
,
lo
g
isti
c
re
g
re
ss
io
n
,
su
p
p
o
r
t
v
e
c
to
r
m
a
c
h
in
e
,
L
ig
h
tG
BM
,
a
n
d
X
G
Bo
o
s
t.
Ex
p
e
rime
n
tal
re
su
lt
s
o
n
t
h
is
d
a
tas
e
t
d
e
m
o
n
stra
te
th
a
t
d
e
e
p
Q
-
l
e
a
rn
in
g
o
u
t
p
e
rf
o
rm
s
th
e
se
trad
it
i
o
n
a
l
m
o
d
e
ls
in
h
a
n
d
li
n
g
im
b
a
lan
c
e
d
d
a
ta
a
n
d
b
o
o
st
in
g
p
re
d
icti
o
n
a
c
c
u
ra
c
y
.
T
h
is
re
se
a
rc
h
h
ig
h
l
ig
h
ts
t
h
e
p
o
te
n
ti
a
l
o
f
DRL
a
s
a
ro
b
u
st
r
isk
m
a
n
a
g
e
m
e
n
t
to
o
l
,
h
e
lp
in
g
f
in
a
n
c
ial
in
sti
tu
ti
o
n
s
m
a
k
e
c
re
d
it
a
ss
e
ss
m
e
n
ts
m
o
re
e
ff
ici
e
n
tl
y
a
n
d
re
d
u
c
in
g
d
e
c
isio
n
-
m
a
k
in
g
c
o
sts.
K
ey
w
o
r
d
s
:
Dee
p
Q
-
l
ea
r
n
i
n
g
Ma
ch
i
n
e
lear
n
i
n
g
No
n
-
p
er
f
o
r
m
i
n
g
lo
a
n
s
R
ein
f
o
r
ce
m
e
n
t le
ar
n
in
g
R
is
k
m
a
n
a
g
e
m
en
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
:
Oan
h
T
h
i T
r
an
I
n
ter
n
atio
n
al
Sc
h
o
o
l,
Vietn
a
m
Natio
n
al
Un
iv
er
s
it
y
1
4
4
Xu
an
T
h
u
y
Stre
et,
C
au
Gi
a
y
W
ar
d
,
Han
o
i
,
Vietn
a
m
E
m
ail:
tr
a
n
th
io
an
h
@
v
n
u
.
ed
u
.
v
n
1.
I
NT
RO
D
UCT
I
O
N
No
n
-
p
er
f
o
r
m
i
n
g
lo
a
n
(
NP
L
)
is
a
lo
an
w
h
er
e
t
h
e
b
o
r
r
o
w
er
h
a
s
n
o
t
p
aid
b
ac
k
p
r
in
cip
al
an
d
i
n
ter
est
f
o
r
at
least
9
0
d
ay
s
.
Un
f
o
r
tu
n
atel
y
,
th
er
e
i
s
n
o
s
ta
n
d
ar
d
w
a
y
to
id
en
ti
f
y
b
ad
lo
an
s
i
n
ad
v
a
n
ce
.
C
u
r
r
en
tl
y
,
i
n
m
o
s
t
f
i
n
an
cia
l a
n
d
b
an
k
i
n
g
co
m
p
an
ies,
th
e
co
n
v
e
n
tio
n
al
ap
p
r
o
ac
h
to
th
e
NP
L
s
p
r
ed
ictio
n
p
r
o
b
le
m
is
to
u
s
e
th
e
r
i
s
k
m
o
d
eli
n
g
m
et
h
o
d
.
I
n
th
is
ap
p
r
o
ac
h
,
th
e
y
f
ir
s
t
clea
n
an
d
p
r
ep
ar
e
th
e
d
ata,
th
en
ap
p
ly
tr
ad
itio
n
al
s
tatis
tical
m
et
h
o
d
s
lik
e
w
ei
g
h
t o
f
e
v
id
en
ce
an
d
lo
g
is
tic
r
e
g
r
ess
io
n
,
an
d
f
in
a
ll
y
t
u
r
n
t
h
e
r
es
u
lt
s
in
to
a
c
r
ed
it sco
r
e
.
A
lo
n
g
w
i
th
t
h
e
d
ev
elo
p
m
e
n
t
o
f
m
ac
h
i
n
e
lear
n
i
n
g
(
M
L
)
an
d
ar
tif
icial
in
telli
g
en
ce
(
AI
)
,
tr
ad
itio
n
al
s
tatis
t
ical
m
et
h
o
d
s
s
h
o
w
i
n
e
f
f
icien
c
y
i
n
m
a
k
i
n
g
f
o
r
ec
asts
a
n
d
p
r
ed
ictio
n
s
in
th
e
d
ec
is
io
n
-
m
ak
i
n
g
p
r
o
ce
s
s
f
o
r
b
u
s
i
n
ess
e
s
.
B
ec
au
s
e
o
f
t
h
at,
t
h
er
e
h
a
v
e
b
ee
n
a
n
u
m
b
er
o
f
r
esear
ch
p
ap
er
s
o
n
t
h
e
ap
p
licatio
n
o
f
M
L
i
n
N
PL
p
r
ed
ictio
n
;
h
o
w
e
v
er
,
th
e
co
m
m
o
n
p
o
in
t
o
f
m
o
s
t
o
f
th
e
s
e
p
ap
er
s
is
th
at
th
e
y
o
n
l
y
f
o
cu
s
o
n
th
e
ap
p
licatio
n
o
f
M
L
al
g
o
r
ith
m
s
to
p
r
ed
ict
th
e
o
u
tp
u
t
w
it
h
o
u
t n
o
tici
n
g
t
h
at
t
h
e
d
ata
in
m
o
s
t f
in
a
n
cial
a
n
d
b
an
k
i
n
g
p
r
o
b
le
m
s
ar
e
f
ac
ed
w
it
h
p
r
o
b
le
m
s
o
f
i
m
b
al
an
ce
d
d
ata
.
T
h
is
m
ea
n
s
t
h
at
t
h
e
n
u
m
b
er
o
f
s
a
m
p
le
s
i
n
o
n
e
g
r
o
u
p
(
li
k
e
ca
n
ce
r
p
atien
ts
)
ca
n
b
e
1
,
0
0
0
ti
m
es
s
m
aller
t
h
a
n
i
n
a
n
o
th
er
g
r
o
u
p
(
l
ik
e
h
ea
lt
h
y
p
atien
t
s
)
,
b
u
t
m
o
s
t
ML
m
e
th
o
d
s
w
o
r
k
b
etter
w
h
e
n
t
h
e
d
ata
i
s
b
al
an
ce
d
.
T
y
p
icall
y
,
i
n
t
h
e
N
PL
p
r
ed
ictio
n
p
r
o
b
le
m
,
t
h
i
s
i
s
a
t
y
p
ical
b
i
n
ar
y
class
i
f
icatio
n
p
r
o
b
le
m
w
it
h
t
wo
class
es,
N
PL
a
n
d
P
L
.
I
n
w
h
ich
t
h
e
n
u
m
b
er
o
f
o
b
s
er
v
atio
n
s
o
f
th
e
N
PL
cla
s
s
(
m
in
o
r
it
y
c
lass
)
is
m
a
n
y
ti
m
e
s
less
t
h
an
t
h
e
n
u
m
b
er
o
f
o
b
s
er
v
atio
n
s
o
f
th
e
p
er
f
o
r
m
i
n
g
lo
an
s
clas
s
(
m
aj
o
r
it
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
P
r
ed
ictin
g
n
o
n
-
p
erfo
r
min
g
lo
a
n
s
in
V
ietn
a
m’
s
fin
a
n
cia
l secto
r
:
a
d
ee
p
…
(
Lu
ye
n
A
n
h
Do
)
701
class
)
,
lead
in
g
to
th
e
p
r
ed
ictio
n
o
f
th
e
m
i
n
o
r
it
y
clas
s
h
a
s
m
a
n
y
er
r
o
r
s
,
w
h
ile
t
h
e
p
r
ed
ictio
n
o
f
th
at
class
i
s
th
e
m
o
s
t i
m
p
o
r
tan
t.
Ov
er
th
e
p
as
t
t
w
o
d
ec
ad
es,
th
er
e
h
a
s
b
ee
n
r
esear
ch
o
n
m
et
h
o
d
s
to
s
o
lv
e
t
h
is
i
m
b
al
an
ce
d
-
d
ata
p
r
o
b
lem
.
T
h
ese
m
et
h
o
d
s
ar
e
d
iv
id
ed
in
to
t
w
o
m
ai
n
tec
h
n
i
q
u
es
:
o
v
er
-
s
a
m
p
li
n
g
(
ad
d
m
o
r
e
s
a
m
p
les
to
t
h
e
s
m
al
ler
g
r
o
u
p
u
n
til
it
h
as
th
e
s
a
m
e
n
u
m
b
er
a
s
t
h
e
b
i
g
g
er
g
r
o
u
p
)
,
u
n
d
er
-
sa
m
p
l
in
g
(
r
ed
u
ce
th
e
s
a
m
p
les
i
n
th
e
b
ig
g
er
g
r
o
u
p
s
o
th
e
y
m
atc
h
,
o
r
a
r
e
f
e
w
er
th
a
n
th
e
s
a
m
p
les
in
th
e
s
m
aller
g
r
o
u
p
)
.
S
y
n
t
h
etic
m
i
n
o
r
it
y
o
v
er
s
a
m
p
li
n
g
tec
h
n
iq
u
e
(
SM
OT
E
)
is
a
t
y
p
ical
a
lg
o
r
it
h
m
u
s
in
g
t
h
e
o
v
er
-
s
a
m
p
li
n
g
tec
h
n
i
q
u
e,
f
ir
s
t
i
n
tr
o
d
u
ce
d
in
2
0
0
2
b
y
[
1
]
.
It
w
o
r
k
s
b
y
cr
ea
tin
g
s
y
n
th
e
tic
s
a
m
p
les
th
a
t
ar
e
s
i
m
i
lar
to
ex
i
s
ti
n
g
m
in
o
r
it
y
clas
s
s
a
m
p
le
s
.
T
o
th
is
e
n
d
,
it
s
elec
t
s
a
r
an
d
o
m
m
i
n
o
r
it
y
clas
s
in
s
ta
n
ce
an
d
f
i
n
d
s
its
k
-
n
ea
r
est
n
ei
g
h
b
o
r
s
(
KNN)
.
I
t
th
en
ch
o
o
s
e
s
o
n
e
o
f
t
h
e
n
eig
h
b
o
r
s
r
an
d
o
m
l
y
,
co
m
p
u
tes
t
h
e
d
i
f
f
er
e
n
ce
b
e
t
w
ee
n
t
h
e
f
ea
t
u
r
e
v
ec
to
r
s
,
a
n
d
m
u
l
tip
lies
i
t
b
y
a
r
an
d
o
m
n
u
m
b
er
b
et
w
ee
n
0
a
n
d
1
.
T
h
is
d
if
f
er
e
n
ce
is
th
e
n
a
d
d
ed
to
th
e
s
elec
ted
m
in
o
r
it
y
in
s
ta
n
ce
to
cr
ea
te
a
n
e
w
s
y
n
t
h
etic
e
x
a
m
p
le.
B
y
cr
ea
tin
g
s
y
n
th
e
tic
e
x
a
m
p
le
s
,
S
MO
T
E
h
elp
s
to
b
alan
ce
t
h
e
p
r
o
p
o
r
ti
o
n
o
f
in
s
tan
ce
s
b
et
w
ee
n
th
e
m
i
n
o
r
it
y
a
n
d
m
aj
o
r
ity
cla
s
s
,
m
ak
in
g
th
e
tr
ai
n
i
n
g
d
ata
m
o
r
e
r
ep
r
esen
tati
v
e.
T
h
is
ca
n
i
m
p
r
o
v
e
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
ML
alg
o
r
it
h
m
s
,
esp
ec
iall
y
in
ca
s
e
s
w
h
er
e
th
e
m
i
n
o
r
it
y
clas
s
h
as
cr
iti
ca
l
in
f
o
r
m
atio
n
an
d
n
ee
d
s
to
b
e
w
ell
-
r
ep
r
esen
ted
[
2
]
.
Nea
r
Miss
i
s
a
t
y
p
ical
al
g
o
r
it
h
m
u
s
i
n
g
t
h
e
un
d
er
-
s
a
m
p
lin
g
t
ec
h
n
iq
u
e
p
r
o
p
o
s
ed
b
y
Ma
n
i
a
n
d
Z
h
a
n
g
[
3
]
.
T
h
e
au
th
o
r
s
o
b
s
er
v
ed
th
a
t
th
e
K
NN
a
lg
o
r
it
h
m
ten
d
s
to
class
i
f
y
e
x
a
m
p
les
f
r
o
m
t
h
e
m
aj
o
r
ity
cla
s
s
m
o
r
e
ac
cu
r
atel
y
t
h
an
t
h
e
m
i
n
o
r
it
y
class
in
i
m
b
a
lan
c
ed
d
atasets
.
T
h
is
is
b
ec
au
s
e
th
e
m
aj
o
r
it
y
class
h
as
a
lar
g
er
r
ep
r
esen
tatio
n
i
n
t
h
e
d
atase
t,
m
ak
in
g
it
m
o
r
e
li
k
el
y
f
o
r
th
e
k
n
ea
r
est
n
ei
g
h
b
o
r
s
to
b
e
m
aj
o
r
ity
clas
s
i
n
s
ta
n
ce
s
.
T
o
ad
d
r
ess
th
is
i
s
s
u
e,
th
e
Nea
r
Miss
al
g
o
r
ith
m
f
o
c
u
s
e
s
o
n
s
el
ec
tin
g
r
ep
r
esen
tati
v
e
ex
a
m
p
le
s
f
r
o
m
t
h
e
m
aj
o
r
it
y
class
t
h
at
ar
e
i
n
clo
s
e
p
r
o
x
i
m
it
y
to
th
e
m
in
o
r
it
y
cla
s
s
in
s
ta
n
ce
s
.
