I
n
t
e
r
n
at
ion
al
Jou
r
n
al
of
E
lec
t
r
ical
an
d
Com
p
u
t
e
r
E
n
gin
e
e
r
in
g
(
I
JE
CE
)
Vol.
15
,
No.
1
,
F
e
br
ua
r
y
20
25
,
pp.
604
~
613
I
S
S
N:
2088
-
8708
,
DO
I
:
10
.
11591/i
jec
e
.
v
15
i
1
.
pp
6
04
-
613
604
Jou
r
n
al
h
omepage
:
ht
tp:
//
ij
e
c
e
.
iaes
c
or
e
.
c
om
A
n
t
l
io
n
a
n
d
a
n
t
c
ol
on
y op
t
im
iz
at
io
n
i
n
t
e
g
r
at
e
d
e
n
se
m
b
le
m
ac
h
i
n
e
l
e
a
r
n
i
n
g m
o
d
e
l
f
o
r
e
f
f
e
c
t
iv
e
c
an
c
e
r
d
ia
g
n
osi
s
P
in
ak
s
h
i
P
an
d
a
1
,
S
u
k
a
n
t
Ki
s
h
or
o
B
is
oy
1
,
Am
r
u
t
an
s
h
u
P
a
n
igrahi
2
,
Abh
il
as
h
P
at
i
2
1
D
e
pa
r
tm
e
nt
of
C
omput
e
r
S
c
ie
nc
e
a
nd
E
ngi
ne
e
r
in
g, C
. V
.
R
a
m
a
n G
lo
ba
l
U
ni
ve
r
s
it
y, B
huba
ne
s
w
a
r
, O
di
s
ha
, I
ndi
a
2
D
e
pa
r
tm
e
nt
of
C
omput
e
r
S
c
ie
nc
e
a
nd E
ngi
ne
e
r
in
g,
I
ns
ti
tu
te
of
T
e
c
hni
c
a
l
E
duc
a
ti
on a
nd
R
e
s
e
a
r
c
h (
I
T
E
R
)
,
S
ik
s
ha
’
O
’
A
nus
a
ndha
n (
D
e
e
me
d t
o be
U
ni
v
e
r
s
it
y)
, O
di
s
ha
, I
n
di
a
Ar
t
icle
I
n
f
o
AB
S
T
RA
CT
A
r
ti
c
le
h
is
tor
y
:
R
e
c
e
ived
J
un
18,
2024
R
e
vis
e
d
Aug
12,
2024
Ac
c
e
pted
Aug
20,
2024
St
at
i
s
t
i
cs
fr
o
m
rep
u
t
ab
l
e
s
o
u
rces
,
i
n
cl
u
d
i
n
g
t
h
e
W
o
rl
d
H
eal
t
h
O
r
g
an
i
zat
i
o
n
(W
H
O
),
d
emo
n
s
t
rat
e
t
h
a
t
can
cer
i
s
a
l
ead
i
n
g
cau
s
e
o
f
d
ea
t
h
g
l
o
b
a
l
l
y
,
acco
u
n
t
i
n
g
fo
r
mi
l
l
i
o
n
s
o
f
d
ea
t
h
s
each
y
ear.
W
h
e
n
i
t
co
mes
t
o
t
h
e
earl
y
i
d
e
n
t
i
fi
ca
t
i
o
n
o
f
ca
n
cer,
mach
i
n
e
l
e
arn
i
n
g
(ML
)
i
s
cru
ci
a
l
.
T
o
a
n
al
y
ze
co
mp
l
ex
d
at
a
an
d
i
d
e
n
t
i
fy
mi
n
u
t
e
p
at
t
er
n
s
t
h
at
may
i
n
d
i
cat
e
t
h
e
p
re
s
en
ce
o
f
can
cer,
i
t
emp
l
o
y
s
r
o
b
u
s
t
c
o
mp
u
t
a
t
i
o
n
a
l
ap
p
ro
a
ch
e
s
.
Imp
ro
v
i
n
g
p
at
i
en
t
o
u
t
co
me
s
rel
i
es
o
n
earl
y
can
cer
d
et
ec
t
i
o
n
s
i
n
ce
i
t
p
av
e
s
t
h
e
w
ay
fo
r
fas
t
er
t
reat
me
n
t
an
d
i
n
t
er
v
en
t
i
o
n
,
w
h
i
c
h
mi
g
h
t
l
ead
t
o
b
et
t
er
p
ro
g
n
o
s
es
an
d
h
i
g
h
er
s
u
r
v
i
v
al
rat
es
.
T
o
ch
o
o
s
e
feat
u
re
s
,
t
h
i
s
s
t
u
d
y
i
n
t
e
n
d
s
t
o
b
u
i
l
d
an
ML
-
b
as
ed
en
s
em
b
l
e
mo
d
el
u
t
i
l
i
z
i
n
g
an
t
c
o
l
o
n
y
o
p
t
i
m
i
zat
i
o
n
(A
CO
)
an
d
an
t
l
i
o
n
o
p
t
i
mi
zat
i
o
n
(A
L
O
).
N
e
x
t
,
ML
cl
a
s
s
i
fi
er
s
are
u
s
e
d
as
t
h
e
i
n
i
t
i
a
l
p
re
d
i
c
t
i
o
n
s
'
b
as
i
s
l
earn
er
s
.
T
h
e
l
a
s
t
f
o
recas
t
i
s
t
h
e
re
s
u
l
t
o
f
co
m
b
i
n
i
n
g
t
w
o
e
n
s
em
b
l
e
met
h
o
d
s
:
v
o
t
i
n
g
a
n
d
a
v
erag
i
n
g
cl
a
s
s
i
fi
er
s
.
Fo
u
r
d
i
s
t
i
n
ct
can
cer
m
i
cro
arra
y
d
at
a
s
et
s
are
u
s
ed
t
o
as
s
e
s
s
t
h
e
ap
p
ro
ac
h
.
W
i
t
h
an
accu
ra
cy
o
f
9
9
.
0
8
%
o
n
t
h
e
L
u
n
g
ca
n
cer
d
a
t
as
e
t
,
t
h
e
v
o
t
i
n
g
e
n
s
em
b
l
e
c
l
as
s
i
f
i
er
o
u
t
p
erfo
rm
s
t
h
e
o
t
h
er
s
,
acco
rd
i
n
g
t
o
t
h
e
emp
i
ri
ca
l
an
a
l
y
s
i
s
.
K
e
y
w
o
r
d
s
:
Ant
c
olony
opti
mi
z
a
ti
on
Ant
li
on
opti
mi
z
a
ti
on
Ave
r
a
ging
c
las
s
if
ier
C
a
nc
e
r
M
a
c
hine
lea
r
ning
Voting
c
las
s
if
ier
Th
i
s
i
s
a
n
o
p
en
a
c
ces
s
a
r
t
i
c
l
e
u
n
d
e
r
t
h
e
CC
B
Y
-
SA
l
i
ce
n
s
e.
C
or
r
e
s
pon
din
g
A
u
th
or
:
Amr
utans
hu
P
a
nigr
a
hi
De
pa
r
tm
e
nt
of
C
omput
e
r
S
c
ienc
e
a
nd
E
nginee
r
ing
,
I
ns
ti
tut
e
of
T
e
c
hnica
l
E
duc
a
ti
on
a
nd
R
e
s
e
a
r
c
h
(
I
T
E
R
)
,
S
iks
ha
‘
O’
Anus
a
ndha
n
(
De
e
med
to
be
Unive
r
s
it
y
)
B
huba
ne
s
wa
r
,
Odis
ha
,
I
ndia
E
mail:
a
mr
u
tans
up89@gmail.
c
om
1.
I
NT
RODU
C
T
I
ON
C
a
nc
e
r
s
igni
f
ica
ntl
y
c
ontr
ibut
e
s
to
the
incr
e
a
s
ing
wor
ldwide
mor
tal
it
y
r
a
te.
I
t
is
a
powe
r
f
ul
c
ompetit
or
,
c
a
us
ing
the
de
a
th
of
a
lm
os
t
10
mi
ll
ion
pe
ople
e
a
c
h
ye
a
r
,
a
s
r
e
por
ted
by
the
W
or
ld
He
a
lt
h
Or
ga
niza
ti
on
(
W
HO
)
[
1]
.
T
he
e
f
f
e
c
t
o
f
c
a
nc
e
r
is
magnif
ied
by
it
s
many
manif
e
s
tations
,
whic
h
incl
ude
lung,
br
e
a
s
t,
c
olor
e
c
tal,
pr
os
tate
,
a
nd
s
tom
a
c
h
c
a
nc
e
r
s
.
Va
r
ious
f
a
c
tor
s
c
ontr
ibut
e
to
the
high
oc
c
ur
r
e
nc
e
of
the
dis
e
a
s
e
,
including
ge
ne
ti
c
pr
e
dis
pos
it
ions
,
li
f
e
s
tyl
e
de
c
is
ions
s
uc
h
a
s
tobac
c
o
int
a
ke
a
nd
s
e
de
ntar
y
be
ha
vior
s
,
e
nvir
onmenta
l
c
ontaminants
li
ke
r
a
diation,
a
nd
inf
e
c
ti
ous
a
ge
nts
li
ke
h
e
pa
ti
ti
s
a
nd
human
pa
pil
lom
a
vir
us
[
2]
.
T
he
dif
f
iculty
incr
e
a
s
e
s
s
ince
e
a
r
ly
diagnos
is
is
c
ha
ll
e
nging,
a
s
many
malignanc
ies
do
no
t
e
xhibi
t
r
e
c
ogniza
ble
s
ympt
oms
unti
l
they
ha
ve
r
e
a
c
he
d
th
e
late
s
tage
s
,
whic
h
incr
e
a
s
e
s
the
ove
r
a
ll
number
of
c
a
nc
e
r
-
r
e
late
d
f
a
talit
ies
wor
ldwide
[
3]
.
M
a
c
hine
lea
r
ning
(
M
L
)
is
e
s
s
e
nti
a
l
in
tr
a
ns
f
or
mi
ng
c
a
nc
e
r
de
tec
ti
on
a
nd
ther
a
py.
I
ntegr
a
ti
ng
thi
s
tec
hnology
int
o
he
a
lt
hc
a
r
e
s
ys
tems
ha
s
f
a
c
il
it
a
ted
t
he
pr
e
c
is
e
a
nd
e
f
f
icie
nt
identif
ica
ti
on
of
malignant
ti
s
s
ue
s
,
a
s
s
is
ti
ng
in
e
a
r
ly
c
a
nc
e
r
diagnos
is
a
nd
the
de
ve
lopm
e
nt
of
pe
r
s
ona
li
z
e
d
tr
e
a
tm
e
nt
plans
[
4]
.
M
a
c
hine
lea
r
ning
models
may
us
e
e
xtens
ive
da
ta
s
e
ts
a
nd
a
dva
nc
e
d
a
lgor
it
hms
to
a
na
lyze
int
r
ica
te
pa
tt
e
r
ns
a
nd
biom
a
r
ke
r
s
that
may
be
ove
r
looked
by
human
obs
e
r
ve
r
s
.
