I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
,
p
p
.
3149
~
3
1
6
1
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
15
i
3
.
pp
3
1
4
9
-
3
1
6
1
3149
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Exploring
t
h
e e
ff
e
ctiveness
of
hy
bri
d artificial
bee
P
y
C
a
ret
cla
ss
ifier
in
dela
y
tolera
nt
ne
twork
a
g
a
inst int
rusio
ns
Ra
j
a
s
hri C
ha
ud
ha
ri,
M
a
no
j
Deshp
a
n
de
D
e
p
a
r
t
me
n
t
o
f
C
o
mp
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
,
A
.
C
.
P
a
t
i
l
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
,
K
h
a
r
g
h
a
r
,
M
a
h
a
r
a
s
h
t
r
a
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
n
2
6
,
2
0
2
4
R
ev
is
ed
Dec
2
0
,
2
0
2
4
Acc
ep
ted
J
an
1
6
,
2
0
2
5
In
c
h
a
ll
e
n
g
in
g
e
n
v
iro
n
m
e
n
ts
wit
h
in
term
it
ten
t
c
o
n
n
e
c
ti
v
it
y
a
n
d
th
e
a
b
se
n
c
e
o
f
e
n
d
-
to
-
e
n
d
p
a
th
s,
d
e
lay
to
lera
n
t
n
e
two
r
k
s
(DTNs
)
re
q
u
ire
ro
b
u
st
se
c
u
rit
y
m
e
a
su
re
s
to
sa
fe
g
u
a
rd
a
g
a
in
st
p
o
ten
ti
a
l
th
re
a
ts.
Th
is
stu
d
y
a
d
d
re
ss
e
s
th
e
se
issu
e
s
b
y
imp
lem
e
n
ti
n
g
a
n
in
t
ru
s
io
n
d
e
tec
ti
o
n
s
y
ste
m
(IDS)
e
n
h
a
n
c
e
d
wit
h
m
a
c
h
in
e
lea
rn
in
g
tec
h
n
iq
u
e
s.
Co
m
m
o
n
th
re
a
ts
su
c
h
a
s
d
istri
b
u
ted
d
e
n
ial
-
of
-
se
rv
ice
(
DD
o
S
)
a
n
d
fl
o
o
d
a
tt
a
c
k
s
a
re
tac
k
led
u
si
n
g
d
a
tas
e
ts
li
k
e
n
e
two
rk
in
tru
si
o
n
d
e
tec
ti
o
n
(NID
)
a
n
d
flo
o
d
a
tt
a
c
k
d
a
tas
e
ts.
M
u
lt
i
p
le
mac
h
in
e
s
lea
rn
in
g
m
e
th
o
d
s,
in
c
l
u
d
i
n
g
k
-
n
e
a
re
st
n
e
ig
h
b
o
rs
(K
-
NN)
,
d
e
c
isio
n
t
r
e
e
s
(DT)
,
lo
g
isti
c
re
g
re
ss
io
n
(LR)
,
a
n
d
o
th
e
rs,
a
re
u
ti
li
z
e
d
to
imp
ro
v
e
d
e
tec
ti
o
n
a
c
c
u
ra
c
y
.
A
P
y
Ca
re
t
-
b
a
se
d
a
p
p
ro
a
c
h
is
d
e
v
e
lo
p
e
d
t
o
in
c
re
a
se
e
fficie
n
c
y
wh
il
e
p
re
se
rv
in
g
a
tt
a
c
k
d
e
tec
ti
o
n
a
c
c
u
ra
c
y
in
DTNs
.
C
o
m
p
a
ra
ti
v
e
re
se
a
rc
h
d
e
m
o
n
stra
tes
th
a
t
P
y
Ca
re
t
o
u
t
p
e
rfo
rm
s
S
c
ik
it
-
lea
rn
m
o
d
e
ls,
a
n
d
t
h
e
a
rti
ficia
l
b
e
e
Py
C
a
re
t
c
las
sifier
(ABPC)
o
p
ti
m
ize
s
h
y
p
e
rp
a
ra
m
e
ters
to
imp
r
o
v
e
m
o
d
e
l
p
e
rfo
rm
a
n
c
e
.
NS2
sim
u
lati
o
n
s
h
o
ws
th
e
sy
ste
m
'
s
a
b
il
it
y
t
o
th
wa
rt
a
tt
a
c
k
s,
o
ffe
rin
g
u
se
fu
l
in
si
g
h
ts
i
n
t
o
DT
N
se
c
u
rit
y
a
n
d
imp
r
o
v
i
n
g
c
o
m
m
u
n
ica
ti
o
n
re
li
a
b
il
it
y
in
v
a
rio
u
s sit
u
a
ti
o
n
s.
K
ey
w
o
r
d
s
:
Ar
tific
ial
b
ee
co
lo
n
y
Dela
y
to
ler
an
t n
etwo
r
k
Dis
tr
ib
u
ted
d
en
ial
-
of
-
s
er
v
ice
attac
k
Flo
o
d
in
g
attac
k
Op
tim
izatio
n
Secu
r
ity
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
R
ajash
r
i Ch
au
d
h
ar
i
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
E
n
g
i
n
ee
r
in
g
,
A.
C
.
Patil
C
o
lleg
e
o
f
E
n
g
in
ee
r
in
g
Kh
ar
g
h
ar
,
Ma
h
ar
ash
tr
a,
I
n
d
ia
E
m
ail:
c.
r
ajash
r
i2
0
2
1
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Dela
y
to
ler
an
t
n
etwo
r
k
in
g
(
DT
N)
co
n
tr
asts
with
co
n
v
en
tio
n
al
n
etwo
r
k
ar
c
h
itectu
r
es
b
y
ac
co
m
m
o
d
atin
g
i
n
ter
m
itten
t
c
o
n
n
ec
tiv
ity
,
a
c
h
ar
ac
ter
is
tic
o
f
ten
ab
s
en
t
in
m
o
d
e
r
n
ar
ch
ite
ctu
r
es
[
1
]
.
DT
N
is
d
esig
n
ed
to
o
p
er
ate
ef
f
ec
tiv
e
ly
u
n
d
er
ch
allen
g
in
g
o
r
s
p
o
r
ad
ic
n
etwo
r
k
co
n
d
itio
n
s
,
ca
teg
o
r
ized
b
y
h
i
g
h
e
r
laten
cy
,
b
an
d
wid
t
h
co
n
s
tr
ain
t
s
,
in
cr
ea
s
ed
er
r
o
r
p
r
o
s
p
ec
t,
p
ath
in
s
tab
ilit
y
,
o
r
v
ar
iab
le
n
o
d
e
lo
n
g
ev
ity
[
2
]
.
I
n
itially
p
r
o
p
o
s
ed
as
an
ab
s
tr
a
ctio
n
o
f
m
ess
ag
e
s
witch
in
g
,
DT
N
ar
ch
itectu
r
e
r
e
v
o
lv
es
ar
o
u
n
d
th
e
co
n
ce
p
t
o
f
b
u
n
d
les,
w
h
er
ein
m
ess
ag
es
ar
e
ag
g
r
e
g
ated
a
n
d
tr
an
s
m
itted
.
T
h
ese
b
u
n
d
les
ar
e
m
an
ag
e
d
b
y
b
u
n
d
le
r
o
u
ter
s
o
r
DT
N
g
atew
ay
s
,
wh
ich
p
lay
a
p
iv
o
tal
r
o
le
in
h
an
d
lin
g
th
e
in
s
tab
ilit
y
an
d
f
r
eq
u
en
t
d
is
co
n
n
ec
tio
n
s
in
h
er
en
t
in
DT
N
en
v
ir
o
n
m
en
ts
.
DT
Ns
a
d
d
r
ess
th
e
c
h
allen
g
es
p
o
s
ed
b
y
in
ter
m
itten
t
co
n
n
ec
tio
n
s
,
lo
n
g
o
r
v
ar
ia
b
le
d
ela
y
s
,
asy
m
m
etr
ic
d
ata
r
ates,
an
d
h
ig
h
er
r
o
r
r
ates th
r
o
u
g
h
th
e
u
tili
za
tio
n
o
f
s
to
r
e
-
a
n
d
-
ca
r
r
y
m
ess
a
g
e
s
witch
in
g
.
DT
N
f
o
llo
ws
th
e
s
to
r
e
an
d
f
o
r
war
d
tech
n
iq
u
e
in
DT
N
[
3
]
.
Fig
u
r
e
1
s
h
o
ws
ch
a
r
ac
ter
is
tics
o
f
d
ela
y
to
ler
an
t
n
etwo
r
k
.
DT
Ns
tr
an
s
f
er
m
ess
ag
es
b
et
wee
n
n
o
d
es
with
o
u
t
r
eq
u
ir
in
g
co
n
s
tan
t
co
n
n
ec
tiv
ity
,
m
ak
in
g
th
em
s
u
ited
f
o
r
d
em
an
d
in
g
en
v
ir
o
n
m
e
n
ts
s
u
ch
as
em
er
g
en
cy
s
er
v
ices
an
d
d
is
aster
m
an
ag
em
en
t
[
4
]
.
Ho
wev
er
,
th
ei
r
d
ec
en
tr
alize
d
s
tr
u
ctu
r
e
r
aises
s
ec
u
r
ity
co
n
ce
r
n
s
,
s
u
ch
as
p
ac
k
et
d
r
o
p
p
i
n
g
an
d
f
lo
o
d
i
n
g
ass
au
lts
.
I
m
p
r
o
v
in
g
DT
N
s
ec
u
r
ity
is
cr
itical
f
o
r
p
r
o
v
id
in
g
r
eliab
le
co
m
m
u
n
icatio
n
in
s
u
ch
s
ettin
g
s
[
5
]
.
DT
Ns,
d
esp
ite
th
eir
d
ec
en
tr
ali
ze
d
f
o
r
m
,
p
o
s
e
s
ec
u
r
ity
r
is
k
s
f
r
o
m
co
m
p
r
o
m
is
ed
n
o
d
es.
Attac
k
s
s
u
ch
as
p
ac
k
et
d
r
o
p
p
in
g
,
f
lo
o
d
in
g
,
an
d
o
th
er
s
d
a
m
ag
e
r
o
u
tin
g
s
y
s
tem
s
,
co
m
p
r
o
m
is
in
g
m
es
s
ag
e
in
teg
r
ity
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
1
4
9
-
3
1
6
1
3150
co
n
f
id
en
tiality
.
Ad
d
r
ess
in
g
th
ese
v
u
ln
er
ab
ilit
ies
b
ef
o
r
e
d
ep
lo
y
m
en
t
is
cr
itical,
as
it
n
ec
ess
itate
s
ad
eq
u
ate
s
ec
u
r
ity
m
eth
o
d
s
to
co
n
s
tr
u
ct
a
r
eliab
le
in
f
r
astru
ct
u
r
e.
C
h
allen
g
es
s
u
ch
as
i
n
co
n
s
is
ten
t
c
o
n
n
ec
tiv
ity
,
lim
ited
b
an
d
wid
th
,
an
d
n
o
d
e
m
o
b
ilit
y
co
m
p
o
u
n
d
s
ec
u
r
ity
c
o
n
ce
r
n
s
,
n
ec
ess
itatin
g
n
o
v
el
s
o
lu
ti
o
n
s
.
An
in
tr
u
s
io
n
d
etec
tio
n
s
y
s
tem
(
I
DS)
au
g
m
en
ted
with
m
ac
h
in
e
lear
n
in
g
is
p
r
esen
ted
to
id
en
tify
f
lo
o
d
an
d
d
is
tr
ib
u
ted
d
en
ial
-
of
-
s
er
v
ice
(
DDo
S)
attac
k
s
[
6
]
.
