I
AE
S
I
nte
rna
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
8
,
No
.
4
,
Dece
m
b
er
201
9
,
p
p
.
4
22
~
4
28
I
SS
N:
2252
-
8938
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ai.
v
8
.i
4
.
p
p
4
22
-
4
28
ļ²
422
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//
ija
i
.
ia
es
co
r
e
.
co
m
H
y
bridisa
tion o
f
RF(
X
g
b
)
to i
m
pr
o
v
e t
he t
ree
-
ba
se
d alg
o
rith
m
s
in learning
style
p
rediction
H
a
ziqa
h
Sh
a
m
s
u
din
,
M
a
zia
ni Sa
bu
d
in
,
U
m
i K
a
ls
o
m
Y
u
s
o
f
S
c
h
o
o
l
o
f
Co
m
p
u
ter S
c
ien
c
e
,
1
1
8
0
0
U
n
iv
e
rsiti
S
a
in
s
M
a
lay
sia
,
P
u
l
a
u
P
in
a
n
g
,
M
a
lay
sia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Au
g
2
1
,
2
0
1
9
R
ev
i
s
ed
No
v
1
0
,
2
0
1
9
A
cc
ep
ted
No
v
2
3
,
2
0
1
9
T
h
is
p
a
p
e
r
p
re
se
n
ts
h
y
b
rid
iza
ti
o
n
o
f
Ra
n
d
o
m
F
o
re
st
(RF
)
a
n
d
Ex
trem
e
G
ra
d
ien
t
Bo
o
stin
g
(
X
g
b
),
n
a
m
e
d
RF
(
X
g
b
)
t
o
im
p
ro
v
e
th
e
tree
-
b
a
se
d
a
lg
o
rit
h
m
s
in
lea
rn
in
g
sty
le
p
re
d
ictio
n
.
L
e
a
rn
in
g
sty
le
o
f
sp
e
c
i
f
ic
u
se
rs
in
a
n
o
n
li
n
e
lea
rn
in
g
sy
ste
m
is
d
e
ter
m
i
n
e
d
b
a
se
d
o
n
th
e
ir
i
n
tera
c
ti
o
n
a
n
d
b
e
h
a
v
io
r
to
w
a
rd
s
th
e
s
y
ste
m
.
T
h
e
m
o
st
c
o
m
m
o
n
o
n
li
n
e
lea
rn
i
n
g
th
e
o
r
y
u
se
d
in
d
e
term
in
in
g
th
e
lea
rn
i
n
g
sty
le
is
th
e
F
e
ld
e
r
-
S
il
v
e
rm
a
n
ās
L
e
a
rn
in
g
S
ty
le
M
o
d
e
l
(F
S
L
S
M
).
M
a
n
y
re
s
e
a
rc
h
e
rs
h
a
v
e
p
ro
p
o
se
d
m
a
c
h
in
e
lea
rn
in
g
a
lg
o
rit
h
m
s
to
e
sta
b
li
sh
lea
rn
i
n
g
sty
le
b
y
u
sin
g
th
e
lo
g
f
il
e
a
tt
rib
u
tes
.
T
h
i
s
h
e
lp
s
i
n
d
e
term
in
in
g
th
e
lea
rn
i
n
g
sty
le
a
u
to
m
a
ti
c
a
ll
y
.
Ho
w
e
v
e
r,
c
u
rre
n
t
re
se
a
rc
h
e
s
stil
l
p
e
rf
o
r
m
p
o
o
rly
,
w
h
e
re
th
e
ra
n
g
e
o
f
a
c
c
u
ra
c
y
is
b
e
twe
e
n
5
8
%
-
8
9
%
.
He
n
c
e
,
RF
(X
g
b
)
is
p
ro
p
o
se
d
t
o
h
e
l
p
in
i
m
p
ro
v
in
g
th
e
lea
rn
in
g
st
y
le
p
re
d
ictio
n
.
T
h
is
h
y
b
rid
a
lg
o
rit
h
m
w
a
s
f
u
rth
e
r
e
n
h
a
n
c
e
d
b
y
o
p
ti
m
izin
g
it
s
p
a
ra
m
e
ter
s
.
F
ro
m
th
e
e
x
p
e
ri
m
e
n
ts,
RF
(X
g
b
)
w
a
s
p
ro
v
e
n
to
b
e
m
o
re
e
ff
e
c
ti
v
e
,
w
it
h
a
c
c
u
ra
c
y
o
f
96%
c
o
m
p
a
re
d
to
J4
8
a
n
d
L
S
ID
-
A
NN
a
lg
o
rit
h
m
f
ro
m
p
re
v
io
u
s li
tera
tu
re
.
K
ey
w
o
r
d
s
:
H
y
b
r
id
H
y
p
er
p
ar
am
e
ter
Op
ti
m
izat
io
n
L
ea
r
n
i
n
g
St
y
le
On
li
n
e
L
ea
r
n
i
n
g
R
an
d
o
m
Fo
r
est
Co
p
y
rig
h
t
Ā©
2
0
1
9
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Ma
zian
i Sab
u
d
in
,
Sch
o
o
l o
f
C
o
m
p
u
ter
Scie
n
ce
,
1
1
8
0
0
Un
iv
er
s
it
i Sai
n
s
Ma
la
y
s
ia
,
P
u
lau
P
in
a
n
g
.
E
m
ail:
m
az
ia
n
i
@
u
s
m
.
m
y
1.
I
NT
RO
D
UCT
I
O
N
L
ea
r
n
i
n
g
s
t
y
le
is
k
n
o
w
n
a
s
w
a
y
o
f
lear
n
i
n
g
o
r
p
r
ef
er
en
ce
b
y
th
e
lear
n
er
o
n
h
o
w
m
a
ter
ials
ar
e
p
r
esen
ted
,
h
o
w
to
w
o
r
k
w
it
h
it
an
d
h
o
w
to
in
ter
n
alize
i
n
f
o
r
m
atio
n
[1
-
2]
.
I
d
en
tify
in
g
a
s
t
u
d
en
t
ā
s
lear
n
i
n
g
s
t
y
le
h
as
s
e
v
er
al
b
en
e
f
it
s
,
s
u
ch
a
s
m
ak
in
g
s
tu
d
e
n
ts
a
w
ar
e
o
f
th
eir
s
t
r
en
g
t
h
an
d
w
ea
k
n
ess
e
s
w
h
en
i
t
co
m
e
s
to
lear
n
i
n
g
.
I
t
is
al
s
o
m
ea
n
t
to
b
e
u
s
ed
i
n
d
eter
m
i
n
i
n
g
t
h
e
lear
n
in
g
p
r
ef
er
en
ce
s
o
f
ea
c
h
s
t
u
d
en
t
ei
t
h
er
in
a
tr
ad
itio
n
al
class
r
o
o
m
o
r
t
h
r
o
u
g
h
a
n
o
n
li
n
e
lear
n
i
n
g
b
ased
s
y
s
te
m
[
3
]
.
An
o
n
l
in
e
lear
n
i
n
g
b
ased
s
y
s
te
m
ca
n
b
e
d
ef
i
n
ed
as
an
o
n
li
n
e
s
y
s
te
m
w
h
er
e
th
er
e
is
an
in
ter
ac
tio
n
b
et
w
ee
n
s
tu
d
e
n
ts
a
n
d
s
y
s
te
m
[
4
]
.
I
n
itiall
y
,
in
an
o
n
lin
e
lear
n
i
n
g
b
ased
s
y
s
te
m
,
th
e
lear
n
i
n
g
s
t
y
l
e
o
f
th
e
u
s
er
is
d
eter
m
i
n
ed
b
y
u
s
i
n
g
av
ailab
le
lear
n
i
n
g
s
t
y
le
q
u
esti
o
n
n
air
es
b
ased
o
n
s
elec
ted
lear
n
in
g
s
t
y
le
m
o
d
el
an
d
th
e
m
o
s
t
co
m
m
o
n
l
y
u
s
ed
lear
n
in
g
s
t
y
le
m
o
d
el
is
th
e
FS
L
SM
w
h
ic
h
also
in
co
r
p
o
r
ates
d
if
f
er
e
n
t
ele
m
e
n
t
s
f
r
o
m
d
if
f
er
e
n
t
lear
n
in
g
s
t
y
le
m
o
d
els
s
u
c
h
a
s
[
5
]
.
Ho
w
e
v
er
,
w
h
e
n
s
tu
d
e
n
ts
ar
e
ask
ed
to
f
i
ll
in
th
e
q
u
e
s
tio
n
n
air
e,
th
e
y
ta
k
e
lo
n
g
er
ti
m
e
t
o
f
ill
it
a
s
t
h
e
q
u
esti
o
n
s
ar
e
lo
n
g
a
n
d
th
e
y
r
ef
u
s
e
s
p
en
d
in
g
to
o
m
u
c
h
ti
m
e
o
n
t
h
e
q
u
esti
o
n
n
air
e
w
h
ich
ca
u
s
e
s
th
e
m
to
p
u
t
in
r
an
d
o
m
a
n
s
w
er
s
[
6
]
.
T
h
er
ef
o
r
e,
r
esear
ch
er
s
ca
m
e
o
u
t
w
it
h
a
n
a
lter
n
ati
v
e
w
h
er
e
t
h
e
y
d
eter
m
i
n
e
th
e
lear
n
i
n
g
s
t
y
le
a
u
to
m
atica
ll
y
[
5
]
.
T
h
is
i
s
d
o
n
e
b
y
co
llecti
n
g
lo
g
f
i
les
o
f
th
e
in
ter
ac
ti
v
e
b
eh
av
io
r
o
f
th
e
u
s
er
w
i
th
th
e
s
y
s
te
m
.
T
h
e
co
n
ten
t
o
f
th
e
lo
g
f
ile
s
co
n
s
is
ts
o
f
s
e
v
er
al
r
elate
d
attr
ib
u
tes
m
atc
h
ed
to
th
e
s
y
s
te
m
s
u
c
h
as
th
e
n
u
m
b
er
o
f
v
i
s
its
,
ch
ar
ac
ter
is
tic
s
an
d
t
y
p
es
o
f
o
b
j
ec
ts
ch
o
s
e
n
,
s
eq
u
en
ce
s
o
f
ac
tio
n
s
an
d
s
e
lecte
d
s
ea
r
ch
ter
m
s
,
n
u
m
b
er
o
f
v
is
i
ts
,
ti
m
e
s
p
en
t
an
d
p
er
f
o
r
m
a
n
ce
.
