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.
624
~
634
I
S
S
N:
2088
-
8708
,
DO
I
:
10
.
11591/i
jec
e
.
v
15
i
1
.
pp
6
24
-
634
624
Jou
r
n
al
h
omepage
:
ht
tp:
//
ij
e
c
e
.
iaes
c
or
e
.
c
om
Hyb
r
id
o
p
t
imiz
at
io
n
al
gor
it
h
m
f
or
a
n
al
ysi
s of
i
n
f
lu
e
n
c
e
p
r
o
p
agat
io
n
i
n
so
c
ia
l
n
e
t
w
or
k
Aks
h
at
a
S
an
d
e
e
p
B
h
ayyar
1
,
2
,
Ki
r
an
P
u
r
u
s
h
ot
h
a
m
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, R
a
ma
i
a
h I
ns
ti
tu
te
of
T
e
c
hnol
ogy, a
f
f
il
ia
te
d t
o
V
is
ve
s
v
a
r
a
ya
T
e
c
hnol
ogi
c
a
l
U
ni
ve
r
s
it
y, B
e
la
ga
vi
, 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, R
N
S
I
ns
ti
t
ut
e
of
T
e
c
hnol
ogy, B
e
ng
a
lu
r
u, a
f
f
il
ia
te
d t
o V
is
ve
s
va
r
a
ya
T
e
c
hn
ol
ogi
c
a
l
U
ni
ve
r
s
it
y, B
e
la
ga
vi
, I
ndi
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
M
a
r
4,
2024
R
e
vis
e
d
J
ul
8,
2024
Ac
c
e
pted
Oc
t
1,
2024
In
fl
u
en
ce
max
i
mi
za
t
i
o
n
(IM)
i
s
d
ef
i
n
e
d
as
t
h
e
p
r
o
b
l
em
o
f
i
d
e
n
t
i
fy
i
n
g
a
n
o
d
e
s
u
b
s
et
i
n
a
s
o
c
i
al
n
et
w
o
r
k
w
h
i
c
h
i
n
creas
es
t
h
e
s
p
rea
d
o
f
i
n
fl
u
e
n
ce.
IM
p
l
a
y
s
a
cru
ci
a
l
ro
l
e
i
n
s
o
ci
a
l
n
e
t
w
o
rk
s
b
y
cat
a
l
y
z
i
n
g
t
h
e
d
i
s
s
em
i
n
a
t
i
o
n
o
f
i
n
f
l
u
e
n
ce,
res
u
l
t
i
n
g
i
n
an
au
g
men
t
ed
co
u
n
t
o
f
i
n
f
l
u
e
n
ced
n
o
d
es
fo
l
l
o
w
i
n
g
t
h
e
p
ro
p
ag
a
t
i
o
n
p
ro
ce
s
s
.
T
h
e
ex
i
s
t
i
n
g
res
earc
h
es
mai
n
l
y
co
n
ce
n
t
ra
t
ed
o
n
i
n
crea
s
i
n
g
t
h
e
s
p
rea
d
o
f
i
n
fl
u
en
ce,
b
u
t
d
i
d
n
o
t
co
n
s
i
d
er
t
h
e
r
u
n
n
i
n
g
t
i
me
o
f
t
h
e
n
e
t
w
o
rk
.
In
t
h
i
s
man
u
s
cr
i
p
t
,
t
h
e
s
al
p
s
w
arm
a
l
g
o
r
i
t
h
m
(SSA
)
an
d
bi
-
a
d
ap
t
i
v
e
s
t
ra
t
eg
y
p
art
i
cl
e
s
w
arm
o
p
t
i
m
i
zat
i
o
n
(Bi
A
S
-
PSO
)
al
g
o
r
i
t
h
ms
are
i
n
t
eg
ra
t
ed
a
n
d
n
ame
d
as
SS
-
Bi
A
S
-
PSO
a
l
g
o
ri
t
h
m
t
o
i
n
creas
e
t
h
e
s
p
rea
d
o
f
i
n
f
l
u
e
n
ce
b
as
e
d
o
n
t
h
e
IM
p
ro
b
l
em
t
o
mi
n
i
m
i
ze
t
h
e
ru
n
n
i
n
g
t
i
me
o
f
t
h
e
n
et
w
o
r
k
.
T
h
e
d
at
a
s
et
s
u
t
i
l
i
ze
d
fo
r
t
h
e
re
s
earch
are
E
g
o
-
Faceb
o
o
k
,
E
p
i
n
i
o
n
s
,
G
o
w
al
l
a,
an
d
H
e
p
T
h
,
w
h
i
l
e
l
i
n
ear
t
h
res
h
o
l
d
(L
T
)
i
s
u
t
i
l
i
zed
a
s
a
d
i
ff
u
s
i
o
n
met
h
o
d
.
T
h
en
,
t
h
e
p
ro
p
o
s
ed
SS
-
B
i
A
S
-
PS
O
al
g
o
r
i
t
h
m
i
s
d
ep
l
o
y
ed
f
o
r
t
h
e
an
al
y
s
i
s
o
f
i
n
f
l
u
e
n
ce
p
r
o
p
a
g
at
i
o
n
.
T
h
e
p
ro
p
o
s
ed
a
l
g
o
r
i
t
h
m
reach
e
s
a
h
i
g
h
i
n
f
l
u
e
n
ce
s
p
rea
d
o
f
6
4
5
,
6
8
0
,
7
1
5
,
an
d
7
5
0
w
i
t
h
l
es
s
ru
n
n
i
n
g
t
i
mes
res
p
ec
t
i
v
el
y
fo
r
1
0
,
2
0
,
3
0
,
an
d
4
0
s
eed
s
et
s
i
zes
i
n
E
g
o
-
Face
b
o
o
k
.
T
h
e
p
ro
p
o
s
ed
al
g
o
r
i
t
h
m
p
r
o
v
e
s
mo
re
effec
t
i
v
e
t
h
an
t
h
e
ex
i
s
t
i
n
g
t
ech
n
i
q
u
e
s
l
i
k
e
t
rad
i
t
i
o
n
al
SSA
a
n
d
p
ar
t
i
c
l
e
s
w
arm
o
p
t
i
m
i
zat
i
o
n
(
PSO
)
.
K
e
y
w
o
r
d
s
:
I
nf
luenc
e
maximi
z
a
ti
on
I
nf
luenc
e
pr
opa
ga
ti
on
L
inea
r
thr
e
s
hold
P
a
r
ti
c
le
s
wa
r
m
opti
mi
z
a
ti
on
S
a
lp
s
wa
r
m
a
lgor
it
hm
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
:
Aks
ha
ta
S
a
nde
e
p
B
ha
yya
r
De
pa
r
tm
e
nt
of
C
omput
e
r
S
c
ienc
e
a
nd
E
nginee
r
ing
,
R
a
maia
h
I
ns
ti
tut
e
o
f
T
e
c
hnology,
a
f
f
il
iate
d
to
Vis
ve
s
va
r
a
ya
T
e
c
hnologi
c
a
l
Unive
r
s
it
y
B
e
laga
vi,
I
ndia
De
pa
r
tm
e
nt
of
C
omput
e
r
S
c
ienc
e
a
nd
E
nginee
r
ing
,
R
NS
I
ns
ti
tut
e
of
T
e
c
hnology
B
e
nga
lur
u
,
a
f
f
il
iate
d
to
Vis
ve
s
va
r
a
ya
T
e
c
hnologi
c
a
l
Unive
r
s
it
y
B
e
laga
vi,
I
ndia
E
mail:
a
ks
ha
taphd@
gmail.
c
om
1.
I
NT
RODU
C
T
I
ON
W
it
h
the
quick
de
ve
lopm
e
nt
o
f
s
oc
ial
ne
twor
ks
,
m
a
ny
pe
ople
a
r
e
us
ing
W
e
C
ha
t,
T
witt
e
r
,
F
a
c
e
book,
a
nd
va
r
ious
s
oc
ial
s
of
twa
r
e
to
c
a
r
r
y
out
da
ta
e
xc
ha
nge
,
pr
omo
ti
on
o
f
pr
oduc
ts
,
opini
on
of
the
publi
c
,
a
nd
f
o
r
va
r
ious
other
a
c
ti
vit
ies
that
br
ing
c
onve
nienc
e
f
or
the
pr
oduc
ti
ve
li
f
e
o
f
pe
ople
[
1]
,
[
2
]
.
T
he
in
ter
a
c
ti
on
a
mong
indi
viduals
c
a
n
in
f
luenc
e
the
s
pr
e
a
d
in
s
o
c
ial
ne
twor
ks
[
3
]
,
a
nd
the
e
videnc
e
c
onf
ir
ms
that
da
ta
or
dis
tr
ibut
ion
of
inf
luenc
e
is
e
f
f
icie
nt
in
c
e
r
tain
pr
a
c
ti
c
a
l
a
ppli
c
a
ti
ons
li
ke
f
r
e
s
h
pr
oduc
t
pr
omot
ion
[
4]
.
T
he
main
pr
oblem
is
to
identi
f
y
a
tar
ge
t
s
e
t
with
be
tt
e
r
c
omm
unica
ti
on
c
ha
r
a
c
ter
is
ti
c
s
with
the
s
uppor
t
of
a
r
e
lations
hip
ne
twor
k
de
ve
loped
be
twe
e
n
us
e
r
s
r
e
c
e
ivi
ng
the
p
r
oduc
t
whi
le
a
tt
a
ini
ng
high
pr
omo
ti
on
of
the
pr
oduc
t,
a
nd
thi
s
p
r
oc
e
dur
e
is
a
n
inf
luenc
e
on
the
maximi
z
a
ti
on
pr
oblem
[
5]
,
[
6]
.
He
nc
e
,
a
na
lys
is
o
f
be
ha
vior
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
Hy
br
id
opti
miz
ati
on
algor
it
hm
for
analys
is
of
inf
lu
e
nc
e
pr
opagati
on
…
(
A
k
s
hata
Sande
e
p
B
hay
y
ar
)
625
a
nd
s
oc
ial
ne
twor
k
s
tr
uc
tur
e
c
ha
r
a
c
ter
is
ti
c
s
pr
ovi
de
s
a
theor
e
ti
c
a
l
ba
s
e
f
or
s
olut
ions
to
va
r
ious
s
oc
ial
a
nd
e
c
onomi
c
is
s
ue
s
[
7]
.
I
nf
luenc
e
maximi
z
a
ti
on
(
I
M
)
thr
e
a
ts
a
r
e
a
v
it
a
l
pa
r
t
of
the
a
na
lys
is
of
s
oc
ial
ne
twor
ks
a
nd
a
r
e
one
of
the
c
r
uc
ial
is
s
ue
s
in
s
oc
ial
ne
twor
k
s
[
8]
,
[
9]
.
I
M
pr
ob
lem
is
de
f
ined
a
s
c
hoos
ing
o
f
a
gr
oup
o
f
us
e
r
s
f
r
om
s
oc
ial
ne
twor
k
to
incr
e
a
s
e
the
pr
e
dicta
ble
number
of
a
f
f
e
c
ted
us
e
r
s
[
10]
.
R
e
c
e
ntl
y,
many
a
lgor
it
hms
f
or
the
p
r
oblem
of
in
f
luenc
e
maximi
z
a
ti
on
r
e
s
e
a
r
c
h
ha
ve
a
bs
tr
a
c
ted
s
oc
ial
ne
twor
ks
a
s
s
tatic
s
tr
uc
tur
e
s
,
a
voidi
ng
the
f
a
c
t
that
int
e
r
a
c
ti
on
a
m
ong
us
e
r
s
c
ha
nge
s
ove
r
ti
me
[
11
]
,
[
12]
.
I
n
the
pr
e
vious
r
e
s
e
a
r
c
he
s
,
the
a
mount
pa
id
f
o
r
mes
s
a
ge
s
pr
e
a
ding
in
the
pr
oc
e
s
s
of
in
f
luenc
e
maxim
iza
ti
on
wa
s
c
ons
ider
e
d
a
s
r
a
r
e
[
13]
.
He
nc
e
,
the
s
e
lec
ti
on
of
a
mi
nim
u
m
-
c
os
t
s
e
e
d
node
gr
oup
to
a
c
quir
e
inf
luenc
e
maximi
z
a
ti
on
of
a
node
is
a
majo
r
is
s
ue
ye
t
to
be
s
olved.