Var
io
u
s
r
o
u
tin
g
alg
o
r
ith
m
s
in
clu
d
in
g
th
e
s
to
r
e
-
an
d
-
f
o
r
war
d
tech
n
iq
u
e
im
p
r
o
v
e
r
esil
ien
ce
in
u
n
ce
r
tain
cir
cu
m
s
tan
ce
s
.
I
m
p
lem
en
tatio
n
s
s
u
ch
as
in
ter
p
lan
etar
y
o
v
e
r
lay
n
et
wo
r
k
-
d
elay
-
to
ler
an
t
n
etwo
r
k
in
g
(
I
ON
-
DT
N
)
h
ig
h
l
ig
h
t
r
esear
ch
ef
f
o
r
ts
to
ad
d
r
ess
s
ec
u
r
ity
in
DT
Ns,
em
p
h
asizi
n
g
th
e
n
ec
ess
ity
o
f
in
teg
r
ativ
e
tech
n
i
q
u
es a
n
d
m
a
ch
in
e
lear
n
in
g
f
o
r
d
ep
e
n
d
ab
le
co
m
m
u
n
icatio
n
[
2
]
.
Fig
u
r
e
1
.
C
h
ar
ac
ter
is
tics
o
f
d
e
lay
to
ler
an
t n
etwo
r
k
[
2
]
T
h
e
o
b
jectiv
es
in
clu
d
e
c
r
ea
tin
g
a
s
y
s
tem
to
d
is
tin
g
u
is
h
b
etwe
en
leg
itima
te
an
d
m
alicio
u
s
p
ac
k
ets
in
n
etwo
r
k
tr
af
f
ic,
d
ev
is
in
g
a
n
alg
o
r
ith
m
to
t
h
war
t
DOS
attac
k
s
in
DT
N,
s
u
g
g
esti
n
g
a
tech
n
i
q
u
e
f
o
r
en
h
a
n
cin
g
n
etwo
r
k
p
er
f
o
r
m
an
ce
,
a
n
d
ass
ess
in
g
th
e
m
o
d
el'
s
ef
f
icac
y
u
s
in
g
th
e
NS2
s
im
u
lato
r
[
7
]
.
Nu
m
er
o
u
s
s
tu
d
ies
in
DT
Ns
h
av
e
ex
p
lo
r
e
d
u
s
in
g
m
ac
h
in
e
lear
n
in
g
(
ML
)
tech
n
iq
u
es
to
im
p
r
o
v
e
n
etwo
r
k
s
ec
u
r
ity
,
with
a
f
o
c
u
s
o
n
in
tr
u
s
io
n
d
etec
tio
n
[
8
]
.
I
n
itially
,
n
etwo
r
k
in
tr
u
s
io
n
d
etec
tio
n
s
y
s
tem
s
(
NI
DS)
r
elied
o
n
k
n
o
wn
attac
k
s
ig
n
atu
r
es,
lim
itin
g
th
eir
ab
ilit
y
to
d
etec
t
n
ew
o
r
m
o
d
i
f
ied
attac
k
s
an
d
leav
in
g
n
etwo
r
k
s
v
u
ln
er
a
b
le
[
9
]
.
ML
-
b
ased
I
DS
m
eth
o
d
s
h
av
e
s
in
ce
em
er
g
e
d
,
an
aly
zin
g
o
v
er
all
b
e
h
av
io
r
al
p
atter
n
s
r
ath
er
th
an
s
p
ec
if
ic
att
ac
k
s
ig
n
atu
r
es,
an
d
o
f
f
er
in
g
in
c
r
ea
s
ed
r
o
b
u
s
tn
ess
i
n
in
tr
u
s
io
n
d
etec
tio
n
[
1
0
]
.
T
h
e
s
e
ML
ap
p
r
o
ac
h
es
h
av
e
b
ee
n
v
alid
ated
in
v
ar
io
u
s
s
tu
d
ies,
d
em
o
n
s
tr
atin
g
th
eir
p
o
ten
tial
to
en
h
an
ce
s
ec
u
r
ity
a
n
d
r
esil
ien
ce
with
in
DT
Ns.
DT
N
s
tu
d
ies
h
av
e
lo
o
k
e
d
at
a
v
ar
iety
o
f
s
ec
u
r
ity
c
o
n
ce
r
n
s
,
s
u
ch
as
f
lo
o
d
attac
k
s
,
B
lack
Ho
le
attac
k
s
,
f
ak
e
p
ac
k
et
att
ac
k
s
,
p
ac
k
et
d
r
o
p
s
,
co
llu
d
in
g
attac
k
s
,
f
au
lty
n
o
d
e
s
,
an
d
DDOS
a
ttack
s
[
1
1
]
,
[
1
2
]
.
C
h
atter
jee
et
a
l.
[
1
]
ex
p
l
o
r
ed
th
e
d
etec
tio
n
o
f
v
ar
io
u
s
r
o
u
tin
g
attac
k
s
in
DT
Ns,
r
ev
ea
lin
g
h
u
r
d
les
to
e
n
h
an
cin
g
p
er
f
o
r
m
a
n
ce
wh
ile
m
ain
t
ain
in
g
r
eliab
ilit
y
a
n
d
s
ec
u
r
ity
.
T
h
eir
r
esear
ch
f
o
cu
s
es
o
n
th
e
im
p
ac
t
o
f
th
ese
ass
au
lts
o
n
n
etwo
r
k
p
er
f
o
r
m
a
n
ce
,
em
p
h
asizin
g
n
o
d
es'
v
u
ln
er
ab
ili
ty
t
o
m
alicio
u
s
ac
ti
o
n
s
.
Flo
o
d
attac
k
s
,
f
o
r
ex
am
p
le,
in
clu
d
e
h
o
s
tile
n
o
d
es
f
lo
o
d
i
n
g
th
e
n
etwo
r
k
with
p
ac
k
ets,
r
ed
u
cin
g
DT
N
r
eso
u
r
ce
s
an
d
r
esu
ltin
g
in
lo
wer
p
ac
k
et
d
eliv
er
y
r
atio
s
(
PDR
)
an
d
h
ig
h
er
p
ac
k
et
lo
s
s
r
atio
s
(
PLR)
[
1
3
]
.
As
d
em
o
n
s
tr
ated
b
y
n
u
m
e
r
o
u
s
s
tu
d
ies
ac
r
o
s
s
v
ar
io
u
s
n
etwo
r
k
ap
p
licatio
n
s
[
1
4
]
,
[
1
5
]
,
th
ey
ad
d
r
ess
s
ec
u
r
ity
co
n
ce
r
n
s
s
u
ch
as
co
n
g
esti
o
n
tr
af
f
ic
m
an
ag
e
m
e
n
t
an
d
in
tr
u
s
io
n
d
etec
tio
n
,
en
ab
lin
g
p
r
o
ac
tiv
e
ap
p
r
o
ac
h
es
to
class
if
y
an
d
m
itig
ate
s
ec
u
r
ity
th
r
ea
ts
[
1
6
]
,
[
1
7
]
.
ML
in
teg
r
atio
n
in
n
etwo
r
k
s
ec
u
r
ity
ex
ten
d
s
to
m
ee
tin
g
th
e
ev
o
lv
in
g
s
ec
u
r
ity
n
ee
d
s
o
f
m
o
d
er
n
n
etwo
r
k
en
v
i
r
o
n
m
en
ts
,
p
la
y
in
g
a
v
ital
r
o
le
i
n
f
o
r
tif
y
in
g
n
etwo
r
k
d
ef
en
s
es
an
d
en
s
u
r
in
g
r
o
b
u
s
t
s
ec
u
r
ity
m
ea
s
u
r
es
ag
ain
s
t
p
o
ten
tial
th
r
ea
ts
[
1
8
]
,
[
1
9
]
.
R
ec
en
t
r
esear
ch
h
as
lo
o
k
ed
in
to
m
ac
h
in
e
lear
n
in
g
m
eth
o
d
s
to
im
p
r
o
v
e
n
etwo
r
k
s
ec
u
r
ity
[
2
0
]
.
On
e
s
tu
d
y
u
s
ed
a
m
ac
h
in
e
lear
n
in
g
-
b
ased
s
tr
ateg
y
to
class
if
y
h
y
p
er
tex
t
tr
an
s
f
er
p
r
o
to
c
o
l
(
HT
T
P)
t
r
a
f
f
ic
as
b
en
i
g
n
o
r
d
an
g
er
o
u
s
t
o
d
etec
t
m
alwa
r
e,
o
b
tain
in
g
a
n
am
az
in
g
9
0
%
ac
cu
r
ac
y
with
th
e
r
an
d
o
m
f
o
r
est
(
R
F)
alg
o
r
ith
m
[
2
1
]
.
An
o
t
h
er
s
tu
d
y
p
r
o
p
o
s
ed
th
e
b
u
f
f
er
m
an
a
g
em
e
n
t
f
o
r
R
PR
T
D
(
B
MRTD)
alg
o
r
ith
m
to
o
p
tim
ize
b
u
f
f
er
m
a
n
ag
em
en
t
in
DT
Ns,
im
p
r
o
v
in
g
m
ess
ag
e
d
eliv
er
y
ef
f
icien
cy
b
y
em
p
lo
y
in
g
s
tr
ateg
ies
s
u
ch
a
s
d
r
o
p
o
ld
est
(
DOA)
,
d
r
o
p
least
r
ec
en
tly
r
ec
eiv
ed
(
DL
R
)
,
an
d
d
r
o
p
lar
g
est
(
DL
A)
to
e
f
f
ec
tiv
ely
m
an
ag
e
m
ess
ag
e
q
u
e
u
es.
Stu
d
ies
h
a
v
e
ex
p
lo
r
ed
v
ar
i
o
u
s
m
eth
o
d
s
to
en
h
an
ce
attac
k
d
etec
tio
n
in
n
etwo
r
k
s
ec
u
r
ity
[
2
2
]
.
On
e
s
tu
d
y
f
o
cu
s
es
o
n
d
etec
tin
g
b
e
h
av
i
o
r
al
b
o
tn
et
attac
k
s
,
em
p
lo
y
in
g
a
s
m
ar
t
s
y
s
tem
th
at
co
m
b
in
es
t
h
e
r
a
n
d
o
m
f
o
r
est
cl
ass
if
ier
with
p
r
in
cip
al
c
o
m
p
o
n
en
t
an
aly
s
is
(
PC
A)
to
ac
h
iev
e
r
o
b
u
s
t
b
o
tn
et
d
etec
tio
n
[
2
3
]
.
An
o
t
h
er
s
tu
d
y
in
tr
o
d
u
ce
s
Sh
ield
r
ec
u
r
r
en
t
n
e
u
r
al
n
etwo
r
k
(
R
NN
)
,
a
f
r
am
ewo
r
k
d
esig
n
ed
to
d
etec
t
DDo
S
attac
k
s
with
in
I
o
T
n
etwo
r
k
s
,
in
co
r
p
o
r
atin
g
1
2
d
i
f
f
er
en
t
class
if
ier
s
to
en
h
an
ce
ac
c
u
r
ac
y
a
n
d
r
eliab
ili
ty
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
E
xp
lo
r
in
g
th
e
effec
tiven
ess
o
f
h
yb
r
id
a
r
tifi
cia
l b
ee
P
yCa
r
et
cla
s
s
ifie
r
in
d
ela
y
…
(
R
a
j
a
s
h
r
i Ch
a
u
d
h
a
r
i
)
3151
T
h
e
s
tu
d
y
ex
p
lo
r
es
m
ac
h
in
e
lear
n
in
g
-
d
r
iv
en
s
elf
-
h
ea
lin
g
m
ec
h
an
is
m
s
in
c
y
b
er
-
p
h
y
s
ica
l
s
y
s
tem
s
,
h
ig
h
lig
h
tin
g
th
eir
p
o
ten
tial
to
im
p
r
o
v
e
s
ec
u
r
it
y
an
d
p
r
ev
e
n
t
s
y
s
tem
f
ailu
r
es
[
2
4
]
.