I
t
al
s
o
in
cl
u
d
es
t
h
e
ac
ti
v
itie
s
tr
ac
k
ed
s
u
c
h
as
th
e
s
ea
r
ch
i
n
g
,
e
n
r
o
ll
i
n
e
x
a
m
,
q
u
iz,
s
elf
-
a
s
s
es
s
m
en
t
test
,
u
s
in
g
f
o
r
u
m
,
s
e
n
d
in
g
e
m
a
il a
n
d
d
is
cu
s
s
io
n
b
o
ar
d
in
clu
d
i
n
g
r
ea
d
in
g
o
r
d
o
w
n
lo
ad
in
g
o
f
m
ater
ials
f
r
o
m
t
h
e
s
y
s
te
m
[
5
,
7
]
.
T
h
ese
attr
ib
u
tes
w
er
e
th
e
n
m
atc
h
ed
w
it
h
th
e
lear
n
in
g
s
t
y
le
m
o
d
el.
T
h
en
,
th
e
r
esu
lt
is
f
u
r
th
er
an
al
y
ze
d
u
s
i
n
g
m
ac
h
i
n
e
lear
n
i
n
g
al
g
o
r
ith
m
s
u
n
til t
h
e
lear
n
i
n
g
s
t
y
le
o
f
t
h
e
u
s
er
is
d
eter
m
i
n
e
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
ļ²
Hyb
r
id
is
a
tio
n
o
f R
F
(
X
g
b
)
to
imp
r
o
ve
th
e
tr
ee
-
b
a
s
ed
a
lg
o
r
ith
ms in
... (
Ha
z
iq
a
h
S
h
a
msu
d
in
)
423
R
esear
ch
er
s
h
av
e
ap
p
lied
s
o
m
e
w
id
el
y
u
s
ed
tech
n
iq
u
es
s
u
ch
as
A
r
ti
f
icial
Ne
u
r
al
Net
w
o
r
k
(
A
NN)
,
NaĆÆv
e
B
a
y
e
s
(
NB
)
an
d
Dec
i
s
i
o
n
T
r
ee
(
D
T
)
,
an
d
f
u
zz
y
lo
g
ic
in
p
r
ed
ictin
g
s
tu
d
e
n
t
ac
ad
e
m
i
c
p
er
f
o
r
m
an
ce
[
8
]
.
T
r
ee
b
ased
lear
n
in
g
alg
o
r
it
h
m
s
ar
e
co
n
s
id
er
ed
to
b
e
o
n
e
o
f
th
e
b
es
t
an
d
m
o
s
tl
y
u
s
ed
s
u
p
er
v
i
s
ed
lear
n
i
n
g
m
et
h
o
d
s
.
T
r
ee
b
ased
m
et
h
o
d
s
e
m
p
o
w
er
p
r
ed
ictiv
e
m
o
d
els
w
it
h
h
i
g
h
ac
cu
r
ac
y
,
s
ta
b
ilit
y
an
d
ea
s
e
o
f
in
ter
p
r
etatio
n
.
Un
l
ik
e
li
n
ea
r
m
o
d
el
s
,
th
e
y
m
ap
n
o
n
-
l
in
ea
r
r
elatio
n
s
h
ip
s
q
u
ite
w
el
l.
Me
th
o
d
s
lik
e
R
F
a
n
d
Xg
b
ar
e
u
s
ed
in
all
k
in
d
s
o
f
d
ata
s
cien
ce
p
r
o
b
lem
s
[
9
]
.
Fro
m
p
r
ev
io
u
s
r
esear
ch
,
th
er
e
ar
e
t
w
o
p
ap
er
s
in
lear
n
in
g
s
t
y
le
p
r
ed
ictio
n
t
h
at
u
s
e
d
ec
is
io
n
tr
ee
alg
o
r
it
h
m
s
[
1
0
-
1
1
]
.
B
o
th
o
f
th
ese
p
ap
er
s
m
an
a
g
e
to
in
cr
ea
s
e
t
h
e
p
er
ce
n
tag
e
o
f
ac
cu
r
ac
y
i
n
lear
n
in
g
s
t
y
le
p
r
ed
ict
io
n
co
m
p
ar
e
d
to
p
r
ev
io
u
s
p
ap
er
s
.
Ho
w
ev
er
,
th
er
e
ar
e
s
ti
ll
a
g
ap
w
it
h
i
n
t
h
e
u
s
a
g
e
o
f
th
e
s
tated
alg
o
r
ith
m
i
n
ter
m
s
o
f
t
h
e
ac
c
u
r
ac
y
o
f
t
h
e
r
es
u
lt o
b
tai
n
ed
.
On
e
o
f
th
e
ap
p
r
o
ac
h
es
to
en
h
a
n
ce
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
al
g
o
r
ith
m
s
,
i
s
b
y
p
er
f
o
r
m
i
n
g
h
y
p
er
p
ar
a
m
eter
o
p
ti
m
iz
atio
n
in
t
h
e
s
elec
ted
alg
o
r
ith
m
s
.
H
y
p
er
p
ar
a
m
e
ter
o
p
ti
m
izatio
n
is
th
e
p
r
o
ce
s
s
o
f
ch
o
o
s
in
g
a
s
et
o
f
o
p
ti
m
al
h
y
p
er
p
ar
am
eter
s
f
o
r
a
lear
n
in
g
a
lg
o
r
it
h
m
.
I
d
en
ti
f
y
i
n
g
a
g
o
o
d
v
alu
e
f
o
r
h
y
p
er
p
ar
am
eter
s
,
Ī»
w
h
er
e
Ī»
=
p
ar
a
m
eter
,
is
ca
lled
h
y
p
er
p
ar
a
m
eter
o
p
ti
m
izat
io
n
[
1
2
]
.
T
h
e
cr
itical
s
tep
in
h
y
p
er
p
ar
am
eter
o
p
ti
m
izat
io
n
is
to
ch
o
o
s
e
th
e
s
et
o
f
tr
ials
Ī»
^1
ā¦.
Ī»
^s.
Ma
ch
i
n
e
lear
n
in
g
s
y
s
te
m
s
ar
e
ab
o
u
n
d
in
g
w
it
h
h
y
p
er
p
ar
a
m
e
ter
s
.
H
y
p
er
p
ar
am
eter
o
p
ti
m
iza
tio
n
is
th
e
m
i
n
i
m
izatio
n
o
f
p
ar
a
m
eter
o
v
er
a
s
u
b
s
et
o
f
p
ar
a
m
eter
.
T
h
is
f
u
n
ctio
n
is
s
o
m
eti
m
es
ca
l
led
th
e
r
esp
o
n
s
e
s
u
r
f
ac
e
in
t
h
e
ex
p
er
i
m
en
t
d
es
ig
n
lite
r
atu
r
e.
Dif
f
er
en
t
d
ata
s
ets,
ta
s
k
s
,
a
n
d
lear
n
i
n
g
al
g
o
r
ith
m
f
a
m
ilie
s
g
i
v
e
r
is
e
to
d
if
f
er
e
n
t
s
et
s
o
f
p
ar
a
m
eter
s
an
d
f
u
n
c
tio
n
s
[
1
3
]
.
C
h
o
o
s
i
n
g
th
e
b
est
h
y
p
er
p
ar
a
m
eter
s
ar
e
b
o
th
cr
u
ci
al
an
d
f
r
u
s
tr
atin
g
l
y
d
if
f
ic
u
lt.
H
y
p
er
p
ar
a
m
eter
s
ar
e
ch
o
s
en
to
o
p
ti
m
ize
th
e
v
alid
atio
n
lo
s
s
af
ter
c
o
m
p
lete
tr
ai
n
i
n
g
o
f
th
e
m
o
d
el
p
ar
am
e
ter
s
[
1
4
]
.
T
h
e
cr
itical
s
tep
in
h
y
p
er
p
ar
a
m
et
er
o
p
tim
izatio
n
is
to
ch
o
o
s
e
th
e
s
et
o
f
tr
ials
Ī»
1
ā¦Ī»
s
.
T
h
e
m
o
s
t
co
m
m
o
n
l
y
u
s
ed
t
ec
h
n
iq
u
e
i
n
h
y
p
er
p
ar
a
m
eter
o
p
ti
m
izatio
n
is
a
g
r
id
s
ea
r
ch
tec
h
n
iq
u
e.
Gr
id
s
ea
r
ch
r
eq
u
ir
es
ch
o
o
s
i
n
g
a
s
et
o
f
v
alu
es
f
o
r
ea
c
h
v
ar
iab
le.
I
t
is
s
i
m
p
l
e
to
i
m
p
le
m
e
n
t
a
n
d
p
ar
alleliza
t
io
n
is
tr
iv
ial.
Ot
h
er
th
an
t
h
at,
it
is
also
is
r
eliab
le
in
lo
w
d
i
m
e
n
s
io
n
al
s
p
ac
es
[
1
2
]
.
T
h
e
o
th
er
cr
u
cial
s
tep
to
f
u
r
th
er
i
m
p
r
o
v
e
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
a
lg
o
r
it
h
m
s
is
b
y
d
o
in
g
a
h
y
b
r
id
.
Nu
m
er
o
u
s
m
et
h
o
d
s
h
a
v
e
b
ee
n
s
u
g
g
est
ed
f
o
r
th
e
cr
ea
tio
n
o
f
h
y
b
r
id
o
f
clas
s
i
f
ier
s
[
1
5
]
.
A
lth
o
u
g
h
m
a
n
y
m
et
h
o
d
s
o
f
h
y
b
r
id
h
av
e
b
ee
n
p
r
o
p
o
s
ed
,
y
et
th
er
e
is
n
o
clea
r
p
ictu
r
e
o
f
w
h
ic
h
m
et
h
o
d
is
th
e
b
est
[
1
6
]
.