T
he
ti
m
e
uti
li
z
e
d
f
or
inf
luenc
e
p
r
opa
ga
ti
on
in
s
oc
ial
ne
t
wor
ks
is
maximum
in
the
pr
e
vious
methods
,
whic
h
a
ls
o
ne
e
ds
to
be
r
e
s
olved
[
14]
,
[
15]
.
T
o
ove
r
c
ome
thes
e
li
mi
tations
a
nd
s
olve
I
M
pr
oblems
,
it
is
r
e
qui
r
e
d
to
de
ve
lop
ne
w
s
wa
r
m
int
e
ll
igenc
e
-
ba
s
e
d
a
lgor
it
hms
f
or
thi
s
pr
oblem.
T
he
s
e
a
lgor
it
hms
leve
r
a
ge
diver
s
it
y
a
nd
loca
l
de
ve
lopm
e
nt
c
a
pa
bil
it
ies
of
a
n
a
lgor
it
hm,
e
f
f
icie
ntl
y
a
ddr
e
s
s
ing
I
M
pr
oblem
while
mi
nim
izing
the
r
un
ning
ti
me.
T
he
r
e
a
r
e
two
wa
ys
in
s
wa
r
m
in
telli
ge
nc
e
-
ba
s
e
d
a
lgor
it
hms
f
or
s
olvi
ng
I
M
pr
oblems
whic
h
a
r
e
de
ve
lopi
ng
of
a
n
objec
ti
ve
f
unc
ti
on
,
a
nd
r
e
nde
r
i
ng
of
a
n
e
nha
nc
e
d
pe
r
f
or
manc
e
by
the
opti
m
iza
ti
on
a
lgo
r
it
hm.
I
n
thi
s
r
e
s
e
a
r
c
h,
the
s
a
lp
s
wa
r
m
a
lgor
i
thm
(
S
S
A)
a
nd
bi
-
a
da
pti
ve
s
tr
a
tegy
pa
r
ti
c
le
s
wa
r
m
opti
mi
z
a
ti
on
(
B
iAS
-
P
S
O)
a
lgor
it
hms
a
r
e
int
e
gr
a
ted
to
incr
e
a
s
e
t
he
s
pr
e
a
d
of
inf
luenc
e
ba
s
e
d
on
the
I
M
p
r
oblem
a
nd
mi
nim
iz
e
the
r
unning
ti
me
o
f
the
ne
twor
k.
L
i
e
t
a
l
.
[
16
]
im
pl
e
m
e
n
te
d
a
n
a
ge
n
t
-
ba
s
e
d
e
vo
lu
t
ion
a
r
y
mo
de
l
(
AB
E
M
)
f
o
r
in
f
lue
nc
in
g
m
a
x
im
iz
a
t
io
n
i
n
s
oc
ia
l
n
e
t
wo
r
k
s
.
I
n
it
ia
ll
y
,
th
e
m
od
e
l
us
e
d
a
di
s
t
r
ib
u
ted
m
e
t
ho
d
t
h
r
o
ug
h
a
n
e
nh
a
nc
e
d
ge
ne
ti
c
a
lg
or
i
thm
(
GA
)
t
o
a
dd
r
e
s
s
I
M
i
n
s
o
c
ia
l
n
e
t
wo
r
k
s
.
T
he
i
mp
lem
e
n
te
d
m
od
e
l
w
a
s
in
te
g
r
a
ted
wi
th
G
A
a
n
d
a
c
t
iv
i
ty
-
b
a
s
e
d
m
od
e
l
(
A
B
M
)
f
o
r
o
pt
i
mi
z
i
ng
s
e
e
ds
s
e
l
e
c
ti
on
f
r
o
m
tw
o
s
ta
ge
s
a
t
th
e
in
d
iv
id
ua
l
a
nd
g
l
oba
l
lev
e
ls
.
T
h
e
i
m
pl
e
m
e
n
ted
me
th
od
no
t
on
l
y
pe
r
f
o
r
me
d
we
ll
bu
t
a
ls
o
ha
nd
led
hu
ge
-
s
c
a
l
e
s
oc
ia
l
ne
tw
o
r
ks
b
y
d
is
tr
ibu
t
in
g
t
he
e
xe
c
u
ti
on
c
os
t
.
Ho
we
v
e
r
,
t
he
i
m
ple
me
nt
e
d
me
th
od
di
d
n
ot
c
o
ns
i
de
r
t
he
e
f
f
e
c
t
o
f
o
ve
r
l
a
p
pi
ng
c
a
us
e
d
by
the
c
hos
e
n
h
i
gh
c
e
n
tr
a
l
no
de
s
i
n
t
he
s
e
e
d
g
r
ou
p
t
ha
t
de
g
r
a
de
th
e
me
t
ho
ds
e
f
f
ic
ie
nt
.
Z
ha
n
g
e
t
a
l
.
[
17
]
in
tr
od
uc
e
d
a
n
o
ve
r
la
pp
in
g
c
o
mm
un
i
ty
-
ba
s
e
d
pa
r
ti
c
l
e
s
wa
r
m
opt
i
mi
z
a
ti
on
(
OC
P
S
O
)
a
lg
o
r
i
th
m
f
o
r
t
he
m
a
x
im
iza
t
io
n
o
f
i
n
f
lue
nc
e
i
n
s
oc
ia
l
n
e
t
wo
r
ks
.
T
he
in
t
r
o
duc
e
d
me
th
od
u
ti
li
z
e
d
ov
e
r
la
pp
i
ng
,
n
on
-
ov
e
r
la
pp
i
ng
a
nd
in
te
r
a
c
t
i
ve
d
a
ta
n
o
de
s
.
P
a
r
t
icu
la
r
ly
,
a
n
a
l
go
r
it
h
m
o
f
ov
e
r
la
pp
i
ng
c
o
mm
un
it
y
de
te
c
t
io
n
wa
s
ut
il
iz
e
d
t
o
a
c
qu
i
r
e
d
a
ta
of
o
ve
r
la
pp
in
g
c
o
mm
un
i
ty
s
t
r
uc
t
u
r
e
s
.
F
u
r
t
he
r
,
de
pe
nd
in
g
on
th
r
e
e
s
t
r
a
teg
ies
o
f
e
vo
lu
t
io
na
r
y
i
ni
t
ia
l
iza
ti
on
,
m
u
ta
ti
on
a
n
d
loc
a
l
s
e
a
r
c
h
de
v
e
l
op
e
d
in
OC
P
S
O
,
the
i
nf
l
ue
n
t
ia
l
n
od
e
s
we
r
e
s
u
pe
r
io
r
ly
id
e
n
ti
f
ie
d
.
H
owe
ve
r
,
th
e
i
n
t
r
o
duc
e
d
a
lg
or
i
th
m
di
d
no
t
c
on
t
r
o
l
t
he
s
ol
u
ti
on
a
c
c
u
r
a
c
y
w
e
l
l
,
a
nd
i
nc
r
e
a
s
e
d
the
n
e
t
wo
r
k’
s
r
u
nn
in
g
t
im
e
.
K
u
ik
ka
e
t
a
l
.
[
1
8
]
p
r
e
s
e
n
te
d
tw
o
pr
og
r
a
m
mi
ng
m
e
th
o
ds
a
nd
t
he
ir
c
or
r
e
s
p
on
di
ng
ps
e
ud
o
-
a
lg
o
r
i
th
ms
to
a
na
l
yz
e
c
om
p
lex
n
e
t
wo
r
ks
f
or
in
f
lue
nc
e
max
i
mi
z
a
ti
on
i
n
s
oc
ia
l
ne
tw
o
r
ks
.
B
o
t
h
me
th
ods
r
e
p
r
e
s
e
n
te
d
t
he
n
e
t
wo
r
k
s
t
r
uc
tu
r
e
o
n
a
d
e
ta
i
led
l
e
ve
l
.
T
he
s
e
t
wo
a
l
go
r
it
h
ms
de
p
e
n
de
d
o
n
s
im
il
a
r
in
f
l
ue
nc
e
s
p
r
e
a
d
in
g
m
e
th
ods
r
e
p
r
e
s
e
nt
e
d
to
be
c
om
b
ine
d
to
mea
s
u
r
e
the
s
p
r
e
a
d
in
g
p
r
oba
b
il
it
ies
a
m
on
g
t
he
no
de
s
pa
i
r
.
T
he
m
e
th
ods
c
on
ta
in
e
d
mu
lt
i
pl
e
u
ni
que
a
n
d
a
c
tu
a
l
f
e
a
t
u
r
e
s
,
but
we
r
e
i
ne
f
f
e
c
ti
ve
f
or
m
os
t
r
e
a
l
-
wo
r
ld
ne
t
wo
r
k
s
iz
e
s
.
T
he
p
r
e
s
e
nte
d
a
lg
o
r
i
th
m
s
how
e
d
s
up
e
r
i
or
s
c
a
lab
i
li
ty
a
n
d
p
e
r
f
o
r
m
a
n
c
e
b
y
t
he
id
e
n
ti
f
ica
t
io
n
o
f
in
f
lue
nc
e
no
de
s
.
H
ow
e
ve
r
,
t
he
ne
tw
o
r
k
s
i
z
e
a
n
d
c
ou
n
t
o
f
i
nd
iv
i
dua
l
s
w
e
r
e
ma
xi
m
ize
d
,
be
c
a
us
e
m
e
a
s
u
r
i
ng
t
he
c
os
t
of
c
e
n
t
r
a
l
v
a
l
ue
wa
s
hu
ge
a
n
d
t
he
me
th
od
wa
s
h
ig
hl
y
ti
me
-
c
o
ns
um
in
g
.
D
ua
n
e
t
a
l
.
[
19
]
s
ug
ge
s
ted
a
mu
lt
i
-
h
op
r
e
mo
ve
(
M
HR
)
a
l
go
r
it
hm
f
o
r
ma
xi
mi
z
a
ti
on
o
f
in
f
lu
e
n
c
e
i
n
s
oc
i
a
l
ne
tw
or
ks
.
T
h
e
s
u
gge
s
te
d
a
lg
or
i
th
m
de
te
r
mi
ne
d
a
h
op
r
a
nge
un
de
r
v
a
r
i
ous
p
r
o
ba
b
i
li
ti
e
s
o
f
p
r
op
a
ga
t
io
n
.
T
he
c
o
mp
le
xi
ty
o
f
t
im
e
c
a
n
be
hi
gh
l
y
m
in
im
iz
e
d
if
nod
e
s
m
e
e
t
t
he
r
e
q
u
ir
e
m
e
n
ts
of
d
i
r
e
c
t
ly
c
h
os
e
n
.
T
h
e
s
ug
ge
s
ted
a
lg
o
r
i
th
m
e
n
ha
nc
e
d
in
f
lue
nc
e
s
p
r
e
a
d
a
n
d
m
in
im
ize
d
t
he
r
ic
h
c
lu
bs
’
in
te
r
f
e
r
e
nc
e
.
H
owe
ve
r
,
th
e
s
u
gg
e
s
t
e
d
a
l
go
r
it
h
m
c
o
ns
u
me
d
a
h
ig
h
r
u
nn
in
g
ti
me
f
o
r
hi
gh
-
s
c
a
l
e
ne
tw
o
r
ks
.
H
e
e
t
a
l
.
[
20
]
de
ve
lo
pe
d
d
yn
a
m
ic
o
pi
ni
on
m
a
x
im
iza
t
io
n
a
lg
o
r
i
th
m
w
it
h
th
e
hy
br
i
d
mo
de
l
f
o
r
m
a
x
im
iza
t
io
n
of
i
n
f
lue
nc
e
in
s
oc
i
a
l
ne
t
wo
r
ks
.
T
he
d
e
ve
l
ope
d
a
lg
or
i
t
hm
c
hos
e
s
e
e
d
no
de
s
t
ha
t
i
nc
l
ud
e
d
c
om
mu
n
it
y
de
te
c
t
io
n
,
a
nd
d
e
t
e
r
m
in
a
t
io
n
o
f
no
de
s
i
n
c
a
nd
ida
te
s
e
e
d
,
a
lo
n
gs
i
de
a
s
e
e
di
ng
a
lg
o
r
i
th
m
w
it
h
dis
c
r
e
t
e
pa
r
ti
c
le
s
w
a
r
m
o
p
ti
mi
z
a
ti
on
.