I
t
id
e
n
tifie
s
th
r
ee
cr
itical
co
m
p
o
n
en
ts
f
o
r
s
elf
-
h
ea
lin
g
:
an
o
m
aly
d
etec
tio
n
,
f
au
lt
war
n
in
g
,
an
d
f
au
lt
au
to
-
r
em
ed
iatio
n
,
em
p
h
asizin
g
th
e
im
p
o
r
tan
ce
o
f
in
teg
r
atin
g
th
e
s
e
elem
en
ts
f
o
r
p
r
ac
tical
u
s
e.
T
h
e
s
tu
d
y
in
tr
o
d
u
ce
s
th
e
ef
f
icac
y
ar
tific
ial
b
ee
co
lo
n
y
o
p
tim
izatio
n
-
b
ased
G
au
s
s
ian
AOM
DV
(
E
AB
C
O
-
GAOM
DV)
r
o
u
tin
g
p
r
o
to
c
o
l
,
wh
ich
a
d
d
r
ess
es
r
o
u
tin
g
p
r
o
b
lem
s
in
s
to
ch
asti
c
v
eh
icu
lar
ad
h
o
c
n
etwo
r
k
s
(
SVANE
T
s
)
[
2
5
]
.
T
h
e
p
r
o
to
c
o
l a
ttem
p
ts
to
im
p
r
o
v
e
r
o
u
te
d
is
co
v
er
y
an
d
r
er
o
u
tin
g
ef
f
icien
cy
b
y
in
teg
r
atin
g
a
r
tific
ial
b
ee
co
lo
n
y
o
p
tim
izat
io
n
(
E
AB
C
O)
an
d
Gau
s
s
ian
AOM
DV
alg
o
r
ith
m
s
.
E
x
ten
s
iv
e
s
im
u
latio
n
s
ass
e
s
s
E
AB
C
O
-
GA
OM
DV
'
s
p
er
f
o
r
m
an
ce
,
r
ev
ea
lin
g
b
etter
r
o
u
te
s
tab
ilit
y
,
p
ac
k
et
d
eliv
er
y
r
atio
,
a
n
d
en
d
-
to
-
en
d
d
elay
.
T
h
e
p
r
o
to
c
o
l
is
a
d
ap
ta
b
le
to
u
n
a
n
ticip
ated
en
v
ir
o
n
m
en
ts
,
im
p
r
o
v
in
g
tr
af
f
ic
r
er
o
u
t
in
g
ef
f
icien
c
y
an
d
n
et
wo
r
k
r
esil
ien
ce
.
C
lo
u
d
co
m
p
u
tin
g
a
n
d
th
e
in
te
r
n
et
o
f
th
in
g
s
(
I
o
T
)
a
r
e
d
r
iv
in
g
f
u
tu
r
e
tech
n
o
lo
g
ies,
n
o
tab
ly
s
m
ar
t
city
d
ev
elo
p
m
e
n
t
[
2
6
]
.
C
lo
u
d
s
er
v
ices
s
u
cc
ess
f
u
lly
h
an
d
le
r
em
o
te
ac
ce
s
s
,
r
esu
ltin
g
in
i
n
cr
ea
s
ed
u
s
ag
e
b
y
o
r
g
an
izatio
n
s
f
o
r
r
eso
u
r
ce
o
p
tim
izatio
n
.
C
lo
u
d
s
er
v
ice
s
elec
tio
n
in
v
o
lv
es
o
p
tim
izatio
n
b
ased
o
n
cu
s
to
m
e
r
o
b
jectiv
es
an
d
s
er
v
ice
q
u
ality
s
tan
d
ar
d
s
.
C
o
m
b
in
in
g
g
en
eti
c
alg
o
r
it
h
m
s
(
GAs)
with
an
t
c
o
lo
n
y
o
p
tim
izatio
n
(
AC
O)
im
p
r
o
v
es
clo
u
d
c
o
m
p
u
tin
g
p
e
r
f
o
r
m
an
ce
.
T
h
e
AC
O+
GA
ap
p
r
o
ac
h
o
u
tp
er
f
o
r
m
s
e
x
is
tin
g
o
p
tim
izatio
n
m
eth
o
d
s
,
in
clu
d
in
g
e
n
er
g
y
-
an
d
r
eliab
ilit
y
-
awa
r
e
m
u
ltio
b
jectiv
e
o
p
tim
izatio
n
a
n
d
h
y
b
r
id
cu
ck
o
o
p
ar
ticle
s
war
m
with
ar
ti
f
icial
b
ee
co
lo
n
y
o
p
tim
izatio
n
.
2.
M
E
T
H
O
D
T
h
e
p
ap
er
d
escr
ib
es
a
m
ac
h
in
e
lear
n
in
g
m
o
d
el
f
o
r
d
etec
tin
g
v
ar
io
u
s
ass
au
lts
in
DT
Ns.
I
t
co
m
p
ar
es
th
e
p
er
f
o
r
m
an
ce
o
f
s
ev
er
al
ML
al
g
o
r
ith
m
s
to
im
p
r
o
v
e
attac
k
d
et
ec
tio
n
,
tr
ain
i
n
g
t
h
e
m
o
d
el
o
n
a
v
ar
iety
o
f
d
atasets
,
in
clu
d
in
g
DDo
S
an
d
f
lo
o
d
attac
k
s
.
I
n
itially
,
s
tan
d
ar
d
ML
m
o
d
els
b
u
ilt
with
Scik
it
-
lear
n
u
s
e
m
eth
o
d
s
s
u
ch
as
r
an
d
o
m
f
o
r
est
(
R
F)
,
XGBo
o
s
t
,
C
atB
o
o
s
t,
lo
g
is
t
ic
r
eg
r
ess
io
n
(
L
R
)
,
d
ec
is
io
n
tr
ee
(
DT
)
,
e
x
tr
a
tr
ee
s
,
an
d
l
ig
h
t
GB
M.
T
h
eir
p
er
f
o
r
m
an
ce
is
ev
alu
ated
u
s
in
g
m
an
y
f
ac
to
r
s
.
T
h
e
m
o
d
el
is
th
en
e
n
h
an
ce
d
u
s
in
g
Py
C
ar
et
to
b
o
o
s
t
p
er
f
o
r
m
an
ce
an
d
r
ed
u
ce
ex
ec
u
tio
n
tim
e.
Acc
u
r
ac
y
,
F1
-
Sco
r
e,
AUC,
p
r
ec
is
io
n
,
r
ec
all
,
Ma
tth
ews
co
r
r
elatio
n
co
ef
f
icien
t
(
MCC
)
,
an
d
Kap
p
a
ar
e
s
o
m
e
o
f
th
e
ev
alu
atio
n
m
e
tr
ics,
as
i
s
tr
ain
in
g
tim
e
with
d
if
f
er
en
t
class
if
ier
s
.
B
y
co
m
p
ar
in
g
Scik
it
-
lear
n
an
d
Py
C
ar
et
m
o
d
els,
t
h
e
s
tu
d
y
s
ee
k
s
to
d
is
co
v
er
t
h
e
b
est
-
p
e
r
f
o
r
m
in
g
m
o
d
el
f
o
r
in
tr
u
s
io
n
d
etec
t
io
n
in
DT
Ns.
Fig
u
r
e
2
d
is
p
lay
s
th
e
ML
m
o
d
el
ar
ch
itectu
r
e,
wh
ich
in
clu
d
es
d
ata
co
llectio
n
,
p
r
ep
r
o
ce
s
s
in
g
,
an
d
m
o
d
el
tr
ai
n
in
g
o
n
s
p
ec
if
ic
attr
ib
u
tes.
T
h
e
p
er
f
o
r
m
an
ce
o
f
th
e
ML
al
g
o
r
ith
m
in
class
if
y
in
g
attac
k
s
is
m
ea
s
u
r
ed
u
s
in
g
th
e
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
an
d
r
ec
all
m
etr
ics.
Fig
u
r
e
2
.
T
r
ain
in
g
a
n
d
d
ep
lo
y
m
en
t
o
f
ML
m
o
d
el
u
s
in
g
s
cik
it
-
lear
n
2
.
1
.
Da
t
a
c
o
llect
io
n
T
h
e
DT
N
-
s
p
ec
if
ic
ML
-
b
ased
a
ttack
d
etec
tio
n
m
o
d
el
m
ak
es
u
s
e
o
f
f
lo
o
d
attac
k
d
atasets
o
b
ta
in
ed
f
r
o
m
p
u
b
licly
av
ailab
le
DDo
S
an
d
f
l
o
o
d
attac
k
r
ec
o
r
d
s
.
T
h
ese
d
ata
s
ets
ar
e
es
s
en
tial
f
o
r
tr
ain
in
g
an
d
test
in
g
th
e
m
o
d
el
s
in
ce
th
ey
o
f
f
er
in
f
o
r
m
ati
o
n
o
n
n
etwo
r
k
tr
af
f
ic
p
atter
n
s
,
co
m
m
u
n
icatio
n
h
ab
its
,
an
d
f
lo
o
d
attac
k
in
d
icato
r
s
.
W
ith
a
d
ataset
o
f
2
5
,
1
9
2
s
am
p
les,
th
e
m
o
d
el
is
ef
f
ec
tiv
ely
tr
ain
ed
.
T
o
e
n
ab
le
s
tr
ict
ev
alu
atio
n
,
th
e
d
ataset
is
d
iv
id
ed
i
n
to
two
s
u
b
s
ets:
1
7
,
6
3
4
f
o
r
tr
ain
i
n
g
a
n
d
7
,
5
5
8
f
o
r
test
in
g
.
T
h
is
p
h
ase
ev
alu
ates
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
o
n
u
n
s
ee
n
d
ata,
p
r
o
v
id
in
g
in
s
ig
h
ts
in
to
its
g
en
er
aliza
tio
n
ca
p
ab
ilit
ies
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
1
4
9
-
3
1
6
1
3152
2
.
2
.
Da
t
a
pre
-
pro
ce
s
s
ing
Af
ter
co
llectin
g
an
d
p
ar
titi
o
n
in
g
th
e
d
ataset,
p
r
e
-
p
r
o
ce
s
s
in
g
p
r
o
ce
d
u
r
es
ar
e
u
s
ed
to
im
p
r
o
v
e
d
ata
q
u
ality
.
Data
clea
n
in
g
d
etec
ts
an
d
elim
in
ates
d
u
p
licate
r
ec
o
r
d
s
,
p
r
eser
v
in
g
in
te
g
r
it
y
an
d
d
ec
r
ea
s
in
g
r
ed
u
n
d
an
cy
.
Miss
in
g
d
ata
is
h
an
d
led
b
y
r
em
o
v
al
o
r
attr
ib
u
tio
n
,
a
n
d
o
u
tlier
s
ar
e
m
a
n
ag
ed
to
av
o
id
u
n
d
u
e
in
f
lu
en
ce
o
n
r
esu
lts
.
PC
A
d
ec
r
ea
s
es
d
im
en
s
io
n
ality
b
y
p
r
eser
v
in
g
u
s
ef
u
l
f
ea
tu
r
es
an
d
r
em
o
v
in
g
ir
r
elev
a
n
t
o
n
es
,
s
o
o
v
er
co
m
in
g
th
e
cu
r
s
e
o
f
c
o
m
p
lex
ity
an
d
im
p
r
o
v
in
g
g
en
er
aliza
tio
n
.
No
r
m
aliza
tio
n
an
d
s
ca
lin
g
n
o
r
m
a
lize
f
ea
tu
r
e
v
al
u
es,
p
r
ev
e
n
tin
g
la
r
g
er
f
ea
tu
r
es
f
r
o
m
d
o
m
in
atin
g
th
e
lear
n
in
g
p
r
o
ce
s
s
wh
ile
also
e
n
h
an
cin
g
alg
o
r
ith
m
s
tab
ilit
y
d
u
r
in
g
tr
ain
in
g
.