T
h
u
s
,
an
ac
tiv
e
ar
ea
o
f
r
esear
ch
in
s
u
p
er
v
is
ed
lear
n
i
n
g
is
t
h
e
s
t
u
d
y
o
f
m
e
th
o
d
s
f
o
r
th
e
co
n
s
tr
u
ctio
n
o
f
g
o
o
d
h
y
b
r
id
al
g
o
r
ith
m
s
.
H
y
b
r
id
al
g
o
r
ith
m
s
is
o
b
tai
n
ed
b
y
co
m
b
in
in
g
a
p
o
r
tio
n
o
f
ele
m
e
n
ts
f
r
o
m
e
x
is
t
in
g
ele
m
e
n
ts
an
d
co
m
p
o
s
i
n
g
a
m
ea
n
in
g
f
u
l
co
m
b
i
n
atio
n
.
T
h
is
r
esu
lts
i
n
s
tr
en
g
t
h
en
i
n
g
th
e
tech
n
iq
u
es
co
m
b
in
ed
to
p
r
o
v
id
e
a
s
tab
le
an
d
ac
cu
r
ate
r
e
s
u
lts
.
Selecti
n
g
th
e
r
ele
v
a
n
t
al
g
o
r
ith
m
s
p
r
o
d
u
ce
d
ef
f
icien
t
co
m
b
i
n
atio
n
s
.
Ma
n
y
r
esear
ch
er
s
h
av
e
ac
tiv
el
y
w
o
r
k
ed
o
n
co
m
b
i
n
i
n
g
m
u
lt
ip
le
alg
o
r
ith
m
s
to
g
et
h
er
f
o
r
m
i
n
in
g
[
1
7
-
18]
.
Alth
o
u
g
h
t
h
er
e
ar
e
m
an
y
m
et
h
o
d
s
p
r
o
p
o
s
ed
f
o
r
h
y
b
r
id
al
g
o
r
ith
m
s
,
y
et
th
er
e
is
n
o
clea
r
p
ictu
r
e
o
f
w
h
ic
h
m
eth
o
d
is
t
h
e
b
est [
1
9
]
.
I
n
th
is
p
ap
er
,
Xg
b
w
as
ch
o
s
e
n
to
b
e
in
co
r
p
o
r
ated
in
t
h
e
R
F
alg
o
r
it
h
m
s
.
X
g
b
is
k
n
o
w
n
t
o
h
av
e
a
n
ab
ilit
y
to
h
elp
a
w
ea
k
lear
n
e
r
g
r
o
w
s
in
to
a
s
tr
o
n
g
lear
n
er
.
T
h
e
ad
v
an
tag
e
o
f
u
s
i
n
g
X
g
b
m
et
h
o
d
,
is
th
at
i
t
i
m
p
r
o
v
es
t
h
e
tr
ee
s
b
y
in
cr
ea
s
i
n
g
t
h
e
w
eig
h
t
o
f
o
n
e
tr
ee
af
te
r
an
o
th
er
[
2
0
]
.
On
e
i
m
p
o
r
tan
t
h
y
p
er
p
ar
a
m
eter
in
Xg
b
is
th
e
lear
n
in
g
r
ate.
C
o
m
m
o
n
l
y
,
in
X
g
b
,
th
e
lo
w
er
t
h
e
lear
n
in
g
r
ate
m
ea
n
s
it is
b
etter
f
o
r
test
in
g
er
r
o
r
,
b
u
t
th
is
w
ill
r
es
u
lt
i
n
i
n
cr
ea
s
i
n
g
m
o
r
e
tr
ee
s
.
W
ith
t
h
at,
t
h
e
h
y
b
r
id
b
etw
ee
n
R
F
an
d
X
g
b
m
a
y
r
esu
lt
i
n
b
etter
p
er
f
o
r
m
a
n
ce
o
f
ac
c
u
r
ac
y
.
T
h
e
o
r
g
an
izatio
n
o
f
th
e
p
ap
er
is
as
f
o
llo
w
s
.
Sectio
n
2
p
r
esen
ts
t
h
e
m
et
h
o
d
o
lo
g
y
o
f
th
e
h
y
b
r
id
alg
o
r
it
h
m
s
p
r
o
p
o
s
e
d
in
th
is
p
ap
er
.
I
n
Sectio
n
3
,
th
e
r
esu
lts
o
f
th
e
h
y
b
r
id
alg
o
r
ith
m
ar
e
ev
al
u
ated
an
d
co
m
p
ar
ed
w
it
h
o
th
er
r
esu
l
ts
r
e
p
o
r
ted
in
th
e
liter
atu
r
e.
Fi
n
all
y
,
Sectio
n
4
co
n
clu
d
es t
h
e
p
ap
er
.
2.
T
H
E
H
YB
RID O
F
O
P
T
I
M
I
Z
E
D
R
ANDO
M
F
O
RE
ST
A
ND
E
X
T
RE
M
E
G
R
ADI
E
N
T
B
O
O
ST
I
N
G
RF
(
Xg
b)
2
.
1
.
Da
t
a
s
elec
t
io
n
T
h
e
d
atasets
u
s
ed
i
n
th
is
r
ese
ar
ch
is
ta
k
en
f
r
o
m
a
r
esear
c
h
d
o
n
e
b
y
[
1
1
]
.
T
h
e
d
ata
is
co
ll
ec
ted
f
r
o
m
th
e
y
ea
r
2
0
1
2
to
2
0
1
6
.
I
t
co
n
tain
s
a
r
ec
o
r
d
of
5
0
7
s
tu
d
en
t
s
en
r
o
lled
in
th
e
C
o
m
p
u
ter
T
ec
h
n
o
lo
g
y
co
u
r
s
es
w
h
ic
h
h
av
e
s
u
cc
es
s
f
u
l
l
y
co
m
p
leted
th
e
C
o
m
p
u
ter
P
r
o
g
r
am
m
i
n
g
1
s
u
b
j
ec
t.
T
h
is
d
ataset
co
n
s
is
ts
o
f
1
5
d
if
f
er
en
t
attr
ib
u
tes.
A
s
m
en
tio
n
ed
b
y
[
1
1
]
th
e
attr
ib
u
tes
s
elec
ted
is
b
ased
o
n
r
elev
an
c
y
a
n
d
th
e
s
u
i
tab
ilit
y
d
esi
g
n
ed
as
r
ef
er
r
ed
f
r
o
m
p
r
ev
io
u
s
r
esear
c
h
b
y
[
2
1
-
2
2
]
.
T
ab
le
1
s
h
o
w
s
t
h
e
s
u
m
m
ar
y
o
f
t
h
e
d
ataset.
T
ab
le
1
.
Su
m
m
ar
y
o
f
d
ataset
.
P
a
r
a
me
t
e
r
V
a
l
u
e
S
o
u
r
c
e
o
f
D
a
t
a
se
t
C
o
mp
u
t
e
r
T
e
c
h
n
o
l
o
g
y
c
o
u
r
se
s fr
o
m U
n
i
v
e
r
si
t
y
o
f
P
h
i
l
l
i
p
i
n
e
s
N
u
mb
e
r
o
f
i
n
st
a
n
c
e
s
5
0
7
N
u
mb
e
r
o
f
a
t
t
r
i
b
u
t
e
s
15
Evaluation Warning : The document was created with Spire.PDF for Python.
ļ²
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
8
,
No
.
4
,
Dec
em
b
er
20
1
9
:
4
22
ā
4
28
424
2
.
2
.
P
er
f
o
rm
a
nce
m
et
rice
s
T
h
e
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
e
s
w
h
ic
h
ar
e
co
n
s
id
er
ed
in
t
h
is
p
ap
er
is
th
e
ef
f
ec
t
iv
e
n
es
s
o
f
t
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
.
I
t
is
m
ea
s
u
r
ed
b
y
th
e
p
er
ce
n
ta
g
e
o
f
ac
cu
r
ac
y
u
s
i
n
g
co
n
f
u
s
io
n
m
atr
i
x
.
C
o
n
f
u
s
io
n
m
atr
i
x
co
n
tain
s
i
n
f
o
r
m
atio
n
ab
o
u
t
a
ctu
al
a
n
d
p
r
ed
icted
class
i
f
ica
tio
n
s
d
o
n
e
b
y
a
clas
s
i
f
icatio
n
s
y
s
te
m
[
2
3
]
.
I
n
co
n
f
u
s
io
n
m
atr
i
x
,
th
e
ev
al
u
at
io
n
is
b
ased
o
n
s
o
m
e
s
ta
n
d
ar
d
s
an
d
ter
m
s
,
s
u
c
h
as
tr
u
e
p
o
s
itiv
e
(
T
P),
Fals
e
n
eg
at
iv
e
(
FN)
,
f
a
ls
e
p
o
s
it
iv
e
(
FP
)
,
an
d
tr
u
e
n
eg
a
tiv
e
(
T
N)
.
Fro
m
th
e
ter
m
s
,
s
o
m
e
eq
u
a
tio
n
ca
n
b
e
d
ed
u
ce
d
s
u
c
h
as
th
e
eq
u
atio
n
f
o
r
ac
cu
r
ac
y
v
alu
e.
A
cc
u
r
ac
y
i
s
n
ee
d
ed
to
d
eter
m
in
e
h
o
w
o
f
ten
th
e
clas
s
if
ier
i
s
co
r
r
ec
t.
T
h
e
eq
u
atio
n
is
s
h
o
w
n
i
n
(
1
)
.
ī£
īæīæ
ī
ī
ī½
īæī
=
īÆīÆ
+
īÆīÆ
īÆīÆ
+
īÆīÆ
+
ī®æīÆ
+
ī®æīÆ
(
1
)
2
.
3
.
I
nco
rpo
ra
t
ing
t
he
ex
t
re
m
e
g
ra
dient
bo
o
s
t
ing
f
un
ct
io
n in t
he
ra
nd
o
m
f
o
re
s
t
B
o
o
s
tin
g
is
b
ased
o
n
w
ea
k
lear
n
er
s
(
h
i
g
h
b
ias,
lo
w
v
ar
ia
n
ce
)
.
