I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
,
p
p
.
3
2
4
1
~
3
2
4
6
I
SS
N:
2088
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
9
i
4
.
p
p
3
2
4
1
-
3246
3241
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e
.
co
m/
jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
I
JE
C
E
Ens
e
m
bl
e learnin
g
f
o
r so
ft
w
a
re f
a
ult
pred
iction p
ro
b
le
m
w
ith
i
m
ba
la
nced
da
ta
T
ha
nh
T
un
g
K
hu
a
t
,
M
y
H
a
nh
L
e
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
l
o
g
y
F
a
c
u
lt
y
,
T
h
e
Un
iv
e
rsit
y
o
f
Da
n
a
n
g
,
Un
iv
e
rsity
o
f
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
lo
g
y
,
V
ietn
a
m
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Oct
10
,
2
0
1
8
R
ev
i
s
ed
Mar
19
,
2
0
1
9
A
cc
ep
ted
A
p
r
3
,
2
0
1
9
F
a
u
lt
p
re
d
icti
o
n
p
r
o
b
lem
h
a
s
a
c
ru
c
ial
ro
le
in
th
e
so
f
t
wa
re
d
e
v
e
lo
p
m
e
n
t
p
ro
c
e
ss
b
e
c
a
u
se
it
c
o
n
tri
b
u
tes
to
re
d
u
c
in
g
d
e
f
e
c
ts
a
n
d
a
ss
isti
n
g
t
h
e
tes
ti
n
g
p
ro
c
e
ss
to
w
a
rd
s
fa
u
lt
-
f
re
e
so
f
t
w
a
re
c
o
m
p
o
n
e
n
ts.
T
h
e
re
f
o
re
,
th
e
re
a
re
a
lo
t
o
f
e
ff
o
rts
a
i
m
in
g
to
a
d
d
re
ss
th
i
s
ty
p
e
o
f
issu
e
s,
in
w
h
ich
sta
ti
c
c
o
d
e
c
h
a
ra
c
teristics
a
re
u
su
a
ll
y
a
d
o
p
ted
to
c
o
n
stru
c
t
f
a
u
lt
c
las
sif
ica
ti
o
n
m
o
d
e
ls.
On
e
o
f
th
e
c
h
a
ll
e
n
g
in
g
p
ro
b
lem
s
in
f
lu
e
n
c
in
g
t
h
e
p
e
rf
o
rm
a
n
c
e
o
f
p
re
d
ictiv
e
c
las
si
f
iers
is
th
e
h
ig
h
i
m
b
a
lan
c
e
a
m
o
n
g
p
a
tt
e
rn
s
b
e
lo
n
g
in
g
to
d
if
f
e
re
n
t
c
las
se
s.
T
h
is
p
a
p
e
r
a
i
m
s
to
in
teg
ra
te
th
e
sa
m
p
li
n
g
te
c
h
n
iq
u
e
s
a
n
d
c
o
m
m
o
n
c
las
si
f
ica
ti
o
n
tec
h
n
i
q
u
e
s
t
o
f
o
rm
a
u
se
f
u
l
e
n
se
m
b
le
m
o
d
e
l
f
o
r
th
e
so
f
t
w
a
re
d
e
f
e
c
t
p
re
d
ictio
n
p
ro
b
lem
.
T
h
e
e
m
p
iri
c
a
l
re
su
lt
s c
o
n
d
u
c
ted
o
n
th
e
b
e
n
c
h
m
a
r
k
d
a
tas
e
ts
o
f
so
f
t
w
a
re
p
ro
j
e
c
ts
h
a
v
e
sh
o
w
n
th
e
p
ro
m
isin
g
p
e
rf
o
r
m
a
n
c
e
o
f
o
u
r
p
ro
p
o
sa
l
i
n
c
o
m
p
a
riso
n
w
it
h
i
n
d
i
v
id
u
a
l
c
las
sif
iers
.
K
ey
w
o
r
d
s
:
C
las
s
i
f
ier
Data
s
a
m
p
li
n
g
E
n
s
e
m
b
le
lear
n
i
n
g
R
an
d
o
m
u
n
d
er
s
a
m
p
li
n
g
So
f
t
w
ar
e
f
a
u
lt
p
r
ed
ictio
n
Co
p
y
rig
h
t
©
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
:
M
y
Han
h
L
e,
I
n
f
o
r
m
a
tio
n
T
ec
h
n
o
lo
g
y
Fac
u
l
t
y
,
Un
i
v
er
s
it
y
o
f
Da
n
an
g
,
U
n
iv
er
s
it
y
o
f
Scie
n
ce
an
d
T
ec
h
n
o
lo
g
y
,
5
4
Ng
u
y
e
n
L
u
o
n
g
B
a
n
g
,
Da
n
a
n
g
,
5
5
0
0
0
0
,
Viet
Nam
.
E
m
ail:
lt
m
h
an
h
@
d
u
t.
u
d
n
.
v
n
1.
I
NT
RO
D
UCT
I
O
N
In
th
e
p
as
t
f
e
w
y
ea
r
s
,
r
esear
ch
er
s
h
av
e
p
u
t
m
o
r
e
e
f
f
o
r
t
in
t
o
s
o
f
t
w
ar
e
p
r
o
p
er
ty
p
r
ed
ictio
n
p
r
o
b
lem
s
s
u
c
h
as
e
f
f
o
r
t
esti
m
atio
n
[
1
]
,
d
ef
ec
t
clas
s
i
f
icatio
n
[
2
]
,
an
d
s
o
f
t
w
ar
e
q
u
alit
y
p
r
ed
ictio
n
[
3
]
.
W
h
ile
t
h
e
r
is
k
o
f
d
ef
ec
ts
w
i
th
in
th
e
s
o
f
t
w
ar
e
m
o
d
u
les
u
n
d
er
d
ev
elo
p
m
e
n
t
ar
e
h
i
g
h
,
test
i
n
g
o
p
er
atio
n
s
[
4
]
ar
e
ti
m
e
-
co
n
s
u
m
i
n
g
an
d
ex
p
en
s
iv
e,
an
d
th
e
y
ca
n
n
o
t
b
e
p
er
f
o
r
m
ed
f
o
r
en
tire
ele
m
en
ts
.
A
s
a
r
esu
l
t
,
ac
cu
r
ate
p
r
ed
ictio
n
o
f
f
a
u
lts
i
n
s
o
f
t
w
ar
e
u
n
its
m
i
g
h
t
h
elp
m
a
n
ag
er
s
to
allo
ca
te
li
m
ited
ti
m
e
an
d
p
r
ec
io
u
s
r
eso
u
r
ce
s
to
d
ep
lo
y
a
n
e
f
f
icie
n
t
s
o
f
t
w
ar
e
test
i
n
g
p
r
o
ce
ss
.
Alo
n
g
w
it
h
th
e
ad
v
a
n
ce
m
en
t
o
f
m
ac
h
i
n
e
lear
n
i
n
g
tec
h
n
iq
u
e
s
,
v
ar
io
u
s
s
o
f
t
w
ar
e
m
etr
ics
h
a
v
e
b
ee
n
u
s
ed
to
co
n
s
tr
u
ct
p
r
ed
ictiv
e
m
o
d
els
f
o
r
id
en
ti
f
y
i
n
g
f
au
l
t
-
p
r
o
n
e
s
o
f
t
w
a
r
e
m
o
d
u
les
s
u
ch
a
s
s
tatic
co
d
e
m
etr
ic
s
,
ex
ec
u
tio
n
tr
ac
es,
an
d
h
is
to
r
ical
co
d
e
ch
an
g
e
s
[
5
]
.
T
h
is
w
o
r
k
also
e
m
p
l
o
y
s
th
e
s
ta
tic
co
d
e
m
etr
ics i
n
cl
u
d
in
g
clas
s
-
lev
el
a
n
d
m
et
h
o
d
-
le
v
el
o
n
es
to
b
u
i
ld
s
o
f
t
w
ar
e
f
au
lt c
la
s
s
i
f
ier
s
.
On
e
o
f
t
h
e
f
ea
tu
r
e
s
o
f
s
o
f
t
w
ar
e
q
u
alit
y
d
ataset
s
is
t
h
e
i
m
b
al
an
ce
b
et
w
ee
n
t
h
e
n
u
m
b
er
o
f
p
atter
n
s
i
n
ea
ch
cla
s
s
lab
el
,
w
h
er
e
m
o
s
t
v
u
ln
er
ab
le
co
m
p
o
n
e
n
ts
o
f
t
h
e
s
o
f
t
w
ar
e
s
y
s
te
m
m
a
y
o
n
l
y
b
e
s
o
u
g
h
t
w
i
th
a
s
m
al
l
r
ati
o
.
T
h
er
ef
o
r
e
,
th
e
q
u
an
tit
y
o
f
f
a
u
lt
y
s
a
m
p
les
i
n
su
ch
s
o
f
t
w
ar
e
d
ataset
s
is
m
u
c
h
lo
w
e
r
th
an
t
h
at
o
f
n
o
n
-
d
ef
ec
tiv
e
p
atter
n
s
[
6
]
.
Un
f
o
r
tu
n
ate
l
y
,
th
e
p
er
f
o
r
m
a
n
ce
o
f
m
o
s
t
co
n
v
e
n
tio
n
al
cla
s
s
i
f
ier
s
,
lik
e
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
s
[
7
]
,
K
-
n
ea
r
est
n
ei
g
h
b
o
r
[
8
]
,
n
eu
r
al
n
et
w
o
r
k
s
[
9
]
,
an
d
B
ay
esia
n
n
et
w
o
r
k
[
1
0
]
,
is
s
ig
n
if
ica
n
tl
y
d
ec
r
ea
s
ed
on
th
e
class
-
i
m
b
ala
n
ce
p
r
o
b
lem
.