T
h
i
s
tech
n
iq
u
e
im
p
r
o
v
es
ML
m
o
d
el
ef
f
icien
cy
,
s
tab
ilit
y
,
an
d
ac
cu
r
ac
y
.
Mo
d
el
in
itializatio
n
co
n
f
ig
u
r
es
alg
o
r
ith
m
s
with
s
p
ec
if
ic
h
y
p
e
r
p
ar
am
eter
s
th
at
a
r
e
f
in
e
-
tu
n
ed
f
o
r
m
ax
im
u
m
p
er
f
o
r
m
an
ce
.
T
h
e
m
o
d
el
is
iter
ativ
ely
tr
ain
ed
a
n
d
test
ed
u
n
til
it
ac
h
iev
es
s
u
f
f
icien
t
ef
f
ic
ien
cy
,
en
s
u
r
in
g
th
at
attac
k
s
ar
e
d
etec
ted
an
d
r
ed
u
c
ed
in
DT
N
.
2
.
3
.
T
ra
ini
ng
Du
r
in
g
t
h
e
tr
ain
i
n
g
p
h
ase,
ea
ch
d
ata
p
o
i
n
t
f
r
o
m
t
h
e
g
iv
en
d
ataset
is
p
r
e
-
p
r
o
ce
s
s
ed
an
d
f
ea
tu
r
es
ar
e
ex
tr
ac
ted
.
T
h
is
s
tag
e
in
v
o
lv
es
tr
ain
in
g
f
ea
tu
r
es
f
o
r
ea
ch
d
at
a
p
o
in
t
in
th
e
d
ataset
an
d
ass
ig
n
in
g
class
es
to
th
e
tr
ain
ed
f
ea
tu
r
es.
T
h
is
alg
o
r
ith
m
h
as two
cl
ass
es: n
o
r
m
al
an
d
DDo
S o
r
f
lo
o
d
attac
k
.
2
.
4
.
T
esting
Du
r
in
g
test
in
g
,
th
e
test
s
am
p
l
es
ar
e
p
ass
ed
in
to
a
m
ac
h
in
e
lear
n
in
g
class
if
ier
,
wh
ich
u
s
es
tr
ain
in
g
f
ea
tu
r
es
to
class
if
y
th
e
test
in
s
tan
ce
in
to
th
e
p
r
o
v
id
ed
class
j.
I
f
th
e
class
if
ier
ac
cu
r
ately
class
if
ies
a
p
ar
ticu
lar
class
,
th
e
p
r
o
ce
s
s
will
y
ield
a
s
u
p
er
io
r
o
u
tco
m
e.
I
f
th
e
m
ac
h
in
e
lear
n
in
g
class
if
ier
's
d
ec
is
io
n
is
n
'
t
f
in
al,
th
e
ch
o
ice
is
d
ec
id
ed
u
s
in
g
a
co
s
t
m
in
im
izatio
n
p
r
o
ce
d
u
r
e.
2
.
5
.
Da
t
a
s
et
T
h
e
d
ataset
is
an
im
p
o
r
tan
t p
a
r
t o
f
th
e
ML
m
o
d
el
th
at
s
h
o
u
ld
in
clu
d
e
a
v
ar
iety
o
f
in
tr
u
s
io
n
d
ata.
h
er
e
u
s
ed
th
e
NI
D
[
1
9
]
d
ataset
av
ai
lab
le
o
n
Kag
g
le
s
o
u
r
ce
to
tr
ain
an
d
test
m
o
d
els.
T
h
e
d
ataset
d
etails
ar
e
s
h
o
wn
in
T
ab
le
1
.
T
h
e
NI
D
d
ataset
co
u
n
ter
f
eit
in
a
m
ilit
ar
y
n
etwo
r
k
en
v
ir
o
n
m
en
t
in
clu
d
es
ex
ten
s
iv
e
d
iv
er
s
ity
o
f
d
at
a
in
clu
d
in
g
i
n
tr
u
s
io
n
s
am
p
les.
T
h
e
US
Air
Fo
r
ce
L
AN
was
b
last
ed
with
m
u
ltip
le
attac
k
s
u
s
in
g
r
aw
T
C
P/IP
d
u
m
p
d
ata.
Fo
r
ea
ch
T
C
P/IP
co
n
n
ec
tio
n
,
d
ata
f
lo
ws
f
r
o
m
a
s
o
u
r
ce
I
P
ad
d
r
ess
to
a
tar
g
et
I
P
ad
d
r
ess
f
o
r
s
o
m
e
tim
e
d
u
r
atio
n
.
T
h
e
d
ataset
co
n
tai
n
s
n
o
r
m
al
an
d
attac
k
d
ata
with
4
1
f
ea
tu
r
es
in
clu
d
in
g
3
q
u
alitativ
e
an
d
3
8
q
u
an
titativ
e
f
ea
tu
r
es.
T
ab
le
1
.
Data
s
et
d
etails
D
a
t
a
s
e
t
N
I
D
S
i
z
e
5
.
2
9
M
B
F
e
a
t
u
r
e
s
e
x
t
r
a
c
t
e
d
41
F
e
a
t
u
r
e
s
sel
e
c
t
e
d
39
C
l
a
s
s
−
N
o
r
mal
−
A
n
o
m
a
l
y
Te
st
d
a
t
a
set
2
.
4
2
M
B
(
4
0
%)
Tr
a
i
n
d
a
t
a
se
t
2
.
8
7
M
B
(
6
0
%)
2
.
6
.
ML
m
o
del a
nd
cla
s
s
if
ica
t
io
n
T
h
e
m
ac
h
in
e
lear
n
in
g
-
b
ased
a
ttack
d
etec
tio
n
m
o
d
el
is
b
u
ilt
o
n
Py
th
o
n
,
wh
ic
h
in
clu
d
es
lib
r
ar
ies
s
u
ch
as
Scik
it
-
lear
n
an
d
Py
C
ar
et.
Sc
ik
it
-
lear
n
h
an
d
les
ca
r
ef
u
lly
d
at
a
b
ef
o
r
e
tr
ea
tm
en
t
a
n
d
m
eth
o
d
s
elec
tio
n
,
allo
win
g
f
o
r
d
etailed
c
o
m
p
ar
is
o
n
s
o
f
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
.
Py
C
ar
et
im
p
r
o
v
es
p
r
o
d
u
ctiv
ity
b
y
au
to
m
atin
g
o
p
er
atio
n
s
lik
e
m
o
d
el
tr
ain
in
g
,
h
y
p
e
r
p
ar
am
ete
r
tu
n
in
g
,
an
d
p
er
f
o
r
m
a
n
ce
ev
al
u
atio
n
.
W
h
ile
b
o
th
lib
r
ar
ies
s
er
v
e
m
ac
h
in
e
lear
n
i
n
g
g
o
als,
Scik
it
-
lear
n
p
r
o
v
id
es
g
r
ea
ter
f
lex
i
b
ilit
y
an
d
co
n
tr
o
l
o
v
er
in
d
iv
id
u
al
co
m
p
o
n
en
ts
o
f
t
h
e
ML
p
ip
elin
e,
wh
er
ea
s
Py
C
ar
et
ac
ce
ler
ates
wo
r
k
f
lo
ws
b
y
au
to
m
atin
g
c
o
m
m
o
n
ac
tiv
ities
lik
e
as
d
ata
p
r
ep
r
o
ce
s
s
in
g
an
d
m
o
d
el
s
elec
tio
n
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el'
s
f
lo
w
u
tili
zin
g
Py
C
ar
et
i
s
d
ep
icted
in
th
e
Fig
u
r
e
3.
T
h
e
p
r
o
ce
s
s
o
f
b
u
ild
in
g
a
Py
C
ar
et
m
o
d
el
f
o
r
DT
N
attac
k
d
etec
tio
n
o
n
a
f
lo
o
d
attac
k
d
ata
s
et
in
v
o
lv
es
s
ev
er
al
k
ey
s
tep
s
as
m
en
tio
n
e
d
b
elo
w.
T
h
e
p
r
o
ce
s
s
b
eg
in
s
with
s
ettin
g
u
p
t
h
e
Py
C
ar
et
en
v
ir
o
n
m
en
t
an
d
a
s
im
p
le
in
ter
f
ac
e
f
o
r
in
itializin
g
th
e
en
v
ir
o
n
m
e
n
t a
n
d
s
p
ec
if
y
in
g
t
h
e
t
ar
g
et
v
ar
ia
b
le,
s
tr
ea
m
lin
in
g
th
e
s
etu
p
p
r
o
ce
s
s
.
−
Data
p
r
ep
r
o
ce
s
s
in
g
:
Py
C
ar
et
a
u
to
m
ates
d
ata
p
r
e
p
r
o
ce
s
s
in
g
t
ask
s
lik
e
h
an
d
lin
g
m
is
s
in
g
v
al
u
es
an
d
f
ea
t
u
r
e
s
ca
lin
g
f
o
r
f
l
o
o
d
attac
k
d
ataset
in
DT
Ns.
−
C
o
m
p
a
r
e
m
o
d
e
l
s
:
M
u
lt
i
p
l
e
m
ac
h
i
n
e
l
e
a
r
n
i
n
g
m
o
d
e
ls
,
l
i
k
e
L
ig
h
t
G
B
M
,
X
GB
o
o
s
t
,
C
atB
o
o
s
t
,
a
n
d
E
x
t
r
a
T
r
e
e
s
,
a
r
e
c
r
o
s
s
-
v
a
li
d
a
t
e
d
t
o
d
e
t
ec
t
D
T
N
a
t
ta
c
k
s
wi
t
h
d
e
f
a
u
l
t
h
y
p
e
r
p
a
r
a
m
e
t
e
r
s
.
P
e
r
f
o
r
m
a
n
c
e
m
e
a
s
u
r
e
s
s
u
c
h
as
p
r
e
c
i
s
i
o
n
,
ac
c
u
r
a
c
y
,
r
e
c
al
l
,
F
1
-
S
c
o
r
e
,
AU
C
,
K
a
p
p
a
,
MCC
,
a
n
d
e
v
a
l
u
a
t
i
o
n
ti
m
e
a
r
e
c
a
l
c
u
l
a
t
ed
f
o
r
e
a
c
h
m
o
d
e
l
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
E
xp
lo
r
in
g
th
e
effec
tiven
ess
o
f
h
yb
r
id
a
r
tifi
cia
l b
ee
P
yCa
r
et
cla
s
s
ifie
r
in
d
ela
y
…
(
R
a
j
a
s
h
r
i Ch
a
u
d
h
a
r
i
)
3153
−
B
est
p
er
f
o
r
m
in
g
m
o
d
el:
A
to
p
m
o
d
el
is
ch
o
s
en
af
ter
c
o
m
p
ar
in
g
r
esu
lts
,
tr
ain
ed
o
n
f
u
ll
d
ataset
to
test
ca
p
ab
ilit
ies.
−
Ar
tific
ial
b
ee
Py
C
ar
et
clas
s
if
ier
(
AB
PC
)
:
An
aly
ze
s
o
lu
tio
n
s
p
ac
e
f
o
r
o
p
tim
izatio
n
is
s
u
es,
f
in
d
h
y
p
er
p
ar
am
eter
s
,
o
p
tim
ize
Py
C
ar
et
m
o
d
el,
p
r
o
tects DT
Ns w
ith
I
DS.
−
Op
tim
al
s
o
lu
tio
n
an
d
E
x
p
lo
i
tatio
n
s
ea
r
ch
:
AB
PC
s
ee
s
f
o
o
d
s
o
u
r
ce
s
as
p
r
o
b
lem
-
s
o
lv
i
n
g
with
n
ec
tar
in
d
icatin
g
v
alu
e
.