Ho
w
ev
er
,
t
h
e
s
e
m
et
h
o
d
s
also
h
a
v
e
w
ea
k
n
ess
e
s
.
A
s
SM
OT
E
m
a
k
es
tr
ain
i
n
g
v
er
y
e
x
p
en
s
i
v
e
b
ec
au
s
e
w
it
h
lar
g
e
d
ata
s
e
ts
,
i
n
cr
ea
s
in
g
th
e
n
u
m
b
er
o
f
o
b
s
er
v
atio
n
s
o
f
t
h
e
m
in
o
r
it
y
cla
s
s
eq
u
al
to
th
e
m
aj
o
r
it
y
class
w
ill
ca
u
s
e
t
h
e
d
ata
to
b
e
g
r
ea
tl
y
i
n
cr
ea
s
ed
in
s
ize
an
d
ti
m
e
co
n
s
u
m
in
g
to
tr
ain
,
lead
in
g
to
m
e
m
o
r
y
la
k
e
[
4
]
.
A
s
Nea
r
Miss
,
d
eletin
g
t
h
e
o
b
s
er
v
atio
n
s
o
f
th
e
m
aj
o
r
ity
clas
s
w
il
l
ca
u
s
e
th
e
d
ata
to
lo
s
e
a
lo
t
o
f
in
f
o
r
m
a
t
io
n
an
d
lead
to
a
d
ec
r
ea
s
e
in
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
t
h
e
m
o
d
el
[
5
]
.
Dee
p
r
ein
f
o
r
ce
m
e
n
t
lear
n
i
n
g
h
as
b
ee
n
s
u
cc
es
s
f
u
ll
y
u
s
ed
i
n
r
ec
en
t
y
ea
r
s
to
ap
p
l
y
i
n
co
m
p
u
ter
g
a
m
e
s
,
r
o
b
o
t
co
n
tr
o
l,
s
elf
-
d
r
i
v
in
g
c
ar
s
,
an
d
o
th
er
s
y
s
te
m
s
.
Dee
p
r
ein
f
o
r
ce
m
e
n
t
lear
n
i
n
g
h
a
s
g
r
ea
tl
y
i
m
p
r
o
v
e
d
class
i
f
icatio
n
p
er
f
o
r
m
a
n
ce
f
o
r
class
if
icatio
n
i
s
s
u
es
b
y
d
el
etin
g
n
o
is
y
d
ata
an
d
s
t
u
d
y
i
n
g
b
etter
f
ea
t
u
r
es.
A
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
o
f
d
ee
p
r
ein
f
o
r
ce
m
en
t
lear
n
i
n
g
i
s
i
n
d
ee
d
th
e
g
r
ea
t
e
f
f
ec
ti
v
e
m
et
h
o
d
f
o
r
lear
n
i
n
g
f
r
o
m
i
m
b
alan
ce
d
d
ata
b
ec
au
s
e
o
f
h
o
w
ea
s
il
y
it
ca
n
f
o
c
u
s
m
o
r
e
atten
tio
n
o
n
th
e
s
m
aller
cl
ass
b
y
it
s
r
e
w
ar
d
s
f
u
n
ctio
n
o
r
p
en
alt
y
’
s
f
u
n
ct
io
n
[
6
]
.
T
h
e
m
ain
id
ea
o
f
d
ee
p
Q
-
l
ea
r
n
in
g
is
t
h
at
tr
y
to
m
e
m
o
r
y
th
e
p
r
ev
io
u
s
s
t
u
d
y
b
y
u
s
i
n
g
r
ep
lay
b
u
f
f
er
an
d
u
s
e
th
at
m
e
m
o
r
y
f
o
r
tr
ain
i
n
g
,
th
e
ag
en
t
w
ill
i
n
ter
ac
t
w
it
h
en
v
i
r
o
n
m
e
n
t
b
y
ac
tio
n
,
ac
tio
n
w
ill
b
e
d
eter
m
in
ed
b
as
ed
on
p
o
licy
.
E
n
v
ir
o
n
m
en
t
w
i
ll
r
etu
r
n
a
g
e
n
t
r
e
w
ar
d
o
r
p
en
alt
y
i
f
ac
tio
n
is
tr
u
e
o
r
f
alse.
T
h
e
g
o
al
o
f
d
ee
p
Q
-
l
ea
r
n
in
g
is
to
ac
h
iev
e
a
s
m
a
n
y
r
e
w
ar
d
s
as
it c
a
n
.
T
h
e
o
b
j
ec
tiv
e
o
f
o
u
r
s
tu
d
y
is
to
b
e
d
esig
n
ed
as
a
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
to
h
an
d
le
i
m
b
alan
ce
d
d
ata
an
d
pr
ed
ictin
g
NP
L
s
u
s
i
n
g
d
ee
p
Q
-
l
ea
r
n
in
g
alg
o
r
it
h
m
.
W
ith
t
h
e
u
s
e
o
f
th
is
m
et
h
o
d
an
d
a
s
p
ec
ial,
ex
clu
s
i
v
e
d
ataset
o
f
a
Vietn
a
m
ese
le
n
d
i
n
g
s
er
v
ice
co
m
p
a
n
y
,
w
e
ar
e
ab
le
to
m
a
k
e
s
ig
n
i
f
ica
n
t
ad
v
a
n
ce
s
to
th
is
f
ield
o
f
s
tu
d
y
.
O
u
r
ex
p
er
i
m
en
tal
r
es
u
lt
s
s
h
o
w
t
h
at
d
ee
p
Q
-
l
ea
r
n
in
g
s
i
g
n
i
f
ica
n
tl
y
i
m
p
r
o
v
e
s
NP
L
d
etec
tio
n
ac
c
u
r
ac
y
b
y
ef
f
ec
tiv
e
l
y
h
an
d
li
n
g
i
m
b
ala
n
c
ed
d
ata
an
d
lear
n
in
g
o
p
ti
m
al
c
lass
i
f
icatio
n
s
tr
ate
g
ies.
I
n
b
r
ief
,
th
i
s
s
t
u
d
y
h
as
t
h
e
co
n
tr
ib
u
tio
n
s
a
s
f
o
llo
w
s
:
i)
I
n
tr
o
d
u
ce
s
d
ee
p
Q
-
l
ea
r
n
i
n
g
as
an
alter
n
ati
v
e
to
tr
ad
itio
n
al
ML
m
o
d
els
f
o
r
p
r
ed
ictin
g
NP
L
,
ad
d
r
ess
in
g
t
h
e
li
m
ita
tio
n
s
o
f
ex
is
tin
g
m
e
th
o
d
s
to
h
an
d
le
i
m
b
alan
ce
d
d
ata
b
y
d
y
n
a
m
icall
y
ad
j
u
s
ti
n
g
its
f
o
cu
s
o
n
t
h
e
m
i
n
o
r
it
y
clas
s
u
s
i
n
g
r
e
w
ar
d
an
d
p
en
alt
y
m
ec
h
a
n
is
m
s
.
ii)
I
n
tr
o
d
u
ce
s
an
ex
cl
u
s
i
v
e
d
ataset
o
f
8
3
,
7
3
2
cu
s
to
m
er
r
ec
o
r
d
s
f
r
o
m
a
lead
in
g
Viet
n
a
m
ese
f
in
a
n
cial
in
s
t
itu
tio
n
(
2
0
1
9
–
2
0
2
2
)
,
en
s
u
r
in
g
p
r
ac
tical
r
ele
v
an
ce
a
n
d
ap
p
licab
ilit
y
.
iii)
E
x
ten
s
i
v
el
y
co
n
d
u
ct
ex
p
er
i
m
en
ts
to
p
r
o
v
e
th
e
ef
f
ec
ti
v
e
n
es
s
o
f
th
e
d
ee
p
Q
-
n
et
w
o
r
k
(
DQN
)
m
et
h
o
d
s
i
n
co
m
p
ar
is
o
n
w
it
h
s
o
m
e
s
tr
o
n
g
b
aselin
es o
f
tr
ad
itio
n
al
ap
p
r
o
ac
h
.
T
h
e
r
em
ain
d
er
o
f
th
is
p
ap
er
is
s
tr
u
ct
u
r
ed
as
f
o
llo
w
s
:
s
ec
tio
n
2
in
tr
o
d
u
ce
s
th
e
d
ataset
u
s
e
d
in
d
o
in
g
ex
p
er
i
m
e
n
ts
.
W
e
p
r
esen
t
th
e
p
r
o
p
o
s
ed
m
e
th
o
d
u
s
in
g
d
ee
p
r
ein
f
o
r
ce
m
en
t
lear
n
i
n
g
i
n
s
ec
ti
o
n
3
.
T
h
en
,
s
e
ctio
n
4
s
h
o
w
s
th
e
ex
p
er
i
m
en
tal
s
et
u
p
s
an
d
r
esu
lts
.
L
astl
y
,
s
ec
tio
n
5
s
u
m
m
ar
izes
t
h
e
p
ap
er
an
d
t
h
en
s
u
g
g
e
s
ts
s
o
m
e
f
u
tu
r
e
r
esear
ch
d
ir
ec
tio
n
s
.
2.
M
E
T
H
O
D
2
.
1
.
D
a
t
a
s
et
T
h
e
d
ataset
h
as
2
3
co
lu
m
n
s
(
2
2
in
d
ep
en
d
en
t
v
ar
iab
le
an
d
1
d
ep
en
d
en
t
v
ar
iab
le)
an
d
8
3
,
7
3
2
o
b
s
er
v
atio
n
s
.
O
f
w
h
ic
h
,
1
5
in
p
u
t
v
ar
iab
les
ar
e
ca
te
g
o
r
ical
v
ar
iab
les
a
n
d
th
e
r
e
m
ai
n
i
n
g
7
ar
e
n
u
m
er
ical
v
ar
iab
les
as
s
h
o
w
n
i
n
T
ab
le
1
.
T
h
e
d
ata
co
llected
f
r
o
m
0
1
/0
1
/2
0
1
9
t
o
3
1
/1
2
/2
0
2
2
.
I
n
d
ep
e
n
d
en
t
v
ar
iab
les:
2
2
o
f
th
e
m
ar
e
lis
ted
i
n
T
ab
le
1
a
n
d
d
ep
en
d
en
t v
a
r
iab
le
s
h
o
w
n
i
n
T
ab
le
2
.
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
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
ar
y
20
2
6
:
7
0
0
-
709
702
T
ab
le
1
.
A
ttrib
u
tes o
f
c
u
s
to
m
e
r
s
co
llected
f
r
o
m
th
e
co
m
p
a
n
y
N
o
.
A
t
t
r
i
b
u
t
e
s
Ty
p
e
Ex
p
l
a
n
a
t
i
o
n
1
A
R
EA
C
a
t
e
g
o
r
i
c
a
l
N
o
r
t
h
-
C
e
n
t
r
a
l
–
S
o
u
t
h
a
r
e
a
o
f
V
i
e
t
n
a
m
2
C
A
T
E
G
O
R
Y
N
A
M
E
C
a
t
e
g
o
r
i
c
a
l
C
a
r
,
mo
t
o
r
b
i
k
e
r
e
g
i
st
r
a
t
i
o
n
3
P
A
P
ER
T
Y
P
E
O
b
j
e
c
t
Ty
p
e
s o
f
v
e
r
i
f
i
c
a
t
i
o
n
d
o
c
u
me
n
t
s
(
Eg
:
I
d
e
n
t
i
t
y
c
a
r
d
,
d
r
i
v
i
n
g
l
i
c
e
n
se
,
e
t
c
.
)
4
P
R
O
V
I
N
C
E_
S
H
O
P
C
a
t
e
g
o
r
i
c
a
l
P
r
o
v
i
n
c
e
,
c
i
t
y
o
f
t
h
a
t
s
t
o
r
e
(
6
3
i
t
e
ms)
5
D
ESCR
I
P
TI
O
N
C
a
t
e
g
o
r
i
c
a
l
L
o
a
n
p
a
c
k
a
g
e
(
Eg
:
C
o
n
su
me
r
l
o
a
n
,
b
u
si
n
e
ss l
o
a
n
,
.
.
.
)
6
K
EN
H
C
a
t
e
g
o
r
i
c
a
l
L
e
n
d
i
n
g
c
h
a
n
n
e
l
(
Eg
:
a
g
e
n
t
,
A
p
p
s
,
W
e
b
si
t
e
,
e
t
c
.
)
7
L
O
A
I
_
K
H
A
C
H
_
H
A
N
G
C
a
t
e
g
o
r
i
c
a
l
C
u
s
t
o
me
r
t
y
p
e
s (i
n
d
i
v
i
d
u
a
l
,
o
r
g
a
n
i
z
a
t
i
o
n
)
8
L
O
A
I
_
H
I
N
H
_
C
U
_
T
R
U
C
a
t
e
g
o
r
i
c
a
l
Ty
p
e
o
f
r
e
si
d
e
n
c
e
(
Eg
:
t
e
mp
o
r
a
r
y
r
e
si
d
e
n
c
e
,
p
e
r
man
e
n
t
r
e
si
d
e
n
c
e
,
…)
9
M
A
R
I
TA
L
C
a
t
e
g
o
r
i
c
a
l
M
a
r
i
t
a
l
s
t
a
t
u
s
(
si
n
g
l
e
,
marr
i
e
d
,
d
i
v
o
r
c
e
d
n
o
c
h
i
l
d
,
d
i
v
o
r
c
e
d
w
i
t
h
c
h
i
l
d
,
w
i
d
o
w
,
u
n
k
n
o
w
n
)
10
W
O
R
K
P
L
A
C
E_
T
Y
P
E
C
a
t
e
g
o
r
i
c
a
l
Ty
p
e
o
f
w
o
r
k
p
l
a
c
e
(
Eg
:
2
.