I
n
ter
m
s
o
f
d
ec
is
io
n
tr
ee
s
,
w
ea
k
lear
n
er
s
ar
e
s
h
allo
w
tr
ee
s
,
s
o
m
eti
m
e
s
e
v
en
as
s
m
all
a
s
d
ec
is
io
n
s
tu
m
p
s
(
tr
ee
s
w
it
h
t
w
o
leav
e
s
)
.
T
h
e
b
o
o
s
tin
g
co
n
ti
n
u
e
to
u
p
d
ate
th
e
w
ei
g
h
ts
o
f
tr
ai
n
i
n
g
s
et
b
ased
o
n
p
r
ev
io
u
s
w
ea
k
er
lear
n
er
to
i
m
p
r
o
v
e
t
h
e
i
m
p
o
r
tan
ce
o
f
d
ata
w
h
ic
h
ar
e
cl
ass
if
ied
w
r
o
n
g
l
y
.
T
h
e
illu
s
tr
atio
n
d
iag
r
a
m
o
f
t
h
e
p
ar
a
m
e
ter
is
s
h
o
w
n
i
n
Fi
g
u
r
e
1
.
I
n
th
i
s
p
ap
er
,
th
e
f
u
n
ctio
n
o
f
n
r
o
u
n
d
s
,
eta,
Ī±
,
an
d
Ī»
is
in
co
r
p
o
r
ated
in
th
e
R
F
alg
o
r
it
h
m
s
to
f
o
r
m
t
h
e
h
y
b
r
id
alg
o
r
it
h
m
,
R
F(X
g
b
)
.
T
h
is
f
u
n
ctio
n
is
s
elec
ted
,
as
it
h
as
t
h
e
ab
i
lit
y
to
co
n
tr
o
l
th
e
n
u
m
b
er
o
f
iter
atio
n
s
n
ee
d
ed
in
b
u
il
d
th
e
tr
ee
in
o
r
d
er
to
g
et
an
o
p
ti
m
al
tr
ee
-
b
ased
m
o
d
el
w
h
ich
r
es
u
lt
s
in
a
b
etter
p
r
e
d
ictio
n
ac
cu
r
ac
y
.
T
h
e
r
o
le
o
f
t
h
is
f
u
n
ctio
n
tak
e
n
in
th
e
Xg
b
is
to
im
p
r
o
v
e
th
e
m
aj
o
r
ity
v
o
te
v
al
u
e
in
R
F
an
d
at
th
e
s
am
e
ti
m
e
h
elp
in
i
m
p
r
o
v
in
g
th
e
f
o
r
m
a
tio
n
o
f
in
d
i
v
id
u
al
tr
ee
i
n
t
h
e
p
r
o
ce
s
s
o
f
b
a
g
g
i
n
g
m
e
th
o
d
s
i
n
R
F.
R
F(X
g
b
)
h
elp
s
i
n
i
m
p
r
o
v
i
n
g
t
h
e
R
F
m
o
d
el
i
n
r
ed
u
cin
g
th
e
OOB
er
r
o
r
v
alu
e
w
h
ic
h
ev
e
n
t
u
all
y
i
n
cr
ea
s
ed
th
e
ac
cu
r
ac
y
v
al
u
e
i
n
th
e
m
o
d
el.
T
h
is
is
b
ec
au
s
e
R
F
ten
d
s
to
o
v
er
f
it
its
m
o
d
el,
as
it
h
as
th
e
p
r
o
b
lem
i
n
d
ec
id
in
g
t
h
e
m
o
s
t
o
p
ti
m
al
n
u
m
b
er
o
f
tr
ee
.
T
h
e
o
v
er
all
f
lo
w
o
f
th
e
p
r
o
p
o
s
ed
R
F(X
g
b
)
is
d
is
cu
s
s
ed
in
Sec
tio
n
2
.
4
.
Fig
u
r
e
1
.
I
llu
s
tr
atio
n
d
ia
g
r
a
m
f
o
r
p
ar
am
eter
i
n
Xg
b
2
.
4
.
T
he
o
v
er
a
ll f
lo
w
o
f
t
he
pro
po
s
ed
RF
(
Xg
b)
Fig
u
r
e
2
s
h
o
w
s
th
e
w
o
r
k
i
n
g
f
l
o
w
d
iag
r
a
m
o
f
R
F(X
g
b
)
alg
o
r
ith
m
.
First,
t
h
e
d
atase
t
is
s
p
ec
if
ied
.
T
h
e
R
F(X
g
b
)
is
ad
j
u
s
ted
ac
co
r
d
in
g
l
y
b
y
u
s
i
n
g
h
y
p
er
p
ar
a
m
eter
o
p
ti
m
izatio
n
to
o
b
tain
th
e
m
o
s
t
o
p
tim
al
p
ar
a
m
eter
.
Nex
t,
f
r
o
m
t
h
e
o
p
ti
m
al
p
ar
am
eter
,
R
F(X
g
b
)
is
u
s
ed
to
d
e
tect
th
e
u
s
er
lear
n
i
n
g
s
t
y
le,
b
ased
o
n
th
e
FS
L
SM
m
o
d
el
a
n
d
ev
alu
ated
u
s
in
g
d
if
f
er
en
t
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
e
w
h
ic
h
ar
e
ac
cu
r
ac
y
an
d
R
OC
cu
r
v
e.
I
n
th
i
s
p
ap
er
,
tr
ain
i
n
g
d
ataset
D
is
u
s
ed
to
s
p
ec
if
y
t
h
e
s
u
p
p
o
r
tin
g
p
ar
a
m
et
er
o
f
m
o
d
el
t
as
s
h
o
w
n
in
A
l
g
o
r
ith
m
1
.
Giv
e
n
a
tr
ain
i
n
g
d
ataset,
D=
x
_
1
,
x
_
2
ā¦.
x
_
n
,
ea
c
h
tr
ai
n
i
n
g
i
n
s
ta
n
c
e
is
r
ep
r
esen
ted
as
x
_
i=x
_
i1
,
x
_
i2
ā¦x
_
i
n
a
n
d
D
co
n
tain
s
th
e
f
o
llo
w
in
g
attr
ib
u
t
es
k
1
,
k
2
ā¦k
n
.
Firs
t,
th
e
tr
ee
i
s
s
p
ec
if
ied
w
it
h
1
0
-
f
o
ld
cr
o
s
s
v
alid
atio
n
a
n
d
th
e
n
tr
ee
b
o
o
ts
tr
ap
s
a
m
p
les
is
d
r
aw
.
Fo
r
ea
ch
b
o
o
ts
tr
ap
s
am
p
le
s
,
u
n
p
r
u
n
ed
tr
ee
is
g
r
e
w
b
y
ch
o
o
s
in
g
th
e
b
est
s
p
lit
b
ased
o
f
r
an
d
o
m
s
a
m
p
les
o
f
m
tr
y
p
r
ed
ictio
n
at
ea
ch
n
o
d
e.
T
h
en
,
v
alu
e
o
f
t
i
s
s
p
ec
if
ied
a
n
d
m
tr
y
i
s
o
p
ti
m
ized
to
r
ed
u
ce
OOB
er
r
o
r
.
T
h
e
o
p
ti
m
ized
v
alu
e
o
f
Xg
b
;
n
r
o
u
n
d
,
c
,
eta,
Ī»
,
a
n
d
Ī±
is
d
eter
m
in
ed
a
n
d
in
s
er
t
in
m
o
d
el
t
alo
n
g
w
i
th
th
e
s
p
ec
if
ied
p
ar
am
eter
f
o
r
R
F
t
h
at
w
a
s
d
eter
m
i
n
ed
ea
r
lier
.
T
h
e
o
p
tim
ized
p
ar
am
eter
v
al
u
es
i
s
s
h
o
w
n
i
n
T
ab
le
2
.
Mo
d
el
t
is
th
en
ap
p
lied
to
a
test
s
e
t
D_
i
w
h
ic
h
co
n
ta
i
n
s
a
s
u
b
s
et
o
f
tr
a
in
i
n
g
d
ata
s
et.
T
h
e
alg
o
r
ith
m
is
s
h
o
w
n
i
n
A
lg
o
r
it
h
m
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
ļ²
Hyb
r
id
is
a
tio
n
o
f R
F
(
X
g
b
)
to
imp
r
o
ve
th
e
tr
ee
-
b
a
s
ed
a
lg
o
r
ith
ms in
... (
Ha
z
iq
a
h
S
h
a
msu
d
in
)
425
Fig
u
r
e
2
.
T
h
e
w
o
r
k
in
g
f
lo
w
d
i
ag
r
a
m
o
f
R
F(X
g
b
)
Alg
o
r
ith
m
1
RF
(X
g
b
)
1
:
p
r
o
c
e
d
u
r
e
RF
(X
g
b
)
(
Ntre
e
,
mt
ry
,
n
ro
u
n
d
s,c
,
e
ta
,
Ī±
,
Ī»
)
2
:
fo
r
e
a
c
h
c
la
s
s
,
ī„
īÆ
ī³ī¦
,
do
3
:
S
p
e
c
ify
th
e
trCo
n
tr
o
l
w
it
h
5
f
o
l
d
o
f
c
ro
ss
-
v
a
li
d
a
ti
o
n
a
n
d
g
rid
se
a
rc
h
4
:
e
n
d
fo
r
5
:
fo
r
RF
f
u
n
c
ti
o
n
s
do
6
:
Dra
w
n
tre
e
b
o
o
tstrap
sa
m
p
les
7
:
F
o
r
e
a
c
h
b
o
o
tstrap
sa
m
p
le,
g
ro
w
u
n
-
p
ru
n
e
d
tree
b
y
c
h
o
o
si
n
g
b
e
st
sp
l
it
b
a
se
d
o
f
ra
n
d
o
m
sa
m
p
le
o
f
mtry
p
re
d
icti
o
n
a
t
e
a
c
h
n
o
d
e
8
:
t:o
p
ti
m
iz
e
mtry
to
r
e
d
u
c
e
O
O
B
e
r
r
o
r
9
:
e
n
d
fo
r
1
0
:
fo
r
t,
u
se
(
D),
do
1
1
:
sp
e
c
ify
th
e
sup
p
o
r
ti
n
g
p
a
r
a
m
e
te
r
o
f
m
o
d
e
l
t
a
n
d
n
tre
e
1
2
:
De
te
r
m
i
n
e
b
e
st
mtry
v
a
l
u
e
1
3
:
e
n
d
fo
r
1
4
:
fo
r
X
g
b
F
u
n
c
ti
o
n
s
do
1
5
:
t:
o
p
ti
m
ize
n
ro
u
n
d
s,
c
,
e
t
a
,
Ī»
,
Ī±
t
o
re
d
u
c
e
OO
B
1
6
:
Cr
e
a
te
n
e
w
m
o
d
e
l
t1
c
o
m
b
i
n
e
th
e
fu
n
c
ti
o
n
fr
o
m
S
te
p
5
a
n
d
S
te
p
6
1
7
:
e
n
d
fo
r
1
8
:
fo
r
Fu
n
c
tio
n
t1
do
1
9
:
Pre
d
ict th
e
r
e
s
u
lt
u
si
n
g
te
stin
g
d
a
ta
b
a
s
e
d
o
n
fi
n
a
l
m
o
d
e
l
t1
2
0
:
e
n
d
fo
r
2
1
:
Prin
t
th
e
r
e
su
lt
2
2
:
r
e
tu
r
n
2
3
:
e
n
d
p
r
o
c
e
d
u
r
e
T
ab
le
2
.