−
R
esu
lts
:
T
h
e
r
esu
lts
f
r
o
m
th
e
v
alid
atio
n
p
h
ase
ar
e
an
aly
ze
d
to
d
eter
m
in
e
th
e
b
est
-
p
e
r
f
o
r
m
i
n
g
m
o
d
el
b
ased
o
n
p
r
ed
ef
in
e
d
e
v
alu
atio
n
m
e
tr
ics.
T
h
e
b
est
-
p
e
r
f
o
r
m
in
g
m
o
d
el
is
s
elec
ted
f
o
r
f
u
r
th
er
r
ef
in
em
en
t
a
n
d
d
ep
lo
y
m
e
n
t.
−
Pre
d
ictio
n
: T
h
e
b
est
-
p
e
r
f
o
r
m
in
g
m
o
d
el
is
k
ep
t
f
o
r
later
u
s
ag
e,
m
ak
in
g
it
ea
s
ily
av
ailab
le
f
o
r
p
r
e
d
ictio
n
s
o
n
n
ew
d
ata.
T
h
is
p
er
s
is
ten
ce
en
ab
les
r
ap
id
d
etec
tio
n
o
f
f
lo
o
d
attac
k
s
in
r
ea
l
-
tim
e
DT
N
s
ce
n
ar
io
s
,
allo
win
g
f
o
r
r
a
p
id
r
ea
ctio
n
an
d
m
itig
atio
n
.
Fig
u
r
e
3
.
Pro
ce
s
s
o
f
tr
ain
in
g
P
y
C
ar
et
m
o
d
el
Ov
er
all,
th
e
Py
C
ar
et
m
o
d
el
-
b
u
ild
in
g
p
r
o
ce
s
s
s
tr
ea
m
lin
es
t
h
e
wo
r
k
f
l
o
w
f
o
r
DT
N
attac
k
d
etec
tio
n
,
au
to
m
atin
g
m
an
y
ted
io
u
s
task
s
an
d
p
r
o
v
id
in
g
a
u
s
er
-
f
r
ien
d
ly
in
ter
f
ac
e
f
o
r
m
o
d
el
tr
ain
i
n
g
,
ev
alu
atio
n
,
an
d
d
ep
lo
y
m
e
n
t.
T
h
e
AB
C
O
m
eth
o
d
is
a
n
atu
r
e
-
in
s
p
ir
ed
o
p
tim
i
za
tio
n
s
tr
ateg
y
b
ased
o
n
h
o
n
e
y
b
ee
b
eh
av
io
r
.
I
t
is
u
s
ed
in
s
war
m
in
tellig
en
ce
al
g
o
r
ith
m
s
to
f
i
n
d
th
e
b
est
s
o
lu
tio
n
s
to
co
m
p
le
x
p
r
o
b
lem
s
.
A
B
C
O
i
s
im
p
o
r
tan
t
in
m
ac
h
in
e
lear
n
in
g
m
o
d
els
f
o
r
s
ettin
g
h
y
p
er
p
a
r
am
eter
s
an
d
i
m
p
r
o
v
i
n
g
s
ec
u
r
ity
in
DT
Ns
th
r
o
u
g
h
I
DSs
.
I
DSs
m
o
n
ito
r
n
etwo
r
k
tr
af
f
ic
f
o
r
s
u
s
p
icio
u
s
ac
tiv
ity
in
DT
Ns,
wh
ich
f
ac
e
co
n
n
ec
tiv
ity
c
h
allen
g
es.
AB
C
O
ad
ju
s
ts
I
DS
h
y
p
er
p
ar
a
m
eter
s
to
en
h
an
ce
d
etec
tio
n
ac
cu
r
ac
y
an
d
r
ed
u
ce
f
alse
p
o
s
itiv
es,
im
p
r
o
v
in
g
DT
Ns
'
s
ec
u
r
ity
.
T
h
e
alg
o
r
ith
m
u
s
es
f
o
o
d
s
o
u
r
ce
s
as
s
o
lu
tio
n
s
,
with
n
ec
tar
co
n
ten
t
r
ep
r
esen
tin
g
th
e
d
esire
d
o
u
tco
m
e.
E
m
p
lo
y
ed
an
d
o
b
s
er
v
er
b
ee
s
f
o
c
u
s
o
n
ex
p
l
o
itatio
n
wh
ile
s
co
u
t
b
ee
s
ex
p
l
o
r
e
n
ew
s
o
lu
tio
n
s
.
T
h
e
alg
o
r
i
th
m
b
eg
i
n
s
with
a
p
o
p
u
latio
n
o
f
f
o
o
d
s
o
u
r
ce
s
.
T
h
e
g
o
al
o
f
th
e
em
p
lo
y
e
d
b
ee
m
ec
h
an
is
m
is
to
e
x
p
lo
it
th
e
ex
is
tin
g
f
o
o
d
s
o
u
r
ce
.
T
o
ac
tu
aliz
e
th
is
id
ea
,
th
ey
cr
ea
te
a
n
ew
s
o
lu
tio
n
n
ea
r
th
e
cu
r
r
en
t
f
o
o
d
s
u
p
p
ly
.
Fo
r
ea
ch
s
o
lu
tio
n
i
,
o
n
e
elem
en
t
j
(
r
ea
l
n
u
m
b
er
)
is
r
an
d
o
m
l
y
ch
o
s
en
an
d
co
u
p
le
d
with
th
e
eq
u
iv
alen
t
elem
en
t
f
r
o
m
an
o
th
er
r
an
d
o
m
ly
s
elec
ted
f
o
o
d
s
o
u
r
ce
k
.
T
ak
e
n
o
te
th
at
th
ey
h
av
e
≠
.
T
h
is
is
ac
co
m
p
lis
h
ed
u
s
in
g
th
e
f
o
llo
win
g
f
o
r
m
u
la.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
1
4
9
-
3
1
6
1
3154
=
+
(
−
)
(
1
)
w
h
er
e
is
j
th
r
ea
l
n
u
m
b
er
o
f
f
o
o
d
s
o
u
r
ce
i
,
is
a
r
an
d
o
m
ly
g
e
n
er
ated
n
u
m
b
er
b
etwe
en
[
-
1
,
1
]
,
an
d
is
th
e
n
ew
r
ea
l n
u
m
b
er
o
f
j
th
elem
en
t o
f
f
o
o
d
s
o
u
r
ce
i
.
T
h
e
h
ir
ed
b
ee
ac
ce
p
ts
a
n
ew
f
o
o
d
s
o
u
r
ce
if
its
n
ec
tar
am
o
u
n
t
ex
ce
ed
s
th
at
o
f
th
e
p
r
esen
t o
n
e.
Oth
er
wis
e,
it c
o
n
tin
u
es wi
th
th
e
p
r
ev
io
u
s
o
n
e.
Af
ter
lo
o
k
in
g
f
o
r
f
o
o
d
s
o
u
r
ce
s
,
h
ir
ed
b
ee
s
to
s
h
ar
e
th
eir
d
is
co
v
er
ies
with
o
b
s
er
v
er
b
ee
s
.
On
l
o
o
k
er
b
ee
s
th
en
ass
ess
th
e
n
ec
tar
in
f
o
r
m
atio
n
f
r
o
m
all
em
p
lo
y
e
d
b
ee
s
b
ef
o
r
e
s
ele
ctin
g
a
f
o
o
d
s
o
u
r
ce
b
ased
o
n
it
s
n
ec
tar
c
o
n
ten
t.
Pro
b
a
b
ilit
y
f
u
n
ctio
n
s
ar
e
u
s
ed
to
ca
lcu
late
th
e
p
o
s
s
ib
ilit
y
o
f
ea
c
h
f
o
o
d
s
o
u
r
ce
b
ein
g
s
elec
ted
b
y
o
n
lo
o
k
er
b
ee
s
.
=
∑
=
1
(
2
)
w
h
er
e
is
th
e
n
ec
tar
am
o
u
n
t
(
o
b
jectiv
e
v
alu
e)
o
f
th
e
f
o
o
d
s
o
u
r
ce
(
s
o
lu
tio
n
i
)
.
Usi
n
g
ea
ch
o
n
lo
o
k
er
s
elec
t a
n
ew
n
ea
r
b
y
f
o
o
d
s
o
u
r
ce
an
d
i
ts
n
ec
tar
am
o
u
n
t.
I
f
a
n
ew
f
o
o
d
s
o
u
r
ce
p
r
o
v
id
es
m
o
r
e
n
ec
t
ar
th
an
th
e
p
r
e
v
io
u
s
o
n
e,
th
e
b
ee
will st
o
r
e
th
e
n
ew
p
o
s
itio
n
an
d
d
is
ca
r
d
th
e
o
l
d
o
n
e.
W
h
en
em
p
lo
y
e
d
an
d
o
b
s
er
v
er
b
ee
s
r
u
n
o
u
t o
f
f
o
o
d
,
th
e
y
b
ec
o
m
e
s
co
u
t
b
ee
s
Sco
u
t
b
ee
s
r
ep
lace
d
e
p
leted
f
o
o
d
s
o
u
r
ce
s
r
a
n
d
o
m
l
y
.
T
h
e
p
r
o
ce
d
u
r
e
e
n
tails
p
r
eser
v
in
g
a
s
p
ec
if
ic
n
u
m
b
er
o
f
r
ec
en
t
f
o
o
d
s
o
u
r
ce
s
p
r
o
d
u
ce
d
b
y
em
p
lo
y
ed
an
d
o
n
lo
o
k
er
b
e
es.
W
h
en
r
ep
lacin
g
an
ex
h
a
u
s
ted
f
o
o
d
s
o
u
r
ce
,
o
n
e
o
f
t
h
e
co
n
s
er
v
ed
f
o
o
d
s
o
u
r
c
e
s
is
ch
o
s
en
u
s
in
g
a
r
o
u
lette
w
h
ee
l
b
ased
o
n
th
ei
r
p
r
o
b
a
b
ilit
y
.
(
−
)
−
1
∑
=
1
(
−
)
−
1
(
3
)
T
h
e
am
o
u
n
t
o
f
n
ec
tar
in
th
e
s
av
ed
an
d
cu
r
r
e
n
t
f
o
o
d
s
o
u
r
c
es
is
r
ep
r
esen
ted
b
y
an
d
,
r
esp
ec
tiv
ely
.
T
h
e
r
o
u
lette
wh
ee
l d
iv
id
es th
e
in
te
r
v
al
[
0
,
1
]
in
t
o
b
ac
k
s
u
b
in
ter
v
als.
[
0
,
(
1
−
)
−
1
]
,
…
,
[
∑
(
−
)
−
1
−
1
=
1
,
∑
(
−
)
−
1
=
1
]
,
…
,
[
∑
(
−
)
−
1
,
1
−
1
=
1
]
(
4
)
T
h
e
s
u
b
in
ter
v
als co
r
r
esp
o
n
d
t
o
s
av
ed
f
o
o
d
s
o
u
r
ce
s
,
with
th
e
i
-
th
s
u
b
in
ter
v
al
d
en
o
tin
g
th
e
m
o
s
t r
ec
en
t
s
av
ed
f
o
o
d
s
o
u
r
ce
.
T
h
e
n
ex
t
s
tep
is
to
g
en
er
ate
a
r
an
d
o
m
r
ea
l
n
u
m
b
er
b
etwe
en
0
an
d
1
.
T
h
e
f
o
o
d
s
o
u
r
ce
is
th
en
ch
o
s
en
d
ep
e
n
d
in
g
o
n
th
e
s
u
b
i
n
ter
v
al
to
wh
ich
th
is
n
u
m
b
er
b
elo
n
g
s
.