0
:
I
n
d
o
o
r
s;
1
.
0
:
O
u
t
d
o
o
r
s)
11
I
N
D
U
S
T
R
Y
_
N
M
C
a
t
e
g
o
r
i
c
a
l
I
n
d
u
st
r
y
t
y
p
e
s (Eg
:
I
n
d
u
st
r
i
a
l
/
F
o
o
d
/
M
e
c
h
a
n
i
c
a
l
,
T
r
a
n
sp
o
r
t
a
t
i
o
n
/
W
a
r
e
h
o
u
si
n
g
/
S
u
p
p
l
y
,
.
.
.
)
12
JO
B
_
N
M
C
a
t
e
g
o
r
i
c
a
l
Jo
b
t
y
p
e
s (Eg
:
S
a
l
e
s,
w
o
r
k
e
r
s…)
13
N
U
M
B
E
R
_
O
F
_
C
H
I
L
D
N
u
mb
e
r
N
u
mb
e
r
o
f
c
h
i
l
d
r
e
n
14
L
O
A
N
_
P
U
R
P
O
S
E
C
a
t
e
g
o
r
i
c
a
l
L
o
a
n
p
u
r
p
o
se
15
R
ESI
D
EN
C
E_
TI
M
E
N
u
mb
e
r
R
e
si
d
e
n
c
e
t
i
me
i
n
y
e
a
r
s
16
D
I
S
TA
N
C
E
F
l
o
a
t
D
i
st
a
n
c
e
f
r
o
m t
h
e
c
e
n
t
e
r
t
o
t
h
e
p
l
a
c
e
o
f
r
e
si
d
e
n
c
e
17
I
S
_
B
A
D
_
D
EB
T
I
n
t
C
u
s
t
o
me
r
s h
a
v
e
a
h
i
st
o
r
y
o
f
b
a
d
d
e
b
t
18
P
A
C
K
A
G
E_
C
O
D
E
O
b
j
e
c
t
L
o
a
n
p
a
c
k
a
g
e
18
I
S
_
C
U
S
T
O
M
ER
_
N
EW
B
o
o
l
e
a
n
I
s c
u
st
o
me
r
n
e
w
20
I
N
C
O
M
E
N
u
mb
e
r
C
u
s
t
o
me
r
’
s i
n
c
o
me
(
i
n
m
i
l
l
i
o
n
V
N
D
)
21
M
O
N
EY
_
A
P
P
R
A
I
S
A
L
N
u
mb
e
r
A
mo
u
n
t
o
f
p
r
o
p
e
r
t
y
v
a
l
u
a
t
i
o
n
22
A
G
E
N
u
mb
e
r
C
u
s
t
o
me
r
’
s a
g
e
T
ab
le
2
.
D
ep
en
d
en
t v
ar
iab
le
V
a
r
i
a
b
l
e
n
a
me
Ty
p
e
D
e
scri
p
t
i
o
n
G
O
O
D
_
B
A
D
I
n
t
G
o
o
d
d
e
b
t
o
r
B
a
d
d
e
b
t
(
>
9
0
d
a
y
s o
v
e
r
d
u
e
d
e
b
t
i
s
b
a
d
d
e
b
t
)
T
h
e
d
ataset
is
co
llected
f
r
o
m
o
n
e
o
f
t
h
e
lar
g
est
f
in
a
n
cial
in
s
titu
tio
n
s
in
Viet
n
a
m
.
T
h
e
d
ata
d
escr
ib
in
g
ea
ch
cu
s
to
m
er
is
s
h
o
w
n
i
n
T
a
b
le
1
.
T
h
e
d
ata
w
a
s
co
llected
in
n
ea
r
l
y
4
y
ea
r
s
f
r
o
m
J
an
2
0
1
9
to
Oct
2
0
2
2
an
d
s
p
lit
in
to
tr
ai
n
/d
ev
elo
p
m
en
t
/t
est
w
it
h
th
e
r
at
io
7
:2
:1
.
Nu
m
b
er
s
o
f
o
b
s
er
v
atio
n
s
a
n
d
o
th
er
s
tati
s
tics
o
f
th
e
d
ataset
ar
e
s
h
o
w
n
i
n
T
ab
le
3
.
T
ab
le
3
.
So
m
e
s
tati
s
tic
s
ab
o
u
t th
e
d
ata
co
llected
D
a
t
a
#
o
f
O
b
se
r
v
a
t
i
o
n
#
o
f
G
o
o
d
D
e
b
t
#
o
f
B
a
d
D
e
b
t
R
a
t
i
o
G
o
o
d
D
e
b
t
R
a
t
i
o
B
a
d
D
e
b
t
T
r
a
i
n
6
0
,
2
8
6
5
2
,
0
8
4
8
,
2
0
2
8
6
.
3
9
%
1
3
.
6
1
%
V
a
l
i
d
a
t
i
o
n
1
5
,
0
7
2
1
3
,
3
8
4
1
,
6
8
8
8
8
.
8
0
%
1
1
.
2
0
%
T
e
st
8
,
3
7
4
7
,
5
2
9
8
4
5
8
9
.
9
1
%
1
0
.
0
9
%
T
o
t
a
l
8
3
,
7
3
2
7
2
,
9
9
7
1
0
,
7
3
5
8
8
.
3
7
%
1
1
.
6
3
%
2
.
2
.
P
r
o
po
s
ed
a
pp
ro
a
ch
t
o
p
re
dict
no
n
-
perf
o
r
m
ing
lo
a
n
2
.
2.
1
.
M
o
delin
g
t
he
t
a
s
k
a
s
a
n RL pro
ble
m
NP
L
p
r
ed
ictio
n
u
s
u
all
y
s
ee
n
as
a
y
es/
n
o
clas
s
i
f
icatio
n
p
r
o
b
lem
.
B
u
t
h
er
e
w
e
n
o
t
u
s
e
n
o
r
m
al
s
u
p
er
v
i
s
ed
lear
n
i
n
g
,
w
e
m
a
k
e
it lik
e
a
R
L
p
r
o
b
le
m
.
I
n
t
h
is
ca
s
e,
an
ag
e
n
t le
ar
n
to
tell if
lo
a
n
is
g
o
o
d
o
r
b
ad
b
y
talk
i
n
g
w
it
h
en
v
ir
o
n
m
e
n
t
an
d
tr
y
to
ch
a
n
g
e
it
s
p
o
li
c
y
to
g
e
t
m
o
r
e
r
e
w
ar
d
s
.
Fo
r
d
ee
p
Q
-
l
ea
r
n
in
g
,
th
e
f
ir
s
t
s
tep
is
p
u
t
t
h
e
s
tar
t
s
tate
in
to
n
e
u
r
al
n
et
w
o
r
k
,
th
e
n
i
t
g
i
v
e
b
a
ck
Q
-
v
al
u
e
s
f
o
r
all
ac
t
io
n
s
.
T
h
e
b
ig
d
if
f
er
en
ce
b
et
w
ee
n
Q
-
l
ea
r
n
i
n
g
a
n
d
d
ee
p
Q
-
l
ea
r
n
in
g
i
s
h
o
w
t
h
e
y
s
h
o
w
a
n
d
f
in
d
Q
-
v
al
u
es.
I
n
Q
-
l
e
ar
n
in
g
,
Q
-
v
a
lu
e
s
is
k
ee
p
in
a
Q
-
tab
le,
w
h
er
e
ev
er
y
s
tate
-
ac
tio
n
h
as
it
s
v
a
lu
e.
B
u
t
i
n
d
ee
p
Q
-
l
ea
r
n
in
g
,
w
e
u
s
e
d
ee
p
n
eu
r
al
n
et
w
o
r
k
(
DNN)
to
g
u
ess
t
h
e
Q
-
v
al
u
es
,
n
o
t
p
u
t
th
e
m
i
n
a
tab
le.
T
h
e
n
eu
r
al
n
et
ta
k
e
th
e
s
tate
as
in
p
u
t
an
d
g
i
v
e
o
u
t
Q
-
v
alu
e
s
f
o
r
ea
ch
ac
tio
n
in
t
h
at
s
tate.
T
h
e
m
ai
n
id
ea
o
f
R
L
i
s
tr
y
an
d
f
ail,
li
k
e
d
o
ag
ain
a
n
d
ag
ain
a
n
d
lear
n
f
r
o
m
ev
er
y
tr
y
.
Dee
p
Q
-
l
ea
r
n
in
g
h
a
v
e
s
i
x
b
ig
th
i
n
g
s
:
ag
en
t,
en
v
ir
o
n
m
e
n
t,
s
tate,
a
ctio
n
,
r
e
w
ar
d
,
an
d
p
o
licy
.
T
h
e
m
ea
n
i
n
g
o
f
t
h
ese
s
ix
t
h
i
n
g
s
ex
p
lai
n
li
k
e
th
is
:
Ass
u
m
e
t
h
a
t
t
h
e
i
m
b
ala
n
ce
d
tr
ain
i
n
g
d
ata
s
et
is
=
{
(
1
,
1
)
,
(
2
,
2
)
,
…
,
(
,
)
}
w
h
er
e
is
th
e
ℎ
s
a
m
p
le
an
d
is
th
e
lab
el
o
f
t
h
e
ℎ
s
a
m
p
le
:
E
n
v
ir
o
n
m
e
n
t:
t
h
e
d
ataset
co
n
t
ain
i
n
g
lo
a
n
ap
p
lican
ts
’
f
ea
tu
r
e
s
an
d
th
eir
ac
t
u
al
lo
an
p
er
f
o
r
m
an
ce
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
P
r
ed
ictin
g
n
o
n
-
p
erfo
r
min
g
lo
a
n
s
in
V
ietn
a
m’
s
fin
a
n
cia
l secto
r
:
a
d
ee
p
…
(
Lu
ye
n
A
n
h
Do
)
703
State
:
a
2
2
-
d
i
m
e
n
s
io
n
al
f
ea
t
u
r
e
v
ec
to
r
r
ep
r
esen
ti
n
g
a
lo
an
a
p
p
lican
t.
W
h
e
n
tr
ai
n
in
g
s
tar
ts
,
th
e
a
g
e
n
t
g
et
th
e
f
ir
s
t
s
a
m
p
le
1
as
th
e
f
ir
s
t
s
t
ate
1
.
T
h
e
s
tate
s
t
o
f
at
ev
er
y
ti
m
e
s
tep
m
ea
n
s
th
e
s
a
m
p
le
.
W
h
en
t
h
e
n
e
w
ep
is
o
d
e
s
tar
ts
,
t
h
e
en
v
ir
o
n
m
e
n
t
m
i
x
u
p
t
h
e
o
r
d
er
o
f
s
am
p
les i
n
tr
ain
i
n
g
d
ata.
A
ctio
n
:
t
h
e
ag
en
t
c
h
o
o
s
es
b
et
w
ee
n
t
w
o
ac
tio
n
s
:
c
lass
if
y
as
NP
L
(
b
ad
lo
an
)
o
r
class
if
y
a
s
P
L
(
g
o
o
d
lo
an
)
.
Fo
r
y
es/
n
o
class
i
f
icatio
n
p
r
o
b
lem
,
A
={
0
,
1
}
w
h
er
e
0
r
ep
r
esen
ts
th
e
s
m
al
ler
class
an
d
1
r
e
p
r
esen
ts
th
e
b
ig
g
er
class
.
R
e
w
ar
d
:
a
r
e
w
ar
d
is
lik
e
th
e
f
ee
d
b
ac
k
f
r
o
m
th
e
en
v
ir
o
n
m
e
n
t,
it
tells
if
ag
e
n
t
’
s
ac
tio
n
is
g
o
o
d
o
r
b
a
d
.
T
o
h
elp
ag
en
t
s
lear
n
b
etter
r
u
le
in
u
n
b
alan
ce
d
d
ata,
th
e
r
e
w
ar
d
v
al
u
e
f
o
r
s
a
m
p
le
in
s
m
a
ll
class
is
b
i
g
g
er
th
an
s
a
m
p
le
i
n
b
ig
class
.
So
,
if
ag
e
n
t
s
a
y
r
i
g
h
t
o
r
w
r
o
n
g
o
n
s
m
al
l
class
s
a
m
p
le,
th
e
en
v
ir
o
n
m
en
t
g
i
v
e
b
ig
g
er
r
e
w
ar
d
o
r
b
ig
g
er
p
u
n
is
h
.
S
m
all
c
lass
s
a
m
p
le
i
s
h
ar
d
t
o
f
in
d
co
r
r
ec
t
i
n
u
n
b
ala
n
ce
d
d
ataset.
T
o
m
a
k
e
ag
en
t
s
ee
s
m
all
cla
s
s
b
etter
,
t
h
e
alg
o
r
it
h
m
m
u
s
t
b
e
m
o
r
e
ca
r
ef
u
l
w
i
th
s
m
all
cl
as
s
.
So
,
w
h
en
a
g
e
n
t
m
ee
t
s
m
al
l c
lass
s
a
m
p
le,
it
g
et
b
ig
r
e
w
ar
d
o
r
b
ig
p
u
n
i
s
h
.
T
h
e
r
e
w
a
r
d
f
u
n
ct
io
n
is
l
ik
e
t
h
is
:
(
,
,
)
=
{
+
1
,
=
an
d
∈
−
1
,
≠
an
d
∈
λ
,
=
an
d
∈
−
λ
,
≠
an
d
∈
w
h
er
e
λ
∈
[
0
,
1
]
,
is
s
m
aller
class
s
a
m
p
le
s
et,
is
b
ig
g
er
class
s
a
m
p
le
s
et.