Ana
lyzin
g
e
xtens
ive
ge
ne
ti
c
a
nd
c
li
nica
l
da
ta
may
a
id
in
id
e
nti
f
ying
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
A
nt
li
on
and
ant
c
olony
opti
miz
ati
on
int
e
gr
ated
e
ns
e
mble
mac
hine
…
(
P
inaks
hi
P
anda
)
605
ne
w
biom
a
r
ke
r
s
,
f
o
r
e
c
a
s
ti
ng
pa
ti
e
nt
pr
ognos
is
,
a
n
d
e
xpe
dit
ing
medic
a
ti
on
de
ve
lopm
e
nt.
M
icr
oa
r
r
a
y
da
ta
is
c
r
uc
ial
in
c
a
nc
e
r
de
tec
ti
on
s
ince
it
a
ll
ows
f
or
tho
r
oughly
e
xa
mi
ning
ge
ne
e
xpr
e
s
s
ion
pa
tt
e
r
ns
in
m
a
li
gna
nt
c
e
ll
s
.
T
his
method
e
na
bles
r
e
s
e
a
r
c
he
r
s
a
nd
phys
i
c
ians
to
de
tec
t
dis
ti
nc
t
ge
ne
s
ignatur
e
s
li
nke
d
to
va
r
ious
f
or
ms
of
c
a
nc
e
r
,
f
a
c
il
it
a
ti
ng
p
r
e
c
is
e
c
a
tegor
iza
ti
on,
pr
ognos
is
pr
e
diction,
a
nd
ther
a
py
s
e
lec
ti
on
[
5]
.
M
icr
oa
r
r
a
y
da
tas
e
ts
of
ten
ha
ve
s
e
ve
r
a
l
ge
ne
s
or
c
ha
r
a
c
ter
is
ti
c
s
,
a
s
igni
f
ica
nt
po
r
ti
on
o
f
whic
h
may
be
e
xtr
a
ne
ous
or
dupli
c
a
ti
ve
f
o
r
pr
e
dictive
modeli
ng.
F
e
a
tur
e
s
e
lec
ti
on
tec
hniques
a
id
in
identi
f
ying
the
mos
t
us
e
f
ul
a
nd
dis
c
r
im
inative
s
ubs
e
t
of
f
e
a
tur
e
s
,
he
nc
e
lowe
r
ing
the
dim
e
ns
ionalit
y
o
f
the
da
ta
while
pr
e
s
e
r
ving
s
igni
f
ica
nt
inf
or
mation
[
6
]
.
T
his
tec
hnique
not
only
boos
ts
the
e
f
f
icie
nc
y
a
nd
pe
r
f
o
r
manc
e
of
mac
hine
lea
r
ning
models
but
a
ls
o
im
pr
ove
s
thei
r
int
e
r
p
r
e
tabili
ty
a
nd
a
bil
it
y
to
ge
ne
r
a
li
z
e
to
ne
w
da
ta.
M
e
ta
he
ur
is
ti
c
a
lgor
it
hms
a
r
e
e
f
f
icie
nt
a
nd
e
f
f
e
c
ti
ve
tec
hniques
f
or
s
e
lec
ti
ng
f
e
a
tur
e
s
while
wor
king
with
m
icr
oa
r
r
a
y
da
ta
in
c
a
nc
e
r
r
e
s
e
a
r
c
h.
Due
to
the
c
ompl
e
x
a
nd
mul
ti
-
dim
e
ns
ional
na
tur
e
of
mi
c
r
oa
r
r
a
y
da
tas
e
ts
,
meta
he
ur
is
ti
c
a
lgor
it
hms
s
uc
h
a
s
ge
ne
ti
c
a
lgor
it
hms
,
pa
r
ti
c
le
s
wa
r
m
opti
mi
z
a
ti
on,
a
nd
s
im
ulate
d
a
nne
a
li
ng
p
r
ovide
a
s
ys
tema
ti
c
method
to
identif
y
im
por
tant
ge
ne
s
or
f
e
a
tur
e
s
that
play
a
c
r
it
ica
l
r
ole
in
c
a
nc
e
r
d
iagnos
is
,
pr
ognos
is
,
a
nd
pr
e
diction
of
tr
e
a
tm
e
nt
r
e
s
pons
e
[
7]
.
Az
iz
[
8]
pr
opos
e
d
a
hyb
r
id
model
with
ind
e
pe
nde
nt
c
omponent
a
na
lys
is
(
I
C
A)
with
two
meta
he
ur
is
ti
c
a
ppr
oa
c
he
s
,
C
uc
koo
s
e
a
r
c
h
(
C
S
)
,
a
r
ti
f
icia
l
be
e
c
olony
(
AB
C
)
,
a
nd
ge
ne
ti
c
a
lgor
it
hm
(
GA
)
,
to
pr
opos
e
two
hybr
id
models
with
na
ïve
B
a
ye
s
(
NB
)
c
las
s
if
ier
.
Ne
kouie
e
t
al.
[
9
]
pr
opos
e
d
a
n
e
ns
e
mbl
e
-
ba
s
e
d
model
f
o
r
c
a
nc
e
r
diagnos
is
.
F
o
r
thi
s
,
the
a
uth
or
us
e
d
a
two
-
s
tage
f
e
a
tur
e
s
e
lec
ti
on
p
r
oc
e
s
s
.
I
nit
ially,
mul
ti
modal
opti
mi
z
a
ti
on
a
nd
the
F
i
r
e
f
ly
a
lgo
r
it
h
m
a
r
e
a
ppli
e
d
to
the
mi
c
r
oa
r
r
a
y
da
ta
in
the
f
i
r
s
t
s
tage
.
T
he
ne
xt
s
tage
of
f
e
a
tur
e
s
e
lec
ti
on
is
pa
r
ti
c
le
s
wa
r
m
o
pti
mi
z
a
ti
on
(
P
S
O)
.
M
a
c
hine
lea
r
ning
c
las
s
if
ier
s
a
r
e
a
ppli
e
d
to
the
s
e
lec
ted
f
e
a
tur
e
s
,
a
nd
then
to
the
ini
ti
a
l
pr
e
diction,
the
s
of
t
voti
ng
e
ns
e
mbl
e
c
las
s
if
ier
is
a
ppli
e
d.
Na
ji
e
t
al
.
[
10]
p
r
opos
e
d
a
M
L
model
f
o
r
c
a
nc
e
r
diagnos
is
.
T
he
model
us
e
s
f
ive
dif
f
e
r
e
nt
M
L
c
las
s
if
ier
s
to
c
las
s
if
y
the
c
a
nc
e
r
dis
e
a
s
e
.
T
o
e
va
luate
the
model's
a
c
c
ur
a
c
y,
the
F
-
1
s
c
or
e
,
s
pe
c
if
icity,
a
nd
s
e
ns
it
ivi
ty
ha
ve
be
e
n
c
a
lcula
ted.
L
u
e
t
al.
[
11]
pr
opos
e
d
a
n
e
ns
e
m
ble
model
f
o
r
e
f
f
e
c
ti
ve
c
a
nc
e
r
diagnos
is
.
F
inally
,
dif
f
e
r
e
nt
M
L
c
las
s
if
ier
s
a
r
e
a
ppli
e
d
to
pr
e
dict
c
a
nc
e
r
dis
e
a
s
e
.
T
he
n,
the
voti
ng
method
is
a
ppli
e
d
a
s
the
e
ns
e
mbl
e
tec
hnique
to
e
nha
nc
e
the
pr
e
diction
r
e
s
ult
.
T
a
va
s
oli
e
t
al.
[
12]
ha
ve
pr
opos
e
d
a
n
e
ns
e
mbl
e
mac
hine
-
lea
r
ning
model
f
or
e
f
f
e
c
ti
ve
c
a
nc
e
r
pr
e
diction
with
wa
ter
c
yc
le
a
lgor
it
hm
a
nd
s
uppor
t
ve
c
tor
mac
hine
(
S
VM
)
f
or
f
e
a
tur
e
s
e
lec
ti
on
a
nd
c
las
s
if
ica
ti
on
pur
pos
e
s
.
S
un
e
t
al
.
[
13]
ha
ve
int
e
gr
a
ted
the
f
uz
z
y
r
oughe
s
t,
e
ntr
opy
-
ba
s
e
d
f
e
a
tur
e
s
e
lec
ti
on
with
the
F
is
he
r
s
c
or
e
.
T
he
de
ve
loped
model
us
e
s
the
f
is
he
r
s
c
or
e
to
r
e
duc
e
th
e
number
of
ge
ne
s
pr
e
s
e
nt
in
the
da
tas
e
t.
S
hukla
e
t
al.
[
14
]
h
a
ve
pr
opos
e
d
a
M
L
-
ba
s
e
d
hybr
id
model
with
two
dif
f
e
r
e
nt
f
e
a
tur
e
s
e
lec
ti
on
s
tage
s
.
I
n
the
ini
ti
a
l
pha
s
e
of
the
f
e
a
tur
e
s
e
lec
ti
on,
the
mi
nim
um
r
e
dunda
nc
y
a
nd
m
a
xim
um
r
e
leva
nc
e
a
r
e
a
ppli
e
d
to
s
e
lec
t
the
r
e
leva
nt
ge
ne
s
f
r
om
the
da
tas
e
t.
T
he
n
,
the
ne
xt
f
e
a
tur
e
s
e
lec
ti
on
pha
s
e
int
e
gr
a
tes
two
meta
he
ur
is
ti
c
a
ppr
oa
c
he
s
,
includi
ng
the
tea
c
hing
lea
r
n
ing
a
lgor
it
hm
a
nd
the
g
r
a
vit
a
ti
ona
l
s
e
a
r
c
h
a
lgor
it
hm.
T
he
na
ïve
B
a
ye
s
c
las
s
if
ier
c
a
lcula
tes
the
f
it
ne
s
s
f
unc
ti
on
a
nd
c
las
s
if
ies
the
c
a
nc
e
r
.
M
e
e
na
c
hi
e
t
al.
[
15]
pr
opos
e
d
a
nother
hybr
id
mod
e
l
ba
s
e
d
on
the
a
nt
c
olony
opti
mi
z
a
ti
on
a
lgor
it
hm
(
AC
O)
,
ge
ne
ti
c
a
lgor
it
hm
(
GA
)
,
a
nd
tabu
s
e
a
r
c
h
a
lgor
it
hm
(
T
S
A)
f
oll
owe
d
by
a
f
uz
z
y
r
ough
s
e
t
c
las
s
if
ier
to
c
las
s
if
y
c
a
nc
e
r
.
T
a
ble
1
s
hows
the
s
umm
a
r
y
of
the
r
e
por
ted
li
ter
a
tur
e
s
.
T
a
ble
1.
L
it
e
r
a
tur
e
s
ur
ve
y
s
umm
a
r
y
R
e
f
e
r
e
nc
e
F
e
a
tu
r
e
s
e
le
c
ti
on a
lg
or
it
hm
C
la
s
s
if
ie
r
E
ns
e
mbl
e
c
la
s
s
if
ie
r
C
a
nc
e
r
da
ta
s
e
t
A
z
iz
[
8]
I
C
A
, C
S
, A
B
C
, G
A
NB
--
C
ol
on, L
ung I
I
, P
r
os
ta
te
, A
c
ut
e
le
uke
mi
a
a
nd L
e
uke
mi
a
N
e
koui
e
e
t
al
.