Ps
eu
d
o
co
d
e
f
o
r
Py
C
ar
et
m
o
d
el
:
a.
L
o
ad
th
e
f
lo
o
d
attac
k
d
atasets
,
wh
ich
co
n
tain
r
ec
o
r
d
s
o
f
DD
o
S a
n
d
f
lo
o
d
attac
k
s
.
b.
Pre
p
r
o
ce
s
s
th
e
d
ata
to
r
em
o
v
e
m
is
s
in
g
v
alu
es,
s
ca
le
f
ea
tu
r
es,
an
d
en
co
d
e
ca
teg
o
r
ical
v
ar
iab
les,
r
esu
ltin
g
in
well
-
f
o
r
m
atted
d
atasets
ap
p
r
o
p
r
iate
f
o
r
m
o
d
el
tr
ai
n
in
g
.
c.
T
o
an
aly
ze
th
e
m
o
d
el,
m
an
y
m
ac
h
in
e
lea
r
n
in
g
tech
n
iq
u
e
s
ar
e
u
s
ed
,
i
n
clu
d
in
g
L
ig
h
tG
B
M,
XGBo
o
s
t,
C
atB
o
o
s
t,
E
x
tr
aT
r
ee
s
,
an
d
o
th
er
s
.
d.
Pre
p
ar
e
th
e
d
ata
f
o
r
tr
ai
n
in
g
v
alid
atio
n
.
e.
U
s
e
m
e
t
r
i
c
s
l
i
k
e
a
c
c
u
r
ac
y
,
AU
C
,
r
e
c
al
l
,
p
r
e
c
is
i
o
n
,
F
1
-
S
c
o
r
e
,
k
a
p
p
a
,
a
n
d
M
C
C
t
o
e
v
a
l
u
a
t
e
t
h
e
m
o
d
e
l'
s
p
e
r
f
o
r
m
a
n
c
e
.
f.
T
h
e
v
alid
atio
n
r
esu
lts
ar
e
e
x
am
in
ed
to
d
eter
m
in
e
th
e
b
est
-
p
er
f
o
r
m
in
g
m
o
d
el
b
ased
o
n
estab
lis
h
ed
ev
alu
atio
n
cr
iter
ia.
g.
Op
tim
izin
g
r
esu
lts
u
s
in
g
th
e
A
B
C
alg
o
r
ith
m
.
h.
C
o
m
p
ar
e
th
e
r
esu
lts
to
ex
is
tin
g
ap
p
r
o
ac
h
es a
n
d
d
em
o
n
s
tr
ate
th
e
ef
f
ec
tiv
en
ess
o
f
t
h
e
n
ew
a
p
p
r
o
ac
h
.
i.
Use th
e
tr
ain
ed
m
o
d
el
to
m
ak
e
p
r
ed
ictio
n
s
.
Alg
o
r
ith
m
1
.
F
o
r
h
y
b
r
id
ar
tifi
cial
b
ee
Py
C
ar
et
class
if
ier
1.
import numpy as np
2.
def measom(variables_values=[0, 0]):
3.
x1, x2 = variables_values
4.
func_value =
-
np.cos(x1) * np.cos(x2) * np.exp(
-
((x1
-
np.pi) * 2 + (x2
-
np.pi) * 2))
5.
return func_value
6.
def artificial_bee_colony_optimization(food_sources=20, iterations=100,
min_values=[5,5], max_values=[5, 5], employed_bees=5, outlookers_bees=5,
limit=10,target_function=measom, verbose=True, start_init=None, target_value=None):
7.
sources = initial_variables(food_sources, min_values, max_values, target_function,
start_init)
8.
fitness = fitness_function(sources, fitness_calc)
9.
best_bee = sources[np.argmin(sources[:,
-
1]), :]
10.
return best_bee
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
E
xp
lo
r
in
g
th
e
effec
tiven
ess
o
f
h
yb
r
id
a
r
tifi
cia
l b
ee
P
yCa
r
et
cla
s
s
ifie
r
in
d
ela
y
…
(
R
a
j
a
s
h
r
i Ch
a
u
d
h
a
r
i
)
3155
11.
def initial_variables(size, min_values, max_values, target_function, start_init):
a.
dim = len(min_values)
12.
if start_init is not None:
a.
population = start_init
13.
else:
14.
population = np.random.uniform(min_values, max_values, (size, dim))
15.
if
hasattr(target_function, 'vectorized'):
a.
fitness_values = target_function(population)
16.
else:
17.
fitness_values = np.apply_along_axis(target_function, 1, population)
18.
population = np.hstack((population, fitness_values[:, np.newaxis]))
19.
return population
20.
def fitness_function(sources, fitness_calc):
21.
return sources[:,
-
1]
22.
def fitness_calc():
23.
pass
24.
custom_grid = {
a.
'max_depth': [None, 10, 20, 30, 40, 50],
b.
'min_samples_split': [2, 5, 10],
c.
'min_samples_leaf': [1, 2, 4],
d.
}
25.
best_params = None
26.
best_min_value = float('inf')
27.
for max_depth in custom_grid['max_depth']:
28.
for min_samples_split in custom_grid['min_samples_split']:
29.
for min_samples_leaf in custom_grid['min_samples_leaf']:
i.
print(f"Running ABCO with max_depth: {max_depth}, min_samples_split:
{min_samples_split}, min_samples_leaf: {min_samples_leaf}")
ii.
abco_result = artificial_bee_colony_optimization()
iii.
variables = abco_result[:
-
1]
iv.
minimum = abco_result[
-
1]
v.
if minimum < best_min_value:
vi.
best_min_value = minimum
vii.
best_params = {'max_depth': max_depth, 'min_samples_split':
min_samples_split, 'min_samples_leaf': min_samples_leaf}
viii.
print("Results:")
ix.
print(f"Optimal Variables: {variables}")
x.
print(f"Minimum Value: {minimum}")
xi.
print("="*50)
30.
print("Best Parameters:")
31.
print(best_params)
print(f"Corresponding Minimum Value: {best_min_value}")
2.
7
.
Sim
ula
t
i
o
n a
nd
ex
periment
a
l set
up
NS2
,
o
r
n
etwo
r
k
s
im
u
lato
r
2
,
is
an
o
p
en
-
s
o
u
r
ce
to
o
l
th
at
is
u
s
ef
u
l
f
o
r
in
v
esti
g
atin
g
DT
Ns
an
d
th
eir
v
u
ln
er
ab
ilit
y
t
o
ass
au
lts
.
I
t
s
im
u
lates
g
en
u
in
e
DT
N
s
itu
atio
n
s
,
in
clu
d
in
g
in
co
m
p
atib
le
co
n
n
e
ctiv
ity
,
d
elay
s
,
an
d
d
is
co
n
n
ec
tio
n
s
.
NS2
en
ab
les
t
h
e
in
jectio
n
o
f
m
alicio
u
s
attac
k
s
s
u
ch
a
s
f
lo
o
d
s
an
d
DDo
S
attac
k
s
to
ev
alu
ate
s
ec
u
r
ity
m
ec
h
an
is
m
s
.
Simu
latio
n
p
ar
am
eter
s
an
d
th
eir
v
al
u
es
u
s
ed
d
u
r
in
g
e
x
p
er
im
e
n
tal
s
etu
p
is
s
h
o
wn
in
T
ab
le
2
.
T
o
s
im
u
late
an
I
DS
with
in
NS2
,
th
e
s
im
u
latio
n
p
ar
am
eter
s
,
in
clu
d
in
g
n
etwo
r
k
to
p
o
lo
g
y
an
d
n
o
d
e
ch
ar
ac
ter
is
tics
,
ar
e
d
ef
in
ed
.
A
r
o
u
tin
g
p
r
o
to
c
o
l
is
estab
lis
h
e
d
f
o
r
a
n
etwo
r
k
o
f
2
0
n
o
d
es,
as
d
ep
icted
in
th
e
f
ig
u
r
e
s
h
o
wca
s
in
g
n
o
d
es with
i
n
th
e
NS2
en
v
ir
o
n
m
en
t.
E
ac
h
n
o
d
e
is
ass
ig
n
ed
u
n
iq
u
e
attr
ib
u
tes,
s
u
ch
as
co
lo
r
s
an
d
s
h
ap
es,
s
p
ec
if
y
i
n
g
th
eir
lo
ca
tio
n
s
an
d
co
n
n
ec
tiv
ity
p
ar
am
eter
s
.
Simu
latio
n
o
f
a
n
etwo
r
k
i
n
NS2
en
v
ir
o
n
m
en
t is sh
o
wn
in
Fig
u
r
e
4.
T
ab
le
2
.
E
x
p
er
im
en
tal
s
etu
p
s
im
u
latio
n
p
ar
a
m
eter
s
an
d
t
h
eir
v
alu
es
P
a
r
a
me
t
e
r
V
a
l
u
e
s
Tr
a
n
sm
i
ssi
o
n
r
a
n
g
e
2
0
0
m
N
u
mb
e
r
o
f
n
o
d
e
s
20
R
o
u
t
i
n
g
p
r
o
t
o
c
o
l
M
a
x
p
r
o
p
,
Ep
i
d
e
mi
c
,
P
r
o
P
h
e
t
S
i
mu
l
a
t
i
o
n
a
r
e
a
1
0
0
0
m
b
y
1
0
0
0
m
Tr
a
v
e
l
s
p
e
e
d
2
ms
A
n
t
e
n
n
a
t
y
p
e
O
mn
i
A
n
t
e
n
n
a
P
r
o
p
a
g
a
t
i
o
n
Tw
o
R
a
y
G
r
o
u
n
d
(
R
a
d
i
o
)
M
A
C
I
EEE
8
0
2
.
1
1
Tr
a
f
f
i
c
t
y
p
e
C
B
R
C
o
n
st
a
n
t
B
i
t
R
a
t
e
-
U
D
P
t
r
a
f
f
i
c
I
n
t
e
r
f
a
c
e
Q
u
e
u
e
P
r
i
o
r
i
t
y
Q
u
e
u
e
Q
u
e
u
e
l
e
n
g
t
h
5
0
(
ma
x
p
a
c
k
e
t
s
i
n
q
u
e
u
e
)
N
e
t
w
o
r
k
i
n
t
e
r
f
a
c
e
W
i
r
e
l
e
ssP
h
y
M
e
ss
a
g
e
g
e
n
e
r
a
t
i
o
n
r
a
t
e
1
0
0
K
b
p
s
P
a
c
k
e
t
si
z
e
1
0
0
0
b
y
t
e
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
1
4
9
-
3
1
6
1
3156
Fig
u
r
e
4
.
Simu
latio
n
o
f
n
o
d
es
in
NS2
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
test
s
d
if
f
er
e
n
t
m
o
d
els
to
id
en
tif
y
th
e
m
o
s
t
ef
f
ec
tiv
e
tech
n
i
q
u
e
f
o
r
attac
k
d
etec
tio
n
in
DT
Ns.
Scik
it
-
lear
n
an
d
Py
C
ar
et
m
o
d
els
ar
e
tr
ain
ed
an
d
co
m
p
ar
ed
f
o
r
p
er
f
o
r
m
an
ce
.
T
h
is
an
aly
s
is
is
cr
u
cial
f
o
r
d
ev
elo
p
i
n
g
a
n
e
f
f
icien
t a
ttack
d
etec
t
io
n
s
y
s
tem
in
th
e
DT
N
en
v
ir
o
n
m
e
n
t.
T
h
e
m
o
d
el
is
e
v
alu
ated
u
s
in
g
th
e
NS2
s
im
u
lato
r
in
a
s
im
u
lated
en
v
ir
o
n
m
en
t
.
3
.
1
.