T
he
de
ve
lo
pe
d
a
lg
or
i
th
m
h
a
d
a
n
a
d
va
n
ta
ge
ov
e
r
t
he
s
e
l
e
c
ted
be
nc
hm
a
r
ks
on
th
e
m
e
a
n
o
pi
n
io
ns
o
f
a
c
t
iv
a
t
e
d
no
de
s
.
No
ne
the
les
s
,
wh
e
n
a
h
ug
e
n
umb
e
r
o
f
no
de
s
we
r
e
id
e
n
ti
f
ie
d
i
n
the
s
e
a
r
c
h
pr
oc
e
s
s
o
f
t
he
d
e
v
e
l
op
e
d
me
th
od
,
it
r
e
s
u
lt
e
d
in
hu
ge
c
om
pu
ta
ti
on
a
l
c
os
ts
.
Kuma
r
e
t
al.
[
21]
im
pleme
nted
a
modi
f
ied
de
gr
e
e
with
e
xc
lus
ion
r
a
ti
o
(
M
DE
R
)
method
f
or
inf
luenc
e
maximi
z
a
ti
on
in
s
oc
ial
ne
twor
ks
.
T
he
i
mpl
e
mente
d
method
identi
f
ied
a
n
inf
luential
node
in
s
oc
ial
ne
twor
ks
by
modi
f
ied
de
gr
e
e
c
e
ntr
a
li
ty
c
onc
e
pt
a
n
d
e
xc
lus
ion
of
mut
ua
l.
T
he
im
pleme
nted
a
lgor
i
thm
uti
li
z
e
d
ten
r
e
a
l
-
li
f
e
ne
twor
ks
of
dif
f
e
r
e
nt
a
ppli
c
a
ti
ons
,
c
ompl
e
xit
y
a
nd
s
ize
.
T
he
im
pleme
nted
method
pr
ovided
s
upe
r
ior
a
nd
highl
y
diver
s
e
s
olut
ions
.
How
e
ve
r
,
the
method
did
not
tes
t
huge
-
s
c
a
le
ne
twor
ks
a
n
d
lac
ke
d
theor
e
ti
c
a
l
r
e
s
is
tanc
e
to
the
r
e
lations
hip
be
twe
e
n
inf
luenc
e
a
nd
f
a
ir
ne
s
s
.
F
a
n
e
t
al.
[
22]
int
r
o
duc
e
d
a
dis
c
r
e
ti
z
e
s
the
Ha
r
r
is
H
a
wka
opti
mi
z
a
ti
on
(
DH
H
O)
a
lgor
it
hm
f
or
I
M
in
s
oc
ial
ne
twor
ks
.
De
pe
ndi
ng
on
the
s
ix
de
gr
e
e
s
of
s
e
pa
r
a
ti
on
theor
y
in
s
oc
ial
ne
two
r
ks
,
huge
a
c
c
ur
a
te
a
nd
c
omm
on
objec
ti
ve
f
unc
ti
o
ns
we
r
e
de
ve
loped
to
mea
s
ur
e
the
s
e
e
d
node
s
'
inf
luenc
e
.
T
he
int
r
oduc
e
d
method
uti
li
z
e
d
dis
c
r
e
te
c
oding
a
nd
e
ne
r
gy,
whe
r
e
in
the
pos
it
ion
pr
e
s
e
ntation
r
ules
we
r
e
r
e
de
f
i
ne
d
a
nd
then
given
to
in
f
luenc
e
the
maximi
z
a
ti
on
pr
oblem.
T
he
int
r
oduc
e
d
method
c
ove
r
e
d
inf
luenc
e
quickly
with
huge
a
c
c
ur
a
c
y.
How
e
ve
r
,
the
inf
luenc
e
s
pr
e
a
d
of
the
int
r
oduc
e
d
a
lgor
it
hm
wa
s
s
mall.
T
a
ng
e
t
al
.
[
23
]
s
ugge
s
ted
a
dis
c
r
e
te
s
c
he
duled
pa
r
ti
c
le
s
w
a
r
m
opti
mi
z
a
ti
on
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
:
624
-
634
626
(
DSP
S
O)
a
lgor
i
thm
f
or
the
maxim
iza
ti
on
of
inf
lu
e
nc
e
in
s
oc
ial
ne
twor
ks
.
T
he
s
ugge
s
ted
a
lgor
it
hm
c
hos
e
a
n
opti
mum
s
ize
of
node
s
to
s
e
e
d
the
g
r
oup
in
e
ve
r
y
r
ound
to
e
ns
ur
e
the
c
onti
nua
ti
on
of
the
s
pr
e
a
ding
pr
oc
e
dur
e
.
T
o
make
the
whole
e
xplor
a
ti
on
of
the
s
olut
ion
s
pa
c
e
,
the
s
tr
a
tegy
of
loca
l
s
e
a
r
c
h,
pa
r
ti
c
ular
ly
f
or
dis
c
r
e
te
ne
twor
k
topol
ogy
wa
s
de
ve
loped
on
the
b
e
s
t
s
wa
r
m
indi
viduals
.
T
he
s
ugge
s
ted
a
lgor
it
hm
ha
d
a
high
inf
luenc
e
s
pr
e
a
d.
Ye
t,
r
e
pe
a
ted
s
e
lec
ti
on
of
nod
e
s
c
ons
umed
mor
e
ti
me
in
the
s
ugge
s
ted
a
lgor
it
hm.
T
he
e
xis
ti
ng
a
lgor
it
hms
ha
ve
li
mi
tations
noted
a
s
:
no
c
ons
ider
a
ti
on
of
the
e
f
f
e
c
t
of
ove
r
lapping
c
a
us
e
d
by
c
hos
e
n
huge
c
e
ntr
a
l
node
s
in
the
s
e
e
d
s
e
t,
ther
e
by
a
f
f
e
c
ti
ng
it
s
e
f
f
icie
nc
y.
T
he
a
lgor
it
hms
we
r
e
no
t
a
ble
to
c
o
ntr
ol
the
s
olut
ion
a
c
c
ur
a
c
y
e
f
f
icie
ntl
y
,
r
e
s
ult
ing
in
incr
e
a
s
e
d
ne
twor
k
r
unning
ti
me
with
s
mall
in
f
luenc
e
s
pr
e
a
d.
Additi
ona
ll
y,
many
e
xis
ti
ng
a
lgo
r
it
hms
ha
ve
dr
a
w
ba
c
ks
in
ba
lanc
ing
e
f
f
icie
nc
y
a
nd
e
f
f
e
c
ti
ve
ne
s
s
to
va
r
ious
e
xtents
.
I
n
thi
s
r
e
s
e
a
r
c
h,
the
p
r
opos
e
d
S
S
-
B
iAS
-
P
S
O
a
lgor
it
hm
ha
s
a
global
s
e
a
r
c
h
a
bil
it
y
to
e
s
c
a
pe
loca
l
opti
mal
in
inf
luenc
e
maximi
z
a
ti
on
pr
opa
ga
ti
on
(
I
M
P
)
.
M
or
e
ove
r
,
the
a
lgor
it
h
m
p
r
ovides
a
n
a
dva
ntage
in
opti
mum
a
c
c
ur
a
c
y
a
nd
t
im
e
c
os
t.
T
his
hyb
r
id
meth
od
take
s
be
ne
f
it
o
f
the
opti
mi
z
a
ti
on
p
r
oc
e
s
s
a
nd
e
f
f
icie
nc
y
of
t
im
e
that
e
na
bles
the
f
e
a
tur
e
s
of
ne
two
r
k
dyn
a
mi
c
s
to
a
dd
r
e
s
s
I
M
P
in
huge
-
s
c
a
le
s
oc
ial
ne
twor
ks
.
T
he
main
c
ontr
ibut
ions
o
f
the
r
e
s
e
a
r
c
h
a
r
e
given
be
low
:
a.
B
a
s
e
d
on
the
a
c
ti
va
ti
on
s
tatus
with
li
ne
a
r
thr
e
s
hol
d
(
L
T
)
model
,
a
n
a
c
ti
va
ted
op
ini
on
model
is
de
ve
loped.
T
he
f
our
r
e
a
l
-
ti
me
da
tas
e
ts
uti
li
z
e
d
f
or
the
a
na
lys
is
of
inf
luenc
e
p
r
opa
ga
ti
on
in
s
oc
ial
ne
twor
ks
a
r
e
:
E
go
-
F
a
c
e
book,
E
pini
ons
,
Gow
a
ll
a
a
nd
He
pT
h
.
b.
S
S
A
a
nd
B
iAS
-
P
S
O
a
lgor
it
hms
a
r
e
in
tegr
a
ted
to
i
nc
r
e
a
s
e
the
s
pr
e
a
d
of
in
f
luenc
e
ba
s
e
d
on
I
M
p
r
obl
e
m
to
identif
y
inf
luential
node
s
in
s
oc
ial
ne
twor
ks
thr
oug
h
modi
f
ying
the
s
e
e
d
s
e
t
va
lues
.
c.
T
he
pe
r
f
or
manc
e
of
the
p
r
opos
e
d
a
lgor
it
hms
is
a
na
lyze
d
by
inf
luenc
e
s
pr
e
a
d
a
nd
r
unning
ti
me
of
the
ne
twor
k
in
te
r
ms
of
va
r
ious
it
e
r
a
ti
ons
.
T
he
r
e
s
t
of
the
r
e
s
e
a
r
c
h
is
or
ga
nize
d
in
the
f
oll
o
wing
f
or
mat:
s
e
c
ti
on
2
e
xplains
the
de
tails
o
f
the
pr
opos
e
d
a
lgor
it
hm.
S
e
c
ti
on
3
de
s
ignate
s
the
r
e
s
ult
s
a
nd
dis
c
us
s
ion
of
the
pr
opos
e
d
a
lgo
r
it
hm
.
F
inally,
s
e
c
ti
on
4
pr
e
s
e
nts
the
c
onc
lus
ion.
2.
P
ROP
OS
E
D
M
E
T
HO
D
I
n
thi
s
s
e
c
ti
on,
the
da
tas
e
t
uti
li
z
e
d
f
or
r
e
s
e
a
r
c
h
a
nd
the
pr
oblem
of
I
M
in
s
oc
ial
ne
twor
ks
is
de
s
c
r
ibed.
T
his
r
e
s
e
a
r
c
h
f
oc
us
e
s
on
the
L
T
method,
one
of
the
two
pr
im
a
r
y
inf
luenc
e
dif
f
us
ion
metho
ds
,
with
the
other
be
ing
the
indepe
nde
nt
c
a
s
c
a
de
method.
At
las
t,
the
p
r
oc
e
s
s
of
the
p
r
opos
e
d
hybr
id
SS
-
B
iAS
-
P
S
O
a
lgor
it
hm
is
de
s
c
r
ibed.
2.
1.
Dat
as
e
t
T
he
da
tas
e
ts
uti
li
z
e
d
f
or
r
e
s
e
a
r
c
h
a
r
e
f
our
r
e
a
l
-
wor
ld
s
oc
ial
ne
twor
ks
with
dis
ti
nc
t
c
ha
r
a
c
ter
is
ti
c
s
.
T
he
f
our
r
e
a
l
-
wor
ld
da
tas
e
ts
a
r
e
:
E
go
-
F
a
c
e
boo
k,
E
pini
ons
,
Gow
a
ll
a
,
a
nd
He
pT
h
[
24]
.
E
go
-
F
a
c
e
book,
Gow
a
ll
a
a
nd
He
pT
h
a
r
e
r
e
f
e
r
r
e
d
to
a
s
“
c
oll
a
bor
a
ti
on
ne
twor
ks
,”
while
E
pini
ons
is
r
e
f
e
r
r
e
d
to
a
s
a
“
tr
us
t
ne
twor
k.
”
T
a
ble
1
gives
the
s
tatis
ti
c
a
l
da
ta
of
f
ou
r
da
tas
e
ts
including
the
type
o
f
ne
twor
k
,
number
o
f
node
s
,
number
of
e
dge
s
in
the
ne
twor
k
,
a
nd
the
de
s
c
r
ipt
io
n
of
ne
two
r
k.
T
a
ble
1.