T
h
e
b
est
p
er
f
o
r
m
an
ce
in
ex
p
er
i
m
e
n
t is
w
h
e
n
λ
eq
u
al
s
t
o
th
e
i
m
b
ala
n
ce
d
r
atio
=
|
|
|
|
T
r
an
s
itio
n
p
r
o
b
a
b
ilit
y
P
:
t
r
an
s
itio
n
p
r
o
b
ab
ilit
y
(
+
1
|
,
)
is
d
eter
m
i
n
i
s
tic.
T
h
e
ag
en
t
g
o
es
f
r
o
m
th
e
cu
r
r
en
t
s
tate
to
th
e
n
e
x
t
s
tate
+1
b
y
f
o
llo
w
i
n
g
t
h
e
o
r
d
er
o
f
s
a
m
p
les
i
n
th
e
tr
ai
n
in
g
d
ata
.
Dis
co
u
n
t
f
ac
to
r
γ
: γ
∈
[
0
,
1
]
is
to
h
elp
b
alan
ci
n
g
t
h
e
cu
r
r
e
n
t a
n
d
f
u
t
u
r
e
r
e
w
ar
d
.
E
p
is
o
d
e
in
R
L
is
j
u
s
t
th
e
p
ath
f
r
o
m
th
e
f
ir
s
t
s
tate
to
th
e
l
ast
s
tate
{
1
,
1
,
1
,
2
,
2
,
2
,
…
,
,
,
}
.
An
ep
is
o
d
e
s
to
p
s
w
h
e
n
all
s
a
m
p
l
es
in
tr
ain
i
n
g
d
ata
ar
e
class
if
i
ed
o
r
w
h
e
n
th
e
ag
e
n
t
w
r
o
n
g
l
y
cla
s
s
i
f
ie
s
th
e
s
a
m
p
le
f
r
o
m
s
m
aller
clas
s
.
P
o
licy
π
θ
:
t
h
e
p
o
lic
y
π
θ
is
li
k
e
a
r
u
le
f
u
n
c
tio
n
π:
S
→
A
w
h
er
e
π
θ
(
)
m
ea
n
s
w
h
at
ac
tio
n
a
g
en
t
s
h
o
u
ld
d
o
w
h
e
n
it in
s
tate
.
T
h
e
p
o
licy
π
θ
ca
n
b
e
s
ee
n
lik
e
a
clas
s
i
f
ier
w
it
h
θ.
Q
-
v
alu
e
(
,
)
:
t
h
e
ex
p
ec
ted
cu
m
u
l
ativ
e
r
e
w
ar
d
f
o
r
tak
in
g
ac
tio
n
in
s
ta
te
,
w
h
ich
t
h
e
ag
e
n
t
lea
r
n
s
to
o
p
tim
ize.
W
ith
t
h
e
m
ea
n
in
g
a
n
d
s
y
m
b
o
l
s
ab
o
v
e,
t
h
e
u
n
b
ala
n
ce
d
clas
s
i
f
icatio
n
p
r
o
b
le
m
is
j
u
s
t
to
f
i
n
d
th
e
b
est
p
o
licy
π
∗
:
S
→
A
,
th
at
m
a
k
e
th
e
cu
m
u
lati
v
e
r
e
w
ar
d
s
as
b
i
g
as
p
o
s
s
ib
le
.
T
h
e
o
v
er
all
s
tr
u
ctu
r
e
i
s
s
h
o
w
n
i
n
Fig
u
r
e
1
.
On
e
k
e
y
s
tr
o
n
g
p
o
in
t
o
f
D
QN
is
t
h
at
it
ca
n
ad
ap
tiv
el
y
f
o
cu
s
o
n
m
i
n
o
r
it
y
cla
s
s
p
r
ed
ictio
n
s
th
r
o
u
g
h
its
r
e
w
ar
d
m
ec
h
a
n
is
m
:
P
en
alt
y
f
o
r
m
is
cla
s
s
i
f
y
i
n
g
NP
L
s
(
f
alse
n
eg
a
tiv
e
s
)
is
lar
g
er
→
e
n
co
u
r
a
g
es c
o
r
r
ec
t d
etec
tio
n
o
f
b
ad
lo
an
s
.
Hig
h
er
r
e
w
ar
d
f
o
r
co
r
r
ec
tl
y
c
l
ass
i
f
y
in
g
NP
L
s
→
f
o
r
ce
s
t
h
e
a
g
en
t to
lear
n
th
e
m
i
n
o
r
it
y
clas
s
b
etter
.
Fig
u
r
e
1
.
T
h
e
ar
ch
itectu
r
e
o
f
m
o
d
ell
in
g
NP
L
p
r
ed
ictio
n
as a
r
ein
f
o
r
ce
m
e
n
t le
ar
n
in
g
p
r
o
b
lem
2
.
2
.
2.
I
m
ple
m
ent
a
t
io
n us
ing
d
ee
p Q
-
lea
rni
ng
a)
State
r
ep
r
esen
tatio
n
: e
ac
h
lo
an
is
r
ep
r
esen
ted
as a
2
2
-
d
i
m
e
n
s
io
n
al
v
ec
to
r
.
b)
DQN:
a
n
e
u
r
al
n
et
w
o
r
k
esti
m
a
tes Q
-
v
al
u
es
f
o
r
ea
ch
ac
tio
n
g
i
v
en
t
h
e
s
ta
te
(
lo
an
d
ata)
.
c)
T
r
ain
in
g
p
r
o
ce
s
s
:
T
h
e
ag
en
t o
b
s
er
v
es a
lo
an
's
f
ea
tu
r
es (
s
tate)
.
I
t
ch
o
o
s
es
an
ac
tio
n
(
clas
s
i
f
y
as
NP
L
/P
L
)
u
s
i
n
g
a
n
ε
-
g
r
ee
d
y
s
tr
ate
g
y
(
b
alan
c
in
g
e
x
p
lo
r
atio
n
an
d
ex
p
lo
itatio
n
)
.
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
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
ar
y
20
2
6
:
7
0
0
-
709
704
I
t r
ec
eiv
es a
r
e
w
ar
d
b
ased
o
n
class
i
f
icatio
n
ac
cu
r
ac
y
.
T
h
e
ex
p
er
ien
ce
is
m
e
m
o
r
ized
in
a
r
ep
la
y
b
u
f
f
er
.
T
h
e
Q
-
n
et
w
o
r
k
i
s
u
p
d
ated
u
s
i
n
g
B
ell
m
a
n
’
s
eq
u
atio
n
:
(
,
)
=
+
ma
x
(
′
,
′
)
w
h
er
e
γ
is
th
e
d
is
co
u
n
t f
ac
to
r
o
f
f
u
t
u
r
e
r
e
w
ar
d
s
.
d)
E
v
alu
a
tio
n
:
t
h
e
m
o
d
el
is
test
ed
o
n
u
n
s
ee
n
lo
an
ap
p
licatio
n
s
to
m
ea
s
u
r
e
p
r
ed
ictio
n
ac
cu
r
ac
y
an
d
r
ec
all
(
esp
ec
iall
y
f
o
r
NP
L
s
)
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
I
n
th
is
s
ec
tio
n
,
w
e
co
n
d
u
c
t e
x
p
er
im
e
n
t
s
to
v
alid
ate
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
an
d
at
t
h
e
s
a
m
e
ti
m
e
we
g
i
v
e
s
o
m
e
co
m
p
r
eh
e
n
s
iv
e
d
is
c
u
s
s
i
o
n
.
3
.
1
.
E
x
peri
m
ent
s
et
u
p
3
.
1
.
1
.
H
y
per
-
pa
ra
m
et
er
t
un
i
ng
T
o
im
p
le
m
e
n
t
DQN
f
o
r
NL
P
p
r
ed
ictio
n
,
w
e
u
s
ed
ε
-
g
r
ee
d
y
p
o
licy
.
W
e
v
ar
ied
th
e
h
y
p
er
-
p
ar
a
m
s
an
d
ch
o
s
e
th
e
b
est
v
al
u
e
u
s
in
g
th
e
d
ev
elo
p
m
e
n
t
s
et.
T
h
e
ex
p
lo
r
atio
n
r
ate
ε
d
ec
r
ea
s
es
lin
ea
r
l
y
f
r
o
m
1
.
0
to
0
.
0
0
1
d
u
r
in
g
th
e
p
r
o
ce
s
s
.
T
h
e
r
ep
lay
m
e
m
o
r
y
s
ize
is
1
,
070
,
0
0
0
an
d
th
e
in
ter
ac
tio
n
s
b
et
w
ee
n
a
g
e
n
t
an
d
en
v
ir
o
n
m
e
n
t
ar
e
ab
o
u
t
1
,
000
,
0
0
0
s
tep
s
.
γ
-
th
e
d
is
co
u
n
t
f
a
cto
r
is
s
et
at
0
.
2
.
T
h
e
Q
-
n
et
w
o
r
k
is
o
p
ti
m
i
ze
d
w
ith
t
h
e
A
d
a
m
alg
o
r
ith
m
w
it
h
its
lear
n
i
n
g
r
ate
at
0
.
0
0
0
1
,
th
e
b
atch
s
ize
at
3
2
.
T
h
e
am
o
u
n
t
o
f
d
ata
co
llected
f
o
r
r
e
p
lay
b
u
f
f
er
ea
ch
ep
is
o
d
e
is
3
,
0
0
0
.
T
h
e
s
te
p
in
ter
v
al
to
co
llect
d
ata
d
u
r
in
g
tr
ai
n
in
g
is
2
,
0
0
0
.
Up
d
ate
th
e
tar
g
et
Q
-
n
et
w
o
r
k
ev
er
y
2
,
0
0
0
ep
is
o
d
es.
T
h
e
n
u
m
b
er
o
f
i
m
b
alan
ce
r
atio
n
is
0
.
1
3
.
Fo
r
o
th
er
ML
m
et
h
o
d
s
an
d
tech
n
iq
u
e
s
to
h
an
d
le
cla
s
s
i
m
b
a
lan
ce
,
w
e
e
x
p
lo
ited
s
k
lear
n
lib
r
ar
ies.
3
.
1
.
2
.
E
v
a
lua
t
io
n
m
et
rics
T
o
m
ea
s
u
r
e
th
e
e
f
f
ec
ti
v
en
e
s
s
o
f
th
e
NP
L
p
r
ed
ictio
n
m
o
d
el,
w
e
f
o
cu
s
o
n
d
etec
ti
n
g
clas
s
1
(
h
ig
h
-
r
is
k
cu
s
to
m
er
s
li
k
el
y
to
d
ef
a
u
lt)
r
ath
er
th
a
n
ev
al
u
ati
n
g
b
o
th
class
es
eq
u
all
y
.
I
d
en
ti
f
y
i
n
g
t
h
ese
c
u
s
to
m
er
s
is
cr
u
cia
l
f
o
r
f
in
a
n
cial
in
s
tit
u
tio
n
s
to
m
i
tig
ate
r
is
k
an
d
r
ed
u
ce
NP
L
g
r
o
w
t
h
.
Fo
r
th
is
,
w
e
m
ea
s
u
r
e
th
e
p
r
ec
is
io
n
,
r
ec
all
an
d
F1
s
co
r
es o
n
th
i
s
clas
s
.
I
n
ad
d
itio
n
,
o
n
th
e
b
est
-
p
er
f
o
r
m
i
n
g
m
o
d
el,
w
e
al
s
o
r
ep
o
r
t th
e
ar
ea
u
n
d
er
th
e
c
u
r
v
e
(
A
U
C
)
s
co
r
e,
Ma
tth
e
w
s
co
r
r
elatio
n
co
ef
f
icie
n
t
(
MC
C
)
an
d
G
-
m
ea
n
w
h
ic
h
ar
e
also
s
t
an
d
ar
d
m
etr
ics
f
o
r
ev
alu
a
tin
g
i
m
b
alan
ce
d
clas
s
i
f
i
ca
tio
n
.
3
.
2
.
E
x
peri
m
ent
re
s
ults
T
h
r
ee
ty
p
es o
f
e
x
p
er
i
m
e
n
ts
:
1)
T
h
e
f
ir
s
t
ex
p
er
i
m
e
n
t
test
s
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
tr
ad
itio
n
al
M
L
m
o
d
el
s
w
it
h
o
u
t
ap
p
l
y
in
g
a
n
y
d
ata
-
b
ala
n
ci
n
g
tech
n
iq
u
es.
W
e
tr
ain
ed
t
h
e
f
o
l
lo
w
i
n
g
m
o
d
els
o
n
t
h
e
d
ataset
u
s
i
n
g
lo
g
i
s
tic
r
eg
r
es
s
io
n
[7
]
-
[
9]
;
d
ec
is
io
n
tr
ee
[
1
0
]
-
[
1
3
]
;
r
an
d
o
m
f
o
r
est
[
1
4
]
-
[
17]
;
SVM
[
1
8
]
-
[
20]
;
L
i
g
h
t
GB
M
[
2
1
]
-
[
2
3
]
an
d
X
GB
o
o
s
t
[
2
4
]
-
[
26]
.
T
h
e
p
u
r
p
o
s
e
is
to
o
b
s
er
v
e
h
o
w
c
l
ass
i
m
b
a
lan
ce
a
f
f
ec
ts
m
o
d
el
p
er
f
o
r
m
a
n
ce
,
esp
ec
iall
y
i
n
d
etec
tin
g
h
i
g
h
-
r
is
k
(
NP
L
)
cu
s
to
m
er
s
.
2)
I
n
th
e
n
ex
t
e
x
p
er
i
m
e
n
t,
w
e
ap
p
ly
r
esa
m
p
li
n
g
tech
n
iq
u
e
s
to
i
m
p
r
o
v
e
clas
s
b
alan
ce
s
u
ch
a
s
o
v
er
-
s
a
m
p
li
n
g
an
d
u
n
d
er
-
s
a
m
p
li
n
g
tec
h
n
iq
u
es.