[
9]
M
ul
ti
moda
l,
F
ir
e
f
ly
a
lg
or
it
hm
(
F
A
)
,
PSO
S
V
M
,
K
-
ne
a
r
e
s
t
ne
ig
hbor
s
(
K
N
N
)
,
e
xt
r
e
me
l
e
a
r
ni
ng
ma
c
hi
ne
(
E
L
M
)
S
of
t
vot
in
g
B
r
a
in
, C
ol
on, L
e
uke
mi
a
, L
ung,
P
r
os
ta
te
, B
r
e
a
s
t,
S
R
B
C
T
,
O
va
r
ia
n
N
a
ji
e
t
al
.
[
10]
--
S
V
M
,
r
a
ndom f
or
e
s
t
(
R
F
)
,
de
c
is
io
n t
r
e
e
(
D
T
)
, K
N
N
--
B
r
e
a
s
t
L
u
e
t
al
.
[
11]
--
L
in
e
a
r
r
e
gr
e
s
s
io
n
(
LR
)
, K
N
N
,
S
V
M
,
mul
ti
la
ye
r
pe
r
c
e
pt
r
on
(
M
L
P
)
, D
T
V
ot
in
g
C
e
r
vi
c
a
l
T
a
va
s
ol
i
e
t
al
.
[
12]
W
a
te
r
c
yc
le
a
lg
or
it
hm (
W
C
A
)
S
V
M
S
of
t
vot
in
g
L
e
uke
mi
a
, C
ol
on, P
r
os
ta
te
,
D
L
B
C
L
S
un
e
t
al
.
[
13]
F
is
he
r
s
c
or
e
F
uz
z
y
r
ough s
e
t
--
S
R
B
C
T
,
C
ol
on, B
r
a
in
,
L
ymphom
a
, L
e
uke
mi
a
S
hukl
a
e
t
al
.
[
14]
T
e
a
c
hi
ng l
e
a
r
ni
ng
-
ba
s
e
d a
lg
or
it
hm
(
T
L
B
O
)
,
g
r
a
vi
ta
ti
on
a
l
s
e
a
r
c
h
a
l
gor
it
hm
(
G
S
A
)
,
m
in
im
u
m r
e
du
nd
a
n
c
y
m
a
xi
m
um
r
e
l
e
v
a
nc
e
(
m
R
M
R
)
NB
--
L
e
uke
mi
a
1
a
nd
2, D
L
B
C
L
,
P
r
os
ta
te
, C
ol
on
M
e
e
na
c
hi
e
t
al
.
[
15]
A
C
O
, G
A
, T
S
F
uz
z
y
r
ough s
e
t
--
D
L
B
C
L
, S
R
B
C
T
, B
r
e
a
s
t,
L
e
uke
mi
a
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
15
,
No.
1
,
F
e
br
ua
r
y
20
25
:
604
-
613
606
T
he
c
ur
r
e
nt
wor
k
a
im
s
to
pr
opos
e
a
n
e
ns
e
mbl
e
M
L
model
f
or
c
a
nc
e
r
diagnos
is
.
T
he
p
r
opos
e
d
wor
k
a
dopts
two
meta
he
ur
is
ti
c
a
ppr
oa
c
he
s
,
a
nt
li
on
op
ti
mi
z
e
r
(
AL
O)
a
nd
a
nt
c
olony
opti
mi
z
e
r
(
AC
O)
,
to
s
e
lec
t
r
e
leva
nt
f
e
a
tur
e
s
f
r
om
the
mi
c
r
oa
r
r
a
y
da
ta.
T
he
n,
thr
e
e
M
L
-
ba
s
e
d
c
las
s
if
ier
s
a
r
e
a
ppli
e
d
a
long
with
the
voti
ng
a
nd
a
ve
r
a
ging
e
ns
e
mbl
e
tec
hnique
f
or
c
las
s
if
ica
ti
on
pur
pos
e
s
.
T
he
objec
ti
ve
s
of
the
c
ur
r
e
nt
wor
k
c
a
n
be
s
umm
a
r
ize
d
a
s
f
oll
ows
:
i)
to
im
pleme
nt
AL
O
a
nd
AC
O
a
s
the
meta
he
ur
is
ti
c
f
e
a
tur
e
s
e
lec
ti
on
method
;
ii
)
to
im
p
leme
nt
M
L
-
ba
s
e
d
c
las
s
if
ier
s
s
uc
h
a
s
S
VM
,
Ada
B
oos
t,
XG
B
oos
t,
a
nd
R
F
c
las
s
if
ier
s
f
or
ini
ti
a
l
pr
e
diction
;
ii
i
)
to
a
pply
the
voti
ng
a
nd
a
ve
r
a
gi
ng
e
ns
e
mbl
e
tec
hnique
to
f
o
r
m
a
n
e
ns
e
mbl
e
M
L
-
ba
s
e
d
diagnos
is
model
;
iv)
to
e
va
luate
the
pr
opos
e
d
mod
e
l
ove
r
th
r
e
e
kinds
o
f
c
a
nc
e
r
mi
c
r
oa
r
r
a
y
da
tas
e
t
;
a
nd
v)
to
mea
s
ur
e
the
model's
pe
r
f
or
manc
e
us
ing
di
f
f
e
r
e
nt
M
L
-
ba
s
e
d
e
va
luative
pa
r
a
mete
r
s
,
including
a
c
c
ur
a
c
y,
pr
e
c
is
ion,
r
e
c
a
ll
,
s
pe
c
if
icity,
a
nd
F
1
-
s
c
or
e
.
2.
M
AT
E
R
I
AL
S
AN
D
M
E
T
HO
DS
T
his
s
e
c
ti
on
de
s
c
r
ibes
the
da
tas
e
t
a
nd
the
methods
us
e
d
to
de
ve
lop
the
pr
opos
e
d
model.
T
o
de
ve
lop
the
model,
a
nt
li
on
opti
mi
z
a
ti
on
a
nd
a
nt
c
olony
o
pti
mi
z
a
ti
on
a
r
e
a
ppli
e
d
s
e
que
nti
a
ll
y
a
s
the
f
e
a
tur
e
s
e
lec
ti
on
a
lgor
it
hm.
T
he
n,
to
the
s
e
lec
ted
f
e
a
tur
e
s
,
thr
e
e
m
a
c
hine
lea
r
ning
-
ba
s
e
d
c
las
s
if
ier
s
,
including
s
uppor
t
ve
c
tor
mac
hine
,
Ada
B
oos
t,
a
nd
XG
B
oos
t,
a
r
e
a
ppli
e
d
a
s
the
ba
s
e
c
la
s
s
if
ier
s
to
pe
r
f
or
m
ini
ti
a
l
pr
e
dicti
on.
T
he
two
e
ns
e
mbl
e
s
tr
a
tegie
s
(
voti
ng
a
nd
a
ve
r
a
ging)
a
r
e
a
ppli
e
d
to
the
ini
ti
a
l
pr
e
diction
to
bu
il
d
the
pr
opos
e
d
model.
2.
1.
Dat
as
e
t
d
e
s
c
r
ip
t
ion
T
he
pr
opos
e
d
a
ppr
oa
c
h
wa
s
e
va
luat
e
d
on
f
our
c
a
nc
e
r
da
tas
e
t
s
[
16]
.
T
a
ble
2
s
hows
the
da
tas
e
t
de
s
c
r
ipt
ion
.
All
thes
e
f
our
da
tas
e
ts
c
ontain
mi
c
r
o
a
r
r
a
y
da
ta
with
binar
y
c
las
s
va
lues
.
A
b
r
ief
de
s
c
r
ipt
ion
o
f
thes
e
da
tas
e
t
s
is
de
picte
d
in
T
a
ble
2.
T
a
ble
2.
Da
tas
e
t
de
s
c
r
ipt
ion
D
a
ta
s
e
t
N
umbe
r
of
F
e
a
tu
r
e
s
N
umbe
r
of
S
a
mpl
e
s
C
la
s
s
O
va
r
ia
n
15,154
253
2
L
ung
12,533
181
2
B
r
e
a
s
t
24,481
97
2
L
e
uke
mi
a
7,129
72
2
2.
2.
Ant
li
on
op
t
i
m
izat
ion
algorit
h
m
T
he
a
nt
l
ion
opti
m
iza
ti
on
(
AL
O)
method
is
de
r
iv
e
d
f
r
om
the
p
r
e
da
tor
y
s
tr
a
tegy
us
e
d
by
a
ntl
ions
,
whe
r
e
they
c
a
ptur
e
a
nts
by
c
ons
tr
uc
ti
ng
c
onica
l
t
r
e
nc
he
s
.
W
it
hin
the
c
ontext
of
AL
O,
the
a
c
ti
ons
of
a
ntl
ions
a
nd
a
nts
a
r
e
s
im
pli
f
ied
a
nd
r
e
pr
e
s
e
nted
a
s
a
n
op
t
im
iza
ti
on
a
lgor
i
thm
[
17]
,
[
18]
.
I
n
thi
s
a
lgor
it
hm
,
a
ntl
ions
s
ymbol
ize
potential
s
olut
ions
,
while
a
nts
s
ymbol
ize
the
pr
oc
e
s
s
of
s
e
a
r
c
hing
a
nd
inves
ti
ga
ti
ng
the
s
pa
c
e
of
pos
s
ibl
e
s
olut
ions
.
r
e
pr
e
s
e
nts
the
a
nt,
a
nd
r
e
pr
e
s
e
nts
the
a
nt
li
on.
T
he
r
a
ndom
wa
lk
of
with
r
a
ndo
m
s
tep
s
ize
t
to
s
e
a
r
c
h
the
f
ood
c
a
n
be
modele
d
a
s
(
1)
:
(
)
=
[
0
,
(
2
Ρ
(
1
)
−
1
,
…
.
(
)
−
1
,
)
]
(
1)
whe
r
e
(
1
)
is
the
s
tocha
s
ti
c
f
unc
ti
on,
is
the
c
umul
a
ti
ve
s
um,
a
nd
n
is
the
maximum
number
of
it
e
r
a
ti
ons
.
(
)
c
a
n
be
de
f
ined
us
ing
(
2)
,
with
r
a
s
a
r
a
ndom
va
r
ia
ble
be
twe
e
n
[
0,
1]
.
Ρ
(
)
=
{
1
>
0
.
5
,
0
≤
0
.
5
(
2)
Us
ing
the
r
a
ndom
wa
lk,
the
ne
xt
pos
it
ion
of
the
a
nt
c
a
n
be
upda
ted
us
ing
(
3)
.
How
e
ve
r
,
in
(
1
)
c
a
n
not
us
e
d
dir
e
c
tl
y
a
s
e
a
c
h
s
e
a
r
c
h
s
pa
c
e
of
a
n
a
nt
ha
s
a
boun
da
r
y.
S
o,
to
upda
te,
the
pos
it
ion
ne
e
ds
to
be
no
r
m
a
li
z
e
d
by
us
ing
(
3)
,
whic
h
is
a
ls
o
us
e
d
to
de
f
ine
the
ne
xt
pos
it
ion
of
.
(
+
1
)
=
(
)
+
(
)
2
(
3)
whe
r
e
(
)
is
the
c
ur
r
e
nt
pos
it
ion
of
the
a
nd
(
)
is
the
be
s
t
c
a
ndidate
s
olut
ion
f
ound
s
o
f
a
r
.