Da
tas
e
t
d
e
s
c
r
ipt
ion
D
a
ta
s
e
ts
N
ode
s
E
dge
s
T
ype
s
D
e
s
c
r
ip
ti
on
E
go
-
F
a
c
e
book
4,039
88,234
U
ndi
r
e
c
te
d
S
oc
ia
l
c
ir
c
le
s
on F
a
c
e
book
E
pi
ni
ons
75,879
508,837
D
ir
e
c
te
d
W
ho t
r
us
ts
w
hom a
ne
twor
k of
E
p.c
om
G
ow
a
ll
a
196,591
950,327
U
ndi
r
e
c
te
d, G
e
o
-
lo
c
a
ti
on
G
ow
a
ll
a
lo
c
a
ti
on
-
ba
s
e
d onli
ne
s
oc
ia
l
ne
twor
k
H
e
pT
h
27,770
352,807
D
ir
e
c
te
d, T
e
mpor
a
l
la
be
le
d
A
r
X
iv
hi
gh e
ne
r
gy phys
ic
s
pa
pe
r
c
it
a
ti
on ne
twor
k
2.
2.
I
n
f
lu
e
n
c
e
m
axim
iza
t
ion
(
I
M
)
I
M
is
the
p
r
oc
e
s
s
of
s
e
lec
ti
ng
a
gr
oup
of
node
s
in
a
s
oc
ial
ne
twor
k
s
o
the
major
indi
viduals
a
r
e
inf
luenc
e
d
by
them
[
25]
.
T
he
s
oc
ial
ne
twor
k
i
s
de
s
c
r
ibed
a
s
a
we
ight
e
d
gr
a
ph
=
(
,
,
)
,
the
r
e
pr
e
s
e
nts
a
gr
oup
of
node
s
,
r
e
pr
e
s
e
nts
a
gr
oup
of
e
dge
s
a
mong
node
s
in
that
e
xpr
e
s
s
r
e
lations
hip
a
mong
two
us
e
r
s
(
,
)
is
int
e
gr
a
ted
with
we
ight
(
,
)
e
dge
a
nd
that
r
e
pr
e
s
e
nts
inf
luenc
e
of
on
.
C
ons
ider
ing
the
s
oc
ial
ne
twor
k
=
(
,
,
)
a
nd
is
pos
it
ive
int
e
ge
r
,
the
I
M
is
s
ue
a
im
s
to
identif
y
the
s
e
e
d
s
e
t
∗
with
node
s
thr
ough
s
e
t
a
s
to
incr
e
a
s
e
the
s
p
e
e
d
of
inf
luenc
e
(
)
be
low
a
given
dif
f
us
ion
method.
T
he
mathe
matica
l
f
or
mul
a
f
o
r
I
M
is
given
a
s
(
1)
.
I
n
(
1)
,
r
e
pr
e
s
e
nts
the
c
hos
e
n
s
e
e
d
s
e
t,
∗
r
e
pr
e
s
e
nts
the
opti
mum
s
e
t
of
s
e
e
d
node
s
f
or
I
M
,
a
nd
(
)
r
e
pr
e
s
e
nts
the
pr
e
dicta
ble
c
ount
o
f
in
f
luenc
e
d
node
s
.
∗
=
(
)
⊆
,
|
|
=
(
1)
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
Hy
br
id
opti
miz
ati
on
algor
it
hm
for
analys
is
of
inf
lu
e
nc
e
pr
opagati
on
…
(
A
k
s
hata
Sande
e
p
B
hay
y
ar
)
627
2.
3.
M
e
t
h
od
s
f
or
d
i
f
f
u
s
ion
T
he
s
pr
e
a
d
methods
ut
il
ize
d
f
o
r
I
M
pr
ob
lems
maj
or
ly
include
indepe
nde
nt
c
a
s
c
a
de
(
I
C
)
a
nd
li
ne
a
r
thr
e
s
hold
(
L
T
)
methods
.
I
n
a
L
T
method
,
e
ve
r
y
dir
e
c
ted
e
dge
(
,
)
∈
of
s
oc
ial
ne
twor
k
is
int
e
gr
a
ted
with
the
r
e
s
pe
c
ti
ve
we
ight
(
,
)
∈
[
0
,
1
]
.
He
r
e
,
(
,
)
de
s
c
r
ibes
the
r
a
t
io
of
us
e
r
’
s
inf
luenc
e
on
us
e
r
be
twe
e
n
their
ne
ighbor
s
.
Additi
ona
ll
y
,
e
ve
r
y
node
is
int
e
gr
a
ted
with
thr
e
s
hold
(
)
∈
[
0
,
1
]
.
T
his
thr
e
s
hold
is
de
ter
mi
ne
d,
with
no
a
bil
it
y
to
modi
f
y
the
pr
oc
e
s
s
of
s
pr
e
a
ding,
mea
ning
that
the
node
is
a
c
ti
va
ted
while
the
we
ight
e
d
s
um
of
whole
a
c
ti
va
ted
node
s
in
the
ne
ig
hbor
is
higher
than
or
e
qua
l
to
thr
e
s
hold
(
)
.
I
ni
ti
a
ll
y,
the
s
e
e
d
node
is
a
c
ti
ve
a
nd
va
r
ious
node
s
a
r
e
inac
ti
ve
.
Af
ter
wa
r
d,
the
node
is
s
ti
mul
a
ted
to
dis
tur
b
the
ne
ighbor
node
s
,
a
nd
the
a
f
o
r
e
mentioned
pr
oc
e
s
s
is
r
e
pe
a
ted.
T
he
p
r
oc
e
s
s
of
s
pr
e
a
d
is
s
topped
whe
n
the
inf
luen
c
e
,
s
um
the
a
c
ti
ve
node
s
in
the
p
r
e
vious
a
c
ti
ve
node
s
of
the
ne
twor
k
that
c
a
n
not
a
c
ti
va
te
the
inac
ti
ve
ne
ighbo
r
node
s
.
I
n
L
T
method,
node
thr
e
s
hold
is
node
a
c
c
e
ptanc
e
to
objec
t
pr
opa
ga
ti
ng
in
the
pr
e
s
e
nt
ne
twor
k.
T
he
objec
t
include
s
inf
o
r
mation,
p
r
oduc
ts
,
whe
r
e
a
s
ingl
e
node
is
not
e
nough
to
a
c
ti
va
te
the
node
,
but
the
whole
inf
luenc
e
of
va
r
ious
node
s
a
c
ti
va
tes
.
W
hil
e
ne
w
objec
ts
a
r
e
pr
opa
ga
ted
in
the
s
oc
ial
ne
twor
ks
,
us
e
r
s
r
e
quir
e
a
s
igni
f
ica
nt
c
ount
of
f
r
iends
a
nd
r
e
latives
to
a
c
c
e
pt
objec
ts
.
He
nc
e
,
node
is
a
c
ti
va
ted
thr
ough
the
ge
ne
r
a
l
inf
luenc
e
of
r
e
c
e
ivi
ng
ne
ighbor
s
.
T
his
ge
ne
r
a
l
inf
luenc
e
tr
a
ns
f
or
mation
is
a
s
e
t
be
ha
vior
that
f
r
e
que
ntl
y
oc
c
ur
s
in
the
s
oc
iety
while
f
a
c
ing
dif
f
ic
ult
s
e
lec
ti
ons
.
2.
4.
P
r
op
os
e
d
B
i
-
ad
ap
t
ive
s
t
r
at
e
gy
-
p
ar
t
icle
s
war
m
op
t
i
m
izat
ion
(
B
iAS
-
P
S
O)
algorit
h
m
I
n
thi
s
r
e
s
e
a
r
c
h,
including
GA
mut
a
ti
on
s
tr
a
tegy
a
nd
metr
opol
is
c
r
it
e
r
ia
of
s
im
ulate
d
a
nne
a
li
ng
(
S
A)
,
a
B
iAS
-
P
S
O
is
p
r
opos
e
d
f
or
im
pr
ov
ing
the
a
bil
it
y
o
f
global
a
nd
loca
l
s
e
a
r
c
h
of
P
S
O.
I
n
P
S
O,
a
ne
w
s
tr
a
tegy
known
a
s
bi
-
a
da
pti
ve
i
s
c
r
e
a
ted
a
mong
s
wa
r
m
a
nd
indi
viduals
,
whe
r
e
e
ve
r
y
pa
r
ti
c
le
be
lo
ng
s
to
a
bi
-
a
da
pti
ve
s
tr
a
tegy.
T
he
de
tailed
e
xplana
ti
on
of
b
i
-
a
da
pti
ve
s
tr
a
tegy
is
de
s
c
r
ibed
in
the
be
low
s
e
c
ti
ons
.
2.
5.
Bi
-
ad
ap
t
ive
s
t
r
at
e
gy
I
n
biol
ogica
l
be
ha
vior
,
the
indi
v
iduals
lea
r
ne
d
the
e
xpe
r
ienc
e
thr
ough
population,
that
tende
d
to
c
r
e
a
te
e
xpe
r
ienc
e
f
r
om
r
e
s
tr
icte
d
c
ount
o
f
g
r
a
ph
n
ode
s
,
a
s
a
n
outcome
,
population
be
c
ome
much
di
ve
r
s
e
.
I
n
P
S
O,
a
ne
w
s
tr
a
tegy
known
a
s
the
bi
-
a
da
pti
ve
c
r
e
a
ted
a
mong
s
wa
r
m
a
nd
indi
viduals
is
pr
opos
e
d.
He
r
e
,
e
ve
r
y
pa
r
ti
c
le
be
long
s
to
a
bi
-
a
da
pti
ve
s
tr
a
tegy.
T
he
mat
he
matica
l
f
or
mul
a
f
or
pa
r
ti
c
le
s
pe
e
d
of
P
S
O
is
giv
e
n
a
s
(
2)
.
+
1
=
+
1
1
(
−
)
+
2
2
(
−
)
+
3
3
(
(
)
−
)
(
2)
W
he
r
e
,
(
)
r
e
pr
e
s
e
nts
whic
h
two
a
da
pti
ve
s
tr
a
tegie
s
t
he
ℎ
pa
r
ti
c
le
be
longed
to
,
a
nd
(
)
r
e
pr
e
s
e
nts
the
opti
mum
loca
ti
on
o
f
bi
-
a
da
pti
ve
s
tr
a
tegy.
T
h
e
(
3
3
)
is
s
a
me
a
s
(
1
1
)
a
nd
(
2
2
)
.
W
it
h
bi
-
a
da
pti
ve
s
tr
a
tegy
a
dde
d,
the
population
main
taine
d
a
c
ha
r
a
c
ter
is
ti
c
of
e
ve
r
y
bi
-
s
tr
a
tegy
on
the
pr
oc
e
s
s
of
it
e
r
a
ti
on
upda
ti
on
whic
h
maximi
z
e
s
the
diver
s
it
y
o
f
popula
ti
on
a
nd
e
a
r
ly
c
onve
r
ge
nc
e
o
f
P
S
O
in
r
e
s
olvi
ng
t
he
huge
-
dim
e
ns
ional
is
s
ue
s
.
T
he
mathe
matica
l
f
or
mul
a
o
f
iner
ti
a
we
ight
im
pr
ove
s
the
loca
l
s
e
a
r
c
h
pe
r
f
o
r
m
a
nc
e
,
a
s
given
in
(
3
)
.
I
n
(
3)
,
a
nd
de
notes
the
maximum
a
nd
mi
nim
um
s
c
or
e
s
.
de
notes
the
highes
t
number
of
it
e
r
a
ti
ons
.
(
)
=
−
−
(
3)
2.
5.
1.
M
u
t
a
t
ion
s
t
r
a
t
e
gy
T
he
ge
ne
mut
a
ti
on
is
pe
r
f
or
med
while
the
pa
r
e
nt
pr
oduc
e
d
ne
xt
ge
ne
r
a
ti
on
f
or
a
ll
owing
c
hil
dr
e
n
ge
ne
r
a
ti
on
that
ha
s
high
s
e
a
r
c
h
a
bil
it
y.
M
utation
ope
r
a
ti
on
is
im
pleme
nted
in
P
S
O
a
lgor
it
hm
th
r
ough
the
mut
a
ti
on
pos
s
ibi
li
ty
ope
r
a
tor
=
−
⁄
.
Af
ter
the
c
onve
nt
ional
upda
te
of
P
S
O,
a
r
a
ndom
number
a
mong
0
a
nd
1
is
pr
oduc
e
d
a
nd
c
ompar
e
d
to
f
o
r
e
ve
r
y
pa
r
ti
c
le.