T
h
ey
ar
e
u
s
u
all
y
to
w
ar
d
s
th
e
d
o
m
in
a
n
t
clas
s
an
d
ten
d
to
d
is
r
eg
ar
d
th
e
m
i
n
o
r
it
y
clas
s
,
an
d
th
i
s
p
h
en
o
m
en
o
n
i
s
p
o
s
s
ib
le
to
lea
d
to
h
i
g
h
f
a
ls
e
n
e
g
ati
v
e
r
ates
[
1
1
]
.
T
o
s
o
lv
e
th
is
p
r
o
b
lem
,
d
ata
s
a
m
p
li
n
g
m
et
h
o
d
s
ar
e
r
eg
u
lar
l
y
ad
o
p
ted
co
m
b
in
ed
w
i
th
p
r
ed
icti
v
e
m
o
d
el
s
.
T
h
is
p
ap
er
m
ak
e
s
u
s
e
o
f
r
an
d
o
m
u
n
d
er
s
a
m
p
li
n
g
(
R
U
S)
to
co
p
e
w
it
h
th
e
i
m
b
a
lan
ce
d
d
ata
p
r
o
b
le
m
.
W
e
f
ir
s
t
p
r
o
d
u
ce
t
h
e
b
alan
ce
d
d
atasets
b
y
u
tili
z
in
g
th
e
R
US
tech
n
iq
u
e
s
f
o
r
an
o
r
ig
in
a
l
i
m
b
ala
n
ce
d
d
ataset.
T
h
ese
b
alan
ce
d
d
atasets
ar
e
th
e
n
p
u
t
to
v
ar
io
u
s
b
ase
p
r
ed
icto
r
s
an
d
f
i
n
all
y
,
a
s
p
ec
if
ic
e
n
s
e
m
b
le
r
u
le
is
d
ep
lo
y
ed
to
co
m
b
i
n
e
th
e
clas
s
i
f
icatio
n
r
es
u
lt
s
o
f
t
h
e
s
e
b
ase
m
o
d
els.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
:
3
2
4
1
-
3246
3242
T
h
eo
r
etica
l
an
d
e
m
p
ir
ical
e
v
i
d
en
ce
h
a
s
i
n
d
icate
d
th
a
t
t
h
e
en
s
e
m
b
le
m
et
h
o
d
o
f
m
u
l
tip
le
class
i
f
ier
s
m
a
y
m
a
k
e
th
e
u
lti
m
ate
p
r
ed
ictiv
e
m
o
d
el
m
o
r
e
ac
c
u
r
ac
y
[
1
2
]
.
Nev
er
th
eles
s
,
th
er
e
ar
e
v
er
y
f
e
w
s
t
u
d
ies
ap
p
l
y
i
n
g
t
h
e
en
s
e
m
b
le
ap
p
r
o
ac
h
to
th
e
s
o
f
t
w
ar
e
d
ef
ec
t
p
r
ed
ictio
n
p
r
o
b
lem
[
1
3
]
.
T
o
th
e
b
est
o
f
o
u
r
k
n
o
w
led
g
e,
th
is
w
o
r
k
is
t
h
e
f
ir
s
t
s
tu
d
y
o
n
e
m
p
ir
ical
ass
e
s
s
m
en
t
o
f
th
e
in
f
l
u
e
n
ce
o
f
s
a
m
p
lin
g
o
n
en
s
e
m
b
l
e
m
o
d
el
s
co
n
ce
r
n
i
n
g
i
m
b
alan
ce
d
tr
ain
i
n
g
d
ataset
s
f
o
r
t
h
e
s
o
f
t
w
ar
e
f
a
u
lt
p
r
ed
ictio
n
p
r
o
b
le
m
.
T
h
e
p
r
in
cip
al
ai
m
o
f
th
is
p
ap
er
is
to
r
ev
ea
l
th
e
v
ital
r
o
le
o
f
th
e
s
am
p
li
n
g
tech
n
i
q
u
e
to
th
e
ac
cu
r
ac
y
o
f
th
e
en
s
e
m
b
le
class
i
f
ier
on
i
m
b
alan
ce
d
d
ata.
W
e
u
s
e
a
s
o
f
t
w
ar
e
d
ef
ec
t
en
s
e
m
b
le
p
r
ed
icto
r
co
n
s
is
ti
n
g
o
f
f
i
v
e
b
ase
class
i
f
ier
s
:
k
-
n
ea
r
es
t
n
eig
h
b
o
r
,
B
ay
e
s
ia
n
n
et
w
o
r
k
s
,
J
4
8
,
m
u
lti
la
y
er
p
er
ce
p
tr
o
n
,
an
d
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
i
n
e
s
.
T
h
e
d
iv
er
s
it
y
i
n
class
i
f
icatio
n
ab
ilit
ies
o
f
th
e
b
ase
class
i
f
ier
s
m
a
y
co
n
tr
ib
u
te
to
ca
p
tu
r
in
g
d
i
f
f
er
en
t
s
tatis
tic
al
ch
ar
ac
ter
is
tic
s
o
f
th
e
u
n
d
er
l
y
in
g
d
ata.
E
m
p
ir
ica
l
r
esu
lt
s
ar
e
p
er
f
o
r
m
ed
o
n
s
e
v
en
s
o
f
t
w
ar
e
d
ef
ec
t
d
ataset
s
f
r
o
m
t
h
e
P
R
OM
I
S
E
r
ep
o
s
ito
r
y
[
1
4
]
.
Ou
r
m
ain
co
n
t
r
ib
u
tio
n
s
i
n
t
h
is
p
ap
er
ca
n
b
e
s
u
m
m
ar
ized
as
f
o
llo
w
s
:
-
W
e
p
r
o
p
o
s
e
a
g
en
er
al
m
eth
o
d
o
f
b
u
ild
in
g
a
n
en
s
e
m
b
le
m
o
d
el
o
f
b
ase
cla
s
s
i
f
ier
s
f
o
r
s
o
f
t
w
ar
e
f
a
u
lt
p
r
ed
ictio
n
u
s
i
n
g
i
m
b
ala
n
ce
d
tr
ain
i
n
g
d
atase
ts
-
W
e
ass
es
s
t
h
e
cr
u
c
ial
r
o
le
o
f
t
h
e
u
n
d
er
-
r
a
n
d
o
m
s
a
m
p
le
tech
n
iq
u
e
o
n
i
m
p
r
o
v
i
n
g
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
en
s
e
m
b
le
m
o
d
els t
h
r
o
u
g
h
e
x
p
er
i
m
en
tal
r
esu
lts
o
n
h
i
g
h
l
y
i
m
b
alan
ce
d
s
o
f
t
w
ar
e
f
a
u
lt d
ata
s
e
ts
T
h
e
r
em
ai
n
d
er
o
f
t
h
is
p
ap
er
is
o
u
tli
n
ed
as
f
o
llo
w
s
:
s
ec
tio
n
2
p
r
esen
ts
t
h
e
b
ac
k
g
r
o
u
n
d
k
n
o
wled
g
e
an
d
r
elate
d
w
o
r
k
o
f
t
h
e
r
an
d
o
m
u
n
d
er
s
a
m
p
lin
g
a
n
d
en
s
e
m
b
le
lea
r
n
in
g
.
Sect
io
n
3
d
is
c
u
s
s
es
o
u
r
p
r
o
p
o
s
ed
m
et
h
o
d
,
w
h
ile
s
ec
tio
n
4
th
e
a
n
al
y
s
i
s
o
f
ex
p
er
i
m
e
n
tal
r
es
u
lts
.
T
h
e
co
n
clu
s
io
n
an
d
f
u
t
u
r
e
w
o
r
k
ar
e
g
i
v
en
i
n
s
ec
t
io
n
5
.
2.
B
ACK
G
RO
UND
2
.
1
.
So
f
t
wa
re
f
a
ult
predict
io
n
Def
ec
t
p
r
ed
ictio
n
is
a
m
eth
o
d
o
f
ea
r
l
y
id
en
ti
f
icatio
n
o
f
f
au
lts
in
s
o
f
t
w
ar
e
m
o
d
u
le
s
.
I
t
in
v
e
s
t
ig
ates
t
h
e
p
r
o
p
er
ties
o
f
in
d
iv
id
u
al
co
d
e
ele
m
e
n
ts
to
d
eter
m
i
n
e
t
h
o
s
e
u
n
its
b
ein
g
f
au
lt
-
p
r
o
n
e
o
r
n
o
t
[
1
5
]
o
r
to
p
r
ed
ict
th
e
n
u
m
b
er
o
f
f
au
lts
i
n
ea
c
h
co
m
p
o
n
en
t
[
1
6
]
.
W
h
ile
th
e
latter
co
n
s
id
er
s
s
o
f
t
w
ar
e
d
e
f
ec
t p
r
ed
ictio
n
as a
r
eg
r
e
s
s
io
n
is
s
u
e,
th
e
f
o
r
m
er
ap
p
r
o
ac
h
r
eg
ar
d
s
it
as
a
class
i
f
icat
io
n
p
r
o
b
le
m
.
T
h
is
s
t
u
d
y
o
n
l
y
d
ea
ls
w
i
t
h
th
e
cla
s
s
i
f
icat
io
n
v
ie
w
p
o
in
t,
w
h
ich
p
r
ed
icts
a
s
o
f
t
w
ar
e
m
o
d
u
le
in
to
f
a
u
lt
-
p
r
o
n
e
o
r
n
o
n
-
f
au
lt
-
p
r
o
n
e.