Op
tim
ized
p
ar
a
m
eter
v
alu
e
f
o
r
R
F(X
g
b
)
P
a
r
a
me
t
e
r
V
a
l
u
e
n
t
r
e
e
3
0
0
m
t
ry
2
n
ro
u
n
d
5
0
0
e
t
a
0
.
0
0
1
Ī»
1
Ī±
0
Evaluation Warning : The document was created with Spire.PDF for Python.
ļ²
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
8
,
No
.
4
,
Dec
em
b
er
20
1
9
:
4
22
ā
4
28
426
3.
CO
M
P
ARATI
VE
ANA
L
YS
I
S O
N
T
H
E
P
E
RF
O
RM
ANCE O
F
AL
G
O
RI
T
H
M
S
3
.
1
.
Acc
ura
cy
v
a
lue
T
h
is
s
ec
tio
n
d
is
cu
s
s
es
th
e
o
v
e
r
all
r
esu
lt
o
n
th
e
u
s
e
o
f
R
F(X
g
b
)
alg
o
r
ith
m
i
n
p
r
ed
ictin
g
th
e
lear
n
in
g
s
t
y
le
o
f
th
e
u
s
er
.
Fro
m
T
ab
le
3
,
it
s
h
o
w
s
t
h
at
th
e
r
esu
lt
o
f
ac
c
u
r
ac
y
is
co
n
s
i
s
te
n
t
in
th
e
r
an
g
e
o
f
0
.
9
1
%
to
0
.
9
8
%
.
I
t
is
n
o
ticed
th
at
w
h
en
d
o
in
g
th
e
h
y
b
r
id
th
e
r
an
g
e
o
f
ac
cu
r
ac
y
f
o
r
all
o
f
th
e
lear
n
in
g
s
t
y
l
e
d
im
e
n
s
io
n
is
in
a
co
n
s
ta
n
t
r
an
g
e.
T
h
is
b
r
ea
k
s
t
h
e
g
ap
f
r
o
m
p
r
ev
io
u
s
r
esear
c
h
i
n
th
e
ar
ea
o
f
p
r
ed
ictin
g
th
e
lea
r
n
in
g
s
t
y
le
u
s
in
g
a
n
au
to
m
ated
ap
p
r
o
ac
h
d
o
n
e
b
y
s
ev
er
al
r
esear
ch
er
s
.
I
n
th
e
p
r
ev
io
u
s
r
esear
ch
e
s
,
s
o
m
e
r
esear
ch
er
s
h
ad
p
r
o
b
lem
o
f
n
o
t
g
ett
in
g
a
g
o
o
d
ac
cu
r
ac
y
f
o
r
ce
r
tain
d
i
m
e
n
s
io
n
w
h
i
le
f
e
w
r
esear
ch
es
ex
c
lu
d
e
s
o
m
e
d
i
m
en
s
io
n
s
as
t
h
e
y
ar
e
n
o
t
co
m
p
atib
le
w
ith
t
h
eir
m
o
d
els.
Ho
w
e
v
er
,
b
y
u
s
i
n
g
R
F(X
g
b
)
,
a
b
etter
ac
cu
r
ac
y
w
as
o
b
tai
n
ed
in
d
etec
tin
g
t
h
e
lear
n
in
g
s
t
y
le
o
f
th
e
u
s
er
.
T
ab
le
3
.
P
er
ce
n
tag
e
o
f
ac
cu
r
a
c
y
f
o
r
R
F(X
g
b
)
F
S
L
S
M
D
i
me
n
si
o
n
R
F
(
X
g
b
)
I
n
p
u
t
0
.
9
7
P
e
r
c
e
p
t
i
o
n
0
.
9
7
P
r
o
c
e
ssi
n
g
0
.
9
8
U
n
d
e
r
st
a
n
d
i
n
g
0
.
9
1
3
.
2
.
RO
C
a
nd
AUC
v
a
lue
I
n
o
r
d
er
to
ev
alu
ate
f
u
r
t
h
er
th
e
ef
f
ec
tiv
e
n
e
s
s
o
f
t
h
e
m
o
d
el,
R
OC
c
u
r
v
e
is
i
n
cl
u
d
ed
in
t
h
is
p
ap
er
.
T
h
e
co
n
ce
p
t
o
f
a
n
R
O
C
c
u
r
v
e
is
b
ased
o
n
th
e
n
o
tio
n
o
f
a
"
s
e
p
ar
ato
r
"
(
o
r
d
ec
is
io
n
)
v
ar
iab
le.
T
h
e
p
lo
t
o
f
T
P
F
(
s
en
s
i
tiv
it
y
)
v
er
s
u
s
FP
F
(
1
-
s
p
ec
if
icit
y
)
ac
r
o
s
s
v
a
r
y
i
n
g
cu
t
-
o
f
f
s
g
en
er
ate
s
a
cu
r
v
e
in
th
e
u
n
it
s
q
u
ar
e
ca
lled
an
R
OC
c
u
r
v
e.
R
O
C
cu
r
v
e
co
r
r
esp
o
n
d
in
g
to
p
r
o
g
r
ess
i
v
el
y
g
r
e
ater
d
is
cr
i
m
i
n
an
t
ar
e
lo
ca
ted
p
r
o
g
r
ess
iv
el
y
clo
s
er
to
th
e
u
p
p
er
lef
t
-
h
an
d
co
r
n
er
i
n
"
R
OC
s
p
ac
e"
.
T
h
e
R
OC
c
u
r
v
e
lie
o
n
t
h
e
d
ia
g
o
n
al
li
n
e
r
ef
l
ec
ts
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
p
r
ed
ictio
n
test
th
a
t
is
n
o
b
etter
th
an
ch
a
n
ce
lev
el,
i.e
.
a
test
w
h
ic
h
y
ield
s
th
e
p
o
s
iti
v
e
o
r
n
eg
ativ
e
r
es
u
lt
s
u
n
r
elate
d
to
th
e
tr
u
e
clas
s
lab
el.
T
h
e
s
lo
p
e
o
f
an
R
OC
c
u
r
v
e
at
an
y
p
o
in
t
is
eq
u
al
to
t
h
e
r
ati
o
o
f
th
e
t
w
o
d
e
n
s
it
y
f
u
n
ctio
n
s
d
escr
ib
i
n
g
,
r
esp
ec
ti
v
el
y
,
t
h
e
d
is
tr
ib
u
tio
n
o
f
t
h
e
s
e
p
ar
ato
r
v
ar
iab
le
in
th
e
clas
s
la
b
el.
T
h
e
ar
ea
u
n
d
er
th
e
cu
r
v
e
(
A
U
C
)
s
u
m
m
ar
izes
th
e
en
tire
lo
ca
tio
n
o
f
th
e
R
OC
cu
r
v
e
r
at
h
er
th
an
d
ep
en
d
in
g
o
n
a
s
p
ec
i
f
icit
y
o
p
er
atin
g
p
o
in
t.
T
h
e
A
UC
i
s
an
e
f
f
ec
ti
v
e
an
d
co
m
b
in
ed
m
ea
s
u
r
e
o
f
s
e
n
s
i
tiv
it
y
an
d
s
p
ec
i
f
icit
y
t
h
at
d
es
cr
ib
es
th
e
in
h
er
e
n
t
v
alid
it
y
o
f
d
eter
m
i
n
i
n
g
t
h
e
cla
s
s
lab
el.
I
f
t
w
o
te
s
ts
ar
e
to
b
e
co
m
p
ar
ed
,
it
is
d
esira
b
le
to
co
m
p
ar
e
th
e
en
tire
R
O
C
cu
r
v
e
r
ath
er
t
h
an
at
a
p
ar
ticu
lar
p
o
in
t.
T
h
e
m
a
x
i
m
u
m
AUC
=
1
m
ea
n
s
t
h
at
t
h
e
m
o
d
el
is
p
er
f
ec
t
i
n
th
e
d
if
f
er
en
t
iatio
n
b
et
w
ee
n
th
e
class
.
T
h
is
h
ap
p
en
s
w
h
e
n
th
e
d
is
tr
ib
u
tio
n
o
f
t
h
e
clas
s
lab
el
d
o
n
o
t
o
v
er
lap
.
A
UC
=
0
.
5
m
ea
n
s
th
e
c
h
an
ce
d
is
cr
i
m
i
n
atio
n
t
h
at
cu
r
v
e
lo
ca
ted
o
n
d
iag
o
n
al
lin
e
in
R
OC
s
p
ac
e.
T
h
e
m
i
n
i
m
u
m
A
U
C
sh
o
u
ld
b
e
co
n
s
id
er
ed
a
ch
an
ce
lev
el
i.e
.
AUC =
0
.
5
w
h
ile
AUC =
0
m
ea
n
s
test
i
n
co
r
r
ec
tl
y
class
i
f
y
a
ll s
u
b
j
ec
ts
w
it
h
cla
s
s
A
to
cla
s
s
B
an
d
cl
ass
B
to
class
A
.