W
he
n
the
r
a
ndom
number
is
les
s
e
r
than
,
2
dim
e
ns
ions
a
r
e
r
a
ndoml
y
c
hos
e
n
f
or
e
ve
r
y
pa
r
t
icle
a
nd
r
a
ndom
ini
t
ializa
ti
on
is
pe
r
f
or
med.
At
las
t
,
the
f
it
ne
s
s
s
c
or
e
of
mut
a
ti
on
pa
r
ti
c
le
is
mea
s
ur
e
d.
I
f
i
mpr
ove
d,
the
mut
a
ti
on
ope
r
a
ti
on
is
ke
pt
;
or
e
ls
e
,
th
e
pa
r
ti
c
le
is
c
a
c
he
d
a
nd
mut
a
ti
on
is
maintaine
d
,
or
e
ls
e
de
ter
mi
ne
d
thr
ough
a
s
s
igni
ng
the
metr
opoli
s
c
r
i
ter
ia.
2.
5.
2.
M
e
t
r
op
oli
s
c
r
it
e
r
ia
T
he
M
e
tr
opoli
s
c
r
it
e
r
ia
is
the
main
s
tr
a
tegy
of
S
A
a
lgor
it
hm.
As
a
ne
w
s
olut
ion
is
p
r
oduc
e
d
f
r
o
m
the
pr
e
vious
one
,
the
c
r
it
e
r
ia
a
r
e
a
c
c
e
pt
e
d
de
pe
nd
ing
on
the
f
it
ne
s
s
va
r
ianc
e
a
mong
the
ne
w
a
nd
pr
e
vious
s
olut
ions
.
M
e
tr
opoli
s
c
r
it
e
r
ia
ha
ve
a
n
a
ddit
ional
pa
r
a
mete
r
tempe
r
a
tur
e
.
At
high
tempe
r
a
tur
e
,
gr
e
a
ter
the
a
c
c
e
pted
pos
s
ibi
li
ty,
the
pa
r
ti
c
le
is
a
ll
owe
d
to
ha
ve
the
high
a
bil
it
y
of
global
s
e
a
r
c
h
a
t
ini
ti
a
l
s
e
a
r
c
h.
T
he
mathe
matica
l
f
or
mul
a
f
o
r
mea
s
ur
ing
the
pos
s
ibi
l
it
y
of
a
c
c
e
pti
ng
mut
a
ti
on
is
given
a
s
(
4)
.
I
n
(
4)
,
(
)
de
notes
a
n
a
s
s
e
s
s
ment
f
unc
ti
on
a
nd
de
notes
the
pos
s
ibi
li
ty.
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
:
624
-
634
628
=
{
1
,
(
,
)
<
(
,
+
1
)
(
,
+
1
)
−
(
,
)
,
(
,
)
<
(
,
+
1
)
(
4)
2.
6.
P
r
op
os
e
d
s
alp
s
war
m
-
bi
-
a
d
ap
t
ive
s
t
r
at
e
gy
p
ar
t
icle
s
war
m
op
t
i
m
izat
ion
(
S
S
-
B
iAS
-
P
S
O)
alg
or
it
h
m
T
he
hybr
idi
z
a
ti
on
of
the
S
S
A
a
nd
B
iAS
-
P
S
O
is
us
e
d
f
or
the
a
na
lys
is
of
in
f
luenc
e
pr
opa
ga
ti
on
in
s
oc
ial
ne
twor
ks
.
T
he
S
S
A
is
mor
e
a
ble
of
ga
ini
n
g
the
a
c
c
ur
a
c
y
in
dif
f
e
r
e
nt
pr
e
s
e
nted
meta
he
ur
is
ti
c
s
.
T
he
dr
a
wba
c
k
of
S
S
A
is
whic
h
ge
ts
t
r
a
ppe
d
to
a
glob
a
l
or
loca
l
opti
mal,
whic
h
is
un
f
it
f
or
high
di
f
f
icu
lt
ies
a
nd
ha
s
a
low
c
onve
r
ge
nc
e
r
a
te,
diver
s
it
y
a
nd
pr
e
-
c
onve
r
ge
nc
e
.
T
o
r
e
duc
e
thes
e
we
a
kne
s
s
es
a
nd
im
p
r
ove
the
s
e
a
r
c
ha
bil
it
y
of
S
S
A
a
lgo
r
it
hm,
the
be
ne
f
it
s
of
the
P
S
O
a
lgor
it
h
m
[
26
]
,
[
27]
a
r
e
c
ombi
ne
d
a
nd
a
ne
w
a
lgor
it
hm
,
SS
-
B
iAS
-
P
S
O
is
pr
opos
e
d.
T
he
e
xploi
t
a
ti
on
a
nd
e
xplor
a
ti
on
of
S
S
A
a
r
e
im
p
r
ove
d
thr
oug
h
B
iAS
-
P
S
O
to
de
ve
lop
a
hybr
id
S
S
A
-
B
iAS
-
P
S
O
a
lgor
it
hm.
T
he
c
ur
r
e
nt
a
lgor
it
hm
f
inds
the
e
f
f
e
c
ti
ve
va
lue
of
dif
f
icult
opti
mi
z
a
ti
on
pr
oc
e
s
s
.
T
he
B
iAS
-
P
S
O
a
lgor
it
hm
pha
s
e
is
p
r
oc
e
s
s
e
d
in
e
xplor
ing
the
opti
mum
s
olut
ion
ve
c
tor
s
.
T
he
r
e
f
or
e
,
SS
-
B
iAS
-
P
S
O
a
lgor
it
hm
is
e
leva
ted
a
s
a
loca
l
s
e
a
r
c
h
tec
hnique
f
o
r
e
nha
nc
ing
the
domi
na
nc
e
of
opti
mal
s
olut
ion
.
T
he
pr
opos
e
d
me
thod
is
he
lpf
ul
in
quickly
t
r
a
pping
s
olut
ion
o
f
th
e
global
opti
ma
a
nd
ignor
ing
a
loca
l
op
ti
mum
in
s
e
a
r
c
h
a
r
e
a
dur
ing
the
s
e
a
r
c
h
pr
oc
e
dur
e
.
T
he
r
e
f
o
r
e
,
the
p
r
opos
e
d
a
lgor
it
hm
s
uppor
ts
the
c
a
pa
bil
it
y
to
s
e
a
r
c
h
a
nd
obt
a
in
c
or
r
e
c
t
c
onve
r
ge
nc
e
s
by
a
c
c
e
ler
a
ti
ng
s
e
a
r
c
h.
2.
6.
1.
I
n
it
ial
izat
ion
I
n
the
s
e
a
r
c
hing
pr
oc
e
s
s
of
a
n
a
lgor
i
thm
,
the
s
e
a
r
c
h
s
tage
r
a
ndoml
y
ini
ti
a
li
z
e
s
the
c
r
owd
f
oll
owing
the
given
c
r
it
e
r
ia
,
whe
r
e
in
the
a
lgor
it
hm
e
mpl
oys
a
r
a
ndom
ve
c
tor
of
d
im
e
ns
ions
f
or
ℎ
s
a
lp
=
~
(
=
1
,
2
,
3
,
…
,
)
.
T
he
pa
r
a
mete
r
s
uti
li
z
e
d
f
or
ini
ti
a
li
z
a
ti
on
a
r
e
po
pulation
s
ize
(
)
,
iner
ti
a
we
ight
(
)
,
lea
r
n
ing
pa
r
a
mete
r
s
(
1
,
2
)
a
nd
maximum
ve
locity
(
)
.
T
a
ble
2
s
hows
the
pa
r
a
mete
r
s
a
nd
their
r
a
nge
of
opti
mi
z
a
ti
on
a
lgor
it
hms
.
T
a
ble
2.
P
a
r
a
mete
r
s
P
a
r
a
me
te
r
s
R
a
nge
P
opul
a
ti
on s
iz
e
[
100, 500]
I
ne
r
ti
a
w
e
ig
ht
[
0.1, 1.0]
L
e
a
r
ni
ng pa
r
a
me
te
r
s
[
0, 2]
M
a
xi
mum
ve
lo
c
it
y
[
0, 3]
2.
6.
2.
E
valu
at
io
n
of
f
i
t
n
e
s
s
f
u
n
c
t
io
n
F
it
ne
s
s
va
lue
of
e
a
c
h
s
e
a
r
c
h
a
ge
nt
is
e
va
luate
d
t
hr
ough
an
objec
ti
ve
f
unc
ti
on
,
while
e
a
c
h
s
e
a
r
c
h
a
ge
nt
f
ur
ther
c
ons
ider
s
ne
w
loca
ti
ons
by
f
it
ne
s
s
va
lue
s
in
a
s
e
a
r
c
h
s
pa
c
e
.
T
he
opt
im
iza
ti
on
a
lg
or
it
hm
is
ini
ti
a
li
z
e
d
with
a
gr
oup
o
f
r
a
ndom
pa
r
ti
c
les
(
i.
e
.
,
s
olut
ions
)
,
then
s
e
a
r
c
h
ing
f
o
r
a
n
op
ti
mum
s
olut
ion
thr
ough
upda
ti
ng
ge
ne
r
a
ti
ons
.
I
n
e
ve
r
y
i
ter
a
ti
on,
a
ll
pa
r
ti
c
l
e
s
a
r
e
upda
ted
with
the
ne
xt
be
s
t
va
lues
.
T
he
ini
ti
a
l
one
is
a
good
s
olut
ion
(
i
.
e
.
,
f
it
ne
s
s
)
,
a
nd
is
r
e
pr
e
s
e
nted
a
s
.
Ne
xt,
it
s
va
lue
is
then
obtaine
d
th
r
ough
the
op
ti
mi
z
e
r
by
a
ny
pa
r
ti
c
le
in
the
popu
lation,
s
igni
f
ied
a
s
.
2.
6.
3.
Upd
a
t
in
g
leader
p
os
it
ion
T
he
loca
ti
on
of
a
main
s
e
a
r
c
h
a
ge
nt
li
ke
a
lea
de
r
i
s
r
e
pr
e
s
e
nted
by
(
5)
a
nd
(
6)
f
or
the
s
e
a
r
c
h
pr
oc
e
s
s
in
the
s
e
a
r
c
h
s
pa
c
e
.
T
he
lea
de
r
loca
ti
on
is
uti
li
z
e
d
by
the
mathe
matica
l
(
2
)
,
=
{
+
1
(
(
−
)
2
+
)
3
≥
0
.
5
−
1
(
(
−
)
2
+
)
3
<
0
.
5
(
5)
w
he
r
e
,
de
s
c
r
ibes
the
loca
ti
on
o
f
s
upe
r
io
r
a
nd
pr
oba
ble
s
olut
ion
s
,
de
s
c
r
ibes
the
uppe
r
bound
of
dim
e
ns
ion
,
de
s
c
r
ibes
the
lowe
r
bound
of
dim
e
ns
ion
,
de
s
c
r
ibes
the
pos
it
ion
of
the
f
ood
s
our
c
e
,
2
a
nd
3
de
s
c
r
ibe
the
2
r
a
ndom
number
s
in
a
r
a
nge
[
0,
1
]
,
1
de
s
c
r
ibes
the
im
por
tant
va
r
iable
in
the
a
lgo
r
it
h
m
that
gr
a
dua
ll
y
mi
nim
ize
s
ove
r
ge
ne
r
a
ti
ons
,
a
ll
owi
ng
a
va
s
t
e
xplo
r
a
ti
on
a
t
a
n
ini
ti
a
l
s
tage
of
the
opti
mi
z
a
ti
on
pr
oc
e
dur
e
.
T
he
numer
ica
l
e
xpr
e
s
s
ion
of
1
is
mentio
ne
d
(
6)
.
He
r
e
,
de
s
c
r
ibes
the
maximum
it
e
r
a
ti
ons
a
nd
r
e
pr
e
s
e
nts
the
c
ur
r
e
nt
it
e
r
a
ti
ons
.
1
=
2
−
(
4
)
2
(
6)
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
Hy
br
id
opti
miz
ati
on
algor
it
hm
for
analys
is
of
inf
lu
e
nc
e
pr
opagati
on
…
(
A
k
s
hata
Sande
e
p
B
hay
y
ar
)
629
2.
6.
4.