I
n
the
f
inal
s
tage
of
hunti
ng,
whe
n
A
r
e
a
c
he
s
the
pit
of
t
he
tr
a
p
of
,
dr
a
gs
to
the
s
a
nd
a
nd
then
c
ons
umes
the
pr
e
y.
T
he
n,
the
pos
it
ion
is
upda
ted
us
ing
(
3)
[
19]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
A
nt
li
on
and
ant
c
olony
opti
miz
ati
on
int
e
gr
ated
e
ns
e
mble
mac
hine
…
(
P
inaks
hi
P
anda
)
607
2.
3.
Ant
c
olon
y
op
t
i
m
izat
ion
algorit
h
m
Ant
c
olony
opti
mi
z
a
ti
on
(
AC
O
)
is
a
he
ur
is
ti
c
o
pti
mi
z
a
ti
on
tec
hnique
that
e
mul
a
tes
the
f
or
a
ging
be
ha
vior
of
a
nts
to
s
olve
opti
mi
z
a
ti
on
is
s
ue
s
.
T
he
method
wor
ks
on
a
gr
a
ph
r
e
pr
e
s
e
nti
ng
the
is
s
ue
,
with
node
s
r
e
pr
e
s
e
nti
ng
potential
s
olut
ions
a
nd
e
dge
s
indi
c
a
ti
ng
c
onne
c
ti
ons
be
twe
e
n
s
olut
ions
.
T
he
AC
O
a
lgor
it
hm
e
mpl
oys
a
r
ti
f
icia
l
a
nts
to
c
r
e
a
te
s
olut
ions
it
e
r
a
ti
ve
ly
[
20
]
.
T
he
pr
oba
bil
it
y
of
moveme
nt
of
a
n
a
nt
(
)
f
r
om
loca
ti
on
to
loca
ti
on
c
a
n
be
mathe
matica
ll
y
modele
d
a
s
(
4)
:
(
)
=
{
(
Ɣ
)
⋅
(
∨
)
∑
(
Ɣ
)
⋅
(
∨
)
∈
<
0
ℎ
(
4)
w
he
r
e
the
Ɣ
is
the
phe
r
o
mone
leve
l
on
the
pa
th
−
a
t
ti
me
t.
∨
is
the
inve
r
s
e
of
the
dis
tanc
e
be
twe
e
n
two
loca
ti
ons
a
nd
,
α
,
a
nd
β
a
r
e
the
c
ons
tant,
is
the
tot
a
l
number
of
a
ll
owe
d
loca
ti
ons
a
n
a
nt
c
a
n
move
,
a
nd
l
is
a
n
int
e
r
media
te
loca
ti
on
be
twe
e
n
a
nd
.
T
he
phe
r
omone
leve
l
c
a
n
be
upda
ted
by
us
ing
the
(
5)
:
Ɣ
(
+
1
)
=
(
1
−
)
Ɣ
+
(
5)
w
he
r
e
is
the
phe
r
omone
e
va
por
a
ti
on
r
a
te,
is
the
t
otal
number
of
a
nts
;
ups
il
on,
is
the
tot
a
l
phe
r
omo
ne
leve
l
de
pos
it
e
d
a
t
the
e
dge
−
a
t
ti
me
.
T
he
s
e
two
pr
oc
e
s
s
e
s
will
be
r
e
pe
a
ted
by
the
a
nt
unti
l
the
it
e
r
a
ti
on
doe
s
not
e
xc
e
e
d
the
maximum
one
[
21]
,
[
22]
.
2.
4.
Vot
i
n
g
an
d
ave
r
agin
g
e
n
s
e
m
b
le
t
e
c
h
n
iq
u
e
Voting
a
nd
a
ve
r
a
ging
a
r
e
of
ten
us
e
d
e
ns
e
mbl
e
methods
in
mac
hine
lea
r
ning
to
e
nha
nc
e
f
o
r
e
c
a
s
t
a
c
c
ur
a
c
y
by
a
ggr
e
ga
ti
ng
the
r
e
s
ult
s
of
many
ba
s
e
models
.
T
he
s
e
s
tr
a
tegie
s
a
r
e
e
s
pe
c
ially
e
f
f
ici
e
nt
whe
n
s
e
pa
r
a
te
models
pos
s
e
s
s
dis
ti
nc
t
s
tr
e
ngths
a
nd
li
mi
tations
s
ince
they
may
mut
ua
ll
y
e
nha
nc
e
e
a
c
h
other
a
nd
r
e
s
ult
in
mor
e
r
e
s
il
ient
a
nd
pr
e
c
is
e
f
or
e
c
a
s
ts
[
23]
.
A
voti
ng
e
ns
e
mbl
e
invol
ve
s
us
ing
numer
ous
ba
s
e
models
,
s
uc
h
a
s
de
c
is
ion
tr
e
e
s
,
s
uppor
t
ve
c
tor
mac
hines
,
o
r
ne
ur
a
l
ne
two
r
ks
,
indepe
nde
ntl
y
p
r
ovidi
ng
p
r
e
dictions
on
the
s
a
me
da
tas
e
t.
T
he
ult
im
a
te
f
or
e
c
a
s
t
o
f
the
e
ns
e
mbl
e
is
e
s
tablis
he
d
by
c
ons
oli
da
ti
ng
the
in
divi
dua
l
e
s
ti
mate
s
us
ing
a
voti
ng
method.
An
a
ve
r
a
ging
e
n
s
e
mbl
e
,
s
ometim
e
s
c
a
ll
e
d
a
n
a
ve
r
a
ging
o
r
mea
n
e
ns
e
mbl
e
,
c
ombi
ne
s
the
pr
e
dictions
of
ba
s
e
models
by
c
a
lcula
ti
ng
their
a
ve
r
a
ge
[
24]
,
[
25]
.
2.
5.
Wor
k
f
low
of
t
h
e
p
r
op
os
e
d
wor
k
T
he
pr
opos
e
d
methods
a
dopt
two
pha
s
e
s
f
or
c
a
nc
e
r
c
las
s
if
ica
ti
on,
including
the
two
-
pha
s
e
f
e
a
tur
e
s
e
lec
ti
on
a
nd
c
las
s
if
ica
ti
on
pr
oc
e
s
s
e
s
.
I
n
the
f
ir
s
t
pha
s
e
,
the
f
e
a
tur
e
s
e
lec
ti
on
methods
AL
O
a
nd
AC
O
a
r
e
a
ppli
e
d
to
s
e
lec
t
the
s
ubs
e
t
of
ini
ti
a
l
f
e
a
tur
e
s
.
T
h
e
n,
the
s
e
c
ond
pha
s
e
s
tar
ts
by
a
pplyi
ng
the
ba
s
e
lea
r
ne
r
s
S
VM
,
R
F
,
Ada
B
oos
t,
a
nd
XG
B
oos
t
to
ha
ve
a
n
ini
ti
a
l
pr
e
diction.
T
he
n,
to
the
ini
ti
a
l
pr
e
diction
,
vot
ing
a
nd
a
ve
r
a
ging
a
r
e
a
ppli
e
d
to
make
the
f
inal
pr
e
dictio
n.
F
igur
e
1
s
hows
the
wor
king
pr
inciple
o
f
the
p
r
opos
e
d
model.
T
he
wor
kings
o
f
the
p
r
opos
e
d
method
c
a
n
be
e
xplaine
d
be
low.
S
tep
1:
C
ons
ider
the
mi
c
r
oa
r
r
a
y
da
ta
f
o
r
the
nor
ma
li
z
a
ti
on
pr
oc
e
s
s
us
ing
the
s
tanda
r
d
s
c
a
ler
method.
S
tep
2:
S
p
li
t
the
da
tas
e
t
int
o
two
s
e
ts
,
the
tr
a
ini
ng
s
e
t
a
nd
the
tes
ti
ng
s
e
t,
with
a
r
a
ti
o
of
80:20
.
S
tep
3:
T
o
the
tr
a
in
da
ta
,
a
pply
a
two
-
s
tage
f
e
a
tur
e
s
e
lec
ti
on
pr
oc
e
s
s
.
S
tep
4:
Apply
the
AL
O
f
e
a
tur
e
s
e
lec
ti
on
a
lgor
it
hm
−
I
nit
iate
the
a
nt
a
nd
a
nt
li
on
populations
with
maxi
mum
it
e
r
a
ti
on
.
−
C
a
lc
ul
a
t
e
t
he
f
i
tn
e
s
s
f
un
c
t
io
n
o
f
b
ot
h
po
pu
la
t
io
ns
us
in
g
t
he
k
-
f
o
ld
c
r
os
s
-
v
a
l
id
a
t
io
n
me
t
ho
d
wi
th
a
c
c
ur
a
c
y
a
s
a
ke
y
f
o
r
f
it
ne
s
s
f
u
nc
t
io
n
c
a
lc
ula
t
io
n
.
−
(
,
)
=
∑
=
1
,
with
n
a
s
the
numbe
r
of
f
olds
.
−
I
de
nti
f
y
the
ne
xt
pos
it
ion
o
f
the
a
nt
a
nd
a
nt
li
on
.
−
Re
-
c
a
lcula
te
the
f
it
ne
s
s
f
unc
ti
on
−
If
>
,
then
r
e
plac
e
(
)
with
−
Upda
te
the
F
it
(
)
unti
l
=
−
Upda
te
the
f
e
a
tur
e
s
e
t
S
tep
5:
Apply
AC
O
f
e
a
tur
e
s
e
lec
ti
on
a
lgor
it
hm
−
I
nit
iate
the
a
nt
population,
maximum
it
e
r
a
ti
on
−
C
a
lc
ul
a
t
e
t
he
f
i
tn
e
s
s
f
un
c
t
io
n
o
f
b
ot
h
po
pu
la
t
io
ns
us
in
g
t
he
k
-
f
o
ld
c
r
os
s
-
v
a
l
id
a
t
io
n
me
t
ho
d
wi
th
a
c
c
ur
a
c
y
a
s
a
ke
y
f
o
r
f
it
ne
s
s
f
u
nc
t
io
n
c
a
lc
ula
t
io
n
−
(
)
=
∑
=
1
,
with
n
a
s
the
numbe
r
of
f
olds
−
I
de
nti
f
y
the
ne
xt
pos
it
ion
o
f
the
a
nt
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
15
,
No.
1
,
F
e
br
ua
r
y
20
25
:
604
-
613
608
−
Re
-
c
a
lcula
te
the
f
it
ne
s
s
f
unc
ti
on
−
If
>
,
then
r
e
plac
e
(
)
with
−
Upda
te
the
(
)
unti
l
=
−
Upda
te
f
inal
f
e
a
tur
e
s
f
or
P
ha
s
e
2
of
the
pr
opos
e
d
model.
S
tep
6:
Apply
ba
s
e
c
las
s
if
ier
s
(
S
VM
,
R
F
,
XG
B
oos
t,
a
nd
Ada
B
oos
t)
.
S
tep
7:
Apply
e
ns
e
mbl
e
c
las
s
if
ier
voti
ng
a
nd
a
ve
r
a
ging.
S
tep
8:
Apply
tes
t
da
ta
to
two
t
r
a
ined
models
obtai
ne
d
f
r
om
S
tep
7
.