O
v
er
all,
t
h
e
R
OC
c
u
r
v
e
f
o
r
ea
ch
o
f
th
e
d
i
m
en
s
io
n
is
s
h
o
w
n
in
Fig
u
r
e
3
.
Fro
m
Fi
g
u
r
e
3
(
a)
,
th
e
A
U
C
v
a
l
u
e
i
s
0
.
9
9
8
3
.
T
h
e
v
alu
e
w
h
ich
s
h
o
w
s
th
e
p
r
ed
icti
o
n
is
al
m
o
s
t
p
er
f
ec
t
as
m
en
t
io
n
in
b
ef
o
r
e
th
at
th
e
m
a
x
i
m
u
m
A
U
C
=1
,
w
h
ich
m
ea
n
s
th
at
t
h
e
d
iag
n
o
s
t
ic
test
is
p
er
f
ec
t
in
th
e
d
if
f
er
e
n
tiatio
n
b
et
w
ee
n
v
is
u
al
an
d
v
er
b
al
class
.
T
h
e
p
er
ce
n
tag
e
o
f
ac
cu
r
ac
y
f
o
r
in
p
u
t
d
i
m
e
n
s
io
n
is
0
.
9
7
%
w
h
ic
h
is
h
i
g
h
en
o
u
g
h
in
ter
m
s
o
f
th
e
class
i
f
icatio
n
ac
cu
r
ac
y
.
As
f
o
r
t
h
e
p
r
o
ce
s
s
in
g
d
i
m
en
s
i
o
n
,
th
e
p
er
ce
n
ta
g
e
o
f
ac
c
u
r
ac
y
is
0
.
9
8
%
w
h
ich
co
n
tr
ib
u
te
i
n
t
h
e
h
i
g
h
v
alu
e
o
f
A
U
C
w
h
ic
h
is
0
.
9
9
8
9
.
T
h
e
cu
r
v
e
o
f
th
e
p
r
o
ce
s
s
in
g
d
im
en
s
io
n
i
s
al
m
o
s
t
p
er
f
ec
t
w
h
e
r
e
th
e
cu
r
v
e
lo
ca
ted
p
r
o
g
r
ess
iv
el
y
clo
s
er
to
th
e
u
p
p
er
-
lef
t
h
a
n
d
co
r
n
er
in
R
O
C
s
p
ac
e.
T
h
e
cu
r
v
e
is
s
h
o
w
n
in
Fig
u
r
e
3
(
b
)
.
Fo
r
p
er
ce
p
tio
n
d
im
e
n
s
io
n
as
s
h
o
wn
in
Fig
u
r
e
3
(
c)
,
th
e
p
er
ce
n
ta
g
e
o
f
ac
c
u
r
ac
y
is
0
.
9
8
%
w
h
il
e
th
e
A
U
C
v
al
u
e
is
1
.
0
.
A
s
m
e
n
tio
n
in
th
e
p
r
ev
io
u
s
p
ar
ag
r
ap
h
,
w
h
en
A
U
C
=
1
,
it
m
ea
n
s
th
a
t
th
e
m
o
d
el
is
p
er
f
ec
t
in
th
e
d
if
f
er
e
n
tiatio
n
b
et
w
ee
n
t
h
e
cla
s
s
w
h
er
e
i
n
t
h
is
d
i
m
e
n
s
io
n
it
m
ea
n
s
f
o
r
th
e
c
lass
s
e
n
s
i
n
g
an
d
in
tu
iti
v
e
lear
n
er
s
.
L
ast
l
y
,
t
h
e
u
n
d
er
s
ta
n
d
i
n
g
d
i
m
en
s
io
n
p
r
o
d
u
ce
a
s
lig
h
tl
y
lo
w
p
er
ce
n
tag
e
o
f
ac
c
u
r
ac
y
co
m
p
ar
ed
to
o
th
er
d
i
m
en
s
io
n
,
b
u
t it
s
ti
ll
m
a
n
ag
e
to
p
r
o
d
u
ce
a
b
etter
r
esu
lt c
o
m
p
ar
ed
t
o
p
r
ev
io
u
s
liter
atu
r
e
[
1
1
,
2
4
-
25]
.
T
h
e
A
UC
v
alu
e
is
0
.
9
8
3
3
as
s
h
o
w
n
i
n
Fi
g
u
r
e
3
(
d
)
w
h
er
e
in
th
i
s
ca
s
e
i
s
s
till
in
t
h
e
ca
te
g
o
r
y
o
f
al
m
o
s
t
p
er
f
ec
t
v
a
lu
e.
T
h
e
R
OC
c
u
r
v
e
f
o
r
u
n
d
er
s
ta
n
d
in
g
d
i
m
en
s
io
n
i
s
also
n
ea
r
l
y
p
er
f
e
ct
as
th
e
cu
r
v
e
lo
ca
ted
p
r
o
g
r
ess
iv
el
y
clo
s
er
to
th
e
u
p
p
er
lef
t
-
h
a
n
d
co
r
n
er
in
th
e
"
R
OC
s
p
ac
e"
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
ļ²
Hyb
r
id
is
a
tio
n
o
f R
F
(
X
g
b
)
to
imp
r
o
ve
th
e
tr
ee
-
b
a
s
ed
a
lg
o
r
ith
ms in
... (
Ha
z
iq
a
h
S
h
a
msu
d
in
)
427
(
a)
I
n
p
u
t d
i
m
en
s
io
n
(
b
)
P
r
o
ce
s
s
in
g
d
i
m
en
s
io
n
(
c)
P
e
r
ce
p
tio
n
d
im
e
n
s
io
n
(
d
)
Un
d
er
s
tan
d
in
g
d
i
m
e
n
s
io
n
Fig
u
r
e
3
.
R
OC
v
alu
e
f
o
r
FS
L
S
M
o
f
R
F (
X
g
b
)
3
.
3
.
Resul
t
s
co
m
pa
riso
n
T
ab
le
4
s
h
o
w
s
th
e
r
es
u
lt
co
m
p
ar
is
o
n
o
f
R
F,
Xg
b
an
d
R
F(
Xg
b
)
.
Fro
m
t
h
e
tab
le,
R
F(X
g
b
)
s
h
o
w
s
a
p
r
o
m
i
s
in
g
r
esu
lt
an
d
co
n
s
is
ten
t
f
o
r
ea
ch
o
f
th
e
d
im
e
n
s
io
n
.
T
h
e
co
n
s
is
te
n
c
y
is
n
o
t
o
n
l
y
b
ased
o
n
th
e
p
er
ce
n
tag
e
o
f
ac
cu
r
ac
y
,
b
u
t a
ls
o
it
ca
n
b
e
p
r
o
v
ed
b
ased
f
r
o
m
t
h
e
R
OC
c
u
r
v
e
as s
h
o
w
n
i
n
Fig
u
r
e
3
.
T
h
e
r
esu
lts
s
h
o
w
n
t
h
at
p
r
o
p
o
s
ed
R
F(X
g
b
)
h
as
h
i
g
h
er
p
er
ce
n
tag
e
o
f
ac
cu
r
ac
y
f
o
r
ev
er
y
d
i
m
e
n
s
io
n
s
.
R
F(X
g
b
)
is
th
en
co
m
p
ar
ed
w
it
h
p
r
ev
io
u
s
li
ter
atu
r
e
i
n
ter
m
s
o
f
its
a
v
er
ag
e
ac
c
u
r
ac
y
v
al
u
e
an
d
th
e
r
es
u
lt
is
s
h
o
w
n
i
n
T
ab
le
5
.
Fro
m
t
h
e
av
er
a
g
e
ac
cu
r
ac
y
v
al
u
e,
R
F(X
g
b
)
w
a
s
s
li
g
h
tl
y
i
m
p
r
o
v
ed
w
h
en
co
m
p
ar
ed
w
i
th
p
r
ev
io
u
s
liter
at
u
r
e
w
it
h
a
r
a
n
g
e
o
f
i
m
p
r
o
v
e
m
en
t
f
r
o
m
0
.
0
3
to
0
.
1
.
Hig
h
er
p
er
ce
n
ta
g
e
o
f
ac
c
u
r
a
c
y
in
cr
ea
s
ed
th
e
ab
ilit
y
to
p
r
ed
ict
a
m
o
r
e
ac
cu
r
ate
lear
n
in
g
s
t
y
les.
T
h
e
r
esu
lt
s
ac
h
iev
ed
is
i
n
li
n
e
w
it
h
[
2
4
]
,
w
h
i
ch
th
e
h
ig
h
er
ac
cu
r
ac
y
i
n
d
etec
tin
g
lear
n
i
n
g
s
t
y
le
,
th
e
h
i
g
h
er
th
e
ad
ap
tiv
e
n
ess
o
f
th
e
o
n
li
n
e
lear
n
in
g
s
y
s
te
m
w
h
ic
h
lead
to
a
b
etter
en
h
an
c
e
m
en
t
o
f
th
e
o
n
lin
e
lear
n
in
g
s
y
s
te
m
t
h
at
ca
n
s
u
its
th
e
u
s
er
n
ee
d
s
.
T
ab
le
4
.
C
o
m
p
ar
is
o
n
o
f
a
cc
u
r
ac
y
v
al
u
e
f
o
r
R
F,
Xg
b
,
an
d
R
F(X
g
b
)
F
S
L
S
M
D
i
me
n
si
o
m
RF
X
g
b
R
F
(
X
g
b
)
I
n
p
u
t
0
.
9
4
0
.
9
1
0
.
9
7
P
e
r
c
e
p
t
i
o
n
0
.
9
5
0
.
8
3
0
.
9
7
P
r
o
c
e
ssi
n
g
0
.
9
5
0
.
9
2
0
.
9
8
U
n
d
e
r
st
a
n
d
i
n
g
0
.
8
6
0
.
7
9
0
.
9
1
A
v
e
r
a
g
e
0
.
9
3
0
.
8
6
0
.
9
6
T
ab
le
5
.
C
o
m
p
ar
is
o
n
o
f
av
er
a
g
e
ac
cu
r
ac
y
o
f
R
F(X
g
b
)
w
it
h
liter
at
u
r
e
M
e
t
h
o
d
A
v
e
r
a
g
e
A
c
c
u
r
a
c
y
R
F
(
X
g
b
)
0
.