Ve
locit
y
in
it
ia
li
z
at
ion
T
he
ve
locity
ini
t
ializa
ti
on
is
a
s
igni
f
ica
nt
pa
r
t
o
f
population
-
ba
s
e
d
a
lgor
it
hms
a
nd
with
the
he
lp
o
f
the
de
ve
loped
a
lgo
r
it
hm,
i
t
is
pos
s
ibl
e
to
mi
n
im
i
z
e
the
e
f
f
or
ts
of
s
e
a
r
c
h
a
ge
nt
on
the
s
e
a
r
c
h
pr
oc
e
dur
e
in
s
e
a
r
c
h
s
pa
c
e
,
a
nd
a
void
incor
r
e
c
t
pos
it
ions
.
M
a
n
y
ti
mes
,
in
a
s
e
a
r
c
h
pr
oc
e
s
s
,
thes
e
s
e
a
r
c
h
a
ge
nts
lea
ve
the
pr
ojec
ted
bounda
r
y
of
the
s
e
a
r
c
h
a
r
e
a
whic
h
obs
tr
uc
ts
the
identif
ying
e
ne
r
gy
a
nd
pr
ovides
les
s
a
lgor
it
hm
a
c
c
ur
a
c
y
to
f
ind
a
global
opti
mum
,
a
nd
the
r
e
f
or
e
,
the
pr
opos
e
d
a
lgor
it
hm
plays
a
s
igni
f
ica
nt
r
ole
in
tr
a
pping
a
n
e
f
f
e
c
ti
ve
global
opti
mal
s
olut
ion
a
nd
a
voidi
ng
l
oc
a
l
opti
mal
in
s
e
a
r
c
h
s
pa
c
e
.
T
he
ini
ti
a
li
z
a
ti
on
of
ve
locity
is
pr
oc
e
s
s
e
d
in
3
dif
f
e
r
e
nt
wa
ys
,
i.
e
.
,
i)
s
mall
r
a
n
dom
va
lue
is
ini
ti
a
li
z
e
d,
ii
)
r
a
ndom
va
lues
ne
a
r
to
z
e
r
o
a
r
e
ini
ti
a
li
z
e
d
a
nd
a
t
las
t
,
a
nd
ii
i
)
z
e
r
o
is
in
it
ialize
d.
V
a
r
ious
ini
ti
a
li
z
a
ti
on
s
tage
s
im
pa
c
t
the
a
lgor
it
hm
a
c
c
ur
a
c
y
in
s
e
ve
r
a
l
wa
ys
.
2.
6.
5.
P
os
it
ion
u
p
d
at
in
g
of
f
oll
owe
r
T
he
f
oll
owe
r
pos
it
ion
in
the
s
e
a
r
c
h
s
pa
c
e
of
the
s
e
a
r
c
h
pr
oc
e
s
s
is
c
ha
nge
d
by
mathe
matica
l
(
7)
.
+
1
=
×
+
1
×
(
−
)
(
7)
whe
r
e
,
ve
locity
plays
a
s
igni
f
ica
nt
pa
r
t
in
tr
a
pp
ing
the
global
opti
mum
quickly
a
nd
a
voidi
ng
in
c
or
r
e
c
t
pos
it
ions
on
the
s
e
a
r
c
h
pr
oc
e
dur
e
.
2.
6.
6.
S
t
op
p
in
g
c
r
it
e
r
ia
At
las
t,
the
s
toppi
ng
c
r
it
e
r
ia
a
r
e
given
to
s
e
a
r
c
h
th
e
global
opti
mum
f
o
r
e
ve
r
y
k
ind
o
f
is
s
ue
s
uc
h
a
s
f
a
ll
ing
int
o
the
loca
l
opti
mal
,
s
low
c
onve
r
ge
nc
e
r
a
te,
a
nd
s
o
on
.
He
r
e
,
the
c
r
it
e
r
ia
a
r
e
uti
li
z
e
d
f
or
e
s
ti
mating
e
ve
r
y
s
e
a
r
c
h
a
ge
nt
us
e
d
in
the
pr
oc
e
dur
e
a
nd
r
e
plac
ing
the
s
upe
r
ior
pos
it
ion
of
the
s
e
a
r
c
h
a
ge
nt,
a
nd
thi
s
is
r
e
pe
a
ted
unti
l
the
s
toppi
ng
c
r
it
e
r
ia
a
r
e
s
a
ti
s
f
ied.
T
he
r
e
maining
pr
oc
e
s
s
is
s
im
il
a
r
to
the
s
a
lp
s
wa
r
m
a
lgor
it
hm.
F
ig
ur
e
1
r
e
p
r
e
s
e
nts
the
pr
oc
e
s
s
of
the
pr
opos
e
d
S
S
-
B
iAS
-
P
S
O
a
lgor
it
hm.
F
igur
e
1.
P
r
oc
e
s
s
of
pr
opos
e
d
SS
-
T
S
APS
O
a
lgor
it
hm
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
:
624
-
634
630
T
he
pr
opos
e
d
S
S
-
B
iAS
-
P
S
O
a
lgor
it
hm
plays
a
s
i
gnif
ica
nt
r
ole
in
tr
a
pping
a
s
olut
ion
in
the
s
e
a
r
c
h
s
pa
c
e
dur
ing
the
pr
oc
e
s
s
of
s
e
a
r
c
h
a
nd
the
be
ne
f
it
s
a
r
e
given
be
low:
a.
T
he
pr
opos
e
d
S
S
-
B
iAS
-
P
S
O
a
lgor
it
hm
holds
a
s
up
e
r
ior
opti
mu
m
s
olut
ion
a
f
ter
e
ve
r
y
it
e
r
a
ti
on
.
b.
T
he
pr
opos
e
d
S
S
-
B
iAS
-
P
S
O
upda
tes
the
pos
it
ion
of
e
ve
r
y
s
e
a
r
c
h
a
ge
nt
in
the
s
e
a
r
c
h
s
pa
c
e
with
the
a
im
of
the
s
e
a
r
c
h
a
ge
nt
be
ing
to
e
xploi
t
a
nd
e
xplo
r
e
in
the
s
e
a
r
c
h
s
pa
c
e
f
or
a
be
tt
e
r
opti
mal
s
olut
ion.
c.
T
he
SS
-
B
iAS
-
P
S
O
a
lgor
it
hm
upda
tes
a
pos
it
ion
o
f
the
f
oll
owe
r
’
s
a
ge
nt
with
a
s
uppor
t
o
f
ve
locity
.
T
his
plays
a
s
igni
f
ica
nt
r
ole
in
a
voidi
ng
a
n
incor
r
e
c
t
loca
ti
on
a
nd
quick
t
r
a
pping
of
global
opti
mum
i
n
the
s
e
a
r
c
h
s
pa
c
e
.
I
t
a
ls
o
im
pr
ove
s
ba
lanc
e
a
mong
e
xpl
oit
a
ti
on
a
nd
e
xplor
a
ti
on
.
d.
T
he
de
li
be
r
a
te
us
e
o
f
s
low
moveme
nts
by
the
f
o
ll
owe
r
s
e
a
r
c
h
a
ge
nt
du
r
ing
s
e
a
r
c
h
in
the
s
e
a
r
c
h
s
pa
c
e
s
e
r
ve
s
to
s
a
f
e
gua
r
d
the
S
S
-
B
iAS
-
P
S
O
a
lgor
it
hm
a
ga
ins
t
de
s
c
e
nding
to
loca
l
opti
ma.
e.
T
he
s
igni
f
ica
nt
pa
r
a
mete
r
of
the
S
S
A
a
lgor
i
thm
s
uppor
ts
the
S
S
-
B
iAS
-
P
S
O
a
lgor
it
hm
in
mi
nim
izin
g
the
c
ompl
e
xit
y,
making
it
e
f
f
or
tl
e
s
s
f
or
e
xe
c
uti
on.
I
n
the
dif
f
us
ion
pr
oc
e
s
s
,
the
pr
opa
ga
ted
da
ta
is
a
dopted
thr
ough
c
e
r
tain
node
s
in
the
ne
two
r
k.
T
he
node
s
’
obtaine
d
da
ta
is
pr
opa
ga
ted
f
ur
ther
a
t
s
om
e
r
a
te
de
f
ined
by
their
popular
i
ty.
T
he
obtaine
d
p
a
r
a
mete
r
r
e
pr
e
s
e
nts
the
e
xtent
to
whic
h
da
ta
s
pr
e
a
d
s
in
the
n
e
twor
k.
3.
RE
S
UL
T
S
AN
D
DI
S
CU
S
S
I
ON
T
he
pe
r
f
or
manc
e
of
the
pr
opos
e
d
S
S
-
B
iAS
-
P
S
O
a
lgor
it
hm
is
im
pleme
nted
by
P
ython
e
nvir
onment
with
the
f
oll
owing
s
ys
tem
r
e
quir
e
ments
:
R
AM
:
16
GB
,
p
r
oc
e
s
s
or
:
I
ntel
c
or
e
i7
,
a
nd
ope
r
a
t
ing
s
ys
tem
:
W
indows
10
(
64
bit
)
.
T
he
pe
r
f
or
manc
e
of
the
pr
o
pos
e
d
a
lgor
it
hm
is
e
s
ti
mate
d
in
ter
ms
o
f
the
pe
r
f
or
manc
e
mea
s
ur
e
s
of
inf
luenc
e
s
pr
e
a
d
a
nd
r
unning
ti
me
o
f
the
ne
twor
k
with
d
if
f
e
r
e
nt
it
e
r
a
ti
ons
.
Va
r
ious
ta
bles
a
nd
gr
a
phs
a
r
e
de
s
c
r
ibed
to
s
how
the
e
f
f
e
c
ti
ve
ne
s
s
of
p
r
opos
e
d
a
lgor
it
hm.
3
.
1.
Qu
an
t
it
a
t
ive
an
d
q
u
al
it
at
ive
an
alys
is
T
he
pe
r
f
or
manc
e
of
the
pr
opos
e
d
a
lgor
it
hm
is
a
na
lyze
d
with
inf
luenc
e
s
pr
e
a
d
a
nd
r
unning
ti
me
of
a
ne
twor
k
on
f
our
da
tas
e
ts
uti
li
z
e
d
f
or
r
e
s
e
a
r
c
h.
T
he
e
xis
ti
ng
a
lgor
it
hms
c
ons
ider
e
d
f
or
e
va
luation
a
r
e
the
wha
le
opti
mi
z
a
ti
on
a
lgor
it
hm
(
W
OA
)
,
g
r
e
y
wolf
opti
mi
z
a
ti
on
(
GW
O)
,
s
a
lp
s
wa
r
m
a
lgor
it
hm
(
S
S
A)
,
a
nd
pa
r
ti
c
le
s
wa
r
m
opti
mi
z
a
ti
on
(
P
S
O)
a
lgor
it
hms
.
Va
r
ious
table
s
a
nd
f
igur
e
s
a
r
e
r
e
pr
e
s
e
nted
be
low
to
s
how
that
the
pr
opos
e
d
a
lgor
it
hm
is
pe
r
f
o
r
med
s
upe
r
ior
ly
.
T
a
ble
3
a
nd
F
igu
r
e
2
s
how
the
pe
r
f
or
manc
e
of
the
pr
opos
e
d
a
lgor
it
hm
,
whic
h
is
a
na
lyze
d
by
inf
luenc
e
s
pr
e
a
d
with
E
go
-
F
a
c
e
book
da
tas
e
t
a
t
s
e
e
d
s
ize
,
r
a
nging
f
r
o
m
10
to
40.
T
he
pr
opos
e
d
a
lgor
it
hm
r
e
a
c
he
s
a
high
in
f
luenc
e
s
pr
e
a
d
o
f
645
,
680,
705,
a
nd
725
f
or
it
e
r
a
ti
ons
of
100,
200
,
300
a
nd
400,
r
e
s
pe
c
ti
ve
ly.
T
he
pr
opos
e
d
a
lgor
it
hm
pe
r
f
or
ms
s
u
pe
r
ior
ly
whe
n
c
ompar
e
d
with
other
e
xis
ti
ng
a
lgo
r
i
thm
s
li
ke
W
OA
,
GW
O,
S
S
A
,
a
nd
P
S
O
a
lgor
it
hms
.
T
a
ble
3.