T
h
is
is
to
ass
e
s
s
w
h
et
h
er
r
esa
m
p
li
n
g
tec
h
n
iq
u
e
s
i
m
p
r
o
v
e
clas
s
1
r
ec
all
an
d
o
v
er
all
p
er
f
o
r
m
an
ce
o
f
M
L
m
o
d
els.
3)
I
n
th
e
th
ir
d
ex
p
er
i
m
en
t,
w
e
i
n
t
r
o
d
u
ce
DQN,
an
R
L
ap
p
r
o
ac
h
th
at
d
y
n
a
m
ical
l
y
ad
j
u
s
ts
d
ec
i
s
io
n
b
o
u
n
d
a
r
ie
s
b
ased
o
n
r
e
w
ar
d
s
i
g
n
al
s
.
T
h
is
is
to
d
eter
m
in
e
if
DQN
o
u
tp
er
f
o
r
m
s
tr
ad
itio
n
al
m
o
d
e
ls
i
n
h
an
d
li
n
g
i
m
b
alan
ce
d
d
ata
w
it
h
o
u
t t
h
e
n
ee
d
f
o
r
r
esam
p
li
n
g
tech
n
iq
u
es.
3
.
2
.
1
.
P
er
f
o
rm
a
nce
o
f
t
ra
ditio
na
l
ML
m
o
del
s
w
it
ho
ut
a
pp
ly
ing
a
ny
da
t
a
-
ba
la
ncing
t
e
chniq
ues
T
ab
le
4
is
th
e
ev
alu
atio
n
m
atr
ix
f
o
r
o
n
l
y
class
1
(
b
ad
d
eb
t
)
b
et
w
ee
n
6
ML
alg
o
r
ith
m
s
in
t
esti
n
g
s
et
.
I
n
o
v
er
v
ie
w
,
X
GB
o
o
s
t
h
as
t
h
e
b
est
p
er
f
o
r
m
an
ce
,
f
o
llo
w
ed
b
y
L
i
g
h
tGB
M
an
d
R
a
n
d
o
m
Fo
r
est.
T
h
e
w
o
r
s
e
p
er
f
o
r
m
a
n
ce
is
t
h
e
d
ec
is
io
n
tr
ee
,
f
o
llo
w
ed
b
y
lo
g
is
tic
r
e
g
r
es
s
io
n
.
T
ab
le
4
.
E
v
alu
atio
n
m
atr
ix
o
f
class
1
f
o
r
m
u
ltip
le
M
L
al
g
o
r
ith
m
s
i
n
t
esti
n
g
s
et
M
e
t
r
i
c
L
o
g
i
st
i
c
r
e
g
r
e
ssi
o
n
D
e
c
i
si
o
n
t
r
e
e
R
a
n
d
o
m fo
r
e
st
S
V
M
L
i
g
h
t
G
B
M
X
G
B
o
o
st
P
r
e
c
i
si
o
n
5
7
.
0
2
4
9
.
7
8
5
9
.
9
8
5
6
.
7
6
5
9
.
6
6
6
0
.
2
1
R
e
c
a
l
l
7
1
.
6
0
6
7
.
5
7
7
2
.
9
0
7
0
.
5
3
7
4
.
2
0
7
5
.
3
8
F1
-
sco
r
e
6
3
.
4
8
5
7
.
3
3
6
5
.
8
1
6
2
.
9
0
6
6
.
1
4
6
6
.
9
5
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
P
r
ed
ictin
g
n
o
n
-
p
erfo
r
min
g
lo
a
n
s
in
V
ietn
a
m’
s
fin
a
n
cia
l secto
r
:
a
d
ee
p
…
(
Lu
ye
n
A
n
h
Do
)
705
T
h
e
t
o
p
3
a
l
g
o
r
ith
m
s
w
i
th
th
e
h
ig
h
es
t
p
e
r
f
o
r
m
an
c
e
in
b
o
th
p
r
e
c
i
s
i
o
n
,
r
ec
a
l
l
,
a
n
d
F
1
s
c
o
r
e
a
r
e
X
G
B
o
o
s
t,
r
a
n
d
o
m
f
o
r
es
t
an
d
L
ig
h
t
G
B
M,
r
e
s
p
e
c
t
iv
ely
.
I
n
p
r
e
c
is
i
o
n
s
co
r
e
,
t
h
e
h
ig
h
es
t
is
X
G
B
o
o
s
t
,
f
o
l
l
o
w
e
d
b
y
ra
n
d
o
m
f
o
r
e
s
t
,
th
e
l
o
w
e
r
is
X
G
B
o
o
s
t
.
I
n
r
e
c
a
l
l
s
c
o
r
e
,
t
h
e
h
ig
h
es
t
is
X
G
B
o
o
s
t
,
f
o
l
l
o
w
e
d
b
y
L
ig
h
tG
B
M
,
t
h
e
l
o
w
e
r
is
r
a
n
d
o
m
f
o
r
e
s
t
.
Fin
a
l
ly
,
F
1
-
s
c
o
r
e
w
i
ll
g
iv
e
an
o
v
e
r
a
l
l
r
a
t
in
g
w
it
h
a
c
o
m
b
in
a
ti
o
n
o
f
p
r
e
c
is
i
o
n
an
d
r
e
c
al
l
,
F
1
-
s
c
o
r
e
o
f
X
G
B
o
o
s
t
is
th
e
h
ig
h
e
s
t
,
f
o
ll
o
w
e
d
b
y
L
ig
h
tG
B
M
,
l
o
w
e
r
i
s
r
a
n
d
o
m
f
o
r
es
t
.
I
n
s
u
m
m
a
r
y
,
it
c
an
b
e
c
o
n
c
lu
d
e
d
t
h
at
th
e
b
o
o
s
tin
g
a
lg
o
r
i
th
m
h
a
s
t
h
e
b
es
t
p
e
r
f
o
r
m
an
c
e
an
d
it
i
s
s
u
p
e
r
i
o
r
t
o
b
a
g
g
i
n
g
alg
o
r
i
th
m
s
.
3
.
2
.
2
.
Co
nv
ent
io
na
l
m
et
ho
ds
t
o
ha
nd
le
cla
s
s
i
m
ba
la
nce
T
a
b
l
e
5
s
h
o
w
s
th
e
F
1
-
s
c
o
r
e
o
f
s
ev
e
r
a
l
h
a
n
d
l
in
g
im
b
al
an
c
e
d
m
e
t
h
o
d
s
i
n
6
M
L
a
lg
o
r
ith
m
s
o
n
te
s
t
in
g
s
et
.
H
o
w
ev
e
r
,
th
e
s
p
e
c
i
a
l
th
in
g
is
t
h
at
r
an
d
o
m
f
o
r
es
t
h
as
b
e
c
o
m
e
th
e
a
lg
o
r
i
th
m
w
ith
th
e
h
ig
h
e
s
t
F
1
-
s
c
o
r
e
ev
en
t
h
o
u
g
h
in
th
e
v
a
l
i
d
a
ti
o
n
s
et
is
lo
w
e
r
th
an
b
o
o
s
ti
n
g
a
lg
o
r
i
th
m
s
.
T
h
i
s
h
as
s
h
o
w
n
th
a
t
w
i
th
th
e
m
e
th
o
d
o
f
h
an
d
l
in
g
im
b
al
an
c
e
d
d
a
t
a
,
s
p
e
c
if
i
c
a
lly
t
h
e
o
v
e
r
-
s
am
p
l
in
g
m
e
th
o
d
s
,
th
es
e
m
e
th
o
d
s
h
av
e
in
c
r
e
as
e
d
t
h
e
a
c
cu
r
a
cy
o
f
th
e
m
o
d
e
l
.
H
o
w
ev
e
r
,
th
e
l
ev
el
o
f
p
e
r
f
o
r
m
an
c
e
im
p
r
o
v
em
en
t
is
th
er
e
b
u
t
n
o
t
h
ig
h
w
h
en
u
s
i
n
g
t
r
a
d
i
t
i
o
n
a
l
im
b
al
an
c
ed
d
a
t
a
p
r
o
c
e
s
s
in
g
m
e
th
o
d
s
.
T
h
e
r
e
s
e
a
r
ch
d
i
r
e
c
t
i
o
n
in
t
h
e
n
ex
t
s
e
c
ti
o
n
w
il
l
s
h
o
w
a
n
e
w
m
eth
o
d
,
w
h
i
ch
i
s
w
i
d
ely
k
n
o
w
n
in
th
e
A
I
c
o
m
m
u
n
i
ty
b
u
t
h
as
n
o
t
b
e
en
a
p
p
l
ie
d
m
u
ch
i
n
th
e
p
r
o
b
l
em
o
f
im
b
a
la
n
ce
d
d
a
t
a
.
I
t'
s
d
e
e
p
Q
-
l
e
a
r
n
in
g
a
lg
o
r
i
th
m
,
w
e
w
i
l
l
a
p
p
ly
an
d
o
p
tim
i
z
e
th
is
a
lg
o
r
it
h
m
t
o
s
h
o
w
th
a
t
i
t
w
i
ll
b
e
am
az
in
g
ly
g
r
e
at
in
h
an
d
li
n
g
t
h
e
im
b
a
l
an
c
e
d
-
d
a
t
a
p
r
o
b
l
em
co
m
p
a
r
e
d
t
o
th
e
a
b
o
v
e
m
e
th
o
d
s
.
T
ab
le
5
.
C
o
m
p
ar
is
o
n
F1
-
s
co
r
e
o
f
h
an
d
li
n
g
i
m
b
ala
n
ce
d
d
ata
m
et
h
o
d
s
in
t
es
tin
g
s
e
t
S
a
mp
l
i
n
g
me
t
h
o
d
L
o
g
i
st
i
c
r
e
g
r
e
ssi
o
n
D
e
c
i
si
o
n
t
r
e
e
R
a
n
d
o
m fo
r
e
st
S
V
M
L
i
g
h
t
G
B
M
X
GB
o
o
st
B
a
se
6
3
.
4
6
0
.
0
5
6
6
.
9
8
6
3
.
7
2
6
6
.
1
4
6
6
.
9
5
S
M
O
T
E
6
2
.
8
3
6
0
.
0
5
6
6
.
5
3
6
2
.
5
0
6
6
.
1
8
6
6
.
2
8
O
V
ER
S
A
M
P
L
I
N
G
6
1
.
8
0
5
8
.
9
0
6
6
.
9
8
6
3
.
7
2
6
4
.
7
2
6
6
.
2
1
U
N
D
ER
S
A
M
P
L
I
N
G
6
1
.
6
9
5
5
.
3
6
6
6
.
1
5
6
2
.
9
9
6
3
.
6
8
6
4
.
0
2
3
.
2
.
3
.
E
f
f
ec
t
iv
eness
o
f
t
he
pro
po
s
ed
a
pp
ro
a
ch
us
ing
d
ee
p Q
-
l
ea
rning
t
o
ha
nd
le
i
m
ba
la
nced
da
t
a
W
e
tr
ied
w
it
h
d
i
f
f
er
e
n
t
ar
ch
itect
u
r
es,
v
ar
ied
f
r
o
m
2
,
3,
4,
5
,
an
d
6
lay
er
s
to
f
i
n
d
th
e
b
est
co
n
f
i
g
u
r
atio
n
.
Fi
n
all
y
,
t
h
e
b
es
t
o
n
e
is
t
h
e
o
n
e
u
s
i
n
g
4
h
id
d
en
la
y
er
s
w
it
h
5
2
6
n
o
d
es
in
ea
ch
la
y
er
.
Fo
r
ea
ch
la
y
er
,
w
e
u
s
e
r
ec
tif
ied
li
n
ea
r
u
n
it
(
R
e
L
U
)
as
t
h
e
ac
ti
v
atio
n
f
u
n
ct
io
n
.
An
d
th
e
o
u
tp
u
t
la
y
er
in
cl
u
d
es
t
w
o
n
o
d
e
s
w
h
ic
h
ar
e
g
o
o
d
an
d
b
ad
.
I
n
to
t
al,
w
e
h
a
v
e
8
6
4
,
7
4
6
p
ar
am
eter
s
.
Fig
u
r
e
2
s
h
o
w
s
t
h
e
p
r
ec
is
io
n
r
ec
all
(
P
R
)
cu
r
v
e
a
n
d
r
ec
eiv
e
r
o
p
er
atin
g
ch
ar
ac
ter
i
s
tic
(
R
O
C
)
cu
r
v
e.
As
w
e
ca
n
s
ee
,
i
n
P
R
cu
r
v
e,
t
h
e
av
er
a
g
e
p
r
ec
is
io
n
s
co
r
e
is
v
er
y
g
o
o
d
in
tr
ain
i
n
g
s
et
b
u
t
q
u
ite
lo
w
i
n
te
s
tin
g
s
et.
I
n
R
OC
c
u
r
v
e,
A
U
C
s
co
r
e
is
q
u
ite
g
o
o
d
f
o
r
th
e
test
s
et,
th
at
m
ea
n
s
th
e
m
o
d
el
h
as
c
lass
i
f
ied
q
u
ite
w
e
ll
b
et
w
ee
n
2
class
es
g
o
o
d
an
d
b
ad
.
I
n
th
e
test
in
g
s
et,
th
e
p
r
ec
is
io
n
s
co
r
e
o
f
d
ee
p
Q
-
l
ea
r
n
in
g
h
as
b
ec
o
m
e
th
e
h
ig
h
e
s
t,
al
m
o
s
t
2
%
h
i
g
h
er
th
an
L
i
g
h
tGB
M
an
d
1
%
h
ig
h
er
th
an
XGB
o
o
s
t.
Dee
p
Q
-
l
ea
r
n
in
g
's
r
ec
all
is
o
n
l
y
r
ed
u
ce
d
b
y
n
ea
r
l
y
1
% c
o
m
p
ar
ed
to
th
e
v
alid
atio
n
s
et
a
n
d
s
til
l h
as t
h
e
h
i
g
h
est
s
co
r
e.