A
lar
g
e
n
u
m
b
er
o
f
s
tatic
co
d
e
ch
ar
ac
ter
is
tics
h
av
e
b
ee
n
p
r
o
p
o
s
ed
f
o
r
th
e
s
o
f
t
w
ar
e
f
au
lt
p
r
ed
ictio
n
r
an
g
i
n
g
f
r
o
m
m
et
h
o
d
lev
el
m
etr
ics
s
u
c
h
as
L
in
e
s
O
f
C
o
d
e
-
b
ased
m
ea
s
u
r
es
[
1
7
]
,
Mc
C
ab
e
[
1
8
]
an
d
Hal
s
tead
[
1
9
]
m
etr
ics
to
class
lev
el
m
etr
ics
lik
e
C
h
id
a
m
b
er
-
Ke
m
er
er
[
2
0
]
an
d
C
o
n
ce
p
t
u
al
C
o
h
e
s
io
n
o
f
C
lass
es
m
ea
s
u
r
e
[
2
1
]
.
B
ased
o
n
s
tatic
co
d
e
m
e
tr
ics,
r
esear
ch
er
s
h
av
e
ad
o
p
ted
d
if
f
e
r
en
t
m
e
th
o
d
s
to
co
n
s
tr
u
c
t
s
o
f
t
w
ar
e
f
au
lt
p
r
ed
ictio
n
m
o
d
els.
I
n
g
en
er
al
,
co
n
v
en
tio
n
al
d
ef
ec
t
p
r
ed
ictio
n
ap
p
r
o
ac
h
es
co
n
s
is
t
o
f
f
o
u
r
m
ai
n
s
tep
s
,
i.e
.
,
co
n
s
tr
u
ct
io
n
o
f
tr
ai
n
i
n
g
d
at
asets
,
f
ea
tu
r
e
ex
tr
ac
tio
n
f
r
o
m
s
o
f
t
w
ar
e
d
ef
ec
t
d
ataset
s
,
d
ev
elo
p
m
e
n
t
o
f
a
p
r
ed
ictiv
e
m
o
d
el,
an
d
th
e
ap
p
licatio
n
o
f
t
h
e
co
n
s
tr
u
cted
m
o
d
el.
2
.
2
.
Cla
s
s
i
m
ba
la
nce
pro
ble
m
a
n
d r
a
nd
o
m
un
der
s
a
m
pl
ing
C
las
s
i
m
b
ala
n
ce
i
s
a
n
in
teg
r
al
attr
ib
u
te
o
f
t
h
e
s
o
f
t
w
ar
e
d
e
f
ec
t d
ata,
w
h
ic
h
co
m
p
r
is
e
o
n
l
y
a
f
e
w
f
au
lt
y
u
n
i
ts
an
d
a
lar
g
e
n
u
m
b
er
o
f
n
o
n
-
f
a
u
lt
y
m
o
d
u
le
s
[
2
2
]
.
T
h
is
ch
ar
ac
ter
is
tic
h
as
a
co
n
s
id
er
a
b
le
im
p
ac
t
o
n
b
o
th
th
e
t
r
ain
i
n
g
o
f
a
m
o
d
el
an
d
th
e
p
r
ed
ictiv
e
p
er
f
o
r
m
a
n
ce
s
in
c
e
m
o
s
t
m
ac
h
i
n
e
lear
n
i
n
g
al
g
o
r
ith
m
s
ten
d
to
f
o
r
m
class
i
f
ier
s
m
a
x
i
m
izi
n
g
t
h
e
o
v
er
all
clas
s
i
f
icatio
n
ac
c
u
r
ac
y
.
C
o
n
s
eq
u
en
tl
y
,
t
h
e
v
a
lu
ab
le
m
i
n
o
r
it
y
cla
s
s
i
s
u
s
u
all
y
i
g
n
o
r
ed
b
y
s
u
c
h
m
o
d
els.
Fo
r
ex
a
m
p
le,
g
i
v
e
n
a
d
ataset
h
av
in
g
o
n
l
y
1
%
o
f
t
h
e
f
au
lt
y
co
m
p
o
n
en
ts
,
an
o
v
er
all
ac
cu
r
ac
y
o
f
9
9
%
m
ig
h
t
b
e
ea
s
il
y
attain
ed
b
y
a
b
in
ar
y
clas
s
i
f
ier
g
r
o
u
p
in
g
all
d
ata
p
atter
n
s
as
n
o
n
-
f
au
lt
y
p
atter
n
s
.
A
s
a
r
es
u
lt,
t
h
e
m
i
n
o
r
it
y
d
e
f
ec
ti
v
e
in
s
ta
n
c
es
ar
e
all
m
is
c
la
s
s
if
ied
w
it
h
th
is
s
i
m
p
le
m
o
d
el.
I
n
th
i
s
ca
s
e,
it
o
u
tp
u
ts
a
v
er
y
h
ig
h
ac
c
u
r
ac
y
,
b
u
t
it
m
a
k
es
n
o
s
en
s
e.
T
h
er
ef
o
r
e,
th
e
class
i
m
b
a
lan
ce
p
r
o
b
le
m
o
f
ten
d
i
m
i
n
is
h
es
t
h
e
b
in
ar
y
p
r
ed
icto
r
s
,
an
d
f
u
r
th
er
m
ak
e
s
th
e
s
e
class
if
icatio
n
m
o
d
els
n
o
t
to
p
r
ed
ict
th
e
m
i
n
o
r
it
y
f
au
l
t
y
s
o
f
t
w
ar
e
u
n
i
ts
ac
cu
r
atel
y
.
Ma
n
y
s
t
u
d
ies
h
av
e
b
ee
n
i
n
tr
o
d
u
ce
d
to
h
an
d
le
t
h
e
cla
s
s
i
m
b
alan
ce
p
r
o
b
le
m
.
A
s
u
r
v
e
y
o
f
t
ec
h
n
iq
u
es
f
o
r
r
ed
u
cin
g
t
h
e
n
e
g
ativ
e
i
m
p
ac
t
o
f
i
m
b
a
lan
ce
o
n
clas
s
i
f
icatio
n
p
er
f
o
r
m
a
n
ce
w
a
s
p
r
o
p
o
s
ed
b
y
W
eiss
et
a
l.
[
2
3
]
.
C
r
u
cial
m
et
h
o
d
s
f
o
r
allev
iatin
g
th
e
i
n
f
l
u
en
ce
o
f
c
lass
i
m
b
alan
ce
m
i
g
h
t
b
e
ca
te
g
o
r
ized
in
to
g
r
o
u
p
s
,
n
a
m
el
y
e
x
ter
n
a
l
an
d
in
ter
n
al
m
et
h
o
d
s
.
I
n
ter
n
al
tech
n
iq
u
e
s
ai
m
to
m
o
d
if
y
e
x
is
tin
g
m
ac
h
i
n
e
lear
n
in
g
al
g
o
r
ith
m
s
f
o
r
r
ed
u
cin
g
th
eir
s
e
n
s
iti
v
en
e
s
s
to
cla
s
s
i
m
b
a
lan
ce
[
2
4
]
,
w
h
ile
th
e
ex
ter
n
al
ap
p
r
o
ac
h
ten
d
s
to
f
o
r
m
a
b
alan
ce
d
tr
ain
i
n
g
d
ataset.
T
h
e
ex
ter
n
al
ap
p
r
o
ac
h
es
ar
e
w
id
el
y
u
s
ed
as
th
e
y
ar
e
in
d
e
p
en
d
en
t
o
f
th
e
u
n
d
er
l
y
in
g
cla
s
s
i
f
icatio
n
alg
o
r
ith
m
s
.
Data
s
a
m
p
li
n
g
b
e
lo
n
g
s
to
th
e
e
x
ter
n
al
g
r
o
u
p
.
T
h
e
u
n
d
er
s
a
m
p
l
in
g
tech
n
iq
u
e
o
f
ten
eli
m
i
n
ate
s
s
a
m
p
les
o
f
th
e
m
aj
o
r
it
y
clas
s
f
o
r
o
b
tain
in
g
a
b
alan
ce
d
d
ataset
b
ef
o
r
e
tr
ain
in
g
t
h
e
class
i
f
ier
s
.
Ma
n
i
an
d
Z
h
a
n
g
[
2
5
]
p
o
in
ted
o
u
t
th
at
th
e
r
an
d
o
m
u
n
d
er
s
a
m
p
l
in
g
tech
n
i
q
u
e
r
eg
u
lar
l
y
o
u
tp
er
f
o
r
m
s
o
th
er
co
m
p
lex
s
a
m
p
l
in
g
s
tr
ateg
ies.
T
h
er
ef
o
r
e,
w
e
u
s
e
r
a
n
d
o
m
u
n
d
er
s
a
m
p
li
n
g
in
co
m
p
ar
is
o
n
w
it
h
b
ase
class
i
f
ier
s
to
b
u
ild
a
n
u
lt
i
m
ate
en
s
e
m
b
le
m
o
d
el.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
E
n
s
emb
le
lea
r
n
in
g
fo
r
s
o
ftw
a
r
e
fa
u
lt p
r
ed
ictio
n
p
r
o
b
lem
w
ith
imb
a
la
n
ce
d
d
a
t
a
(
My
Ha
n
h
Le
)
3243
3.