P
er
f
o
rma
nce
co
m
pa
riso
n w
hil
e
us
i
ng
Scik
it
-
lea
rn
m
o
del
s
T
h
e
class
ically
tr
ain
ed
m
o
d
el
is
test
ed
with
R
F,
XG
B
o
o
s
t,
C
atB
o
o
s
t,
L
R
,
DT
,
E
x
tr
a
T
r
ee
s
,
an
d
L
ig
h
t
GB
M
cla
s
s
if
ier
s
.
L
ig
h
tGB
M
lead
s
th
e
co
m
p
etitio
n
in
ter
m
s
o
f
ac
cu
r
ac
y
,
AUC,
r
ec
all,
an
d
F1
-
S
co
r
e.
DT
d
is
tin
g
u
is
h
es
o
u
t
f
o
r
its
s
h
o
r
t
ti
m
e
in
v
estme
n
t.
C
atB
o
o
s
t
p
er
f
o
r
m
s
s
im
ilar
ly
to
o
th
er
h
ig
h
-
p
er
f
o
r
m
a
n
ce
m
o
d
els,
alth
o
u
g
h
it
tak
es
s
ig
n
if
ican
tly
lo
n
g
er
to
ex
ec
u
te.
L
R
r
an
k
s
last
in
to
tal
m
o
d
el
p
er
f
o
r
m
an
c
e
wh
ile
tak
in
g
less
tim
e.
R
F,
XG
B
o
o
s
t,
DT
,
an
d
ex
tr
a
tr
ee
s
all
p
er
f
o
r
m
well,
a
lb
eit
th
eir
ex
ec
u
tio
n
tim
es
v
ar
y
.
R
F
h
as
a
lo
n
g
er
ex
ec
u
tio
n
tim
e
t
h
an
L
R
,
L
ig
h
t
GB
M
,
an
d
ex
tr
a
tr
e
es.
Fig
u
r
e
5
s
h
o
ws
co
m
p
ar
is
o
n
o
f
ML
alg
o
r
ith
m
u
s
in
g
Scik
it
-
lear
n
.
Fig
u
r
e
5
.
Mo
d
el
co
m
p
ar
is
o
n
i
n
ML
m
o
d
el
u
s
in
g
Scik
it
-
lear
n
3
.
2
.
P
er
f
o
rma
nce
co
m
pa
riso
n w
hil
e
us
i
ng
P
y
Ca
re
t
Af
ter
ap
p
ly
in
g
th
e
Py
C
ar
et
m
o
d
el
to
th
e
f
lo
o
d
attac
k
d
ataset,
th
e
m
o
d
el
is
tes
ted
with
a
v
ar
iety
o
f
m
eth
o
d
s
,
i
n
clu
d
in
g
GB
,
k
-
n
e
ar
est
n
eig
h
b
o
r
s
(
K
-
NN)
,
Ad
a
B
o
o
s
t,
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
es
(
SVM)
,
lin
ea
r
0
0.5
1
1.5
2
2.5
3
3.5
4
R
a
n
d
o
m
F
o
r
es
t
XG
B
o
o
s
t
Ca
t
B
o
o
s
t
L
o
g
i
s
t
i
c
R
eg
r
es
s
i
o
n
D
ec
i
s
i
o
n
T
r
ee
E
xt
r
a
Tr
e
e
s
L
i
g
h
t
G
B
M
S
c
o
re
s
M
o
de
l
Mo
del
co
m
pa
r
is
o
n
A
c
c
u
racy
A
U
C
R
e
c
all
F
1 s
c
o
re
T
i
me
T
a
k
e
n
(
s
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
E
xp
lo
r
in
g
th
e
effec
tiven
ess
o
f
h
yb
r
id
a
r
tifi
cia
l b
ee
P
yCa
r
et
cla
s
s
ifie
r
in
d
ela
y
…
(
R
a
j
a
s
h
r
i Ch
a
u
d
h
a
r
i
)
3157
d
is
cr
im
in
an
t
an
aly
s
is
(
L
DA)
,
R
id
g
e,
q
u
a
d
r
atic
d
is
cr
im
in
an
t
an
aly
s
is
(
QDA)
,
Naïv
e
B
ay
es
(
NB
)
,
an
d
Du
m
m
y
class
if
ier
s
.
L
ig
h
tGB
M
p
er
f
o
r
m
s
well
ac
r
o
s
s
m
u
ltip
le
m
etr
i
cs,
with
g
o
o
d
Acc
u
r
ac
y
(
0
.
9
9
7
4
)
,
AUC
(
0
.
9
9
9
9
)
,
R
ec
all
(
0
.
9
9
7
4
)
,
Pre
cisi
o
n
(
0
.
9
9
7
5
)
,
F1
-
Sco
r
e
(
0
.
9
9
7
4
)
,
an
d
Kap
p
a
(
0
.
9
9
4
9
)
.
Ad
d
itio
n
al
ly
,
XGBo
o
s
t,
ex
tr
a
t
r
ee
s
,
R
F,
an
d
DT
wo
r
k
s
im
ila
r
ly
to
L
ig
h
tGB
M.
Ho
wev
er
,
C
atB
o
o
s
t
h
as
th
e
lo
n
g
est
ex
ec
u
tio
n
tim
e
as
s
h
o
wn
in
T
ab
le
3
.
T
ab
le
3
.
C
o
m
p
a
r
in
g
p
er
f
o
r
m
a
n
ce
o
f
ML
m
o
d
els in
Py
C
ar
et
M
o
d
e
l
A
c
c
u
r
a
c
y
AUC
R
e
c
a
l
l
P
r
e
c
i
s
i
o
n
F1
-
S
c
o
r
e
K
a
p
p
a
M
C
C
TT
(
S
)
Li
g
h
t
G
B
M
0
.
9
9
7
4
0
.
9
9
9
9
0
.
9
9
7
4
0
.
9
9
7
5
0
.
9
9
7
4
0
.
9
9
4
9
0
.
9
9
4
9
0
.
4
2
5
0
X
G
B
o
o
st
0
.
9
9
7
2
0
.
9
9
9
9
0
.
9
9
7
2
0
.
9
9
7
2
0
.
9
9
7
2
0
.
9
9
4
3
0
.
9
9
4
3
0
.
1
1
5
0
C
a
t
B
o
o
s
t
0
.
9
9
6
9
0
.
9
9
9
8
0
.
9
9
6
9
0
.
9
9
6
9
0
.
9
9
6
9
0
.
9
9
3
8
0
.
9
9
3
9
1
.
7
4
3
0
Ex
t
r
a
Tr
e
e
s
0
.
9
9
6
8
0
.
9
9
9
9
0
.
9
9
6
8
0
.
9
9
6
8
0
.
9
9
6
8
0
.
9
9
3
5
0
.
9
9
3
5
0
.
1
6
4
0
RF
0
.
9
9
6
0
0
.
9
9
9
9
0
.
9
9
6
0
0
.
9
9
6
0
0
.
9
9
6
0
0
.
9
9
2
0
0
.
9
9
2
0
0
.
1
8
7
0
DT
0
.
9
9
4
7
0
.
9
9
4
6
0
.
9
9
4
7
0
.
9
9
4
7
0
.
9
9
4
7
0
.
9
8
9
3
0
.
9
8
9
3
0
.
0
7
0
0
GB
0
.
9
9
2
1
0
.
9
9
9
6
0
.
9
9
2
1
0
.
9
9
2
1
0
.
9
9
2
1
0
.
9
8
4
2
0
.
9
8
4
2
0
.
5
5
9
0
K
-
NN
0
.
9
9
1
6
0
.
9
9
7
5
0
.
9
9
1
6
0
.
9
9
1
6
0
.
9
9
1
6
0
.
9
8
3
1
0
.
9
8
3
2
0
.
1
3
0
0
A
d
a
B
o
o
st
0
.
9
8
3
0
0
.
9
9
8
7
0
.
9
8
3
0
0
.
9
8
3
1
0
.
9
8
3
0
0
.
9
6
5
9
0
.
9
6
6
0
0
.
2
0
0
0
S
V
M
0
.
9
7
1
5
0
.
0
0
0
0
0
.
9
7
1
5
0
.
9
7
1
7
0
.
9
7
1
5
0
.
9
4
2
6
0
.
9
4
2
9
0
.
0
6
2
0
LR
0
.
9
7
0
4
0
.
9
9
4
7
0
.
9
7
0
4
0
.
9
7
0
6
0
.
9
7
0
4
0
.
9
4
0
4
0
.
9
4
0
7
0
.
1
0
2
0
LD
A
0
.
9
6
3
9
0
.
9
9
3
6
0
.
9
6
3
9
0
.
9
6
4
1
0
.
9
6
3
8
0
.
9
2
7
3
0
.
9
2
7
6
0
.
0
7
4
0
R
i
d
g
e
0
.
9
6
3
4
0
.
0
0
0
0
0
.
9
6
3
4
0
.
9
6
3
7
0
.
9
6
3
4
0
.
9
2
6
4
0
.
9
2
6
7
0
.
0
5
7
0
QDA
0
.
9
5
3
1
0
.
9
7
0
2
0
.
9
5
3
1
0
.
9
5
4
9
0
.
9
5
3
0
0
.
9
0
5
5
0
.
9
0
7
5
0
.
0
6
7
0
NB
0
.
9
2
4
8
0
.
9
6
9
7
0
.
9
2
4
8
0
.
9
2
5
2
0
.
9
2
4
8
0
.
8
4
9
0
0
.
8
4
9
4
0
.
0
6
4
0
D
u
mm
y
0
.
5
3
3
9
0
.
5
0
0
0
0
.
5
3
3
9
0
.
2
8
5
0
0
.
3
7
1
6
0
.
0
0
0
0
0
.
0
0
0
0
0
.
0
7
3
0
L
ig
h
tGB
M
o
u
tp
er
f
o
r
m
s
XGBo
o
s
t
s
lig
h
tly
with
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
in
ac
cu
r
ac
y
,
A
UC
,
r
ec
all,
p
r
ec
is
io
n
,
F1
,
Kap
p
a
,
a
n
d
MCC
.
DT
h
as
f
aster
ex
ec
u
tio
n
b
u
t
w
o
r
s
e
o
v
er
all
p
er
f
o
r
m
an
ce
co
m
p
ar
e
d
to
L
ig
h
tGB
M,
XGBo
o
s
t,
C
atB
o
o
s
t,
E
x
tr
a
T
r
ee
s
,
an
d
R
F
m
o
d
els.
R
id
g
e,
SVM
,
LR
,
L
DA
,
QD
A
,
NB
,
an
d
Du
m
m
y
class
if
ier
p
er
f
o
r
m
p
o
o
r
ly
with
s
h
o
r
ter
e
x
ec
u
tio
n
tim
e.
Fig
u
r
e
6
d
e
p
icts
co
m
p
ar
is
o
n
o
f
M
L
alg
o
r
ith
m
s
u
s
in
g
Py
C
ar
et.
Fig
u
r
e
6
.
Mo
d
el
C
o
m
p
ar
is
o
n
in
ML
m
o
d
el
u
s
in
g
Py
C
ar
et
3
.
3
.
Art
if
icia
l
bee
co
lo
ny
o
ptim
iz
er
T
h
ese
s
co
r
es
ar
e
th
e
o
u
tco
m
e
o
f
a
10
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
.
T
ab
le
4
s
h
o
ws
p
er
f
o
r
m
an
ce
m
etr
ics
f
o
r
ea
ch
o
f
th
e
te
n
f
o
ld
s
.
Par
ticu
l
ar
ly
,
f
o
ld
0
h
as
t
h
e
b
est
p
e
r
f
o
r
m
an
ce
,
with
h
ig
h
e
r
m
etr
ics
a
cr
o
s
s
all
ca
teg
o
r
ies,
in
clu
d
in
g
ac
cu
r
ac
y
(
0
.
9
9
7
7
)
,
AUC
(
0
.