Fig
u
r
e
2
.
R
OC
c
u
r
v
e
a
n
d
P
R
cu
r
v
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
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
ar
y
20
2
6
:
7
0
0
-
709
706
Fin
all
y
,
i
n
Fi
g
u
r
e
3
,
d
ee
p
Q
-
l
ea
r
n
in
g
h
as
s
h
o
w
n
g
r
ea
t
p
er
f
o
r
m
an
ce
co
m
p
ar
ed
to
b
o
o
s
tin
g
o
r
b
ag
g
in
g
alg
o
r
ith
m
s
i
n
F1
-
s
co
r
e.
Dee
p
Q
-
l
ea
r
n
i
n
g
's
F1
-
s
co
r
e
is
8
%
h
i
g
h
er
t
h
an
L
i
g
h
tGB
M
an
d
7
%
h
ig
h
er
t
h
a
n
XGB
o
o
s
t a
n
d
r
an
d
o
m
f
o
r
est.
Fig
u
r
e
3
.
C
o
m
p
ar
is
o
n
d
ee
p
Q
-
l
ea
r
n
in
g
w
it
h
o
th
er
M
L
m
o
d
el
s
in
te
s
tin
g
s
e
t
Fo
r
th
e
b
est
m
o
d
el,
w
e
al
s
o
ca
lcu
lated
t
h
e
f
i
n
al
r
es
u
lt
s
as
s
h
o
w
n
i
n
T
ab
le
6
.
T
h
e
m
o
d
el
p
er
f
o
r
m
s
q
u
ỉe
w
e
ll
o
n
t
h
e
m
aj
o
r
it
y
cl
ass
,
ac
h
ie
v
i
n
g
a
n
F1
-
s
co
r
e
o
f
9
6
.
2
5
%
an
d
p
r
ec
is
io
n
o
f
9
9
.
2
5
%,
w
h
ile
s
t
ill
id
en
ti
f
y
i
n
g
m
in
o
r
it
y
NP
L
ca
s
es
ef
f
ec
ti
v
el
y
w
it
h
a
h
i
g
h
r
ec
all
o
f
9
3
.
7
3
%.
A
lth
o
u
g
h
p
r
ec
is
io
n
f
o
r
class
1
is
lo
w
er
(
6
1
.
5
4
%)
d
u
e
to
clas
s
i
m
b
alan
ce
,
t
h
e
o
v
er
all
m
etr
i
cs:
A
UC
(
0
.
8
5
6
)
,
G
-
m
ea
n
(
9
3
.
4
8
%),
an
d
MCC
(
0
.
4
8
)
in
d
icate
th
at
th
e
m
o
d
el
m
a
in
ta
in
s
g
o
o
d
d
is
cr
i
m
in
a
tiv
e
ab
ilit
y
an
d
h
an
d
le
s
th
e
i
m
b
alan
ce
ef
f
ec
ti
v
el
y
.
Ov
er
all,
th
e
m
o
d
el
y
ield
ed
b
alan
ce
d
an
d
r
eliab
le
p
er
f
o
r
m
a
n
c
e
f
o
r
NP
L
p
r
ed
ict
io
n
.
T
ab
le
6
.
Deta
iled
m
etr
ics
f
o
r
m
ea
s
u
r
in
g
p
er
f
o
r
m
a
n
ce
o
f
d
ee
p
Q
-
l
ea
r
n
i
n
g
o
n
i
m
b
ala
n
ce
d
d
ataset
L
a
b
e
l
0
L
a
b
e
l
1
M
a
c
r
o
a
v
g
W
e
i
g
h
t
e
d
a
v
g
P
r
e
c
i
si
o
n
9
9
.
2
5
6
1
.
5
4
8
0
.
4
0
9
5
.
4
5
R
e
c
a
l
l
9
3
.
4
3
9
3
.
7
3
9
3
.
5
8
9
3
.
4
6
F1
9
6
.
2
5
7
4
.
3
8
5
.
2
7
9
4
.
0
4
A
U
C
sco
r
e
0
.
8
5
6
M
C
C
0
.
4
8
G
-
m
e
a
n
9
3
.
4
8
3
.
2
.
4
.
M
o
del
d
is
cus
s
io
n
Dee
p
Q
-
le
ar
n
i
n
g
is
a
b
lack
-
b
o
x
m
o
d
el.
Ho
w
ev
er
,
w
e
ca
n
s
til
l
ex
p
lai
n
th
e
m
o
d
el
b
y
p
r
o
v
id
in
g
in
tr
i
n
s
ic
f
ea
tu
r
e
i
m
p
o
r
tan
ce
s
co
r
es
u
s
i
n
g
a
s
u
r
r
o
g
ate
-
m
o
d
el
ex
p
lain
ab
ilit
y
ap
p
r
o
ac
h
w
id
el
y
u
s
ed
in
th
e
liter
atu
r
e.
W
e
co
n
d
u
cted
a
n
e
x
p
lain
ab
ili
t
y
a
n
al
y
s
is
u
s
i
n
g
S
HA
P
.
W
e
tr
ai
n
ed
a
n
XGB
o
o
s
t
s
u
r
r
o
g
ate
m
o
d
el
to
i
m
itate
th
e
D
QN
’
s
p
r
ed
ictio
n
s
an
d
co
m
p
u
ted
SH
A
P
v
al
u
es
o
n
th
e
s
u
r
r
o
g
ate.
T
h
e
r
esu
lts
s
h
o
w
:
h
i
g
h
-
i
m
p
r
ac
t
in
d
icato
r
s
ar
e
h
is
to
r
y
o
f
p
ast
b
ad
d
eb
t,
in
co
m
e
lev
el,
c
u
s
to
m
er
’
s
ag
e,
j
o
b
ty
p
es,
lo
an
p
ac
k
a
g
e
t
y
p
es,
an
d
y
ea
r
s
at
cu
r
r
en
t
r
esid
en
ce
;
m
o
d
er
ate
in
d
icato
r
s
ar
e
ass
et
v
alu
atio
n
,
a
m
o
u
n
t
o
f
p
r
o
p
er
ty
v
alu
at
i
o
n
,
in
d
u
s
tr
y
s
ec
to
r
,
m
ar
ital sta
tu
s
,
in
d
u
s
tr
y
t
y
p
es,
w
o
r
k
p
lace
t
y
p
e
; a
n
d
th
e
r
e
m
ai
n
in
g
f
ea
t
u
r
es a
r
e
at
lo
w
er
-
i
m
p
ac
t in
d
icato
r
s
.
T
h
e
ex
p
er
im
e
n
tal
r
esu
lts
i
n
d
icate
th
at
th
e
p
r
o
p
o
s
ed
m
o
d
el
is
f
ea
s
ib
le
f
o
r
r
ea
l
-
w
o
r
ld
d
ep
lo
y
m
e
n
t.
T
r
ain
in
g
r
e
m
ai
n
s
s
tab
le
ev
e
n
w
it
h
a
lar
g
e
r
ep
lay
b
u
f
f
er
,
as
th
e
av
er
ag
e
Q
in
cr
ea
s
e
s
s
te
ad
il
y
an
d
th
e
lo
s
s
d
ec
r
ea
s
es
o
v
er
tim
e.
T
h
e
m
o
d
el
is
tr
ain
ed
o
n
a
lo
ca
l
m
ac
h
in
e
w
ith
li
m
ited
r
eso
u
r
ce
s
(
C
P
U
o
n
ly
w
it
h
8
GB
R
A
M
)
,
w
h
ile
th
e
tes
tin
g
p
h
a
s
e
is
v
er
y
f
a
s
t,
m
a
k
in
g
it
s
u
itab
le
f
o
r
p
r
ac
tical
ap
p
licatio
n
s
r
eq
u
ir
in
g
q
u
ic
k
r
esp
o
n
s
es.
4.
CO
NCLU
SI
O
N
T
h
is
p
ap
er
p
r
o
p
o
s
ed
a
n
o
v
el
ap
p
r
o
ac
h
to
NP
L
p
r
ed
ic
tio
n
b
y
le
v
er
ag
in
g
d
ee
p
Q
-
l
ea
r
n
i
n
g
,
a
r
ein
f
o
r
ce
m
e
n
t
lear
n
i
n
g
tec
h
n
iq
u
e,
in
s
tead
o
f
tr
ad
itio
n
al
ML
m
e
th
o
d
s
.
Un
li
k
e
co
n
v
en
ti
o
n
al
ap
p
r
o
ac
h
es
th
at
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
P
r
ed
ictin
g
n
o
n
-
p
erfo
r
min
g
lo
a
n
s
in
V
ietn
a
m’
s
fin
a
n
cia
l secto
r
:
a
d
ee
p
…
(
Lu
ye
n
A
n
h
Do
)
707
tr
ea
t
NP
L
p
r
ed
ictio
n
as
a
s
t
atic
class
if
icatio
n
p
r
o
b
le
m
,
o
u
r
m
e
th
o
d
m
o
d
els
th
e
p
r
o
b
le
m
as
a
s
eq
u
e
n
tia
l
d
ec
is
io
n
-
m
a
k
i
n
g
tas
k
,
w
h
er
e
th
e
a
g
en
t
lear
n
s
o
p
ti
m
al
lo
an
class
i
f
icat
io
n
s
tr
at
eg
ie
s
t
h
r
o
u
g
h
in
ter
ac
tio
n
w
i
th
th
e
en
v
ir
o
n
m
e
n
t.
O
u
r
r
esear
ch
ad
d
r
ess
ed
k
e
y
ch
a
llen
g
e
s
in
NP
L
p
r
ed
ictio
n
,
in
cl
u
d
in
g
cla
s
s
i
m
b
a
lan
ce
,
w
h
ic
h
o
f
ten
lead
s
to
p
o
o
r
r
ec
all
f
o
r
NP
L
ca
s
es.
B
y
d
esig
n
i
n
g
a
r
e
w
ar
d
f
u
n
ctio
n
t
h
at
p
en
a
lizes
m
i
s
clas
s
i
f
icati
o
n
s
m
o
r
e
h
ea
v
i
l
y
f
o
r
th
e
m
in
o
r
it
y
cla
s
s
(
NP
L
)
,
o
u
r
m
o
d
el
lear
n
s
to
f
o
cu
s
o
n
co
r
r
ec
tl
y
id
en
tify
i
n
g
h
i
g
h
-
r
i
s
k
lo
an
s
,
i
m
p
r
o
v
in
g
d
ec
is
io
n
-
m
a
k
i
n
g
ac
cu
r
ac
y
.
W
e
also
e
x
p
er
i
m
e
n
t
ed
w
it
h
d
ata
b
alan
c
in
g
tec
h
n
iq
u
es,
s
u
c
h
as o
v
er
-
s
a
m
p
li
n
g
(
S
MO
T
E
)
an
d
u
n
d
er
-
s
a
m
p
li
n
g
(
Nea
r
Miss
)
,
a
n
d
c
o
m
p
ar
ed
d
ee
p
Q
-
l
ea
r
n
in
g
with
tr
ad
itio
n
a
l
ML
m
o
d
els
(
lo
g
is
tic
r
e
g
r
ess
io
n
,
d
ec
is
io
n
tr
ee
s
,
SVM,
r
an
d
o
m
f
o
r
est
,
L
ig
h
tGB
M,
an
d
XGB
o
o
s
t)
.
Ou
r
r
esu
lts
,
b
ased
o
n
8
3
,
7
3
2
c
u
s
to
m
er
r
ec
o
r
d
s
f
r
o
m
a
lead
in
g
f
in
a
n
cial
i
n
s
t
itu
tio
n
i
n
Viet
n
a
m
(
2
0
1
9
-
2
0
2
2
)
,
d
em
o
n
s
tr
ate
t
h
at
d
ee
p
Q
-
l
ea
r
n
in
g
o
u
tp
er
f
o
r
m
s
tr
ad
iti
o
n
al
m
o
d
els,
p
ar
ticu
lar
l
y
i
n
h
an
d
lin
g
i
m
b
ala
n
ce
d
d
atasets
.
T
h
e
p
r
o
p
o
s
ed
a
p
p
r
o
a
ch
o
f
f
er
s
a
m
o
r
e
ad
ap
tiv
e
a
n
d
au
to
m
ated
NP
L
p
r
ed
ictio
n
f
r
a
m
e
w
o
r
k
,
r
ed
u
ci
n
g
m
an
u
al
r
i
s
k
ass
e
s
s
m
e
n
t
e
f
f
o
r
ts
an
d
i
m
p
r
o
v
i
n
g
lo
an
ap
p
r
o
v
al
ef
f
icie
n
c
y
.
Fo
r
th
e
f
u
tu
r
e
w
o
r
k
,
w
e
w
ill
e
x
p
lo
r
e
m
o
r
e
ad
v
a
n
ce
d
r
ein
f
o
r
ce
m
e
n
t
lear
n
in
g
tec
h
n
iq
u
es
lik
e
p
o
licy
g
r
ad
ien
t
m
et
h
o
d
s
to
en
h
an
c
e
s
tab
ilit
y
o
r
ap
p
l
y
ex
p
lain
ab
ili
t
y
tec
h
n
iq
u
es to
i
n
cr
ea
s
e
m
o
d
el
tr
an
s
p
ar
en
c
y
f
o
r
f
i
n
an
cia
l in
s
tit
u
tio
n
s
.
F
UNDIN
G
I
NF
O
RM
AT
I
O
N
T
h
is
r
esear
ch
is
f
u
n
d
ed
b
y
I
n
ter
n
at
io
n
al
Sch
o
o
l,
Viet
n
a
m
Natio
n
al
Un
iv
er
s
it
y
,
Han
o
i
(
VNU
-
I
S)
u
n
d
er
p
r
o
j
ec
t n
u
m
b
er
C
S.2
0
2
4
-
05.