P
RO
P
O
SE
D
E
N
SE
M
B
L
E
M
O
DE
L
I
n
o
u
r
p
r
o
p
o
s
ed
m
o
d
el,
ea
ch
b
ase
class
i
f
ier
i
s
tr
ain
ed
o
n
a
d
if
f
er
e
n
t
b
alan
ce
d
d
atase
t
f
o
r
m
ed
f
r
o
m
th
e
s
a
m
p
l
in
g
s
tep
,
a
n
d
t
h
e
m
o
d
el
i
n
cl
u
d
es
th
r
ee
co
m
p
o
n
en
ts
:
d
ata
b
ala
n
cin
g
,
class
i
f
ier
s
tr
ain
in
g
,
an
d
class
i
f
y
i
n
g
.
T
h
e
d
etails ar
e
s
h
o
w
n
i
n
Fi
g
u
r
e
1
.
Fig
u
r
e
1
.
P
r
o
p
o
s
ed
en
s
e
m
b
le
class
i
f
ier
Du
r
in
g
t
h
e
tr
ai
n
i
n
g
p
r
o
ce
s
s
,
t
h
e
m
aj
o
r
ity
cla
s
s
s
a
m
p
le
s
i
n
t
h
e
o
r
ig
i
n
al
i
m
b
alan
ce
d
d
atase
t
ar
e
s
p
lit
in
to
s
ev
er
al
b
in
s
b
y
ad
o
p
tin
g
th
e
r
an
d
o
m
u
n
d
er
s
a
m
p
li
n
g
m
et
h
o
d
.
E
ac
h
b
in
in
clu
d
es
t
h
e
eq
u
al
n
u
m
b
er
o
f
p
atter
n
s
to
t
h
at
o
f
th
e
m
i
n
o
r
it
y
clas
s
,
an
d
t
h
e
n
all
m
i
n
o
r
it
y
class
p
at
ter
n
s
ar
e
p
u
t
in
to
ea
ch
b
in
to
f
o
r
m
t
h
e
b
alan
ce
d
tr
ain
i
n
g
d
ataset.
Af
t
er
th
at,
ea
ch
b
ase
clas
s
i
f
ier
will
b
e
tr
ain
ed
o
n
a
s
ep
ar
ated
b
alan
ce
d
d
ataset
b
y
a
s
p
ec
if
ic
c
lass
if
ica
tio
n
al
g
o
r
ith
m
.
Fi
n
all
y
,
t
h
e
f
in
a
l
clas
s
i
f
ier
is
b
u
ilt
b
y
co
m
b
i
n
i
n
g
th
e
o
u
tco
m
e
s
o
f
b
as
e
p
r
ed
icto
r
s
r
elied
o
n
th
e
m
aj
o
r
it
y
v
o
ti
n
g
r
u
le.
T
h
e
en
s
e
m
b
le
m
o
d
el
w
o
u
ld
th
e
n
b
e
d
ep
lo
y
ed
to
class
if
y
n
e
w
d
ata.
T
h
er
e
ar
e
v
ar
io
u
s
c
lass
if
icatio
n
tec
h
n
iq
u
e
s
p
o
s
s
ib
le
to
b
e
u
s
ed
f
o
r
b
ase
cla
s
s
i
f
ier
s
.
T
h
e
d
iv
er
s
it
y
o
f
b
ase
p
r
ed
icto
r
s
m
i
g
h
t
r
es
u
lt
i
n
t
h
e
p
er
f
o
r
m
a
n
ce
i
m
p
r
o
v
e
m
en
t
o
f
th
e
f
in
a
l
en
s
e
m
b
le
m
o
d
el.
I
n
th
is
s
t
u
d
y
,
w
e
u
s
e
f
i
v
e
co
m
m
o
n
clas
s
i
f
icatio
n
al
g
o
r
ith
m
s
,
i
n
cl
u
d
in
g
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
i
n
es
(
SVM)
[
7
]
,
m
u
ltil
a
y
er
p
er
ce
p
tr
o
n
(
ML
P
)
[
9
]
,
B
a
y
es
ian
n
et
w
o
r
k
s
[
1
0
]
,
K
-
n
ea
r
est
n
ei
g
h
b
o
r
(
K
NN)
[
2
6
]
,
an
d
d
ec
is
io
n
tr
ee
J
4
8
[
2
7
]
.
Div
er
s
it
y
is
a
cr
u
cial
f
ac
to
r
i
n
t
h
e
e
n
s
e
m
b
le
m
e
m
b
er
s
'
d
ec
i
s
io
n
s
.
I
t
ca
n
b
e
s
ee
n
th
at
b
ase
lear
n
er
s
ar
e
tr
ain
ed
o
n
d
i
f
f
er
en
t
d
atasets
,
a
n
d
t
h
is
w
ill
co
n
tr
ib
u
te
to
t
h
e
d
i
v
er
s
it
y
o
f
t
h
e
f
i
n
al
en
s
e
m
b
le
m
o
d
el
f
o
r
m
ed
f
r
o
m
th
e
m
aj
o
r
it
y
v
o
tin
g
r
u
le
f
o
r
o
u
tco
m
e
s
o
f
b
ase
clas
s
if
ier
s
.
4.
RE
SU
L
T
S AN
D
ANA
L
Y
SI
S
4
.
1
.
E
m
p
irica
l e
v
a
lua
t
io
n
cr
it
er
ia
a
nd
da
t
a
s
et
E
ac
h
b
in
ar
y
cla
s
s
i
f
icatio
n
is
s
u
e
is
as
s
o
ciate
d
w
it
h
f
o
u
r
p
o
s
s
ib
le
p
r
ed
ictio
n
ca
s
es,
i.e
.
,
tr
u
e
p
o
s
itiv
e
s
(
T
P
)
,
tr
u
e
n
eg
ati
v
es
(
T
N)
,
f
alse
p
o
s
iti
v
es
(
FP
)
,
an
d
f
a
ls
e
n
eg
a
tiv
e
s
(
FN)
.
As
f
o
r
t
h
e
s
o
f
t
w
ar
e
d
e
f
ec
t
p
r
ed
ictio
n
,
if
a
s
a
m
p
le
is
clas
s
if
ied
as
"
f
au
l
t
y
"
an
d
is
ac
tu
a
ll
y
"
f
au
lt
y
"
,
it
is
a
tr
u
e
p
o
s
iti
v
e;
if
a
n
o
n
-
f
a
u
lt
y
p
atter
n
is
m
is
c
lass
if
ied
a
s
"
f
au
lt
y
"
,
it
is
a
f
a
u
lt
p
o
s
iti
v
e.
I
n
a
s
i
m
ilar
w
a
y
,
tr
u
e
n
e
g
ati
v
e
s
h
o
w
s
t
h
at
t
h
e
n
o
n
-
f
au
lt
y
s
a
m
p
le
is
p
r
ed
icted
to
"
n
o
n
-
f
au
l
t
y
,
"
w
h
ile
f
a
u
lt
n
e
g
ativ
e
in
d
icate
s
a
n
er
r
o
r
s
it
u
at
io
n
w
h
er
e
a
b
u
g
g
y
p
r
o
g
r
am
u
n
it
i
s
i
n
co
r
r
ec
tl
y
g
r
o
u
p
ed
as
"
n
o
n
-
b
u
g
g
y
"
.
B
ased
o
n
t
h
ese
f
o
u
r
v
ar
iab
les,
m
ea
s
u
r
es
s
u
c
h
a
s
P
r
ec
is
io
n
,
R
ec
all,
an
d
F1
-
s
co
r
e
ar
e
co
m
p
u
ted
as
f
o
llo
w
s
:
=
+
=
+
1
−
=
2
⋅
⋅
+
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
:
3
2
4
1
-
3246
3244
T
o
ev
alu
ate
th
e
e
f
f
ec
ti
v
en
e
s
s
o
f
th
e
e
n
s
e
m
b
le
cla
s
s
i
f
ier
,
w
e
co
n
d
u
cted
ex
p
er
i
m
en
t
s
o
n
a
co
llectio
n
o
f
s
ev
e
n
h
ig
h
l
y
i
m
b
a
lan
ce
d
b
i
n
ar
y
d
ataset
s
f
r
o
m
th
e
P
R
OM
I
SE
r
ep
o
s
ito
r
y
o
f
s
o
f
t
w
ar
e
d
ef
ec
t
d
atab
ases
[
1
4
]
T
h
ese
s
ev
e
n
o
p
en
s
o
u
r
ce
d
at
asets
h
av
e
t
h
e
d
i
f
f
er
e
n
t
n
u
m
b
er
o
f
p
atter
n
s
,
f
ea
tu
r
es,
a
n
d
t
h
e
clas
s
i
m
b
ala
n
ce
r
atio
.
T
ab
le
1
s
h
o
w
s
t
h
e
at
t
r
ib
u
tes
o
f
ea
c
h
s
elec
ted
i
m
b
alan
ce
d
d
ataset,
in
cl
u
d
i
n
g
t
h
e
to
tal
n
u
m
b
er
o
f
attr
ib
u
tes
(
#
A
t
tr
.
)
,
th
e
n
u
m
b
er
o
f
p
atter
n
s
(
#
P
ats.)
,
th
e
n
u
m
b
er
o
f
d
ef
ec
ti
v
e
co
m
p
o
n
en
t
s
(
#
Def
ec
t)
,
th
e
n
u
m
b
er
o
f
n
o
n
-
d
ef
ec
ti
v
e
u
n
i
ts
(
#
No
n
-
d
ef
ec
t)
,
t
h
e
r
atio
o
f
f
a
u
lt
y
m
o
d
u
les
to
all
m
o
d
u
les
i
n
ea
c
h
d
ataset
(
%
Def
ec
t)
.