9
6
RF
0
.
9
3
X
g
b
0
.
8
6
L
S
I
D
-
A
N
N
[
2
4
]
0
.
8
1
J4
8
[
1
1
]
0
.
8
9
4.
CO
NCLU
SI
O
N
I
n
co
n
cl
u
s
io
n
,
t
h
is
p
ap
er
p
r
esen
t
s
t
h
e
h
y
b
r
id
o
f
R
F
a
n
d
Xg
b
al
g
o
r
ith
m
s
w
it
h
h
y
p
e
r
p
ar
am
eter
o
p
tim
izatio
n
n
a
m
ed
R
F(X
g
b
)
.
R
F(X
g
b
)
g
iv
es
a
p
r
o
m
is
in
g
r
esu
l
ts
in
te
r
m
s
o
f
i
m
p
r
o
v
i
n
g
th
e
p
er
ce
n
tag
e
o
f
ac
cu
r
ac
y
i
n
le
ar
n
in
g
s
t
y
le
d
etec
tio
n
.
T
o
ev
alu
ate
t
h
e
e
f
f
ec
tiv
e
n
e
s
s
o
f
th
e
alg
o
r
ith
m
s
,
f
e
w
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
es
w
er
e
tak
en
i
n
to
co
n
s
id
er
atio
n
w
h
i
ch
ar
e
ac
cu
r
ac
y
an
d
R
O
C
v
alu
e.
B
ased
o
n
th
e
co
m
p
ar
is
o
n
,
R
F(X
g
b
)
s
h
o
w
s
a
b
etter
ac
cu
r
ac
y
v
al
u
e
i
n
t
h
e
lear
n
i
n
g
s
t
y
le
d
etec
tio
n
.
T
h
e
in
cr
ea
s
i
n
g
v
a
lu
e
o
f
ac
cu
r
ac
y
h
elp
s
i
n
i
m
p
r
o
v
i
n
g
t
h
e
lear
n
in
g
s
t
y
le
d
etec
tio
n
w
h
ich
lead
s
to
a
b
etter
ad
ap
tiv
it
y
o
f
t
h
e
o
n
li
n
e
lear
n
in
g
s
y
s
te
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ļ²
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
8
,
No
.
4
,
Dec
em
b
er
20
1
9
:
4
22
ā
4
28
428
ACK
NO
WL
E
D
G
E
M
E
NT
S
T
h
e
au
th
o
r
s
w
i
s
h
to
th
a
n
k
Un
i
v
er
s
iti
Sai
n
s
Ma
la
y
s
ia
(
USM)
f
o
r
t
h
e
s
u
p
p
o
r
t
it
h
a
s
ex
te
n
d
ed
in
t
h
e
co
m
p
letio
n
o
f
th
e
p
r
ese
n
t
r
esear
ch
t
h
r
o
u
g
h
I
n
ce
n
t
iv
e
Gr
an
t
(
3
0
4
/P
KOM
P
/6
3
1
6
3
8
1
)
.
T
h
e
au
th
o
r
s
also
w
i
s
h
to
th
a
n
k
A
s
s
t.
P
r
o
f
.
Dr
.
R
en
ato
R
ac
eli
s
Ma
ali
w
I
I
I
f
r
o
m
So
u
t
h
er
n
L
u
zo
n
S
tate
U
n
iv
er
s
it
y
,
P
h
ilip
p
i
n
es
f
o
r
d
ataset
s
h
ar
i
n
g
.
RE
F
E
R
E
NC
E
S
[1
]
F
e
ld
e
r,
R.
M
.
,
a
n
d
S
il
v
e
rm
a
n
,
L
.
K,
ā
L
e
a
rn
in
g
a
n
d
tea
c
h
in
g
sty
les
in
e
n
g
in
e
e
rin
g
e
d
u
c
a
ti
o
n
ā
.
E
n
g
i
n
e
e
rin
g
e
d
u
c
a
ti
o
n
,
v
o
l.
7
8
,
n
o
.
7
,
p
p
.
6
7
4
-
6
8
1
,
1
9
8
8
.
[2
]
L
it
z
in
g
e
r,
T
h
o
m
a
s
A
,
S
a
n
g
Ha
Lee
,
a
n
d
Jo
h
n
C
W
ise
,
ā
A
stu
d
y
o
f
th
e
re
li
a
b
i
li
ty
a
n
d
v
a
li
d
it
y
o
f
th
e
f
e
ld
e
r
-
so
lo
m
a
n
in
d
e
x
o
f
lea
rn
in
g
sty
les
.
ā
In
Ame
ric
a
n
S
o
c
iety
fo
r
En
g
in
e
e
rin
g
Ed
u
c
a
ti
o
n
,
v
o
l.
1
,
n
o
.
1
,
2
0
0
5
.
[3
]
Ro
m
a
n
e
ll
i,
F
.
,
Bird
,
E.
,
a
n
d
Ry
a
n
,
M
,
ā
L
e
a
rn
in
g
st
y
les
:
a
re
v
ie
w
o
f
t
h
e
o
ry
,
a
p
p
li
c
a
ti
o
n
,
a
n
d
b
e
st
p
ra
c
ti
c
e
sā
.
Ame
ric
a
n
J
o
u
rn
a
l
Of
P
h
a
rm
a
c
e
u
ti
c
a
l
Ed
u
c
a
ti
o
n
,
v
o
l
.
7
3
,
p
p
.
1
-
5
,
2
0
0
9
.
[4
]
S
u
,
B.
,
Bo
n
k
,
C.
J.,
M
a
g
ju
k
a
,
R.
J.,
L
iu
,
X
.
,
a
n
d
L
e
e
,
S
.
-
h
,
ā
T
h
e
i
m
p
o
rtan
c
e
o
f
in
tera
c
ti
o
n
in
w
e
b
-
b
a
se
d
e
d
u
c
a
ti
o
n
:
A
p
ro
g
ra
m
lev
e
l
c
a
se
stu
d
y
o
f
o
n
l
in
e
M
BA
c
o
u
rse
sā
.
J
o
u
rn
a
l
o
f
I
n
t
e
ra
c
ti
v
e
On
li
n
e
L
e
a
r
n
in
g
,
v
o
l.
4
,
p
p
.
1
-
1
9
,
2
0
0
5
.
[5
]
T
ru
o
n
g
,
H.
M
.
,
ā
In
teg
ra
ti
n
g
lea
r
n
in
g
sty
les
a
n
d
a
d
a
p
ti
v
e
e
-
lea
rn
i
n
g
sy
ste
m
:
Cu
rre
n
t
d
e
v
e
lo
p
m
e
n
t
s,
p
r
o
b
lem
s
a
n
d
o
p
p
o
rt
u
n
it
ies
ā
.
Co
mp
u
ter
s in
H
u
ma
n
Beh
a
v
io
r
,
v
o
l.
5
5
,
p
p
.
1
1
8
5
-
1
1
9
3
,
2
0
1
6
.
[6
]
M
o
k
h
tar,
R
.
,
Z
in
,
N.
A
.
M
.
,
a
n
d
A
b
d
u
ll
a
h
,
S
.
N.
H.
S
,
ā
Ru
le
-
b
a
se
d
k
n
o
w
led
g
e
re
p
re
se
n
tatio
n
f
o
r
m
o
d
a
li
ty
lea
rn
in
g
st
y
le i
n
a
iw
b
e
sā
.
In
Kn
o
wled
g
e
m
a
n
a
g
e
me
n
t
i
n
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
v
o
l.
1
,
p
p
.
6
1
4
-
6
1
7
,
2
0
1
0
.
[7
]
Cil
o
g
lu
g
il
,
B,
ā
A
d
a
p
ti
v
it
y
b
a
se
d
o
n
f
e
ld
e
r
-
silv
e
r
m
a
n
lea
rn
in
g
st
y
le
s
m
o
d
e
l
in
e
-
lea
rn
i
n
g
s
y
ste
m
sā
.
V
o
l.
1
,
p
p
.
1
5
2
3
-
153
2
,
2
0
1
6
.
[8
]
A
n
il
e
t.
a
l.
,
ā
A
c
a
d
e
m
ic
P
e
rf
o
r
m
a
n
c
e
P
re
d
icti
o
n
A
lg
o
rit
h
m
b
a
se
d
o
n
F
u
z
z
y
Da
ta
M
in
in
g
ā
.
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Arti
fi
c
ia
l
I
n
telli
g
e
n
c
e
,
v
o
l.
8
,
n
o
.
1
,
p
p
.
2
6
-
3
2
,
2
0
1
9
.
[9
]
No
rsy
e
la
e
t.
a
l,
ā
P
e
rf
o
rm
a
n
c
e
A
n
a
l
y
si
s
o
f
S
u
p
e
rv
ise
d
L
e
a
rn
in
g
M
o
d
e
ls
f
o
r
P
r
o
d
u
c
t
T
it
le
Clas
sif
ic
a
ti
o
n
ā
.
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
Art
if
icia
l
In
telli
g
e
n
c
e
,
v
o
l.
8
,
n
o
.
3
,
p
p
.
2
9
9
-
3
0
6
,
2
0
1
9
.
[1
0
]
Ćz
p
o
lat,
E.
,
a
n
d
A
k
a
r,
G
.
B,
ā
A
u
to
m
a
ti
c
d
e
tec
ti
o
n
o
f
lea
rn
in
g
st
y
les
f
o
r
a
n
e
-
lea
rn
in
g
s
y
ste
m
ā
.
Co
mp
u
ter
s
&
Ed
u
c
a
ti
o
n
,
v
o
l
.
5
3
,
p
p
.
3
5
5
-
3
6
7
,
2
0
0
9
.
[1
1
]
M
a
a
li
w
III,
R.
R,
ā
Clas
si
f
ica
ti
o
n
o
f
lea
rn
in
g
st
y
les
in
v
irt
u
a
l
lea
r
n
in
g
e
n
v
iro
n
m
e
n
t
u
sin
g
d
a
ta
m
in
in
g
:
A
b
a
sis
f
o
r
a
d
a
p
ti
v
e
c
o
u
rse
d
e
sig
n
ā
.