S
tep
9:
E
va
luate
the
pe
r
f
or
manc
e
of
the
t
r
a
ined
mo
de
ls
.
F
igur
e
1.
W
or
kf
low
of
the
p
r
opos
e
d
model
3.
RE
S
UL
T
AN
D
DI
S
CU
S
S
I
ON
T
h
e
a
bo
ve
-
s
ta
ted
p
r
op
os
e
d
m
od
e
l
is
e
v
a
l
ua
ted
o
n
th
e
s
ys
t
e
m
,
i
nc
lu
di
ng
8
GB
R
AM
,
W
in
do
ws
11
O
S
,
a
nd
a
n
I
nt
e
l
i3
p
r
o
c
e
s
s
or
w
it
h
a
2
.
6
GH
z
c
l
oc
k
s
p
e
e
d
.
T
he
p
r
o
pos
e
d
mo
de
l
is
a
na
l
yz
e
d
in
a
P
y
th
on
e
n
vir
o
nm
e
n
t
u
s
i
ng
A
na
c
on
da
na
vi
ga
to
r
.
F
o
r
e
va
l
ua
t
i
ng
t
he
pe
r
f
o
r
ma
nc
e
,
the
p
r
op
os
e
d
a
pp
r
oa
c
h
a
d
op
ts
f
iv
e
di
f
f
e
r
e
nt
M
L
-
b
a
s
e
d
e
v
a
l
ua
ti
ve
pa
r
a
me
te
r
s
,
inc
lu
d
in
g
a
c
c
u
r
a
c
y
,
p
r
e
c
is
i
on
,
r
e
c
a
ll
,
F
1
-
s
c
o
r
e
,
a
n
d
s
pe
c
i
f
ic
it
y
,
w
hi
c
h
c
a
n
be
d
e
f
i
ne
d
us
i
ng
(
6
)
-
(
10
)
w
i
th
T
P
O
,
T
N
E
,
F
P
O
,
a
n
d
F
N
E
a
s
t
r
u
e
pos
it
i
ve
,
t
r
u
e
ne
g
a
t
iv
e
,
f
a
ls
e
p
os
it
iv
e
a
nd
f
a
ls
e
n
e
g
a
t
iv
e
r
e
s
pe
c
ti
ve
ly
.
T
a
b
le
3
s
ho
ws
t
he
a
n
a
l
ys
is
o
f
t
he
hy
b
r
i
d
m
od
e
l
us
in
g
AL
O
a
nd
AC
O
f
e
a
tu
r
e
s
e
le
c
t
ion
m
e
c
ha
nis
ms
.
T
a
b
les
4
a
nd
5
r
e
pr
e
s
e
nt
t
he
a
na
ly
s
is
o
f
t
he
m
od
e
l
us
i
ng
v
ot
in
g
a
n
d
a
ve
r
a
g
in
g
e
ns
e
mb
le
t
e
c
hn
iq
ue
s
wi
th
AL
O
a
n
d
AC
O
f
e
a
tu
r
e
s
e
l
e
c
ti
on
a
lg
o
r
i
t
hms
.
F
ig
ur
e
s
2
t
o
6
s
ho
w
t
he
pe
r
f
o
r
man
c
e
c
om
pa
r
is
o
n
o
f
th
e
v
ot
in
g
a
nd
a
ve
r
a
g
in
g
e
n
s
e
m
b
le
t
e
c
hn
iq
ue
to
th
e
a
bo
ve
-
d
is
c
us
s
e
d
e
v
a
l
ua
ti
ve
pa
r
a
me
te
r
s
.
=
+
+
+
+
(
6)
=
+
(
7)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
A
nt
li
on
and
ant
c
olony
opti
miz
ati
on
int
e
gr
ated
e
ns
e
mble
mac
hine
…
(
P
inaks
hi
P
anda
)
609
=
+
(
8)
−
1
=
+
1
2
(
+
)
(
9)
=
+
(
10)
T
he
e
mpi
r
ica
l
a
na
lys
is
s
hows
that
the
e
ns
e
mbl
e
model
with
a
vot
ing
c
las
s
if
ier
a
nd
we
ight
e
d
a
ve
r
a
ging
outper
f
o
r
ms
a
ll
c
onve
nti
ona
l
M
L
-
ba
s
e
d
c
las
s
if
ier
s
in
the
c
ur
r
e
nt
wor
k
.
F
or
ova
r
ian
c
a
nc
e
r
,
the
pr
opos
e
d
e
ns
e
mbl
e
a
ppr
oa
c
h
with
a
vot
ing
c
las
s
if
ier
s
hows
a
n
a
c
c
ur
a
c
y
of
98.
44
%
,
a
nd
with
the
we
ight
e
d
a
ve
r
a
ging,
the
model
s
hows
96.
33%
.
T
he
voti
ng
a
nd
we
ight
e
d
a
ve
r
a
ging
c
las
s
if
ica
ti
ons
f
or
lung
c
a
nc
e
r
s
how
99.
08%
a
nd
97
.
36%
,
r
e
s
pe
c
ti
ve
ly.
F
or
the
b
r
e
a
s
t
c
a
nc
e
r
da
tas
e
t,
the
voti
ng
c
las
s
if
ier
s
hows
a
98.
06%
a
c
c
ur
a
c
y
leve
l,
a
nd
the
we
ight
e
d
a
ve
r
a
ging
c
las
s
if
ier
s
hows
a
n
a
c
c
ur
a
c
y
leve
l
of
96.
20
%
a
c
c
ur
a
c
y
leve
l.
S
im
il
a
r
ly,
the
pr
opos
e
d
a
ppr
oa
c
h
with
the
voti
ng
c
las
s
if
ier
s
hows
a
n
a
c
c
ur
a
c
y
le
ve
l
of
97.
27%
,
a
nd
we
ight
e
d
a
ve
r
a
ging
s
hows
a
n
a
c
c
ur
a
c
y
leve
l
of
95
.
5%
.
B
a
s
e
d
on
a
c
ompar
is
on
of
the
voti
ng
a
nd
a
ve
r
a
ging
tec
hniques
,
it
c
a
n
be
obs
e
r
ve
d
f
r
o
m
T
a
bles
4
a
nd
5
that
the
vo
ti
ng
tec
hnique
outper
f
o
r
ms
the
a
ve
r
a
ging
tec
hniqu
e
.
S
o,
a
s
a
n
e
va
luative
method,
the
AU
C
va
lue
of
the
voti
ng
tec
hnique
in
c
ontr
a
s
t
to
dif
f
e
r
e
nt
da
tas
e
ts
is
s
hown
in
F
igur
e
s
7
to
10
.
T
he
AU
C
va
lues
f
or
the
ova
r
ian
c
a
nc
e
r
,
lung
c
a
nc
e
r
,
b
r
e
a
s
t
c
a
nc
e
r
,
a
nd
leuke
mi
a
da
t
a
s
e
ts
a
r
e
0.
986,
0
.
99,
0.
978
,
a
nd
0.
973
,
r
e
s
pe
c
ti
ve
ly.
T
he
a
na
lys
is
s
how
s
that
the
de
ve
loped
model
e
xhibi
ts
high
a
c
c
ur
a
c
y,
s
e
ns
it
ivi
ty,
a
nd
s
pe
c
if
icity,
whic
h
indi
c
a
tes
that
the
de
ve
loped
model
c
a
n
be
e
f
f
e
c
ti
ve
ly
us
e
d
to
c
las
s
if
y
c
a
nc
e
r
a
nd
non
-
c
a
nc
e
r
pa
ti
e
nts
.
T
he
model
s
hows
r
obus
tne
s
s
a
nd
s
c
a
labili
ty
a
c
r
os
s
dif
f
e
r
e
nt
c
a
nc
e
r
da
tas
e
ts
.
S
o,
the
de
ve
loped
h
igh
-
pe
r
f
or
mi
ng
M
L
-
ba
s
e
d
model
c
a
n
be
us
e
d
to
de
ve
lop
a
mor
e
r
e
li
a
ble
diagnos
ti
c
tool
that
c
a
n
he
lp
c
li
nicia
ns
make
mo
r
e
inf
or
med
de
c
is
ions
r
e
ga
r
ding
c
a
nc
e
r
diagnos
is
.
I
n
or
de
r
to
s
how
the
e
f
f
ica
c
y,
the
pr
opos
e
d
model
is
c
om
pa
r
e
d
with
s
ome
e
xis
ti
ng
li
ter
a
tur
e
.
T
a
ble
6
s
h
ows
the
pe
r
f
or
manc
e
c
ompar
is
on
of
pr
opos
e
d
model
with
e
xis
ti
ng
li
ter
a
tur
e
s
in
ter
ms
of
a
c
c
ur
a
c
y.
I
t
c
a
n
b
e
c
lea
r
ly
obs
e
r
ve
d
that
the
pr
opos
e
d
model
out
pe
r
f
or
ms
a
ll
e
xis
ti
ng
li
ter
a
tur
e
a
c
r
os
s
a
ll
da
tas
e
ts
.
T
a
ble
3.
P
e
r
f
o
r
manc
e
of
indi
vidual
model
with
AL
O
a
nd
AC
O
f
e
a
tur
e
s
e
lec
ti
on
D
a
ta
s
e
t
H
ybr
id
m
ode
l
w
it
h A
L
O
a
nd A
C
O
A
c
c
ur
a
c
y
P
r
e
c
is
io
n
R
e
c
a
ll
F1
-
s
c
or
e
S
pe
c
if
ic
it
y
O
va
r
ia
n
S
V
M
88.89
91.43
86.49
88.89
91.43
RF
84.72
91.67
80.49
85.71
90.32
X
G
B
oos
t
86.11
82.86
87.88
85.29
84.62
A
da
B
oos
t
87.50
87.23
93.18
90.11
78.57
L
ung
S
V
M
87.35
89.88
90.96
90.42
80.46
RF
86.96
90.66
91.16
90.91
76.39
X
G
B
oos
t
90.51
91.72
93.94
92.81
84.09
A
da
B
oos
t
89.33
92.99
90.12
91.54
87.91
B
r
e
a
s
t
S
V
M
83.98
89.15
88.46
88.80
72.55
RF
84.21
87.79
87.12
87.45
79.22
X
G
B
oos
t
88.63
90.16
93.75
91.92
77.22
A
da
B
oos
t
84.32
89.15
88.46
88.80
74.55
L
e
uke
mi
a
S
V
M
84.25
87.67
85.33
86.49
82.69
RF
87.60
86.96
90.91
88.89
83.64
X
G
B
oos
t
86.67
93.94
86.11
89.86
87.88
A
da
B
oos
t
88.37
91.03
89.87
90.45
86.00
T
a
ble
4.
P
e
r
f
o
r
manc
e
of
voti
ng
e
ns
e
mbl
e
tec
hniqu
e
with
AL
O
a
nd
AC
O
f
e
a
tur
e
s
e
lec
ti
on
D
a
ta
s
e
t
A
c
c
ur
a
c
y
P
r
e
c
is
io
n
R
e
c
a
ll
F1
-
S
c
or
e
S
pe
c
if
ic
it
y
O
va
r
ia
n
98.44
98.73
98.73
98.73
97.96
L
ung
99.08
99.32
99.32
99.32
98.59
B
r
e
a
s
t
98.06
99.14
98.29
98.71
97.37
L
e
uke
mi
a
97.27
96.97
98.46
97.71
95.56
T
a
ble
5.