A
ll
s
e
v
e
n
s
o
f
t
w
ar
e
s
y
s
te
m
s
h
a
v
e
b
ee
n
w
r
it
ten
i
n
J
av
a
p
r
o
g
r
a
m
m
i
n
g
la
n
g
u
a
g
e.
E
ac
h
in
s
tan
ce
i
n
th
ese
d
atasets
r
ep
r
esen
t
s
a
s
i
n
g
le
J
av
a
clas
s
.
T
h
e
f
ea
t
u
r
e
s
et
o
f
ea
c
h
d
ata
s
et
co
n
s
i
s
ts
o
f
2
0
s
o
f
t
w
ar
e
m
etr
ic
s
s
u
c
h
as c
o
m
p
le
x
it
y
,
co
u
p
lin
g
,
co
h
esio
n
,
s
ize
an
d
d
ef
ec
t p
r
o
n
en
es
s
ch
ar
ac
ter
is
t
ic
s
o
f
a
J
av
a
class
.
T
ab
le
1
.
Su
m
m
ar
y
o
f
s
e
v
e
n
h
i
g
h
l
y
i
m
b
alan
ce
d
d
ataset
s
D
a
t
a
se
t
#
A
t
t
r
.
#
P
a
t
s
#
D
e
f
e
c
t
#
N
o
n
-
d
e
f
e
c
t
%De
f
e
c
t
A
n
t
1
.
7
20
7
4
5
1
6
6
5
7
9
2
2
.
2
8
%
C
a
me
l
1
.
6
20
9
6
5
1
8
8
7
7
7
1
9
.
4
8
%
I
v
y
2
.
0
20
3
5
2
40
3
1
2
1
1
.
3
6
%
P
o
i
2
.
0
20
3
1
4
37
2
7
7
1
1
.
7
8
%
T
o
mca
t
20
8
5
8
77
7
8
1
8
.
9
7
%
X
a
l
a
n
2
.
4
20
7
2
3
1
1
0
6
1
3
1
5
.
2
1
%
S
y
n
a
p
se
1
.
2
20
2
5
6
86
1
7
0
3
3
.
5
9
%
4
.
2
.
E
x
peri
m
ent
a
l r
esu
lt
s
4
.
2
.
1
.
Co
m
pa
riso
n o
f
t
he
ens
e
m
ble
m
o
del
s
w
it
h a
nd
w
it
ho
ut
us
ing
ra
nd
o
m
un
dersa
mp
lin
g
T
h
is
p
ar
t
is
to
u
n
co
v
er
i
f
t
h
e
u
n
d
er
s
a
m
p
li
n
g
-
b
ased
en
s
e
m
b
l
e
m
o
d
el
ca
n
h
a
n
d
le
t
h
e
class
i
m
b
alan
c
e
p
r
o
b
lem
m
o
r
e
e
f
f
icien
t
co
m
p
ar
ed
w
i
th
o
n
e
w
it
h
o
u
t
u
s
in
g
t
h
e
u
n
d
er
s
a
m
p
l
in
g
tec
h
n
iq
u
e.
No
n
-
s
a
m
p
li
n
g
en
s
e
m
b
le
m
ea
n
s
th
at
b
ase
c
la
s
s
i
f
ier
s
ar
e
tr
ain
ed
o
n
t
h
e
e
n
ti
r
e
o
r
ig
in
al
i
m
b
a
lan
ce
d
d
ataset
.
T
a
b
le
2
s
h
o
w
s
t
h
e
av
er
ag
e
r
esu
lts
o
f
F1
-
s
co
r
e
o
v
er
3
0
ex
ec
u
tio
n
ti
m
es
f
o
r
th
e
n
o
n
-
s
a
m
p
lin
g
an
d
u
n
d
e
r
s
a
m
p
li
n
g
en
s
e
m
b
le
m
o
d
el
s
.
I
n
th
e
tab
le,
t
h
e
b
est v
alu
e
o
f
ea
c
h
d
ataset
is
h
i
g
h
li
g
h
ted
in
b
o
ld
.
Fro
m
T
ab
le
2
,
it
is
o
b
s
er
v
e
d
th
at
t
h
e
i
n
te
g
r
atio
n
o
f
t
h
e
r
an
d
o
m
u
n
d
er
s
a
m
p
li
n
g
m
e
t
h
o
d
w
ith
en
s
e
m
b
le
lear
n
i
n
g
o
u
tp
er
f
o
r
m
s
t
h
e
e
n
s
e
m
b
le
class
if
ier
w
it
h
o
u
t
u
s
in
g
t
h
e
s
a
m
p
li
n
g
tech
n
iq
u
e
in
all
i
m
b
alan
ce
d
d
ata
s
ets,
esp
ec
iall
y
f
o
r
th
e
p
o
i
2
.
0
d
ataset.
I
n
t
h
is
d
ataset,
th
e
en
s
e
m
b
le
p
r
ed
icto
r
tr
ain
ed
o
n
th
e
o
r
ig
in
al
i
m
b
a
lan
ce
d
d
ata
o
u
tp
u
ts
t
h
e
F1
-
s
co
r
e
v
al
u
e
b
ein
g
c
o
m
p
lete
l
y
i
n
ac
cu
r
ate,
w
h
ile
th
e
en
s
e
m
b
le
m
o
d
el
u
s
i
n
g
t
h
e
r
an
d
o
m
u
n
d
er
s
a
m
p
l
in
g
al
g
o
r
ith
m
s
ig
n
i
f
ica
n
tl
y
e
n
h
an
ce
s
t
h
e
ac
c
u
r
ac
y
o
f
F1
-
s
c
o
r
e.
T
h
ese
r
es
u
lts
in
d
icate
t
h
at
s
a
m
p
lin
g
tech
n
iq
u
e
co
n
tr
ib
u
tes
to
t
h
e
co
n
s
id
er
ab
le
i
m
p
r
o
v
e
m
e
n
t
o
f
t
h
e
ac
cu
r
ac
y
o
f
th
e
en
s
e
m
b
le
clas
s
if
ier
r
eg
ar
d
in
g
t
h
e
class
i
m
b
ala
n
ce
tr
ain
i
n
g
d
atasets
.
T
ab
le
2
.
A
v
er
ag
e
F1
-
s
co
r
e
v
al
u
es o
f
i
m
b
ala
n
ce
d
d
atasets
f
o
r
th
e
en
s
e
m
b
le
m
o
d
els
D
a
t
a
se
t
N
o
n
-
samp
l
i
n
g
e
n
se
mb
l
e
U
n
d
e
r
S
a
m
p
l
i
n
g
e
n
se
mb
l
e
A
n
t
1
.
7
0
.
5
2
7
8
0
.
6
2
6
1
C
a
me
l
1
.
6
0
.
2
3
2
1
0
.
4
4
1
3
I
v
y
2
.
0
0
.
2
7
5
9
0
.
3
9
3
7
P
o
i
2
.
0
0
0
.
3
3
5
4
T
o
mca
t
0
.
2
2
2
2
0
.
3
8
9
9
X
a
l
a
n
2
.
4
0
.
2
5
3
5
0
.
4
5
3
5
S
y
n
a
p
se
1
.
2
0
.
5
3
3
3
0
.
6
4
8
7
4
.
2
.
2
.
Co
m
pa
riso
n o
f
t
he
ens
e
m
ble
m
o
del a
n
d it
s
ba
s
e
cla
s
s
if
iers
T
h
e
p
u
r
p
o
s
e
o
f
th
i
s
ex
p
er
i
m
en
t
is
to
v
alid
ate
w
h
et
h
er
th
e
e
n
s
e
m
b
le
m
o
d
el
u
s
i
n
g
th
e
r
an
d
o
m
u
n
d
er
s
a
m
p
li
n
g
lead
to
b
etter
a
v
er
ag
e
F1
-
s
co
r
e
v
al
u
es c
o
m
p
a
r
ed
to
th
eir
b
ase
clas
s
i
f
ier
s
.
T
ab
le
3
d
escr
ib
es th
e
r
esu
lt
s
o
f
th
e
en
s
e
m
b
le
m
o
d
el
an
d
its
b
ase
class
i
f
ier
s
w
h
en
tr
ain
ed
o
n
th
e
o
r
ig
i
n
al
i
m
b
alan
ce
d
d
ata
an
d
b
alan
ce
d
d
atasets
.
T
h
e
b
est r
esu
lt
s
f
o
r
ea
ch
d
atase
t
ar
e
h
ig
h
l
ig
h
ted
in
b
o
ld
.
Gen
er
all
y
,
b
ase
p
r
ed
icto
r
s
tr
ai
n
ed
o
n
b
ala
n
ce
d
d
ata
o
u
tp
u
t
m
u
ch
b
etter
av
er
a
g
e
F1
-
s
co
r
e
r
e
s
u
lt
s
o
v
er
all
d
atasets
co
m
p
ar
ed
to
th
o
s
e
tr
ain
ed
o
n
o
r
ig
in
al
i
m
b
alan
ce
d
d
atasets
,
esp
ec
iall
y
k
NN,
ML
P
,
an
d
SVM.
I
t is ea
s
y
to
o
b
s
er
v
e
t
h
at
s
e
v
er
al
class
i
f
ier
s
s
u
c
h
as J
4
8
,
SVM,
an
d
k
NN
ar
e
v
er
y
s
e
n
s
iti
v
e
to
i
m
b
alan
ce
d
d
ata,
an
d
th
e
y
g
en
er
ate
i
n
cr
ed
ib
l
y
i
n
ac
cu
r
ate
F1
-
s
co
r
e
v
al
u
es.