9
9
7
7
)
,
r
ec
all
(
0
.
9
9
7
7
)
,
p
r
ec
is
io
n
(
0
.
9
9
7
7
)
,
F1
-
S
co
r
e
(
0
.
9
9
7
7
)
,
Kap
p
a
co
ef
f
icien
t
(
0
.
9
9
5
4
)
,
an
d
MCC
(
0
.
9
9
5
4
)
.
As
a
r
esu
lt,
th
e
o
t
h
er
f
o
ld
s
p
er
f
o
r
m
s
ig
n
if
ican
tl
y
wo
r
s
e
th
an
f
o
l
d
0
.
Fo
ld
1
h
as
th
e
lo
west
ac
c
u
r
ac
y
co
m
p
ar
ed
to
th
e
o
th
e
r
f
o
ld
s
.
T
h
e
a
v
er
ag
e
p
er
f
o
r
m
a
n
ce
o
f
all
f
o
ld
s
is
ac
cu
r
ac
y
(
0
.
9
9
5
7
)
,
AUC
(
0
.
9
9
5
7
)
,
r
ec
al
l
(
0
.
9
9
5
7
)
,
p
r
ec
is
io
n
(
0
.
9
9
5
7
)
,
F1
(
0
.
9
9
5
7
)
,
k
ap
p
a
(
0
.
9
9
1
3
)
,
a
n
d
MCC
(
0
.
9
9
1
3
)
.
T
h
e
s
tan
d
ar
d
d
e
v
iatio
n
o
f
all
f
o
ld
s
is
ac
cu
r
ac
y
,
AUC,
r
ec
all,
p
r
ec
is
io
n
,
F1
is
0
.
0
0
1
9
an
d
Ka
p
p
a,
MC
C
is
0
.
0
0
3
9
.
0
0
.
2
0
.
4
0
.
6
0
.
8
1
1
.
2
1
.
4
1
.
6
1
.
8
2
S
c
o
re
s
Mo
d
el
P
y
C
ar
et
Mode
l
C
om
par
i
son
Ac
c
u
ra
c
y
AUC
Re
c
a
ll
P
re
c
isio
n
F
1
-sc
o
re
Ka
p
p
a
M
CC
TT
(S
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
15
,
No
.
3
,
J
u
n
e
20
25
:
3
1
4
9
-
3
1
6
1
3158
T
ab
le
4
.
Fo
ld
c
r
o
s
s
-
v
alid
atio
n
s
co
r
e
A
c
c
u
r
a
c
y
AUC
R
e
c
a
l
l
P
r
e
c
i
s
i
o
n
F1
-
S
c
o
r
e
K
a
p
p
a
M
C
C
F
o
l
d
0
0
.
9
9
7
7
0
.
9
9
7
7
0
.
9
9
7
7
0
.
9
9
7
7
0
.
9
9
7
7
0
.
9
9
5
4
0
.
9
9
5
4
1
0
.
9
9
0
9
0
.
9
9
1
1
0
.
9
9
0
9
0
.
9
9
1
0
0
.
9
9
0
9
0
.
9
8
1
8
0
.
9
8
1
8
2
0
.
9
9
6
0
0
.
9
9
6
1
0
.
9
9
6
0
0
.
9
9
6
0
0
.
9
9
6
0
0
.
9
9
2
0
0
.
9
9
2
0
3
0
.
9
9
4
3
0
.
9
9
4
2
0
.
9
9
4
3
0
.
9
9
4
3
0
.
9
9
4
3
0
.
9
8
8
6
0
.
9
8
8
6
4
0
.
9
9
6
0
0
.
9
9
6
2
0
.
9
9
6
0
0
.
9
9
6
0
0
.
9
9
6
0
0
.
9
9
2
0
0
.
9
9
2
0
5
0
.
9
9
4
3
0
.
9
9
4
2
0
.
9
9
4
3
0
.
9
9
4
3
0
.
9
9
4
3
0
.
9
8
8
6
0
.
9
8
8
6
6
0
.
9
9
7
2
0
.
9
9
7
1
0
.
9
9
7
2
0
.
9
9
7
2
0
.
9
9
7
2
0
.
9
9
4
3
0
.
9
9
4
3
7
0
.
9
9
7
2
0
.
9
9
7
0
0
.
9
9
7
2
0
.
9
9
7
2
0
.
9
9
7
2
0
.
9
9
4
3
0
.
9
9
4
3
8
0
.
9
9
6
0
0
.
9
9
6
0
0
.
9
9
6
0
0
.
9
9
6
0
0
.
9
9
6
0
0
.
9
9
2
0
0
.
9
9
2
0
9
0
.
9
9
7
2
0
.
9
9
7
2
0
.
9
9
7
2
0
.
9
9
7
2
0
.
9
9
7
2
0
,
9
9
4
3
0
.
9
9
4
3
M
e
a
n
0
.
9
9
5
7
0
.
9
9
5
7
0
.
9
9
5
7
0
.
9
9
5
7
0
.
9
9
5
7
0
.
9
9
1
3
0
.
9
9
1
3
S
t
d
0
.
0
0
1
9
0
.
0
0
1
9
0
.
0
0
1
9
0
.
0
0
1
9
0
.
0
0
1
9
0
.
0
0
3
9
0
.
0
0
3
9
3
.
4
Co
m
pa
ring
t
he
m
e
a
n v
a
lue
bef
o
re
a
nd
a
f
t
er
o
ptim
iza
t
io
n
Fig
u
r
e
7
s
h
o
ws
th
e
co
m
p
ar
is
o
n
o
f
m
ea
n
v
alu
es
b
ef
o
r
e
an
d
a
f
ter
o
p
tim
izatio
n
o
n
v
ar
i
o
u
s
p
er
f
o
r
m
a
n
ce
m
etr
ics.
A
co
m
p
ar
is
o
n
is
d
o
n
e
b
etwe
en
th
e
m
o
d
el'
s
p
er
f
o
r
m
a
n
ce
b
ef
o
r
e
a
n
d
a
f
ter
o
p
tim
izatio
n
.
Fig
u
r
e
7
d
ep
icts
h
o
w
th
e
ac
cu
r
ac
y
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el
im
p
r
o
v
es
af
ter
o
p
tim
izatio
n
.
A
d
d
itio
n
ally
,
m
u
ltip
le
p
er
f
o
r
m
a
n
ce
in
d
icato
r
s
s
u
ch
as
AUC,
r
ec
a
ll,
p
r
ec
is
io
n
,
F1
-
S
co
r
e,
Kap
p
a,
an
d
Ma
tth
ews
co
r
r
elatio
n
co
ef
f
icien
t
(
MCC
)
s
h
o
ws im
p
r
o
v
em
en
t a
f
ter
ad
ju
s
tin
g
v
alu
es.
B
ef
o
r
e
o
p
tim
izati
o
n
,
th
e
m
o
d
el'
s
T
T
(
S)
was 0
.
2
5
5
8
,
in
d
ic
atin
g
th
e
co
m
p
u
tatio
n
al
ef
f
o
r
t
in
v
o
l
v
ed
.
Ho
wev
er
,
af
ter
o
p
tim
izatio
n
,
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
im
p
r
o
v
es
p
er
f
o
r
m
an
ce
ac
r
o
s
s
v
ar
io
u
s
cr
iter
ia,
in
d
icatin
g
a
m
o
r
e
ef
f
icien
t a
n
d
e
f
f
ec
ti
v
e
m
o
d
el.
Fig
u
r
e
7
.
C
o
m
p
a
r
in
g
t
h
e
m
ea
n
v
alu
e
b
ef
o
r
e
a
n
d
af
ter
o
p
tim
i
za
tio
n
3
.
5
.
P
er
f
o
rma
nce
co
m
pa
riso
n o
f
t
he
pro
po
s
ed
m
o
del w
it
h e
x
is
t
ing
m
o
dels
T
ab
le
5
s
h
o
ws co
m
p
ar
is
o
n
o
f
ar
tific
ial
b
ee
Py
C
ar
et
C
lass
if
ie
r
m
o
d
el
p
er
f
o
r
m
an
ce
with
o
th
er
ex
is
tin
g
m
o
d
els.
T
h
e
s
u
g
g
ested
ar
tific
i
al
b
ee
Py
C
ar
et
c
las
s
if
ier
o
u
tp
er
f
o
r
m
s
v
ar
io
u
s
cu
r
r
en
t
m
o
d
els,
in
clu
d
in
g
Py
C
ar
et
u
s
in
g
th
e
r
an
d
o
m
f
o
r
est
(
R
F)
alg
o
r
ith
m
,
Py
C
ar
et
u
s
in
g
e
x
tr
em
e
g
r
ad
ien
t
b
o
o
s
tin
g
,
th
e
NP
tech
n
iq
u
e,
an
d
Au
to
ML
u
s
in
g
th
e
Py
C
ar
et
m
o
d
el.
T
h
is
d
em
o
n
s
tr
ates
th
e
ef
f
icac
y
o
f
th
e
tec
h
n
iq
u
e
in
en
h
an
cin
g
p
r
ed
ictio
n
p
er
f
o
r
m
an
ce
.
T
h
e
a
r
tific
ial
B
e
e
Py
C
ar
et
c
las
s
if
ier
tak
es
ad
v
an
tag
e
o
f
th
e
ca
p
a
b
ilit
ies
o
f
b
o
th
Py
C
ar
et
an
d
AB
C
alg
o
r
ith
m
s
,
r
esu
ltin
g
in
m
o
r
e
ac
cu
r
ate
an
d
r
o
b
u
s
t m
o
d
el
p
r
e
d
ictio
n
s
.
Fig
u
r
e
8
d
ep
icts
co
m
p
ar
is
o
n
o
f
p
r
o
p
o
s
ed
m
o
d
el
an
d
ex
is
tin
g
alg
o
r
ith
m
s
as a
b
o
v
e.
T
ab
le
5
.
C
o
m
p
a
r
is
o
n
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el
an
d
o
th
e
r
ex
is
tin
g
alg
o
r
ith
m
s
R
e
f
e
r
e
n
c
e
M
o
d
e
l
A
l
g
o
r
i
t
h
m
A
c
c
u
r
a
c
y
TT
(
S
)
Th
e
p
r
o
p
o
s
e
d
m
o
d
e
l
H
y
b
r
i
d
A
B
P
C
Li
g
h
t
G
B
M
0
.
9
9
5
7
0
.
4
2
5
0
P
y
C
a
r
e
t
RF
0
.
9
9
6
0
0
.
1
8
7
0
B
a
j
p
a
i
a
n
d
S
h
a
r
m
a
[
2
7
]
P
y
C
a
r
e
t
Ex
G
B
o
o
st
0
.
8
3
7
1
-
P
y
C
a
r
e
t
RF
0
.
8
3
5
5
-
B
a
r
o
u
m
a
n
d
e
t
a
l
.
[
2
8
]
N
P
a
p
p
r
o
a
c
h
RF
0
.
7
5
7
-
H
a
ssan
a
n
d
A
l
s
h
a
r
e
e
f
[
2
9
]
A
u
t
o
M
L
u
si
n
g
P
y
C
a
r
e
t
RF
0
.
9
9
1
6
-
0
0.2
0.4
0.6
0.8
1
1.2
A
c
c
u
r
a
c
y
AUC
R
e
c
a
l
l
Pr
e
c
i
si
o
n
F
1
-
sc
o
r
e
K
a
p
p
a
M
C
C
TT
(
S
)
V
a
l
u
e
s
P
er
f
o
r
m
an
ce
Me
tr
ics
C
om
par
i
ng t
he
m
ean
v
al
ue bef
or
e
and
af
t
er
opt
i
m
i
zat
i
on
Befo
re op
ti
m
izati
o
n
Afte
r
Op
t
im
i
z
a
ti
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.