P
e
r
f
o
r
manc
e
of
a
ve
r
a
ging
e
ns
e
mbl
e
tec
hnique
with
AL
O
a
nd
AC
O
f
e
a
tur
e
s
e
lec
ti
on
D
a
ta
s
e
t
A
c
c
ur
a
c
y
P
r
e
c
is
io
n
R
e
c
a
ll
F1
-
S
c
or
e
S
pe
c
if
ic
it
y
O
va
r
ia
n
96.33
97.10
97.10
97.10
95.00
L
ung
97.36
97.42
98.69
98.05
94.59
B
r
e
a
s
t
96.20
98.29
96.64
97.46
94.87
L
e
uke
mi
a
95.54
95.52
96.97
96.24
93.48
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
15
,
No.
1
,
F
e
br
ua
r
y
20
25
:
604
-
613
610
F
igur
e
2.
Ac
c
ur
a
c
y
c
ompar
is
on
a
mong
voti
ng
a
nd
a
ve
r
a
ging
tec
hnique
F
igur
e
3.
P
r
e
c
is
ion
c
ompar
is
on
a
mong
voti
ng
a
nd
a
ve
r
a
ging
tec
hnique
F
igur
e
4.
R
e
c
a
ll
c
ompar
is
on
a
mong
vot
ing
a
nd
a
ve
r
a
ging
tec
hnique
F
igur
e
5.
F
1
-
s
c
or
e
c
ompar
is
on
a
mong
voti
ng
a
nd
a
ve
r
a
ging
tec
hnique
F
igur
e
6.
S
pe
c
if
icity
c
ompar
is
on
a
mong
vo
ti
ng
a
n
d
a
ve
r
a
ging
tec
hnique
Fi
g
u
r
e
7
.
RO
C
cu
r
v
e
fo
r
o
v
ari
a
n
can
cer
d
a
t
as
e
t
Fi
g
u
r
e
8
.
RO
C
cu
r
v
e
fo
r
l
u
n
g
can
cer
d
a
t
as
e
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
A
nt
li
on
and
ant
c
olony
opti
miz
ati
on
int
e
gr
ated
e
ns
e
mble
mac
hine
…
(
P
inaks
hi
P
anda
)
611
F
igur
e
9.
R
OC
c
ur
ve
f
o
r
b
r
e
a
s
t
C
a
nc
e
r
da
tas
e
t
F
igur
e
10.
R
OC
c
ur
ve
f
o
r
L
e
uke
mi
a
da
tas
e
t
T
a
ble
6.
Ac
c
ur
a
c
y
c
ompar
is
on
of
the
pr
opos
e
d
mo
de
l
with
e
xis
ti
ng
wor
k
R
e
f
e
r
e
nc
e
O
va
r
ia
n
L
ung
B
r
e
a
s
t
L
e
uke
mi
a
[
8]
--
--
--
92.33
[
9]
95.65
93.56
84
91.36
[
10]
--
--
97.2
--
[
12]
--
--
--
92.7
[
13]
--
--
--
86.17
P
r
op
os
e
d
98.44
99.08
98.06
97.27
4.
CONC
L
USI
ON
T
he
c
ur
r
e
nt
wor
k
a
im
s
t
he
de
ve
lop
a
n
e
ns
e
mbl
e
-
ba
s
e
d
c
a
nc
e
r
diagnos
is
model
with
AL
O
a
nd
AC
O
a
s
the
f
e
a
tur
e
s
e
lec
ti
on
a
lgor
it
hm
.
T
o
pr
opos
e
d
method
e
quips
f
our
types
of
mac
hine
lea
r
ning
c
l
a
s
s
if
ier
s
:
s
uppor
t
ve
c
tor
mac
hine
,
r
a
ndom
f
or
e
s
t
,
Ada
B
oos
t,
a
nd
XG
B
oos
t
with
two
e
ns
e
mbl
e
c
las
s
if
ier
s
including
voti
ng
a
nd
a
ve
r
a
ging
.
T
he
e
xpe
r
im
e
ntal
s
tudy
don
e
in
thi
s
wor
k
s
hows
that
the
voti
ng
c
las
s
if
ier
out
pe
r
f
or
ms
the
a
ve
r
a
ging
c
las
s
if
ier
s
with
a
n
a
c
c
ur
a
c
y
of
98.
44%
,
99.
08%
,
98.
06
%
,
a
nd
97
.
27%
f
o
r
the
o
va
r
i
a
n,
lung
,
br
e
a
s
t,
a
nd
leuke
mi
a
C
a
nc
e
r
da
tas
e
t
s
,
r
e
s
pe
c
ti
ve
ly.
As
pe
r
the
R
OC
a
na
lys
i
s
,
the
AU
C
va
lues
of
the
p
r
opos
e
d
model
with
voti
ng
c
las
s
if
ier
a
r
e
0
.
985,
0
.
99,
0.
9
78,
a
nd
0.
973
f
o
r
the
o
va
r
ian
,
lung
,
br
e
a
s
t,
a
nd
l
e
uke
mi
a
c
a
nc
e
r
da
tas
e
ts
.
Our
r
e
s
e
a
r
c
h
on
a
model
f
or
c
a
nc
e
r
de
tec
ti
on
us
ing
mac
hine
lea
r
ning
s
ugge
s
ts
that
thi
s
f
ield
of
s
tudy
a
nd
the
medic
a
l
c
omm
unit
y
may
unde
r
go
a
pa
r
a
digm
c
ha
nge
.
Our
tec
hnique
im
p
r
ove
s
the
r
e
li
a
bil
it
y
a
nd
c
a
pa
c
it
y
to
identif
y
c
a
nc
e
r
e
a
r
ly.
T
he
high
a
c
c
ur
a
c
y,
s
e
ns
it
ivi
ty,
a
nd
s
pe
c
if
icity
leve
ls
s
ugge
s
t
that
thi
s
mi
ght
r
e
s
ult
in
be
tt
e
r
pa
ti
e
nt
outcome
s
.
T
o
f
ull
y
gr
a
s
p
the
int
r
ica
c
ies
of
c
a
nc
e
r
,
it
is
ne
c
e
s
s
a
r
y
to
us
e
mul
ti
-
modal
a
ppr
oa
c
he
s
that
include
mul
ti
ple
da
t
a
s
e
ts
,
including
ge
ne
ti
c
a
nd
c
li
nica
l
da
ta.
Ulti
mat
e
ly,
our
r
e
s
ult
s
im
ply
that
mac
hine
lea
r
ning
mi
ght
im
pr
o
ve
c
a
nc
e
r
de
tec
ti
on,
whic
h
would
be
ne
f
it
s
oc
iety
a
nd
the
r
e
s
e
a
r
c
h
c
omm
unit
y
via
be
tt
e
r
r
e
s
our
c
e
a
ll
oc
a
ti
o
n,
lowe
r
mi
s
diagnos
is
r
a
tes
,
a
nd
mor
e
e
f
f
icie
nt
h
e
a
lt
hc
a
r
e
de
li
ve
r
y.
RE
F
E
RE
NC
E
S
[
1]
A
.
P
a
ni
gr
a
hi
e
t
al
.
,
“
E
n
-
M
in
W
ha
le
:
a
n
e
ns
e
mbl
e
a
ppr
oa
c
h
ba
s
e
d
on
M
R
M
R
a
nd
w
ha
le
opt
im
iz
a
ti
on
f
or
c
a
nc
e
r
di
a
gno
s
is
,”
I
E
E
E
A
c
c
e
s
s
, vol
. 11, pp. 113526
–
113542, 2023, doi:
10.1109/AC
C
E
S
S
.2023.3318261.
[
2]
M
.
K
ha
ls
a
n
e
t
al
.
,
“
A
s
ur
ve
y
of
ma
c
hi
ne
le
a
r
ni
ng
a
ppr
oa
c
he
s
a
ppl
ie
d
to
ge
ne
e
xpr
e
s
s
io
n
a
na
ly
s
i
s
f
or
c
a
nc
e
r
pr
e
di
c
ti
on,”
I
E
E
E
A
c
c
e
s
s
, vol
. 10, pp. 27522
–
27534, 2022, doi:
10.1109/AC
C
E
S
S
.2022.3146312.
[
3]
B
.
M
a
,
F
.
M
e
ng,
G
.
Y
a
n,
H
.
Y
a
n,
B
.
C
ha
i,
a
nd
F
.
S
ong,
“
D
ia
gnos
ti
c
c
la
s
s
if
ic
a
ti
on
of
c
a
nc
e
r
s
u
s
in
g
e
xt
r
e
me
gr
a
di
e
nt
boos
ti
ng
a
lg
or
it
hm
a
nd
mul
ti
-
omi
c
s
da
ta
,”
C
om
put
e
r
s
in
B
io
lo
gy
and
M
e
di
c
in
e
,
vol
.
121
,
A
r
t.
no.
103761
,
J
un.
20
20,
doi
:
10.1016/j
.c
ompbi
ome
d.2020.103761.
[
4]
S
.
O
s
a
ma
,
H
.
S
ha
ba
n,
a
nd
A
.
A
.
A
li
,
“
G
e
ne
r
e
duc
ti
on
a
nd
ma
c
hi
ne
le
a
r
ni
ng
a
lg
or
it
hms
f
or
c
a
nc
e
r
c
la
s
s
if
ic
a
ti
on
ba
s
e
d
on
mi
c
r
oa
r
r
a
y
ge
ne
e
xpr
e
s
s
io
n
da
ta
:
a
c
ompr
e
he
ns
iv
e
r
e
vi
e
w
,”
E
x
pe
r
t
Sy
s
te
m
s
w
it
h
A
ppl
ic
at
io
ns
,
vol
.
213,
M
a
r
.
2
023,
doi
:
10.1016/j
.e
s
w
a
.2022.118946.
[
5]
B
. S
a
hu, A. P
a
ni
gr
a
hi
, S
. K
. R
out
, a
nd A
.
P
a
ti
, “
H
ybr
id
mul
t
ip
l
e
f
il
te
r
e
mbe
dde
d poli
ti
c
a
l
opt
im
iz
e
r
f
or
f
e
a
tu
r
e
s
e
le
c
ti
on,”
in
2
022
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on
I
nt
e
ll
ig
e
nt
C
ont
r
ol
le
r
and
C
om
put
in
g
fo
r
Sm
ar
t
P
ow
e
r
(
I
C
I
C
C
SP
)
,
H
yde
r
a
ba
d,
I
ndi
a
,
2022,
pp.
1
-
6,
doi
:
10.1109/I
C
I
C
C
S
P
53532.2022.9862419.
[
6]
A
. P
a
ti
, M
. P
a
r
hi
, B
. K
. P
a
tt
a
na
ya
k, B
. S
a
hu, a
nd S
. K
ha
s
im
, “
C
a
nD
ia
g:
f
og e
mpowe
r
e
d t
r
a
ns
f
e
r
de
e
p l
e
a
r
ni
ng ba
s
e
d a
ppr
oa
c
h f
o
r
c
a
nc
e
r
di
a
gnos
i
s
,”
D
e
s
ig
ns
, vol
. 7, no. 3, 2023, doi:
10.3390/de
s
ig
ns
7030057.