W
h
en
ad
o
p
tin
g
t
h
e
o
r
ig
in
al
i
m
b
alan
ce
d
d
ataset
to
tr
ain
m
o
d
el
s
,
t
h
e
en
s
e
m
b
le
m
o
d
el
ca
n
n
o
t
o
u
tp
er
f
o
r
m
all
b
ase
clas
s
if
ier
s
o
n
all
e
x
p
er
i
m
en
tal
d
ataset
s
.
Ho
w
e
v
er
,
th
e
r
an
d
o
m
u
n
d
er
s
a
m
p
lin
g
tec
h
n
iq
u
e
ass
i
s
ts
t
h
e
en
s
e
m
b
le
clas
s
i
f
ier
to
p
er
f
o
r
m
b
etter
t
h
an
t
h
eir
b
ase
lear
n
er
s
o
n
all
d
ata
s
ets
.
I
t
is
co
n
clu
d
ed
t
h
at
t
h
e
u
s
e
o
f
r
an
d
o
m
u
n
d
er
s
a
m
p
li
n
g
co
n
tr
ib
u
tes
to
t
h
e
s
ig
n
i
f
ica
n
t
i
m
p
r
o
v
e
m
e
n
t
o
f
th
e
p
er
f
o
r
m
a
n
ce
o
f
b
ase
class
i
f
ier
s
a
n
d
th
e
f
i
n
al
en
s
e
m
b
le
m
o
d
el.
Ob
tain
ed
r
es
u
lts
h
a
v
e
s
h
o
wn
t
h
e
cr
itical
r
o
le
o
f
b
ala
n
c
ed
tr
ain
in
g
d
ata
s
e
ts
o
n
t
h
e
ac
cu
r
ac
y
o
f
b
i
n
ar
y
class
i
f
icatio
n
alg
o
r
it
h
m
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
E
n
s
emb
le
lea
r
n
in
g
fo
r
s
o
ftw
a
r
e
fa
u
lt p
r
ed
ictio
n
p
r
o
b
lem
w
ith
imb
a
la
n
ce
d
d
a
t
a
(
My
Ha
n
h
Le
)
3245
T
ab
le
3
.
A
v
er
ag
e
F1
-
s
co
r
e
o
v
er
d
atasets
o
f
th
e
e
n
s
e
m
b
le
m
o
d
el
an
d
its
b
ase
class
i
f
ier
s
Ty
p
e
C
l
a
ssi
f
i
e
r
D
a
t
a
se
t
A
n
t
1
.
7
C
a
me
l
1
.
6
I
v
y
2
.
0
P
o
i
2
.
0
T
o
mca
t
X
a
l
a
n
2
.
4
S
y
n
a
p
se
1
.
2
O
r
i
g
i
n
a
l
i
mb
a
l
a
n
c
e
d
d
a
t
a
se
t
En
se
mb
l
e
0
.
5
2
7
8
0
.
2
3
2
1
0
.
2
7
5
9
0
0
.
2
2
2
2
0
.
2
5
3
5
0
.
5
3
3
3
k
N
N
0
.
4
9
6
1
0
.
2
1
7
1
0
.
2
6
6
7
0
.
0
8
3
3
0
.
1
1
5
4
0
.
1
9
1
8
0
.
5
5
5
6
BN
0
.
6
0
9
5
0
.
2
4
4
9
0
.
3
2
5
6
0
0
.
3
6
3
4
0
.
4
3
4
8
0
.
5
3
8
5
J4
8
0
.
5
5
7
6
0
.
2
8
5
7
0
.
3
1
2
5
0
0
.
3
0
7
7
0
.
2
6
3
2
0
.
5
5
7
M
L
P
0
.
4
4
9
3
0
.
3
1
7
2
0
.
1
9
3
5
0
.
0
6
6
7
0
.
2
7
4
0
.
3
0
5
9
0
.
5
7
8
9
S
V
M
0
.
3
1
8
6
0
.
0
2
0
8
0
.
1
9
0
5
0
.
0
8
3
3
0
0
.
0
3
5
1
0
.
4
7
0
6
R
a
n
d
o
m
u
n
d
e
r
samp
l
i
n
g
En
se
mb
l
e
0
.
6
2
6
1
0
.
4
4
1
3
0
.
3
9
3
7
0
.
3
3
5
4
0
.
3
8
9
9
0
.
4
5
3
5
0
.
6
4
8
7
k
N
N
0
.
5
3
8
2
0
.
4
0
2
1
0
.
2
6
9
5
0
.
3
1
9
1
0
.
3
4
5
5
0
.
3
8
6
5
0
.
5
9
4
3
BN
0
.
6
0
5
9
0
.
2
4
3
9
0
.
3
7
0
9
0
.
3
2
0
6
0
.
3
6
8
6
0
.
4
0
6
7
0
.
6
1
0
5
J4
8
0
.
5
8
3
3
0
.
3
8
0
4
0
.
3
5
3
4
0
0
.
3
4
1
8
0
.
3
9
8
4
0
.
5
5
9
M
L
P
0
.
5
0
1
5
0
.
3
9
5
4
0
.
2
9
3
4
0
.
3
1
8
4
0
.
3
3
9
3
0
.
3
8
0
7
0
.
5
8
3
3
S
V
M
0
.
5
8
8
4
0
.
3
7
7
6
0
.
3
2
2
7
0
.
3
0
7
8
0
.
3
5
3
8
0
.
3
9
3
1
0
.
5
9
2
4
5.
CO
NCLU
SI
O
N
T
h
is
p
ap
er
s
h
o
w
ed
t
h
e
e
f
f
ic
ie
n
c
y
o
f
i
n
te
g
r
ati
n
g
th
e
r
an
d
o
m
u
n
d
er
s
a
m
p
li
n
g
to
t
h
e
e
n
s
e
m
b
le
lear
n
i
n
g
o
n
th
e
i
m
b
ala
n
ce
d
s
o
f
t
w
ar
e
d
ef
ec
t
d
atasets
.
E
x
p
er
i
m
e
n
t
al
o
u
tco
m
es
p
o
in
ted
o
u
t
t
h
a
t
b
alan
ce
d
tr
ain
in
g
d
atasets
allo
w
t
h
e
s
i
g
n
i
f
ica
n
t
en
h
a
n
ce
m
en
t
o
f
p
er
f
o
r
m
a
n
ce
o
f
b
o
th
th
e
e
n
s
e
m
b
le
m
o
d
el
a
n
d
b
ase
class
if
ier
s
.
As
a
r
es
u
lt,
t
h
e
co
m
b
in
at
io
n
o
f
t
h
e
s
a
m
p
li
n
g
tec
h
n
iq
u
e
an
d
en
s
e
m
b
le
lear
n
i
n
g
co
n
tr
i
b
u
tes
to
f
o
r
m
i
n
g
a
p
r
o
m
i
s
in
g
cla
s
s
i
f
ier
f
o
r
t
h
e
s
o
f
t
w
ar
e
f
au
l
t
p
r
ed
ictio
n
p
r
o
b
le
m
.
T
h
e
en
s
e
m
b
le
m
o
d
el
i
n
t
h
i
s
p
ap
er
ad
o
p
ts
o
n
l
y
a
s
i
m
p
le
m
aj
o
r
it
y
v
o
tin
g
r
u
l
e
.
T
h
er
ef
o
r
e
,
w
e
i
n
ten
d
to
p
r
o
d
u
ce
a
v
ar
iet
y
o
f
e
n
s
e
m
b
l
e
class
i
f
ier
s
u
s
in
g
d
if
f
er
e
n
t
r
u
les
i
n
th
e
f
u
t
u
r
e
.
Mo
r
eo
v
er
,
s
ev
er
al
o
th
er
s
am
p
lin
g
m
et
h
o
d
s
s
u
c
h
as
o
v
er
s
a
m
p
lin
g
tec
h
n
iq
u
es
an
d
ev
o
lv
i
n
g
s
a
m
p
lin
g
s
tr
ate
g
ie
s
w
il
l
b
e
ap
p
lied
to
b
in
ar
y
clas
s
i
f
icatio
n
m
o
d
els.
ACK
NO
WL
E
D
G
E
M
E
NT
S
T
h
is
w
o
r
k
w
as
s
u
p
p
o
r
ted
b
y
Un
i
v
er
s
i
t
y
o
f
Da
n
a
n
g
,
U
n
iv
er
s
it
y
o
f
Scie
n
ce
an
d
T
e
ch
n
o
lo
g
y
,
co
d
e
n
u
m
b
er
o
f
P
r
o
j
ec
t:
T
2
0
1
8
-
02
-
50,
an
d
Min
is
tr
y
o
f
E
d
u
ca
tio
n
a
n
d
T
r
ain
in
g
Vie
tn
a
m
f
o
r
th
e
r
esear
ch
p
r
o
j
ec
t in
th
e
p
er
io
d
2
0
1
9
–
2
0
2
0
,
co
d
e
n
u
m
b
er
o
f
P
r
o
j
ec
t:
B
2
0
1
9
-
DNA
-
03
.
RE
F
E
R
E
NC
E
S
[1
]
T
.
T
.
Kh
u
a
t
a
n
d
M
.
H.
L
e
,
"
A
No
v
e
l
H
y
b
rid
A
BC
-
P
S
O
A
lg
o
rit
h
m
f
o
r
Eff
o
rt
Esti
m
a
ti
o
n
o
f
S
o
f
tw
a
re
P
r
o
jec
ts
Us
in
g
Ag
il
e
M
e
th
o
d
o
lo
g
ies
,
"
J
o
u
r
n
a
l
o
f
In
telli
g
e
n
t
S
y
ste
ms
,
v
o
l.