AUTHO
R
CO
NT
RIB
UT
I
O
NS ST
A
T
E
M
E
NT
T
h
is
jo
u
r
n
al
u
s
e
s
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT
)
t
o
r
ec
o
g
n
ize
in
d
i
v
id
u
al
au
t
h
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
t
h
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
lla
b
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
L
u
y
e
n
An
h
Do
Hu
o
n
g
T
h
i V
iet
P
h
a
m
T
h
in
h
D
u
c
L
e
Oan
h
T
h
i T
r
an
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
si
s
I
:
I
n
v
e
st
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
si
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
t
h
o
r
s
s
tate
n
o
co
n
f
lic
t o
f
i
n
t
er
est.
DATA AV
AI
L
AB
I
L
I
T
Y
Der
iv
ed
d
ata
s
u
p
p
o
r
tin
g
th
e
f
in
d
in
g
s
o
f
t
h
is
s
t
u
d
y
ar
e
av
ai
lab
le
f
r
o
m
t
h
e
co
r
r
esp
o
n
d
in
g
au
th
o
r
o
n
r
eq
u
est.
RE
F
E
R
E
NC
E
S
[
1
]
N
.
V
.
C
h
a
w
l
a
,
K
.
W
.
B
o
w
y
e
r
,
L
.
O
.
H
a
l
l
,
a
n
d
W
.
P
.
K
e
g
e
l
mey
e
r
,
“
S
M
O
T
E:
sy
n
t
h
e
t
i
c
mi
n
o
r
i
t
y
o
v
e
r
-
samp
l
i
n
g
t
e
c
h
n
i
q
u
e
,
”
J
o
u
rn
a
l
o
f
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
Re
s
e
a
rc
h
,
v
o
l
.
1
6
,
p
p
.
3
2
1
–
3
5
7
,
J
u
n
.
2
0
0
2
,
d
o
i
:
1
0
.
1
6
1
3
/
j
a
i
r
.
9
5
3
.
[
2
]
D
.
El
r
e
e
d
y
a
n
d
A
.
F
.
A
t
i
y
a
,
“
A
c
o
mp
r
e
h
e
n
si
v
e
a
n
a
l
y
si
s
o
f
sy
n
t
h
e
t
i
c
mi
n
o
r
i
t
y
o
v
e
r
samp
l
i
n
g
t
e
c
h
n
i
q
u
e
(
S
M
O
T
E)
f
o
r
h
a
n
d
l
i
n
g
c
l
a
ss i
m
b
a
l
a
n
c
e
,
”
I
n
f
o
rm
a
t
i
o
n
S
c
i
e
n
c
e
s
,
v
o
l
.
5
0
5
,
p
p
.
3
2
–
6
4
,
D
e
c
.
2
0
1
9
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
n
s.
2
0
1
9
.
0
7
.
0
7
0
.
[
3
]
I
.
M
a
n
i
a
n
d
I
.
Z
h
a
n
g
,
“
k
N
N
a
p
p
r
o
a
c
h
t
o
u
n
b
a
l
a
n
c
e
d
d
a
t
a
d
i
st
r
i
b
u
t
i
o
n
s:
A
c
a
se
st
u
d
y
i
n
v
o
l
v
i
n
g
i
n
f
o
r
mat
i
o
n
e
x
t
r
a
c
t
i
o
n
,
”
2
0
0
3
.
[
4
]
I
.
M
.
A
l
k
h
a
w
a
l
d
e
h
,
I
.
A
l
b
a
l
k
h
i
,
a
n
d
A
.
J.
N
a
sw
h
a
n
,
“
C
h
a
l
l
e
n
g
e
s
a
n
d
l
i
mi
t
a
t
i
o
n
s
o
f
sy
n
t
h
e
t
i
c
m
i
n
o
r
i
t
y
o
v
e
r
samp
l
i
n
g
t
e
c
h
n
i
q
u
e
s
i
n
mac
h
i
n
e
l
e
a
r
n
i
n
g
,
”
Wo
r
l
d
J
o
u
rn
a
l
o
f
Me
t
h
o
d
o
l
o
g
y
,
v
o
l
.
1
3
,
n
o
.
5
,
p
p
.
3
7
3
–
3
7
8
,
D
e
c
.
2
0
2
3
,
d
o
i
:
1
0
.
5
6
6
2
/
w
j
m.v
1
3
.
i
5
.
3
7
3
.
[
5
]
M
.
C
a
r
v
a
l
h
o
,
A
.
J.
P
i
n
h
o
,
a
n
d
S
.
B
r
á
s,
“
R
e
sam
p
l
i
n
g
a
p
p
r
o
a
c
h
e
s
t
o
h
a
n
d
l
e
c
l
a
ss
i
m
b
a
l
a
n
c
e
:
a
r
e
v
i
e
w
f
r
o
m
a
d
a
t
a
p
e
r
sp
e
c
t
i
v
e
,
”
J
o
u
rn
a
l
o
f
Bi
g
D
a
t
a
,
v
o
l
.
1
2
,
n
o
.
1
,
p
.
7
1
,
M
a
r
.
2
0
2
5
,
d
o
i
:
1
0
.
1
1
8
6
/
s
4
0
5
3
7
-
0
2
5
-
0
1
1
1
9
-
4.
[
6
]
E.
L
i
n
,
Q
.
C
h
e
n
,
a
n
d
X
.
Q
i
,
“
D
e
e
p
r
e
i
n
f
o
r
c
e
me
n
t
l
e
a
r
n
i
n
g
f
o
r
i
mb
a
l
a
n
c
e
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
A
p
p
l
i
e
d
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
5
0
,
n
o
.
8
,
p
p
.
2
4
8
8
–
2
5
0
2
,
A
u
g
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
4
8
9
-
0
2
0
-
0
1
6
3
7
-
z.
[
7
]
“
L
o
g
i
st
i
c
r
e
g
r
e
ssi
o
n
,
”
i
n
E
n
c
y
c
l
o
p
e
d
i
a
o
f
M
a
c
h
i
n
e
L
e
a
rn
i
n
g
,
B
o
st
o
n
,
M
A
:
S
p
r
i
n
g
e
r
U
S
,
2
0
1
1
,
p
p
.
6
3
1
–
6
3
1
.
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
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
,
Vo
l.
41
,
No
.
2
,
Feb
r
u
ar
y
20
2
6
:
7
0
0
-
709
708
[
8
]
C.
-
Y
.
J
.
P
e
n
g
,
K
.
L
.
L
e
e
,
a
n
d
G
.
M
.
I
n
g
e
r
s
o
l
l
,
“
A
n
i
n
t
r
o
d
u
c
t
i
o
n
t
o
l
o
g
i
st
i
c
r
e
g
r
e
ssi
o
n
a
n
a
l
y
si
s
a
n
d
r
e
p
o
r
t
i
n
g
,
”
T
h
e
J
o
u
r
n
a
l
o
f
Ed
u
c
a
t
i
o
n
a
l
Re
s
e
a
r
c
h
,
v
o
l
.
9
6
,
n
o
.
1
,
p
p
.
3
–
1
4
,
S
e
p
.
2
0
0
2
,
d
o
i
:
1
0
.
1
0
8
0
/
0
0
2
2
0
6
7
0
2
0
9
5
9
8
7
8
6
.
[
9
]
S
.
S
p
e
r
a
n
d
e
i
,
“
U
n
d
e
r
st
a
n
d
i
n
g
l
o
g
i
st
i
c
r
e
g
r
e
ssi
o
n
a
n
a
l
y
si
s,”
Bi
o
c
h
e
m
i
a
M
e
d
i
c
a
,
p
p
.
1
2
–
1
8
,
2
0
1
4
,
d
o
i
:
1
0
.
1
1
6
1
3
/
B
M
.
2
0
1
4
.
0
0
3
.
[
1
0
]
Y
.
Y
.
S
o
n
g
a
n
d
Y
.
L
u
,
“
D
e
c
i
si
o
n
t
r
e
e
me
t
h
o
d
s:
a
p
p
l
i
c
a
t
i
o
n
s
f
o
r
c
l
a
ssi
f
i
c
a
t
i
o
n
a
n
d
p
r
e
d
i
c
t
i
o
n
,
”
S
h
a
n
g
h
a
i
Arc
h
i
v
e
s
o
f
P
syc
h
i
a
t
ry
,
v
o
l
.
2
7
,
n
o
.
2
,
p
p
.
1
3
0
–
1
3
5
,
2
0
1
5
,
d
o
i
:
1
0
.
1
1
9
1
9
/
j
.
i
ss
n
.
1
0
0
2
-
0
8
2
9
.
2
1
5
0
4
4
.
[
1
1
]
I
.
D
.
M
i
e
n
y
e
a
n
d
N
.
Je
r
e
,
“
A
su
r
v
e
y
o
f
d
e
c
i
si
o
n
t
r
e
e
s:
c
o
n
c
e
p
t
s
,
a
l
g
o
r
i
t
h
ms,
a
n
d
a
p
p
l
i
c
a
t
i
o
n
s
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
8
6
7
1
6
–
8
6
7
2
7
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
4
1
6
8
3
8
.
[
1
2
]
H
.
B
l
o
c
k
e
e
l
,
L
.
D
e
v
o
s,
B
.
F
r
é
n
a
y
,
G
.
N
a
n
f
a
c
k
,
a
n
d
S
.
N
i
j
sse
n
,
“
D
e
c
i
si
o
n
t
r
e
e
s:
f
r
o
m
e
f
f
i
c
i
e
n
t
p
r
e
d
i
c
t
i
o
n
t
o
r
e
sp
o
n
si
b
l
e
A
I
,
”
Fro
n
t
i
e
rs
i
n
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
6
,
Ju
l
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
8
9
/
f
r
a
i
.
2
0
2
3
.
1
1
2
4
5
5
3
.
[
1
3
]
J.
.
Q
U
I
N
L
A
N
,
“
S
i
mp
l
i
f
y
i
n
g
d
e
c
i
si
o
n
t
r
e
e
s,”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
rn
a
l
o
f
H
u
m
a
n
-
C
o
m
p
u
t
e
r
S
t
u
d
i
e
s
,
v
o
l
.
5
1
,
n
o
.
2
,
p
p
.
4
9
7
–
5
1
0
,
A
u
g
.
1
9
9
9
,
d
o
i
:
1
0
.
1
0
0
6
/
i
j
h
c
.
1
9
8
7
.
0
3
2
1
.
[
1
4
]
M
.
S
c
h
o
n
l
a
u
a
n
d
R
.
Y
.
Z
o
u
,
“
Th
e
r
a
n
d
o
m
f
o
r
e
st
a
l
g
o
r
i
t
h
m
f
o
r
st
a
t
i
st
i
c
a
l
l
e
a
r
n
i
n
g
,
”
T
h
e
S
t
a
t
a
J
o
u
rn
a
l
:
Pr
o
m
o
t
i
n
g
c
o
m
m
u
n
i
c
a
t
i
o
n
s
o
n
s
t
a
t
i
s
t
i
c
s
a
n
d
S
t
a
t
a
,
v
o
l
.
2
0
,
n
o
.
1
,
p
p
.
3
–
2
9
,
M
a
r
.
2
0
2
0
,
d
o
i
:
1
0
.
1
1
7
7
/
1
5
3
6
8
6
7
X
2
0
9
0
9
6
8
8
.
[
1
5
]
Z
.
S
u
n
,
G
.
W
a
n
g
,
P
.
L
i
,
H
.
W
a
n
g
,
M
.
Z
h
a
n
g
,
a
n
d
X
.
L
i
a
n
g
,
“
A
n
i
mp
r
o
v
e
d
r
a
n
d
o
m
f
o
r
e
st
b
a
se
d
o
n
t
h
e
c
l
a
ssi
f
i
c
a
t
i
o
n
a
c
c
u
r
a
c
y
a
n
d
c
o
r
r
e
l
a
t
i
o
n
me
a
su
r
e
me
n
t
o
f
d
e
c
i
si
o
n
t
r
e
e
s,”
Ex
p
e
rt
S
y
st
e
m
s
w
i
t
h
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
2
3
7
,
p
.
1
2
1
5
4
9
,
M
a
r
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
sw
a
.
2
0
2
3
.
1
2
1
5
4
9
.
[
1
6
]
H
.
A
.
S
a
l
man
,
A
.
K
a
l
a
k
e
c
h
,
a
n
d
A
.
S
t
e
i
t
i
,
“
R
a
n
d
o
m
f
o
r
e
st
a
l
g
o
r
i
t
h
m
o
v
e
r
v
i
e
w
,
”
Ba
b
y
l
o
n
i
a
n
J
o
u
r
n
a
l
o
f
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
,
v
o
l
.
2
0
2
4
,
p
p
.
6
9
–
7
9
,
Ju
n
.
2
0
2
4
,
d
o
i
:
1
0
.
5
8
4
9
6
/
B
J
M
L
/
2
0
2
4
/
0
0
7
.
[
1
7
]
L
.
B
r
e
i
man
,
“
R
a
n
d
o
m
f
o
r
e
st
s,”
Ma
c
h
i
n
e
L
e
a
r
n
i
n
g
,
v
o
l
.
4
5
,
n
o
.
1
,
p
p
.
5
–
3
2
,
O
c
t
.
2
0
0
1
,
d
o
i
:
1
0
.
1
0
2
3
/
A
:
1
0
1
0
9
3
3
4
0
4
3
2
4
.
[
1
8
]
T
.
Ev
g
e
n
i
o
u
a
n
d
M
.
P
o
n
t
i
l
,
“
S
u
p
p
o
r
t
v
e
c
t
o
r
mac
h
i
n
e
s
:
t
h
e
o
r
y
a
n
d
a
p
p
l
i
c
a
t
i
o
n
s,”
i
n
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
a
n
d
I
t
s
A
p
p
l
i
c
a
t
i
o
n
s
,
2
0
0
1
,
p
p
.