In
ter
n
a
ti
o
n
a
l
Res
e
a
rc
h
J
o
u
rn
a
l
o
f
En
g
i
n
e
e
rin
g
a
n
d
T
e
c
h
n
o
lo
g
y
(
IRJET
)
,
v
o
l.
3
,
n
o
.
7
,
p
p
.
56
-
6
1
,
2
0
1
6
.
[1
2
]
Be
rg
stra
,
J.,
a
n
d
Be
n
g
io
,
Y,
ā
R
a
n
d
o
m
se
a
rc
h
f
o
r
h
y
p
e
r
-
p
a
ra
m
e
t
e
r
o
p
ti
m
iza
ti
o
n
ā
.
J
o
u
rn
a
l
o
f
M
a
c
h
in
e
L
e
a
rn
i
n
g
Res
e
a
rc
h
,
v
o
l.
1
3
,
p
p
.
2
8
1
-
3
0
5
,
2
0
1
2
.
[1
3
]
Be
rg
stra
,
J.,
Ya
m
in
s,
D.,
a
n
d
C
o
x
,
D.
D,
ā
M
a
k
in
g
a
sc
ien
c
e
o
f
m
o
d
e
l
se
a
rc
h
:
H
y
p
e
rp
a
ra
m
e
ter
o
p
ti
m
iza
ti
o
n
i
n
h
u
n
d
re
d
s
o
f
d
im
e
n
sio
n
s f
o
r
v
isio
n
a
rc
h
it
e
c
tu
re
sā
.
J
M
L
R:
W
&
CP
,
v
o
l.
2
8
,
n
o
.
1
,
p
p
.
1
-
9
,
2
0
1
3
.
[1
4
]
M
a
c
lau
rin
,
D.,
D
u
v
e
n
a
u
d
,
D.,
a
n
d
A
d
a
m
s,
R,
ā
G
ra
d
ien
t
-
ba
se
d
h
y
p
e
rp
a
ra
m
e
t
e
r
o
p
ti
m
iza
ti
o
n
t
h
r
o
u
g
h
re
v
e
rsib
le
lea
rn
in
g
ā
.
In
I
n
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
o
n
ma
c
h
in
e
le
a
rn
i
n
g
,
v
o
l.
1
,
n
o
.
1
,
p
p
.
2
1
1
3
-
2
1
2
2
,
2
0
1
5
.
[1
5
]
Die
tt
e
rich
,
T
.
G
,
ā
A
n
e
x
p
e
ri
m
e
n
tal
c
o
m
p
a
riso
n
o
f
th
re
e
m
e
th
o
d
s
f
o
r
c
o
n
stru
c
ti
n
g
e
n
se
m
b
les
o
f
d
e
c
isio
n
tree
s:
Ba
g
g
in
g
,
b
o
o
stin
g
,
a
n
d
ra
n
d
o
m
iz
a
ti
o
n
ā
.
M
a
c
h
in
e
lea
r
n
in
g
,
v
o
l.
4
0
,
p
p
.
1
3
9
-
1
5
7
,
2
0
0
0
.
[1
6
]
V
il
a
lt
a
,
R.
,
a
n
d
Driss
i,
Y,
ā
A
p
e
rsp
e
c
ti
v
e
v
ie
w
a
n
d
su
rv
e
y
o
f
m
e
ta
-
l
e
a
rn
in
g
ā
.
Arti
fi
c
i
a
l
I
n
telli
g
e
n
c
e
R
e
v
iew
,
v
o
l.
1
8
,
p
p
.
7
7
-
9
5
,
2
0
0
2
.
[1
7
]
A
h
la
w
a
t,
A
.
,
a
n
d
S
u
ri,
B,
ā
Im
p
ro
v
in
g
c
las
sif
ic
a
ti
o
n
in
d
a
ta
m
in
in
g
u
si
n
g
h
y
b
rid
a
lg
o
rit
h
m
ā
.
In
In
f
o
rm
a
ti
o
n
p
ro
c
e
ss
in
g
(
IICIP
),
2
0
1
6
1
st I
n
d
i
a
in
ter
n
a
ti
o
n
a
l
c
o
n
fer
e
n
c
e
,
v
o
l.
1
,
n
o
.
1
,
p
p
.
1
-
4
,
2
0
1
6
.
[1
8
]
Hu
n
g
,
Y
u
Hs
i
n
,
Ra
y
I
Ch
a
n
g
,
a
n
d
Ch
u
n
F
u
L
in
,
ā
Hy
b
rid
lea
rn
i
n
g
sty
le
id
e
n
ti
f
ica
ti
o
n
a
n
d
d
e
v
e
l
o
p
i
n
g
a
d
a
p
t
iv
e
p
ro
b
lem
-
so
lv
in
g
lea
rn
in
g
a
c
ti
v
it
ie
s.ā
Co
mp
u
ter
s i
n
Hu
m
a
n
Beh
a
v
i
o
r
,
v
o
l.
1
,
n
o
.
5
5
,
p
p
.
5
5
2
ā
5
6
1
,
2
0
1
6
.
[1
9
]
V
il
a
lt
a
,
Rica
rd
o
a
n
d
Yo
u
ss
e
f
Driss
i
,
ā
A
p
e
rsp
e
c
ti
v
e
v
ie
w
a
n
d
su
rv
e
y
o
f
m
e
ta
-
lea
rn
in
g
.
ā
Arti
fi
c
ia
l
In
tell
ig
e
n
c
e
Rev
iew
,
v
o
l.
1
8
,
n
o
.
1
,
p
p
.
7
7
ā
9
5
,
2
0
0
2
.
[2
0
]
Ch
e
n
,
T
.
,
a
n
d
G
u
e
strin
,
C,
ā
X
g
b
o
o
st:
A
sc
a
lab
le
tree
b
o
o
stin
g
sy
ste
m
ā
.
In
Pro
c
e
e
d
in
g
s
o
f
t
h
e
2
2
n
d
ACM
sig
k
d
d
In
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
K
n
o
wled
g
e
Disc
o
v
e
ry
a
n
d
Da
t
a
M
in
i
n
g
,
v
o
l.
1
,
n
o
.
1
,
p
p
.
7
8
5
-
7
9
4
,
2
0
1
6
.
[2
1
]
Ch
a
,
Hy
u
n
Jin
,
e
t.
a
l,
ā
L
e
a
r
n
in
g
sty
le
s
d
iag
n
o
sis
b
a
se
d
o
n
u
se
r
in
te
rf
a
c
e
b
e
h
a
v
io
rs
f
o
r
th
e
c
u
sto
m
iza
ti
o
n
o
f
lea
rn
in
g
in
terf
a
c
e
s
in
a
n
i
n
telli
g
e
n
t
tu
t
o
rin
g
s
y
ste
m
.
ā
In
In
ter
n
a
ti
o
n
a
l
Co
n
fe
re
n
c
e
o
n
In
telli
g
e
n
t
T
u
to
ri
n
g
S
y
st
e
ms
,
v
o
l.
4
0
5
3
,
n
o
.
1
,
p
p
.
5
1
3
ā
5
2
4
,
2
0
0
6
.
[2
2
]
G
ra
f
,
S
.
,
V
io
la,
S
.
R.
,
L
e
o
,
T
.
,
a
n
d
Kin
sh
u
k
,
ā
In
-
d
e
p
t
h
a
n
a
ly
sis
o
f
th
e
f
e
ld
e
r
-
silv
e
r
m
a
n
lea
rn
in
g
st
y
l
e
d
ime
n
sio
n
sā
.
J
o
u
rn
a
l
o
f
Res
e
a
rc
h
o
n
T
e
c
h
n
o
lo
g
y
in
Ed
u
c
a
ti
o
n
,
v
o
l.
4
0
,
p
p
.
7
9
-
9
3
,
2
0
0
7
.
[2
3
]
T
o
rg
o
,
L
u
is,
Da
ta
M
in
in
g
w
it
h
R,
L
e
a
rn
in
g
w
it
h
Ca
se
S
tu
d
ies
.
Ne
w
Y
o
rk
:
Ch
a
p
ma
n
a
n
d
H
a
ll
/CR
C
,
v
o
l.
2
,
2
nd
e
d
it
i
o
n
,
2
0
1
7
.
[2
4
]
Be
rn
a
rd
,
J.,
C
h
a
n
g
,
T
.
-
W
.
,
P
o
p
e
sc
u
,
E.
,
a
n
d
G
ra
f
,
S
.
,
ā
L
e
a
rn
in
g
sty
l
e
id
e
n
ti
f
ier:
Im
p
ro
v
in
g
th
e
p
re
c
isio
n
o
f
lea
rn
in
g
st
y
le
id
e
n
ti
f
ica
ti
o
n
th
ro
u
g
h
c
o
m
p
u
tatio
n
a
l
in
telli
g
e
n
c
e
a
lg
o
rit
h
m
sā
.
Exp
e
rt
S
y
ste
ms
wit
h
Ap
p
l
ica
ti
o
n
s
,
v
o
l.
7
5
,
n
o
.
1
,
p
p
.
9
4
-
1
0
8
,
2
0
1
7
.
[2
5
]
G
ra
f
,
L
.
T
.
-
C.
,
S
a
b
in
e
,
e
t
a
l.
,
ā
S
u
p
p
o
rti
n
g
tea
c
h
e
rs
in
id
e
n
ti
f
y
in
g
s
tu
d
e
n
ts'
lea
rn
in
g
sty
les
in
lea
rn
in
g
m
a
n
a
g
e
m
e
n
t
s
y
ste
m
s:
A
n
a
u
to
m
a
ti
c
stu
d
e
n
t
m
o
d
e
ll
i
n
g
a
p
p
ro
a
c
h
ā
.
J
o
u
rn
a
l
o
f
E
d
u
c
a
t
io
n
a
l
T
e
c
h
n
o
l
o
g
y
&
S
o
c
iety
,
v
o
l.
1
2
,
n
o
.
4
,
p
p
.
3
-
1
4
,
2
0
0
9
.
Evaluation Warning : The document was created with Spire.PDF for Python.