1
7
,
n
o
.
3
,
p
p
.
4
8
9
-
5
0
6
,
2
0
1
7
.
[2
]
I.
H.
L
a
ra
d
ji
,
M
.
A
lsh
a
y
e
b
,
a
n
d
L
.
G
h
o
u
ti
,
"
S
o
f
tw
a
re
d
e
f
e
c
t
p
re
d
ictio
n
u
sin
g
e
n
se
m
b
le
lea
rn
in
g
o
n
se
lec
ted
f
e
a
tu
re
s,"
In
fo
rm
a
ti
o
n
a
n
d
S
o
ft
w
a
re
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
5
8
,
p
p
.
3
8
8
-
4
0
2
,
2
0
1
5
.
[3
]
X
.
Y
u
a
n
,
T
.
M
.
K
h
o
sh
g
o
f
taa
r,
E.
B.
A
ll
e
n
,
a
n
d
K.
G
a
n
e
sa
n
,
"
A
n
a
p
p
li
c
a
ti
o
n
o
f
f
u
z
z
y
c
lu
ste
rin
g
to
so
f
twa
re
q
u
a
li
ty
p
re
d
ictio
n
,
"
in
Pro
c
e
e
d
in
g
s
o
f
th
e
3
rd
IEE
E
S
y
mp
o
si
u
m
o
n
Ap
p
li
c
a
ti
o
n
-
S
p
e
c
if
ic
S
y
ste
ms
a
n
d
S
o
ft
w
a
re
En
g
in
e
e
rin
g
T
e
c
h
n
o
l
o
g
y
,
p
p
.
8
5
-
9
0
,
2
0
0
0
.
[4
]
T
.
M
.
H.
Le,
T
.
B.
Ng
u
y
e
n
,
a
n
d
T
.
T
.
Kh
u
a
t,
"
S
u
rv
e
y
o
n
M
u
tatio
n
-
b
a
se
d
T
e
st
D
a
ta
G
e
n
e
r
a
ti
o
n
,
"
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
C
o
mp
u
t
e
r E
n
g
i
n
e
e
rin
g
(
IJ
ECE
),
v
o
l.
5
,
n
o
.
5
,
p
p
.
1
1
6
4
-
1
1
7
3
,
2
0
1
5
.
[5
]
M
.
D’A
m
b
ro
s,
M
.
L
a
n
z
a
,
a
n
d
R.
Ro
b
b
e
s,
"
Ev
a
lu
a
ti
n
g
d
e
f
e
c
t
p
re
d
ictio
n
a
p
p
r
o
a
c
h
e
s:
a
b
e
n
c
h
m
a
rk
a
n
d
a
n
e
x
ten
siv
e
c
o
m
p
a
riso
n
,
"
Emp
irica
l
S
o
f
twa
re
En
g
i
n
e
e
rin
g
,
jo
u
rn
a
l
a
rticle
,
v
o
l.
1
7
,
n
o
.
4
,
p
p
.
5
3
1
-
5
7
7
,
2
0
1
2
.
[6
]
Z.
S
u
n
,
Q.
S
o
n
g
,
a
n
d
X
.
Z
h
u
,
"
Us
in
g
Co
d
i
n
g
-
Ba
se
d
En
se
m
b
le
L
e
a
rn
in
g
to
Im
p
ro
v
e
S
o
f
t
w
a
re
D
e
fe
c
t
P
re
d
icti
o
n
,
"
IEE
E
T
ra
n
s
a
c
ti
o
n
s
o
n
S
y
ste
ms
,
M
a
n
,
a
n
d
Cy
b
e
rn
e
ti
c
s,
Pa
rt
C
(
Ap
p
li
c
a
ti
o
n
s
a
n
d
Rev
iews
)
,
v
o
l.
4
2
,
n
o
.
6
,
p
p
.
1
8
0
6
-
1
8
1
7
,
2
0
1
2
.
[7
]
R.
Ak
b
a
n
i,
S
.
Kw
e
k
,
a
n
d
N.
Ja
p
k
o
w
icz
,
"
A
p
p
ly
in
g
S
u
p
p
o
r
t
V
e
c
to
r
M
a
c
h
in
e
s
to
Im
b
a
lan
c
e
d
Da
tas
e
ts,
"
in
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
1
5
th
Eu
r
o
p
e
a
n
Co
n
fer
e
n
c
e
o
n
M
a
c
h
i
n
e
L
e
a
r
n
in
g
,
p
p
.
3
9
-
50
,
2
0
0
4
.
[8
]
H.
He
a
n
d
E.
A
.
G
a
rc
ia,
"
Le
a
rn
in
g
f
ro
m
I
m
b
a
lan
c
e
d
Da
ta,
"
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
K
n
o
wl
e
d
g
e
a
n
d
D
a
t
a
En
g
i
n
e
e
rin
g
,
v
o
l.
2
1
,
n
o
.
9
,
p
p
.
1
2
6
3
-
1
2
8
4
,
2
0
0
9
.
[9
]
N.
Ja
p
k
o
w
i
c
z
a
n
d
S
.
S
tep
h
e
n
,
"
Th
e
c
las
s
i
m
b
a
lan
c
e
p
ro
b
lem
:
A
s
y
ste
m
a
ti
c
stu
d
y
,
"
In
tell.
Da
ta
An
a
l
.
,
v
o
l.
6
,
n
o
.
5
,
p
p
.
4
2
9
-
4
4
9
,
2
0
0
2
.
[1
0
]
N.
Bo
u
g
u
il
a
,
W
.
Jia
n
Ha
n
,
a
n
d
A
.
B.
Ha
m
z
a
,
"
A
Ba
y
e
sia
n
a
p
p
r
o
a
c
h
f
o
r
so
f
tw
a
r
e
q
u
a
li
ty
p
re
d
icti
o
n
,
"
i
n
Pro
c
e
e
d
in
g
s
o
f
t
h
e
4
th
I
n
ter
n
a
ti
o
n
a
l
I
EE
E
C
o
n
fer
e
n
c
e
In
tell
ig
e
n
t
S
y
ste
ms
,
p
p
.
1
1
4
9
-
1
1
5
4
,
2
0
0
8
.
[1
1
]
Y.
S
u
n
,
M
.
S
.
Ka
m
e
l,
A
.
K.
C.
W
o
n
g
,
a
n
d
Y.
W
a
n
g
,
"
Co
st
-
se
n
siti
v
e
b
o
o
st
in
g
f
o
r
c
las
si
f
ica
ti
o
n
o
f
im
b
a
lan
c
e
d
d
a
ta,"
Pa
tt
e
rn
Rec
o
g
n
i
ti
o
n
,
v
o
l
.
4
0
,
n
o
.
1
2
,
p
p
.
3
3
5
8
-
3
3
7
8
,
2
0
0
7
.
[1
2
]
L
.
Ro
k
a
c
h
,
"
Tax
o
n
o
m
y
f
o
r
c
h
a
ra
c
teriz
in
g
e
n
se
m
b
le
m
e
th
o
d
s
in
c
las
sif
ic
a
ti
o
n
tas
k
s:
A
re
v
ie
w
a
n
d
a
n
n
o
tate
d
b
ib
li
o
g
ra
p
h
y
,
"
Co
mp
u
t
a
ti
o
n
a
l
S
t
a
ti
stics
&
Da
ta
An
a
lys
is,
v
o
l.
5
3
,
n
o
.
1
2
,
p
p
.
4
0
4
6
-
4
0
7
2
,
2
0
0
9
.
[1
3
]
T
.
Wan
g
,
W
.
L
i,
H.
S
h
i,
a
n
d
Z
.
L
iu
,
"
S
o
f
twa
re
De
fe
c
t
P
re
d
icti
o
n
Ba
se
d
o
n
Clas
sifi
e
rs
En
se
m
b
le,"
J
o
u
rn
a
l
o
f
In
fo
rm
a
t
io
n
a
n
d
Co
mp
u
t
a
ti
o
n
a
l
S
c
ien
c
e
,
v
o
l.
8
,
n
o
.
1
6
,
p
p
.
4
2
4
1
-
4
2
5
4
,
2
0
1
2
.
[1
4
]
T
.
M
e
n
z
ies
,
R.
Krish
n
a
,
a
n
d
D
.
P
ry
o
r.
"
T
h
e
P
ro
m
ise
Re
p
o
sito
ry
o
f
E
m
p
iri
c
a
l
S
o
f
t
w
a
r
e
En
g
in
e
e
rin
g
Da
ta
,"
[
On
li
n
e
]
.
A
v
a
il
a
b
le:
h
tt
p
:/
/
o
p
e
n
sc
ien
c
e
.
u
s/re
p
o
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
:
3
2
4
1
-
3246
3246
[1
5
]
T
.
M
e
n
z
ies
,
J.
Gre
e
n
w
a
ld
,
a
n
d
A.
F
ra
n
k
,
"
Da
t
a
M
in
in
g
S
tatic
Co
d
e
A
tt
rib
u
tes
to
L
e
a
rn
De
f
e
c
t
P
r
e
d
icto
rs,"
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
S
o
ft
w
a
re
En
g
in
e
e
rin
g
,
v
o
l
.
3
3
,
n
o
.
1
,
p
p
.
2
-
1
3
,
2
0
0
7
.
[1
6
]
T
.
J.
Os
tran
d
,
E.
J.
W
e
y
u
k
e
r,
a
n
d
R
.
M
.