[
7]
A
.
P
a
ni
gr
a
hi
,
S
.
B
hut
ia
,
B
.
S
a
hu,
M
.
G
.
G
a
le
ty
,
a
nd
S
.
N
.
M
oha
nt
y,
“
B
P
S
O
-
PSO
-
S
V
M
:
a
n
in
te
gr
a
te
d
a
ppr
oa
c
h
f
or
c
a
n
c
e
r
di
a
gnos
is
,”
L
e
c
tu
r
e
N
ot
e
s
i
n E
le
c
tr
ic
al
E
ngi
ne
e
r
in
g
, vol
. 905, p
p. 571
–
579, 2022, doi:
10.1007/978
-
981
-
19
-
2177
-
3_53.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
15
,
No.
1
,
F
e
br
ua
r
y
20
25
:
604
-
613
612
[
8]
R
.
M
.
A
z
i
z
,
“
N
a
tu
r
e
-
in
s
pi
r
e
d
me
ta
h
e
ur
is
ti
c
s
mode
l
f
or
ge
ne
s
e
le
c
ti
on
a
nd
c
la
s
s
if
ic
a
ti
on
of
bi
ome
di
c
a
l
mi
c
r
oa
r
r
a
y
da
ta
,”
M
e
d
ic
al
& B
io
lo
gi
c
al
E
ngi
ne
e
r
in
g & C
om
put
in
g
, vol
. 60, no. 6, pp. 162
7
–
1646, J
un. 2022, doi:
10.1007/s
11517
-
022
-
02555
-
7.
[
9]
N
.
N
e
koui
e
,
M
.
R
omooz
i,
a
nd
M
.
E
s
ma
e
il
i,
“
A
n
e
w
e
vol
u
ti
ona
r
y
e
ns
e
mbl
e
le
a
r
ni
ng
of
mul
ti
moda
l
f
e
a
tu
r
e
s
e
le
c
ti
on
f
r
om
mi
c
r
oa
r
r
a
y da
ta
,”
N
e
ur
al
P
r
oc
e
s
s
in
g L
e
tt
e
r
s
, vol
. 55, no. 5, pp.
6753
–
6780, 2023, doi:
10.1007/s
11063
-
023
-
11159
-
7.
[
10]
M
.
A
. N
a
j
i,
S
.
E
l
F
i
la
li
, K
. A
a
r
i
k
a
,
E
.
H
. B
e
n
la
hm
a
r
, R
.
A
. A
b
d
e
l
ou
h
a
h
id
, a
n
d O
.
D
e
b
a
uc
h
e
,
“
M
a
c
h
in
e
l
e
a
r
ni
ng
a
lg
or
i
th
m
s
f
or
b
r
e
a
s
t
c
a
n
c
e
r
pr
e
di
c
ti
on
a
nd
d
ia
gn
o
s
i
s
,”
P
r
o
c
e
d
ia
C
o
m
p
ut
e
r
S
c
i
e
n
c
e
,
v
ol
.
19
1,
p
p.
4
87
–
4
92
,
20
21
,
do
i:
1
0.
10
16
/j
.
pr
oc
s
.2
02
1.
07
.0
62
.
[
11]
J
.
L
u,
E
.
S
ong,
A
.
G
hone
im
,
a
nd
M
.
A
lr
a
s
houd,
“
M
a
c
hi
n
e
le
a
r
ni
ng
f
or
a
s
s
is
ti
ng
c
e
r
vi
c
a
l
c
a
n
c
e
r
di
a
gno
s
is
:
a
n
e
ns
e
mbl
e
a
ppr
oa
c
h,”
F
ut
ur
e
G
e
ne
r
at
io
n C
om
put
e
r
Sy
s
te
m
s
, vol
. 106, pp.
199
–
205, M
a
y 2020, doi:
10.1016/j
.f
ut
ur
e
.2019.12.033.
[
12]
N
. T
a
va
s
ol
i,
K
. R
e
z
a
e
e
, M
. M
ome
nz
a
de
h, a
nd M
. S
e
hha
ti
, “
A
n
e
ns
e
mbl
e
s
of
t
w
e
ig
ht
e
d ge
ne
s
e
le
c
ti
on
-
ba
s
e
d a
ppr
oa
c
h a
nd c
a
n
c
e
r
c
la
s
s
if
ic
a
ti
on
us
in
g
modi
f
ie
d
me
ta
he
ur
is
ti
c
le
a
r
ni
ng,”
J
our
nal
of
C
om
put
at
io
nal
D
e
s
ig
n
and
E
ngi
ne
e
r
in
g
,
vol
.
8,
no
.
4,
pp. 1172
–
1189, J
ul
. 2021, doi:
10.1093/j
c
de
/q
w
a
b039.
[
13]
L
.
S
un,
X
.
-
Y
.
Z
ha
ng,
Y
.
-
H
.
Q
ia
n,
J
.
-
C
.
X
u,
S
.
-
G
.
Z
ha
ng,
a
nd
Y
.
T
ia
n,
“
J
oi
nt
n
e
ig
hbor
hood
e
nt
r
opy
-
ba
s
e
d
ge
ne
s
e
le
c
ti
on
me
th
od
w
it
h
f
is
he
r
s
c
or
e
f
or
tu
mor
c
la
s
s
if
ic
a
ti
on,”
A
ppl
ie
d
I
nt
e
ll
ig
e
nc
e
,
vol
.
49,
no.
4,
pp.
1245
–
1259,
A
pr
.
2019,
doi
:
10.1007/s
10
489
-
018
-
1320
-
1.
[
14]
A
.
K
.
S
hukl
a
,
P
.
S
in
gh,
a
nd
M
.
V
a
r
dha
n,
“
G
e
ne
s
e
l
e
c
ti
on
f
or
c
a
nc
e
r
ty
pe
s
c
la
s
s
if
ic
a
ti
on
us
in
g
nove
l
hybr
id
me
ta
h
e
ur
is
ti
c
s
a
ppr
oa
c
h,”
Sw
ar
m
and E
v
ol
ut
io
nar
y
C
om
put
at
io
n
, vol
. 54, M
a
y 2020, doi:
10.1016/j
.s
w
e
vo.2020.100661.
[
15]
L
.
M
e
e
na
c
hi
a
nd
S
.
R
a
ma
kr
is
hn
a
n,
“
M
e
ta
he
ur
is
ti
c
s
e
a
r
c
h
ba
s
e
d
f
e
a
tu
r
e
s
e
l
e
c
ti
on
me
th
od
s
f
or
c
la
s
s
if
ic
a
ti
on
of
C
a
nc
e
r
,”
P
at
te
r
n R
e
c
ogni
ti
on
, vol
. 119, Nov. 2021, do
i:
10.1016/j
.pa
tc
og.2021.108079.
[
16]
Z
.
Z
hu,
Y
.
-
S
.
O
ng,
a
nd
M
.
D
a
s
h,
“
M
a
r
kov
bl
a
nke
t
-
e
mbe
dde
d ge
ne
ti
c
a
lg
or
it
hm
f
or
ge
ne
s
e
le
c
ti
on,”
P
at
te
r
n
R
e
c
ogni
ti
on
,
vol
.
40,
no. 11, pp. 3236
–
3248, Nov. 2007, do
i:
10.1016/j
.pa
tc
og.2007.02.007.
[
17]
S
.
M
ir
ja
li
li
,
“
T
he
a
nt
li
on
opt
im
iz
e
r
,”
A
dv
anc
e
s
in
E
ngi
ne
e
r
in
g
Sof
tw
ar
e
,
vol
.
83,
pp.
80
–
98,
M
a
y
2
015,
doi
:
10.1016/j
.a
dve
ngs
of
t.
2015.01.010.
[
18]
E
.
E
ma
r
y,
H
.
M
.
Z
a
w
ba
a
,
a
nd
A
.
E
.
H
a
s
s
a
ni
e
n,
“
B
in
a
r
y
a
nt
li
on
a
ppr
oa
c
he
s
f
or
f
e
a
tu
r
e
s
e
le
c
ti
on,”
N
e
ur
oc
om
put
in
g
,
vol
.
213,
pp. 54
–
65, Nov. 2016, doi:
10.1016/j
.ne
uc
om.2016.03.101.
[
19]
H
.
M
.
Z
a
w
ba
a
,
E
.
E
ma
r
y,
a
nd
B
.
P
a
r
v,
“
F
e
a
tu
r
e
s
e
le
c
ti
on
ba
s
e
d
on
a
nt
li
on
opt
im
iz
a
ti
on
a
lg
or
it
hm,”
in
2015
T
hi
r
d
W
or
ld
C
onf
e
r
e
nc
e
on C
om
pl
e
x
S
y
s
te
m
s
(
W
C
C
S)
, N
ov. 2015, pp. 1
–
7,
doi
:
10.1109/I
C
oC
S
.2015.7483317.
[
20]
H
.
P
e
ng,
C
.
Y
in
g,
S
.
T
a
n,
B
.
H
u,
a
nd
Z
.
S
un,
“
A
n
im
pr
ove
d
f
e
a
tu
r
e
s
e
le
c
ti
on
a
lg
or
it
hm
ba
s
e
d
on
a
nt
c
ol
ony
opt
im
iz
a
ti
on,”
I
E
E
E
A
c
c
e
s
s
, vol
. 6, pp. 69203
–
69209, 2018, doi:
10.1109/AC
C
E
S
S
.2018.2879583.
[
21]
M
.
P
a
ni
r
i,
M
.
B
.
D
ow
la
ts
ha
hi
,
a
nd H
.
N
e
z
a
m
a
ba
di
-
pour
,
“
M
L
A
C
O
:
a
mul
ti
-
la
be
l
f
e
a
tu
r
e
s
e
le
c
ti
on
a
lg
or
it
hm
ba
s
e
d
on
a
nt
c
o
lo
ny
opt
im
iz
a
ti
on,”
K
now
le
dge
-
B
as
e
d Sy
s
te
m
s
, vol
. 192,
M
a
r
. 2020,
doi
:
10.1016/j
.knos
ys
.2019.105285.
[
22]
B
.
M
a
nd
S
.
V
,
“
F
a
ul
t
-
ty
pe
c
ove
r
a
ge
ba
s
e
d
a
nt
c
ol
ony
opt
im
iz
a
ti
on
a
lg
or
it
hm
f
or
a
tt
a
in
in
g
s
ma
ll
e
r
te
s
t
s
ui
te
,”
I
A
E
S
I
nt
e
r
nat
i
onal
J
our
nal
of
A
r
ti
fi
c
ia
l
I
nt
e
ll
ig
e
nc
e
, vol
. 9, no. 3, pp. 507
–
519, S
e
p. 2020, doi:
10.11591/i
ja
i.
v9.i
3.pp507
-
519.
[
23]
H
.
E
.
K
iz
il
oz
,
“
C
la
s
s
if
ie
r
e
ns
e
mbl
e
me
th
od
s
in
f
e
a
tu
r
e
s
e
le
c
ti
on,”
N
e
u
r
oc
om
put
in
g
,
vol
.
419,
pp.
97
–
107,
J
a
n.
20
21,
doi
:
10.1016/j
.ne
uc
om.2020.07.113.
[
24]
G
.
S
a
h
oo
e
t
al
.