2
4
9
–
2
5
7
.
[
1
9
]
J.
C
e
r
v
a
n
t
e
s,
F
.
G
a
r
c
i
a
-
L
a
mo
n
t
,
L
.
R
o
d
r
í
g
u
e
z
-
M
a
z
a
h
u
a
,
a
n
d
A
.
L
o
p
e
z
,
“
A
c
o
mp
r
e
h
e
n
si
v
e
su
r
v
e
y
o
n
su
p
p
o
r
t
v
e
c
t
o
r
ma
c
h
i
n
e
c
l
a
ssi
f
i
c
a
t
i
o
n
:
A
p
p
l
i
c
a
t
i
o
n
s,
c
h
a
l
l
e
n
g
e
s
a
n
d
t
r
e
n
d
s,
”
N
e
u
r
o
c
o
m
p
u
t
i
n
g
,
v
o
l
.
4
0
8
,
p
p
.
1
8
9
–
2
1
5
,
S
e
p
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
n
e
u
c
o
m.
2
0
1
9
.
1
0
.
1
1
8
.
[
2
0
]
C
.
C
o
r
t
e
s
a
n
d
V
.
V
a
p
n
i
k
,
“
S
u
p
p
o
r
t
-
v
e
c
t
o
r
n
e
t
w
o
r
k
s,”
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
,
v
o
l
.
2
0
,
n
o
.
3
,
p
p
.
2
7
3
–
2
9
7
,
S
e
p
.
1
9
9
5
,
d
o
i
:
1
0
.
1
0
2
3
/
A
:
1
0
2
2
6
2
7
4
1
1
4
1
1
.
[
2
1
]
G
K
e
e
t
a
l
.
,
“
L
i
g
h
t
g
b
m:
A
h
i
g
h
l
y
e
f
f
i
c
i
e
n
t
g
r
a
d
i
e
n
t
b
o
o
st
i
n
g
d
e
c
i
si
o
n
t
r
e
e
,
”
A
d
v
a
n
c
e
s
i
n
n
e
u
r
a
l
i
n
f
o
rm
a
t
i
o
n
p
r
o
c
e
ssi
n
g
s
y
st
e
m
s
,
v
o
l
.
3
0
,
p
p
.
3
1
4
6
–
3
1
5
4
,
2
0
1
7
.
[
2
2
]
M
.
R
.
M
a
c
h
a
d
o
,
S
.
K
a
r
r
a
y
,
a
n
d
I
.
T
.
d
e
S
o
u
sa,
“
L
i
g
h
t
G
B
M
:
a
n
e
f
f
e
c
t
i
v
e
d
e
c
i
si
o
n
t
r
e
e
g
r
a
d
i
e
n
t
b
o
o
st
i
n
g
me
t
h
o
d
t
o
p
r
e
d
i
c
t
c
u
s
t
o
me
r
l
o
y
a
l
t
y
i
n
t
h
e
f
i
n
a
n
c
e
i
n
d
u
st
r
y
,
”
i
n
2
0
1
9
1
4
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
&
Ed
u
c
a
t
i
o
n
(
I
C
C
S
E)
,
A
u
g
.
2
0
1
9
,
p
p
.
1
1
1
1
–
1
1
1
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
C
S
E.
2
0
1
9
.
8
8
4
5
5
2
9
.
[
2
3
]
L.
-
H
.
L
i
,
A
.
K
.
S
h
a
r
ma
,
a
n
d
S
.
-
T
.
C
h
e
n
g
,
“
Ex
p
l
a
i
n
a
b
l
e
A
I
b
a
se
d
L
i
g
h
t
G
B
M
p
r
e
d
i
c
t
i
o
n
mo
d
e
l
t
o
p
r
e
d
i
c
t
d
e
f
a
u
l
t
b
o
r
r
o
w
e
r
i
n
so
c
i
a
l
l
e
n
d
i
n
g
p
l
a
t
f
o
r
m,”
I
n
t
e
l
l
i
g
e
n
t
S
y
st
e
m
s
w
i
t
h
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
2
6
,
p
.
2
0
0
5
1
4
,
J
u
n
.
2
0
2
5
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
sw
a
.
2
0
2
5
.
2
0
0
5
1
4
.
[
2
4
]
T
.
C
h
e
n
e
t
a
l
.
,
“
X
G
B
o
o
st
:
e
x
t
r
e
me
g
r
a
d
i
e
n
t
b
o
o
st
i
n
g
,
”
CR
A
N
:
C
o
n
t
ri
b
u
t
e
d
Pa
c
k
a
g
e
s
.
S
e
p
.
0
1
,
2
0
1
4
,
d
o
i
:
1
0
.
3
2
6
1
4
/
C
R
A
N
.
p
a
c
k
a
g
e
.
x
g
b
o
o
st
.
[
2
5
]
Z
.
A
r
i
f
A
l
i
,
Z
.
H
.
A
b
d
u
l
j
a
b
b
a
r
,
H
.
A
.
T
a
h
i
r
,
A
.
B
i
b
o
S
a
l
l
o
w
,
a
n
d
S
.
M
.
A
l
mu
f
t
i
,
“
e
X
t
r
e
me
g
r
a
d
i
e
n
t
b
o
o
st
i
n
g
a
l
g
o
r
i
t
h
m
w
i
t
h
mac
h
i
n
e
l
e
a
r
n
i
n
g
:
a
r
e
v
i
e
w
,
”
Ac
a
d
e
m
i
c
J
o
u
rn
a
l
o
f
N
a
w
ro
z
U
n
i
v
e
rs
i
t
y
,
v
o
l
.
1
2
,
n
o
.
2
,
p
p
.
3
2
0
–
3
3
4
,
M
a
y
2
0
2
3
,
d
o
i
:
1
0
.
2
5
0
0
7
/
a
j
n
u
.
v
1
2
n
2
a
1
6
1
2
.
[
2
6
]
D
.
T
a
r
w
i
d
i
,
S
.
R
.
P
u
d
j
a
p
r
a
se
t
y
a
,
D
.
A
d
y
t
i
a
,
a
n
d
M
.
A
p
r
i
,
“
A
n
o
p
t
i
m
i
z
e
d
X
G
B
o
o
st
-
b
a
se
d
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
me
t
h
o
d
f
o
r
p
r
e
d
i
c
t
i
n
g
w
a
v
e
r
u
n
-
u
p
o
n
a
sl
o
p
i
n
g
b
e
a
c
h
,
”
Me
t
h
o
d
sX
,
v
o
l
.
1
0
,
p
.
1
0
2
1
1
9
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
me
x
.
2
0
2
3
.
1
0
2
1
1
9
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Lu
y
e
n
A
n
h
D
o
g
o
t
h
is
b
a
c
h
e
l
o
r
d
e
g
re
e
o
n
b
u
sin
e
ss
d
a
ta
a
n
a
ly
t
ics
in
2
0
2
3
.
No
w
,
h
e
is
w
o
rk
in
g
a
s
a
sc
ien
ti
st
a
t
Big
Da
ta
De
p
a
rt
m
e
n
t
o
f
V
ietin
Ba
n
k
in
V
iet
n
a
m
.
His
m
a
in
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
d
a
ta
sc
ien
c
e
,
AI
,
a
n
d
ML
to
w
a
rd
s
th
e
a
p
p
li
c
a
ti
o
n
s
in
b
u
si
n
e
ss
f
ield
s.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
lu
y
e
n
3
1
1
2
2
0
0
1
@g
m
a
il
.
c
o
m
.
H
u
o
n
g
T
h
i
V
iet
Ph
a
m
o
b
tain
e
d
h
e
r
B.
S
c
.
i
n
e
lec
tri
c
a
l
e
n
g
in
e
e
rin
g
f
ro
m
H
a
n
o
i
Un
iv
e
rsit
y
o
f
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
lo
g
y
in
2
0
0
7
.
S
h
e
g
o
t
h
e
r
M
.
Sc
.
a
n
d
P
h
.
D
.
,
b
o
t
h
i
n
e
lec
tri
c
a
l
e
n
g
in
e
e
rin
g
,
f
ro
m
Un
iv
e
r
sit
y
o
f
M
a
ss
a
c
h
u
se
tt
s
L
o
w
e
ll
in
th
e
U
n
it
e
d
S
tate
s,
in
2
0
1
0
a
n
d
2
0
1
2
.
F
ro
m
2
0
1
2
to
2
0
1
5
,
sh
e
w
a
s
a
re
se
a
rc
h
e
r
in
th
e
M
a
n
n
in
g
S
c
h
o
o
l
o
f
Bu
sin
e
ss
,
L
o
w
e
ll
,
M
a
ss
a
c
h
u
se
tt
s.
F
ro
m
2
0
1
7
-
20
2
0
,
sh
e
w
a
s
th
e
fa
c
u
lt
y
o
f
V
NU
Un
iv
e
rsit
y
o
f
En
g
in
e
e
rin
g
a
n
d
T
e
c
h
n
o
lo
g
y
,
V
ietn
a
m
(V
NU
-
UET
).
S
in
c
e
2
0
2
0
,
s
h
e
w
o
rk
s in
In
ter
n
a
ti
o
n
a
l
S
c
h
o
o
l
–
V
NU
.
S
h
e
is
in
tere
ste
d
in
d
a
ta
m
in
in
g
a
n
d
a
n
a
ly
ti
c
s,
m
a
c
h
in
e
lea
rn
in
g
m
e
th
o
d
o
lo
g
ies
,
w
it
h
a
p
p
li
c
a
ti
o
n
s
in
b
io
m
e
d
ica
l
e
n
g
in
e
e
rin
g
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
h
u
o
n
g
p
v
@v
n
u
.
e
d
u
.
v
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2
5
0
2
-
4
7
52
P
r
ed
ictin
g
n
o
n
-
p
erfo
r
min
g
lo
a
n
s
in
V
ietn
a
m’
s
fin
a
n
cia
l secto
r
:
a
d
ee
p
…
(
Lu
ye
n
A
n
h
Do
)
709
Th
in
h
D
u
c
Le
is
a
lec
tu
re
r
a
t
In
tern
a
ti
o
n
a
l
S
c
h
o
o
l,
V
iet
n
a
m
Na
ti
o
n
a
l
Un
iv
e
rsity
,
Ha
No
i,
V
iet
n
a
m
.
He
tea
c
h
e
s
m
a
th
e
m
a
ti
c
s
in
E
n
g
li
sh
f
o
r
b
u
sin
e
ss
m
a
jo
rs.
He
o
b
tai
n
e
d
h
is
Ba
c
h
e
lo
r
d
e
g
re
e
in
2
0
0
1
a
n
d
M
a
ste
r
d
e
g
re
e
in
2
0
0
4
,
b
o
t
h
i
n
m
a
th
e
m
a
ti
c
s
a
t
Ha
No
i
Na
ti
o
n
a
l
Un
iv
e
rsit
y
o
f
Ed
u
c
a
ti
o
n
,
V
ietn
a
m
.
H
e
o
b
tain
e
d
h
is
P
h
.
D
.
d
e
g
re
e
in
2
0
1
2
a
t
t
h
e
P
e
n
n
sy
lv
a
n
ia
S
tate
Un
iv
e
rsity
,
USA
.
His
re
se
a
rc
h
i
n
tere
sts
in
c
l
u
d
e
m
a
c
h
in
e
lea
rn
in
g
,
m
a
th
e
m
a
ti
c
a
l
f
in
a
n
c
e
,
a
n
d
e
c
o
n
o
m
ic sta
ti
stics
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
th
in
h
ld
@v
n
u
.
e
d
u
.
v
n
.
O
a
n
h
T
h
i
Tr
a
n
g
o
t
th
e
b
a
c
h
e
lo
r
a
n
d
m
a
ste
r
d
e
g
re
e
s
in
c
o
m
p
u
ter
sc
ien
c
e
a
t
t
h
e
Un
iv
e
rsit
y
o
f
En
g
in
e
e
rin
g
a
n
d
T
e
c
h
n
o
l
o
g
y
,
V
ietn
a
m
Na
ti
o
n
a
l
Un
iv
e
rsit
y
,
Ha
n
o
i
i
n
2
0
0
6
a
n
d
2
0
0
9
,
re
sp
e
c
ti
v
e
ly
.
S
h
e
w
a
s
a
w
a
rd
e
d
a
Ja
p
a
n
e
se
G
o
v
e
rn
m
e
n
t
S
c
h
o
lars
h
i
p
to
p
u
rsu
e
P
h
.
D.
i
n
Co
m
p
u
ter
S
c
ien
c
e
a
t
Ja
p
a
n
A
d
v
a
n
c
e
d
In
stit
u
te
o
f
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
lo
g
y
(J
A
IS
T
)
f
ro
m
2
0
1
1
to
2
0
1
4
.
Cu
rre
n
tl
y
,
sh
e
is a l
e
c
tu
re
r
a
t
th
e
In
tern
a
ti
o
n
a
l
S
c
h
o
o
l
o
f
V
ietn
a
m
Na
ti
o
n
a
l
Un
iv
e
rsity
,
Ha
n
o
i
(
V
NU
-
IS
).
He
r
m
a
in
re
se
a
rc
h
in
tere
sts
a
re
AI
a
n
d
ML
.
He
r
c
o
n
tri
b
u
ti
o
n
s
t
o
t
h
e
f
ield
in
c
lu
d
e
5
0
p
u
b
li
c
a
ti
o
n
s
in
e
ste
e
m
e
d
jo
u
r
n
a
ls
a
n
d
c
o
n
f
e
re
n
c
e
s.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
o
a
n
h
t
t@g
m
a
il
.
c
o
m
o
r
tran
th
io
a
n
h
@v
n
u
.
e
d
u
.
v
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.