I
nf
luenc
e
s
pr
e
a
d
o
f
p
r
opos
e
d
method
f
o
r
E
go
-
F
a
c
e
book
da
tas
e
t
S
e
e
d
s
e
t
s
iz
e
(
k)
I
nf
lu
e
nc
e
s
pr
e
a
d f
or
E
go
-
F
a
c
e
book
W
O
A
G
W
O
SSA
PSO
SS
-
B
iAS
-
P
S
O
10
572
584
603
620
645
20
589
606
620
644
680
30
597
618
647
695
715
40
627
658
684
702
750
F
igur
e
2.
I
n
f
luenc
e
s
pr
e
a
d
of
pr
opos
e
d
method
f
or
E
go
-
F
a
c
e
book
da
tas
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
Hy
br
id
opti
miz
ati
on
algor
it
hm
for
analys
is
of
inf
lu
e
nc
e
pr
opagati
on
…
(
A
k
s
hata
Sande
e
p
B
hay
y
ar
)
631
T
a
ble
4
a
nd
F
igur
e
3
dis
play
the
pe
r
f
or
manc
e
of
the
s
ugge
s
ted
a
lgor
it
hm
whic
h
is
a
na
lyze
d
by
inf
luenc
e
s
pr
e
a
d
with
E
pini
ons
da
tas
e
t
a
t
s
e
e
d
s
ize
r
a
nging
f
r
om
10
to
40.
T
he
pr
opos
e
d
a
lgor
it
hm
r
e
a
c
he
s
a
high
inf
luenc
e
s
pr
e
a
d
of
70,
95,
107
a
nd
128
f
or
it
e
r
a
ti
ons
of
100,
200
,
300
,
a
nd
400,
r
e
s
pe
c
ti
v
e
ly.
T
he
pr
opos
e
d
a
lgor
it
hm
pe
r
f
or
ms
c
omm
e
nda
bly
,
a
s
op
pos
e
d
to
other
e
xis
ti
ng
a
lgo
r
it
hms
.
T
a
ble
4.
I
nf
luenc
e
s
pr
e
a
d
o
f
p
r
opos
e
d
method
f
o
r
E
pini
ons
da
tas
e
t
S
e
e
d
s
e
t
s
iz
e
(
k)
I
nf
lu
e
nc
e
s
pr
e
a
d f
or
E
pi
ni
ons
W
O
A
G
W
O
SSA
PSO
SS
-
B
iAS
P
S
O
10
40
47
65
80
100
20
49
55
90
110
150
30
57
63
76
115
190
40
69
74
88
147
220
F
igur
e
3.
I
n
f
luenc
e
s
pr
e
a
d
of
pr
opos
e
d
method
f
or
E
pini
ons
da
tas
e
t
T
a
ble
5
a
nd
F
igur
e
4
de
mons
tr
a
te
the
pe
r
f
or
manc
e
of
the
s
ugge
s
ted
a
lgor
it
hm,
whic
h
is
a
na
lyze
d
by
the
inf
luenc
e
s
pr
e
a
d
with
Gow
a
ll
a
da
tas
e
t
a
t
s
e
e
d
s
ize
k
r
a
nging
f
r
om
10
to
40.
T
he
SS
-
Bi
A
S
-
PSO
a
l
gor
it
hm
r
e
a
c
he
s
a
high
inf
luenc
e
s
pr
e
a
d
of
2090,
2200
,
23
50,
a
nd
2500
f
o
r
c
or
r
e
s
ponding
it
e
r
a
ti
ons
of
100
,
200,
300
,
a
nd
400,
the
r
e
by
outper
f
o
r
mi
ng
the
pr
e
vious
a
lgo
r
i
thm
s
.
T
a
ble
5.
I
nf
luenc
e
s
pr
e
a
d
o
f
p
r
opos
e
d
method
f
o
r
G
owa
ll
a
da
tas
e
t
S
e
e
d
s
e
t
s
iz
e
(
k)
I
nf
lu
e
nc
e
s
pr
e
a
d f
or
G
ow
a
ll
a
W
O
A
G
W
O
SSA
PSO
SS
-
B
iAS
-
P
S
O
10
1538
1610
1720
1950
2090
20
1605
1728
1944
2070
2200
30
1748
1869
2015
2160
2350
40
1874
1983
2103
2340
2500
F
igur
e
4.
I
n
f
luenc
e
s
pr
e
a
d
of
pr
opos
e
d
method
f
or
Gow
a
ll
a
da
tas
e
t
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
:
624
-
634
632
T
a
ble
6
a
nd
F
igur
e
5
de
mons
tr
a
te
the
outcome
s
of
the
p
r
opos
e
d
a
lgor
it
hm
a
na
lyze
d
by
mea
ns
of
inf
luenc
e
s
pr
e
a
d
with
the
He
pT
h
da
tas
e
t
a
t
s
e
e
d
s
ize
k
r
a
nging
f
r
om
10
to
40.
T
he
SS
-
Bi
A
S
-
PSO
a
lgor
it
hm
a
tt
a
ins
a
high
inf
luenc
e
s
pr
e
a
d
o
f
438
,
460,
472
,
a
nd
493
c
or
r
e
s
pondingl
y
f
or
it
e
r
a
ti
ons
of
100
,
200,
300
a
nd
400,
a
s
oppos
e
d
to
the
pr
e
vious
a
lgor
i
thm
s
.
I
n
T
a
ble
7
,
the
pe
r
f
or
manc
e
of
the
p
r
opos
e
d
a
lgor
it
hm
is
e
va
luate
d
by
r
unning
t
im
e
f
o
r
f
ou
r
dif
f
e
r
e
nt
da
tas
e
ts
.
T
he
pr
opos
e
d
method
uti
l
ize
s
the
r
unning
ti
me
of
0
.
2
×
102,
0.
5
×
102,
0
.
7
×
102
,
a
nd
103
f
o
r
100,
200
,
300
,
a
nd
400
it
e
r
a
ti
ons
in
the
E
go
-
F
a
c
e
book
da
tas
e
t.
T
he
p
r
opos
e
d
method
c
ons
umes
a
r
unning
ti
me
of
102
,
0
.
2
×
102,
0.
6
×
102
,
a
nd
103
s
im
ult
a
ne
ous
ly
f
or
100,
200
,
300
,
a
nd
400
it
e
r
a
ti
ons
on
the
E
pi
nions
da
tas
e
t.
T
he
s
ugge
s
ted
method
c
ons
umes
a
r
unning
ti
me
of
0.
3
×
103,
0
.
5
×
103,
0.
8
×
103
,
a
nd
104
s
im
ult
a
ne
ous
ly
f
or
100
,
200
,
300,
a
nd
400
it
e
r
a
t
ions
on
the
Gow
a
ll
a
da
tas
e
t.
On
the
othe
r
ha
nd
,
on
the
He
pT
h
da
tas
e
t,
it
c
ons
umes
a
r
unning
ti
me
of
0
.
7
×
103,
104,
0
.
5
×
104
,
a
nd
0.
8
×
104
indi
vidually
f
o
r
1
00,
200,
300
,
a
nd
400
it
e
r
a
ti
ons
.
T
a
ble
6.
I
nf
luenc
e
s
pr
e
a
d
o
f
p
r
opos
e
d
method
f
o
r
He
pT
h
da
tas
e
t
S
e
e
d S
e
t
S
iz
e
(
k)
I
nf
lu
e
nc
e
S
pr
e
a
d f
or
H
e
pT
h
W
O
A
G
W
O
SSA
PSO
SS
-
B
iAS
-
P
S
O
10
364
387
402
417
438
20
383
404
424
442
460
30
401
417
439
460
472
40
419
438
467
480
493
F
igur
e
5.
I
n
f
luenc
e
s
pr
e
a
d
of
pr
opos
e
d
method
f
or
He
pT
h
da
tas
e
t
T
a
ble
7.
R
unning
ti
me
f
or
a
pr
opos
e
d
method
with
f
our
da
tas
e
ts
S
e
e
d S
e
t
S
iz
e
(
k)
R
unni
ng t
im
e
f
or
va
r
io
us
da
ta
s
e
ts
(
s
e
c
)
E
go
-
F
a
c
e
book
E
pi
ni
ons
G
ow
a
ll
a
H
e
pT
h
10
0.2
×
102
102
0.3
×
103
0.7
×
103
20
0.5
×
102
0.2
×
102
0.5
×
103
104
30
0.7
×
102
0.6
×
102
0.8
×
103
0.5
×
104
40
103
103
104
0.8
×
104
3
.
2.
Com
p
ar
a
t
ive
an
alys
is
T
he
pe
r
f
or
manc
e
of
the
p
r
opos
e
d
S
S
-
B
iAS
P
S
O
a
lgor
it
hm
is
c
ompar
e
d
with
other
e
xis
ti
ng
tec
hniques
li
ke
AB
E
M
[
16]
,
OC
P
S
O
[
17
]
a
nd
DH
HO
[
22]
a
t
s
e
e
d
s
ize
k
r
a
nging
f
r
o
m
10
to
40.
T
he
SS
-
B
iAS
-
P
S
O
a
lgor
it
hm
is
c
ompar
e
d
in
ter
ms
of
inf
luenc
e
s
pr
e
a
d
a
nd
r
unning
ti
me
of
the
ne
twor
k
on
the
E
go
-
F
a
c
e
book,
E
pini
ons
a
nd
Gow
a
ll
a
da
tas
e
ts
.
T
a
ble
8
r
e
pr
e
s
e
nts
the
c
ompar
a
ti
ve
a
na
lys
is
of
the
pr
opos
e
d
a
lgor
it
hm.
F
r
om
T
a
ble
8,
it
is
c
lea
r
that
the
p
r
opos
e
d
a
lgor
it
hm
pe
r
f
or
ms
p
r
e
f
e
r
a
bly
in
r
e
lation
to
the
e
xis
ti
ng
a
lgor
it
hms
li
ke
AB
E
M
[
16]
,
OC
P
S
O
[
17]
a
nd
DH
HO
[
22]
.
T
he
S
S
-
B
iAS
-
P
S
O
a
lgor
it
hm
a
tt
a
i
ns
a
high
inf
luenc
e
s
pr
e
a
d
of
680
with
a
mi
nim
ize
d
ti
me
o
f
0.
5
×
102
f
or
the
E
go
-
F
a
c
e
book
da
tas
e
t,
a
longs
ide
a
high
inf
luenc
e
s
pr
e
a
d
of
100
with
a
mi
nim
ize
d
ti
me
of
0.
2
×
102
f
o
r
the
E
pini
ons
da
tas
e
t,
a
nd
a
c
omm
e
nda
ble
inf
luenc
e
s
pr
e
a
d
of
2
,
200
with
mi
ni
mi
z
e
d
ti
me
of
0.
5
×
103
on
the
Gow
a
ll
a
da
tas
e
t
a
t
s
e
e
d
s
ize
k=
10,
t
he
r
e
by
outper
r
mi
ng
the
pr
e
e
xis
ti
ng
tec
hniques
.
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
Hy
br
id
opti
miz
ati
on
algor
it
hm
for
analys
is
of
inf
lu
e
nc
e
pr
opagati
on
…
(
A
k
s
hata
Sande
e
p
B
hay
y
ar
)
633
T
a
ble
8.
C
ompar
a
ti
ve
a
na
lys
is
D
a
ta
s
e
t
M
e
th
ods
I
nf
lu
e
nc
e
S
pr
e
a
d
k=
10
k=
20
k=
30
k=
40
E
go
-
F
a
c
e
book
A
B
E
M
[
16]
540
650
700
740
O
C
P
S
O
[
17]
320
340
360
N
/A
DHHO
[
22]
320
350
370
400
P
r
opos
e
d S
S
-
B
iAS
-
PSO
645
680
715
750
E
pi
ni
ons
O
C
P
S
O
[
17]
80
140
170
N
/A
P
r
opos
e
d S
S
-
B
iAS
-
PSO
100
150
190
220
G
ow
a
ll
a
O
C
P
S
O
[
17]
2060
2100
2160
N
/A
P
r
opos
e
d S
S
-
B
iAS
-
PSO
2090
2200
2350
2500
3
.
3.
Dis
c
u
s
s
ion
I
n
thi
s
s
e
c
ti
on,
the
be
ne
f
it
s
of
the
pr
opos
e
d
a
lg
or
it
hm
a
nd
d
r
a
wba
c
ks
of
e
xis
ti
ng
a
lgor
it
hms
is
e
xplaine
d.