,
“
B
r
e
a
s
t
c
a
n
c
e
r
r
e
l
a
ps
e
di
s
e
a
s
e
p
r
e
d
ic
ti
on
im
p
r
o
ve
me
nt
s
w
i
th
e
ns
e
mb
le
le
a
r
ni
ng
a
pp
r
oa
c
he
s
,”
I
n
do
ne
s
i
a
n
J
our
na
l
o
f
E
le
c
tr
i
c
a
l
E
n
gi
ne
e
r
in
g
a
nd
C
o
m
p
ut
e
r
Sc
ie
nc
e
,
v
ol
.
3
5,
no
.
1
,
p
p.
33
5
–
34
2,
J
ul
.
2
02
4,
d
o
i:
1
0
.1
15
91
/i
je
e
c
s
.
v3
5.
i1
.p
p3
35
-
34
2.
[
25]
T
.
E
.
M
a
th
e
w
,
“
A
c
os
t
s
e
ns
it
iv
e
S
V
M
a
nd
n
e
ur
a
l
ne
twor
k
e
ns
e
mbl
e
mode
l
f
or
br
e
a
s
t
C
a
nc
e
r
c
la
s
s
if
ic
a
ti
on,”
I
ndone
s
ia
n
J
ou
r
nal
of
E
le
c
tr
ic
al
E
ngi
ne
e
r
in
g and I
nf
or
m
at
ic
s
(
I
J
E
E
I
)
, vol
. 11, no.
2, A
pr
. 2023, doi:
10.52549/i
je
e
i.
v11i
2.3934.
B
I
OG
RA
P
HI
E
S
OF
AU
T
HO
RS
P
i
na
ks
hi
P
a
nd
a
co
mp
l
et
e
d
h
er
Mas
t
er
o
f
Co
m
p
u
t
er
A
p
p
l
i
ca
t
i
o
n
s
(M.
C.
A
.
)
d
eg
ree
at
Bi
j
u
Pat
n
ai
k
U
n
i
v
er
s
i
t
y
o
f
T
ec
h
n
o
l
o
g
y
(BPU
T
)
i
n
Ro
u
rk
el
a,
In
d
i
a,
i
n
2
0
1
2
.
Su
b
s
e
q
u
en
t
l
y
,
s
h
e
earn
ed
a
Mas
t
er
o
f
T
ech
n
o
l
o
g
y
(M.
T
ech
.
)
d
eg
ree
i
n
co
mp
u
t
er
s
ci
e
n
ce
an
d
i
n
f
o
rmat
i
cs
fro
m
Si
k
s
h
a
'O
'
A
n
u
s
a
n
d
h
a
n
(SO
A
)
U
n
i
v
er
s
i
t
y
,
Bh
u
b
an
e
s
w
ar,
In
d
i
a,
i
n
2
0
1
4
.
Sh
e
i
s
p
u
rs
u
i
n
g
a
D
o
ct
o
r
o
f
Ph
i
l
o
s
o
p
h
y
(P
h
.
D
.
)
d
e
g
ree
i
n
c
o
mp
u
t
er
s
ci
e
n
ce
an
d
en
g
i
n
eeri
n
g
a
t
C.
V
.
Raman
G
l
o
b
al
U
n
i
v
er
s
i
t
y
(CG
U
),
Bh
u
b
a
n
es
w
ar,
In
d
i
a.
Sh
e
h
as
accu
mu
l
at
e
d
mo
re
t
h
an
t
h
ree
y
ear
s
o
f
p
r
o
fes
s
i
o
n
a
l
ex
p
er
i
en
ce
i
n
t
h
e
acad
emi
c
fi
el
d
.
Sh
e
can
b
e
c
o
n
t
act
e
d
at
emai
l
:
p
i
n
ak
s
h
i
p
a
n
d
a@
g
ma
i
l
.
co
m.
Suka
nt
Ki
s
ho
ro
Bi
s
o
y
recei
v
ed
a
D
o
ct
o
r
o
f
Ph
i
l
o
s
o
p
h
y
(P
h
.
D
.
)
d
eg
ree,
a
Mas
t
er
o
f
T
ec
h
n
o
l
o
g
y
(M.
T
ech
.
)
d
eg
ree,
an
d
a
b
ac
h
el
o
r's
d
eg
ree
i
n
c
o
mp
u
t
er
e
n
g
i
n
eer
i
n
g
i
n
2
0
1
7
,
2
0
0
3
,
an
d
2
0
0
0
,
res
p
ec
t
i
v
el
y
.
H
e
p
res
e
n
t
l
y
o
ccu
p
i
e
s
t
h
e
p
ro
fe
s
s
o
r
an
d
d
ea
n
ro
l
e
w
i
t
h
i
n
t
h
e
D
ep
ar
t
men
t
o
f
Co
m
p
u
t
er
Sci
en
ce
a
n
d
E
n
g
i
n
eer
i
n
g
at
C.
V
.
Raman
G
l
o
b
al
U
n
i
v
er
s
i
t
y
i
n
In
d
i
a.
T
h
e
c
u
rren
t
res
earc
h
i
n
t
eres
t
s
o
f
t
h
e
i
n
d
i
v
i
d
u
al
en
c
o
mp
a
s
s
t
h
e
f
i
el
d
s
o
f
n
eu
r
o
-
ro
b
o
t
i
c
s
,
mach
i
n
e
l
earn
i
n
g
,
cl
o
u
d
co
m
p
u
t
i
n
g
,
an
d
s
o
f
t
w
are
d
efi
n
e
d
n
et
w
o
r
k
s
.
H
e
h
a
s
b
een
d
es
i
g
n
at
e
d
as
a
Marg
d
ars
h
ak
b
y
t
h
e
A
l
l
-
In
d
i
a
Co
u
n
c
i
l
fo
r
T
ec
h
n
i
cal
E
d
u
ca
t
i
o
n
/
Mi
n
i
s
t
r
y
o
f
H
u
man
Res
o
u
rce
D
ev
e
l
o
p
men
t
.
H
e
h
as
o
v
er
1
0
0
p
u
b
l
i
cat
i
o
n
s
i
n
SCI
E
,
Sco
p
u
s
j
o
u
rn
a
l
s
,
a
n
d
c
o
n
fere
n
ces
.
H
e
can
b
e
co
n
t
ac
t
ed
at
ema
i
l
:
s
u
k
a
n
t
a
b
i
s
o
y
i
@
y
a
h
o
o
.
co
m.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
A
nt
li
on
and
ant
c
olony
opti
miz
ati
on
int
e
gr
ated
e
ns
e
mble
mac
hine
…
(
P
inaks
hi
P
anda
)
613
A
m
rut
a
ns
hu
P
a
n
i
g
r
a
hi
i
s
cu
rre
n
t
l
y
w
o
r
k
i
n
g
as
an
as
s
i
s
t
a
n
t
p
ro
fe
s
s
o
r
at
t
h
e
D
ep
t
art
me
n
t
o
f
C
o
mp
u
t
er
Sc
i
en
ce
an
d
E
n
g
i
n
eer
i
n
g
,
Si
k
s
h
a
O
A
n
u
s
a
n
d
h
an
(D
eemed
t
o
b
e
U
n
i
v
er
s
i
t
y
),
Bh
u
b
a
n
es
w
ar,
O
d
i
s
h
a,
I
n
d
i
a.
H
e
o
b
t
a
i
n
e
d
M.
T
ech
.
i
n
i
n
fo
rma
t
i
o
n
t
ech
n
o
l
o
g
y
fro
m
t
h
e
C
o
l
l
eg
e
o
f
E
n
g
i
n
eeri
n
g
a
n
d
T
ec
h
n
o
l
o
g
y
,
G
o
v
t
.
o
f
O
d
i
s
h
a,
an
d
a
B.
T
ec
h
.
fro
m
BPU
T
O
d
i
s
h
a
i
n
2
0
1
4
an
d
2
0
1
1
,
res
p
ec
t
i
v
e
l
y
.
H
e
co
m
p
l
e
t
e
d
h
i
s
Ph
.
D
.
i
n
co
m
p
u
t
er
s
c
i
en
ce
an
d
en
g
i
n
eer
i
n
g
w
i
t
h
S
i
k
s
h
a
‘
O
’
A
n
u
s
an
d
h
a
n
U
n
i
v
er
s
i
t
y
i
n
2
0
2
3
.
H
i
s
re
s
earch
i
n
t
eres
t
s
i
n
c
l
u
d
e
b
l
o
ck
c
h
ai
n
,
ML
,
an
d
D
L
.
H
e
h
as
o
v
er
5
0
p
u
b
l
i
ca
t
i
o
n
s
i
n
SCIE
,
Sco
p
u
s
j
o
u
r
n
al
s
,
an
d
co
n
fere
n
ces
.
H
e
can
b
e
co
n
t
ac
t
ed
at
amr
u
t
a
n
s
u
p
8
9
@
g
m
ai
l
.
co
m.
A
bhi
l
a
s
h
P
a
t
i
i
s
cu
rre
n
t
l
y
w
o
r
k
i
n
g
as
an
a
s
s
i
s
t
an
t
p
r
o
fes
s
o
r
i
n
t
h
e
D
ep
ar
t
men
t
o
f
Co
mp
u
t
er
Sci
en
ce
an
d
E
n
g
i
n
eer
i
n
g
,
FE
T
-
I
T
E
R,
Si
k
s
h
a
‘O
’
A
n
u
s
an
d
h
a
n
(D
eemed
t
o
b
e
U
n
i
v
er
s
i
t
y
),
Bh
u
b
a
n
es
w
ar,
In
d
i
a.
H
e
c
o
mp
l
et
e
d
h
i
s
B.
T
ech
.
an
d
M.
T
ec
h
.
i
n
co
m
p
u
t
er
s
c
i
en
ce
an
d
en
g
i
n
eeri
n
g
w
i
t
h
t
h
e
BPU
T
,
O
d
i
s
h
a,
In
d
i
a,
i
n
2
0
0
9
an
d
2
0
1
2
,
res
p
ec
t
i
v
el
y
,
an
d
h
i
s
Ph
.
D
.
in
c
o
mp
u
t
er
s
ci
e
n
ce
an
d
en
g
i
n
eeri
n
g
w
i
t
h
S
i
k
s
h
a
‘
O
’
A
n
u
s
a
n
d
h
an
U
n
i
v
er
s
i
t
y
,
Bh
u
b
a
n
es
w
ar,
In
d
i
a
i
n
2
0
2
3
.
H
i
s
res
earc
h
i
n
t
ere
s
t
s
i
n
cl
u
d
e
t
h
e
I
o
T
,
fo
g
co
mp
u
t
i
n
g
,
ML
,
an
d
D
L
,
an
d
h
e
h
as
mo
re
t
h
a
n
5
0
p
u
b
l
i
ca
t
i
o
n
s
i
n
SCI
E
,
E
SCI,
Sco
p
u
s
i
n
d
e
x
ed
j
o
u
r
n
al
s
an
d
/
o
r
co
n
feren
ces
t
o
h
i
s
cred
i
t
.
H
e
can
b
e
co
n
t
a
ct
e
d
at
er.
ab
h
i
l
a
s
h
.
p
at
i
@
g
mai
l
.
co
m.
Evaluation Warning : The document was created with Spire.PDF for Python.