Be
ll
,
"
P
re
d
icti
n
g
th
e
l
o
c
a
ti
o
n
a
n
d
n
u
m
b
e
r
o
f
f
a
u
lt
s
in
larg
e
so
f
t
w
a
r
e
s
y
ste
m
s,"
IEE
E
T
ra
n
s
a
c
ti
o
n
s o
n
S
o
ft
w
a
re
En
g
in
e
e
rin
g
,
v
o
l
.
3
1
,
n
o
.
4
,
p
p
.
3
4
0
-
3
5
5
,
2
0
0
5
.
[1
7
]
N.
E.
F
e
n
to
n
a
n
d
M
.
Ne
il
,
"
S
o
ftw
a
r
e
m
e
tri
c
s:
su
c
c
e
ss
e
s,
f
a
il
u
re
s
a
n
d
n
e
w
d
irec
ti
o
n
s,"
J
o
u
rn
a
l
o
f
S
y
ste
ms
a
n
d
S
o
ft
w
a
re
,
v
o
l.
4
7
,
n
o
.
2
,
p
p
.
1
4
9
-
1
5
7
,
1
9
9
9
.
[1
8
]
T
.
J.
M
c
Ca
b
e
,
"
A
Co
m
p
lex
it
y
M
e
a
su
re
,
"
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
S
o
ft
wa
re
En
g
in
e
e
rin
g
,
v
o
l
.
S
E
-
2
,
n
o
.
4
,
p
p
.
3
0
8
-
3
2
0
,
1
9
7
6
.
[1
9
]
D.
N.
Ca
rd
a
n
d
W
.
W
.
A
g
r
e
sti,
"
M
e
a
su
rin
g
so
f
t
w
a
re
d
e
si
g
n
c
o
m
p
lex
it
y
,
"
J
o
u
rn
a
l
o
f
S
y
ste
ms
a
n
d
S
o
ft
wa
re
,
v
o
l.
8
,
n
o
.
3
,
p
p
.
1
8
5
-
1
9
7
,
1
9
8
8
.
[2
0
]
S
.
R.
Ch
id
a
m
b
e
r
a
n
d
C.
F
.
Ke
m
e
re
r,
"
A
m
e
tri
c
s
su
it
e
f
o
r
o
b
jec
t
o
rien
ted
d
e
sig
n
,
"
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
S
o
ft
wa
re
En
g
i
n
e
e
rin
g
,
v
o
l.
2
0
,
n
o
.
6
,
p
p
.
4
7
6
-
4
9
3
,
1
9
9
4
.
[2
1
]
A
.
M
a
r
c
u
s,
D.
P
o
sh
y
v
a
n
y
k
,
a
n
d
R.
F
e
re
n
c
,
"
Us
in
g
th
e
Co
n
c
e
p
t
u
a
l
Co
h
e
sio
n
o
f
Clas
s
e
s
f
o
r
F
a
u
lt
P
re
d
ictio
n
in
Ob
jec
t
-
Orie
n
ted
S
y
ste
m
s,
"
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
S
o
ft
w
a
re
En
g
in
e
e
rin
g
,
v
o
l
.
3
4
,
n
o
.
2
,
p
p
.
2
8
7
-
3
0
0
,
2
0
0
8
.
[2
2
]
D.
Bo
w
e
s,
T
.
Ha
ll
,
a
n
d
D.
G
ra
y
,
"
DCo
n
f
u
sio
n
:
a
tec
h
n
iq
u
e
t
o
a
ll
o
w
c
ro
ss
stu
d
y
p
e
rf
o
rm
a
n
c
e
e
v
a
l
u
a
ti
o
n
o
f
f
a
u
lt
p
re
d
ictio
n
s
t
u
d
ies
,
"
Au
to
ma
ted
S
o
ft
wa
re
En
g
in
e
e
rin
g
,
j
o
u
r
n
a
l
a
rticl
e
.
v
o
l.
2
1
,
n
o
.
2
,
p
p
.
2
8
7
-
3
1
3
,
2
0
1
4
.
[2
3
]
G
.
M
.
W
e
iss,
"
M
in
in
g
w
it
h
ra
rit
y
:
a
u
n
ify
in
g
f
ra
m
e
w
o
r
k
,
"
ACM
S
IGKD
D
Exp
lo
r
a
ti
o
n
s
Ne
wsle
tt
e
r,
v
o
l.
6
,
n
o
.
1
,
p
p
.
7
-
1
9
,
2
0
0
4
.
[2
4
]
L
.
G
o
n
z
a
lez
-
A
b
ril
,
H.
Nu
ñ
e
z
,
C.
A
n
g
u
lo
a
n
d
F
.
V
e
las
c
o
,
"
G
S
V
M
:
A
n
S
V
M
f
o
r
h
a
n
d
li
n
g
im
b
a
lan
c
e
d
a
c
c
u
ra
c
y
b
e
t
w
e
e
n
c
las
s
e
s
in
b
i
-
c
las
sif
ic
a
ti
o
n
p
ro
b
lem
s,
"
Ap
p
li
e
d
S
o
ft
C
o
mp
u
ti
n
g
,
v
o
l.
1
7
,
n
o
.
S
u
p
p
lem
e
n
t
C,
p
p
.
2
3
-
3
1
,
2
0
1
4
.
[2
5
]
I.
M
a
n
i
a
n
d
J.
Zh
a
n
g
,
"
KN
N
Ap
p
r
o
a
c
h
to
Un
b
a
lan
c
e
d
Da
ta
Distrib
u
t
io
n
s:
A
Ca
se
S
tu
d
y
In
v
o
lv
in
g
In
f
o
rm
a
ti
o
n
Ex
trac
ti
o
n
,
"
in
Pro
c
e
e
d
in
g
s o
f
In
t
e
rn
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
M
a
c
h
in
e
L
e
a
rn
in
g
,
2
0
0
3
.
[2
6
]
J.
Ha
n
,
M
.
Ka
m
b
e
r,
a
n
d
J.
P
e
i,
D
a
ta
M
i
n
in
g
:
Co
n
c
e
p
ts
a
n
d
T
e
c
h
n
i
q
u
e
s
.
M
o
rg
a
n
Ka
u
f
m
a
n
n
,
2
0
1
2
.
[2
7
]
G
.
E.
Ba
ti
sta
,
R.
C.
P
ra
ti
,
a
n
d
M
.
C.
M
o
n
a
rd
,
"
A
stu
d
y
o
f
th
e
b
e
h
a
v
io
r
o
f
se
v
e
ra
l
m
e
th
o
d
s
f
o
r
b
a
lan
c
in
g
m
a
c
h
in
e
lea
rn
in
g
train
i
n
g
d
a
ta,"
S
IGKD
D
Exp
lo
r.
Ne
wsl.,
v
o
l.
6
,
n
o
.
1
,
p
p
.
2
0
-
2
9
,
2
0
0
4
.
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
T
h
a
n
h
T
u
n
g
K
h
u
a
t
c
o
m
p
lete
d
th
e
B.
S
d
e
g
re
e
in
S
o
f
t
w
a
r
e
En
g
in
e
e
rin
g
f
ro
m
Un
iv
e
rsit
y
o
f
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
D
a
n
a
n
g
,
V
iet
n
a
m
,
in
2
0
1
4
.
Cu
rre
n
tl
y
,
h
e
is
w
o
rk
in
g
to
w
a
rd
s
th
e
P
h
.
D.
d
e
g
re
e
a
t
th
e
A
d
v
a
n
c
e
d
A
n
a
l
y
ti
c
s
In
stit
u
te,
F
a
c
u
lt
y
o
f
En
g
in
e
e
rin
g
a
n
d
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
l
o
g
y
,
Un
iv
e
rsit
y
o
f
T
e
c
h
n
o
lo
g
y
S
y
d
n
e
y
,
A
u
stra
li
a
.
His
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
m
a
c
h
in
e
lea
rn
in
g
,
k
n
o
w
led
g
e
d
isc
o
v
e
r
y
,
e
v
o
lu
ti
o
n
a
ry
c
o
m
p
u
tatio
n
,
i
n
telli
g
e
n
t
o
p
ti
m
iza
ti
o
n
tec
h
n
iq
u
e
s
a
n
d
a
p
p
li
c
a
ti
o
n
s in
so
f
tw
a
r
e
e
n
g
in
e
e
rin
g
.
My
H
a
n
h
L
e
is
c
u
rre
n
tl
y
a
lec
tu
re
r
o
f
th
e
I
n
f
o
rm
a
ti
o
n
T
e
c
h
n
o
lo
g
y
F
a
c
u
lt
y
,
Un
iv
e
rsit
y
o
f
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
Da
n
a
n
g
,
V
iet
n
a
m
.
S
h
e
g
a
in
e
d
M
.
S
c
.
d
e
g
re
e
in
2
0
0
4
a
n
d
P
h
.
D.
d
e
g
re
e
in
Co
m
p
u
ter
S
c
ien
c
e
f
ro
m
T
h
e
Un
iv
e
rsit
y
o
f
Da
n
a
n
g
in
2
0
1
6
.
H
e
r
re
se
a
rc
h
in
tere
sts
a
re
a
b
o
u
t
so
f
t
w
a
re
tes
ti
n
g
a
n
d
m
o
re
g
e
n
e
ra
ll
y
a
p
p
li
c
a
ti
o
n
o
f
h
e
u
rist
ic
tec
h
n
i
q
u
e
s
t
o
p
ro
b
lem
s
in
so
f
tw
a
r
e
e
n
g
in
e
e
rin
g
.
Evaluation Warning : The document was created with Spire.PDF for Python.