T
he
AB
E
M
[
16
]
method
ha
s
the
li
m
it
a
ti
ons
noted
a
s
f
oll
ows
:
no
c
ons
ider
a
ti
on
of
the
e
f
f
e
c
t
of
ove
r
lapping
c
a
us
e
d
by
the
c
hos
e
n
high
c
e
ntr
a
l
no
de
s
in
s
e
e
d
s
e
t
that
a
f
f
e
c
t
the
e
f
f
icie
nc
y.
T
he
OC
P
S
O
[
17
]
a
lgor
it
hm
c
a
nnot
c
ontr
ol
the
s
olut
ion
a
c
c
ur
a
c
y
we
l
l,
whe
r
e
by
the
ne
two
r
k’
s
r
unning
ti
me
is
incr
e
a
s
e
d.
He
nc
e
,
the
s
e
lec
ti
on
of
a
mi
nim
um
-
c
os
t
s
e
e
d
node
gr
oup
to
a
c
quir
e
the
inf
luenc
e
maxi
mi
z
a
ti
on
of
a
node
is
a
major
is
s
ue
to
be
s
olved.
T
he
ti
me
uti
li
z
e
d
f
o
r
inf
luenc
e
pr
opa
ga
ti
on
in
the
s
oc
ial
ne
twor
ks
is
the
maximu
m
in
the
pr
e
vious
method
that
ne
e
ds
to
be
r
e
s
olved.
I
n
t
his
manus
c
r
ipt
,
the
S
S
A
a
nd
B
iAS
-
P
S
O
a
lgo
r
it
hms
a
r
e
int
e
gr
a
ted
to
incr
e
a
s
e
the
s
pr
e
a
d
of
inf
luenc
e
ba
s
e
d
on
the
I
M
pr
oblem
a
nd
mi
nim
ize
the
r
unning
ti
me
of
the
ne
twor
k.
T
he
S
S
-
B
iAS
-
P
S
O
a
lgor
it
hm
is
e
va
luat
e
d
with
f
our
da
tas
e
ts
:
E
go
-
F
a
c
e
book,
E
pini
ons
,
Gow
a
ll
a
,
a
nd
He
pT
h.
T
his
a
lgo
r
it
hm
a
tt
a
ins
a
notew
or
thy
s
p
r
e
a
d
of
in
f
luenc
e
with
les
s
r
unning
ti
me
of
s
oc
ial
n
e
twor
k,
pr
oving
s
upe
r
ior
to
the
e
xis
ti
ng
a
lgor
it
hms
.
4.
CONC
L
USI
ON
T
he
e
xis
ti
ng
r
e
s
e
a
r
c
h
is
mainly
c
onc
e
ntr
a
ted
only
on
incr
e
a
s
ing
the
s
pr
e
a
d
of
in
f
luenc
e
a
nd
doe
s
not
c
ons
ider
the
r
unning
ti
me
o
f
the
ne
twor
k.
I
n
thi
s
r
e
s
e
a
r
c
h,
the
S
S
A
a
nd
B
iAS
-
P
S
O
a
lgor
it
h
ms
a
r
e
in
tegr
a
ted
a
nd
na
med
a
s
S
S
P
S
O
a
lgor
it
hm
to
incr
e
a
s
e
the
s
pr
e
a
d
of
inf
luenc
e
ba
s
e
d
on
the
I
M
pr
oblem
with
mi
nim
ize
d
ne
twor
k
r
unning
t
im
e
.
T
he
da
tas
e
ts
uti
li
z
e
d
f
or
the
r
e
s
e
a
r
c
h
a
r
e
E
go
-
F
a
c
e
book,
E
pini
ons
,
Gow
a
ll
a
a
nd
He
pT
h
da
tas
e
ts
,
while
a
nd
L
T
is
de
ployed
a
s
a
dif
f
us
ion
method.
T
he
n,
the
pr
opos
e
d
S
S
-
P
S
O
a
lg
or
it
hm
is
uti
li
z
e
d
f
or
the
a
na
lys
is
of
inf
luenc
e
pr
opa
ga
t
ion.
T
he
pr
opos
e
d
a
lgor
it
h
m
is
a
na
lys
e
d
in
ter
ms
of
pe
r
f
or
manc
e
mea
s
ur
e
s
of
in
f
luenc
e
s
pr
e
a
d
a
nd
r
u
nning
ti
me
o
f
the
ne
two
r
k.
T
he
S
S
-
B
iAS
-
P
S
O
a
lgor
it
hm
r
e
a
c
he
s
a
high
inf
luenc
e
s
pr
e
a
d
of
645
,
680,
715
,
a
nd
750
with
les
s
r
unning
ti
me
a
t
s
e
e
d
s
ize
k
r
a
ng
ing
f
r
om
10
to
40
in
E
go
-
F
a
c
e
book.
I
t
a
ls
o
a
c
c
ompl
is
he
s
a
high
in
f
luenc
e
s
pr
e
a
d
of
2090
,
2200
,
2350
a
nd
2
500
with
les
s
r
unning
ti
me
a
t
s
e
e
d
s
ize
k
r
a
nging
f
r
om
10
to
40
in
Gow
a
ll
a
.
M
or
e
ove
r
,
a
high
inf
luenc
e
s
pr
e
a
d
of
100,
150,
190
a
nd
220
with
les
s
r
unning
ti
me
is
obtai
ne
d
a
t
s
e
e
d
s
ize
k
r
a
nging
f
r
om
10
to
40
in
E
go
-
E
pini
ons
,
while
a
n
inf
luenc
e
s
pr
e
a
d
of
438,
460,
472
,
a
nd
49
3
with
les
s
r
unning
ti
me
is
witnes
s
e
d
a
t
s
e
e
d
s
ize
k
r
a
nging
f
r
om
10
to
40
in
He
pT
h
.
T
he
pr
opos
e
d
a
lgor
it
hm
pe
r
f
or
ms
c
omm
e
nda
bly
in
r
e
lation
to
other
e
xis
ti
ng
a
lgor
it
hms
s
uc
h
a
s
W
OA
,
GW
O,
S
S
A
,
a
nd
P
S
O
.
I
n
the
f
u
tur
e
,
va
r
ious
meta
he
ur
is
ti
c
op
ti
mi
z
a
ti
on
a
l
gor
it
hms
c
a
n
be
us
e
d
to
f
ur
ther
e
nha
nc
e
the
pe
r
f
or
manc
e
of
inf
luenc
e
pr
opa
ga
ti
on.
T
he
p
r
opos
e
d
method
c
a
n
be
us
e
d
f
or
va
r
ious
pu
r
pos
e
s
li
ke
identif
ying
the
s
pr
e
a
d
of
dis
e
a
s
e
,
c
r
im
e
r
a
te
p
r
e
diction,
a
nd
identif
yi
ng
s
oc
ial
moveme
nts
.
T
he
a
na
lys
is
of
s
oc
ial
ne
twor
ks
is
us
e
d
a
s
a
tool
to
unde
r
s
tand
a
nd
p
r
e
dict
human
be
ha
vior
.
RE
F
E
RE
NC
E
S
[
1]
Y
.
Z
ha
o,
S
.
B
in
,
a
nd
G
.
S
un,
“
R
e
s
e
a
r
c
h
on
in
f
or
ma
ti
on
pr
op
a
ga
ti
on
mode
l
in
s
oc
ia
l
ne
twor
k
ba
s
e
d
on
B
lo
c
kc
ha
in
,”
D
is
c
r
e
te
D
y
nam
ic
s
i
n N
at
u
r
e
and Soc
ie
ty
, vol
. 2022, no. 1,
J
a
n. 2022, do
i:
10.1155/2022/
7562848.
[
2]
W.
-
C
.
Y
e
h,
W
.
Z
hu,
C
.
-
L
.
H
ua
ng,
T
.
-
Y
.
H
s
u,
Z
.
L
iu
,
a
nd
S
.
-
Y
.
T
a
n,
“
A
n
e
w
B
A
T
a
nd
P
a
g
e
R
a
nk
a
lg
or
it
hm
f
or
pr
opa
ga
ti
on
pr
oba
bi
li
ty
i
n s
oc
ia
l
ne
twor
ks
,”
A
ppl
ie
d Sc
ie
nc
e
s
, vol
. 12, no. 14, J
ul
. 2022, doi:
10.3390/app121468
58.
[
3]
P
.
W
a
ng
a
nd
R
.
Z
ha
ng,
“
A
mul
ti
-
ob
je
c
ti
ve
c
r
ow
s
e
a
r
c
h
a
lg
or
it
hm
f
or
in
f
lu
e
nc
e
ma
xi
mi
z
a
ti
on
in
s
oc
ia
l
ne
twor
ks
,”
E
le
c
tr
oni
c
s
(
Sw
it
z
e
r
la
nd)
, vol
. 12, no. 8, Apr
. 2023, doi:
10.3390/ele
c
tr
oni
c
s
12081790.
[
4]
B
.
F
u,
J
.
Z
ha
ng,
H
.
B
a
i,
Y
.
Y
a
ng,
a
nd
Y
.
H
e
,
“
A
n
in
f
lu
e
nc
e
ma
xi
mi
z
a
ti
on
a
lg
or
it
hm
f
o
r
dyna
mi
c
s
oc
ia
l
ne
twor
ks
ba
s
e
d
on
e
f
f
e
c
ti
ve
l
in
ks
,”
E
nt
r
opy
, vol
. 24, no. 7, J
un. 2022, doi:
10.3390
/e
24070904.
[
5]
H
.
R
ogha
ni
a
nd
A
.
B
ouy
e
r
,
“
A
f
a
s
t
lo
c
a
l
b
a
la
nc
e
d
l
a
be
l
di
f
f
us
io
n
a
lg
or
it
hm
f
or
c
omm
uni
ty
de
te
c
ti
on
in
s
oc
ia
l
ne
twor
ks
,”
I
E
E
E
T
r
ans
ac
ti
ons
on K
no
w
le
dge
and Data E
ngi
ne
e
r
in
g
, vol
. 35, no.
6, pp. 5472
–
5484, J
un. 2023, doi:
10.1109/T
K
D
E
.2022.3162161.
[
6]
J
.
T
a
ng,
Y
.
Z
hu,
X
.
T
a
ng,
a
nd
K
.
H
a
n,
“
D
is
tr
ib
ut
e
d
in
f
l
ue
nc
e
ma
xi
mi
z
a
ti
on
f
or
la
r
ge
-
s
c
a
le
onl
in
e
s
oc
ia
l
ne
twor
ks
,
”
in
P
r
oc
e
e
di
ngs
-
I
nt
e
r
nat
io
nal
C
onf
e
r
e
nc
e
on Data E
ngi
ne
e
r
in
g
,
2022, vol. 2022, pp.
81
–
95, doi:
10.1109/I
C
D
E
53745.2022.000
11.
[
7]
J
.
C
he
r
iy
a
n
a
nd
J
.
J
.
N
a
ir
,
“
I
nf
lu
e
nc
e
mi
ni
mi
z
a
ti
on
w
it
h
node
s
ur
ve
il
la
nc
e
in
onl
in
e
s
oc
i
a
l
ne
twor
ks
,”
I
E
E
E
A
c
c
e
s
s
,
vol
.
10,
pp. 103610
–
103618, 2022, doi:
10.1109/AC
C
E
S
S
.2022.32101
26.
[
8]
M
.
T
a
h
e
r
in
ia
,
M
.
E
s
ma
e
il
i,
a
nd
B
.
M
in
a
e
i
-
B
id
gol
i,
“
O
pt
im
iz
in
g
C
E
L
F
a
lg
or
it
hm
f
o
r
in
f
lu
e
nc
e
ma
xi
mi
z
a
ti
on
pr
obl
e
m
in
s
oc
ia
l
ne
twor
ks
,”
T
e
c
hnol
ogy
J
ou
r
nal
of
A
r
ti
fi
c
ia
l
I
nt
e
ll
ig
e
nc
e
and Data M
in
in
g
, vol
. 10, no. 1, pp. 25
–
41, 2022.
[
9]
Q.
-
W
.
Z
.
Q
i
-
W
e
n
Z
ha
ng
a
nd
Q
.
-
H
.
B
.
Q
i
-
W
e
n
Z
ha
ng,
“
A
d
is
c
r
e
te
pa
r
ti
c
le
s
w
a
r
m
opt
im
iz
a
ti
on
a
lg
or
it
hm
ba
s
e
d
on
ne
ig
h
bor
Evaluation Warning : The document was created with Spire.PDF for Python.