I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
1
5
,
No
.
5
,
Octo
b
er
2
0
2
6
,
p
p
.
4
3
6
6
~
4
3
8
0
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
15
.i
5
.
p
p
4
3
6
6
-
4
3
8
0
4366
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
H
y
brid
feature
se
lection st
ra
tegies
for strong
ne
twor
k int
rusio
n
detec
tion
J
o
s
ephin
e
Ra
t
hin
a
m
,
Ana
nd
a
ra
j
Sh
a
nthi P
icha
nd
i
P
r
e
si
d
e
n
c
y
S
c
h
o
o
l
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
P
r
e
s
i
d
e
n
c
y
U
n
i
v
e
r
si
t
y
,
B
e
n
g
a
l
u
r
u
,
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
Sep
3
0
,
2
0
2
5
R
ev
is
ed
J
u
l 2
5
,
2
0
2
6
Acc
ep
ted
Au
g
1
2
,
2
0
2
6
Hig
h
d
ime
n
sio
n
a
l
n
e
two
r
k
traffic
d
a
tas
e
ts
p
o
se
g
re
a
t
c
h
a
ll
e
n
g
e
s
t
o
n
e
two
r
k
in
tru
si
o
n
d
e
tec
ti
o
n
s
y
ste
m
s
(NID
S
)
d
u
e
t
o
re
d
u
n
d
a
n
t
a
n
d
irrele
v
a
n
t
fe
a
tu
re
s
,
wh
ich
d
e
c
re
a
se
th
e
d
e
tec
ti
o
n
a
c
c
u
ra
c
y
a
n
d
in
c
re
a
se
th
e
c
o
m
p
u
tati
o
n
a
l
c
o
st.
Trad
it
io
n
a
l
fe
a
tu
re
se
lec
ti
o
n
m
e
th
o
d
s
o
ften
c
a
n
n
o
t
a
c
h
iev
e
a
b
a
lan
c
e
b
e
twe
e
n
c
las
sifica
ti
o
n
p
e
rfo
rm
a
n
c
e
a
n
d
p
r
o
c
e
ss
in
g
e
fficie
n
c
y
.
Th
is
p
a
p
e
r
p
ro
p
o
se
s
h
y
b
rid
fe
a
tu
re
se
lec
ti
o
n
f
o
r
n
e
two
r
k
in
tr
u
sio
n
d
e
tec
ti
o
n
s
y
ste
m
s
(
Hy
b
F
S
-
NID
S
)
,
a
h
y
b
ri
d
fe
a
tu
r
e
se
lec
ti
o
n
fra
m
e
wo
rk
u
sin
g
f
i
lt
e
r
-
a
n
d
wra
p
p
e
r
-
b
a
se
d
o
p
ti
m
iza
ti
o
n
m
e
th
o
d
s
t
o
d
e
tec
t
th
e
m
o
st
sig
n
ifi
c
a
n
t
in
tru
si
o
n
d
e
tec
ti
o
n
fe
a
tu
re
s
t
o
d
o
t
h
is.
Th
e
p
ro
p
o
se
d
m
o
d
e
l
wa
s
e
v
a
l
u
a
ted
o
n
th
e
CIC
-
IDS2
0
1
7
d
a
tas
e
t
u
sin
g
p
e
rfo
rm
a
n
c
e
m
e
tri
c
s
su
c
h
a
s
a
c
c
u
ra
c
y
,
p
re
c
isio
n
,
re
c
a
ll
,
F
1
-
sc
o
re
,
a
n
d
M
a
tt
h
e
w’s
c
o
rre
latio
n
c
o
e
fficie
n
t
(M
C
C).
T
h
e
e
x
p
e
rime
n
tal
re
su
lt
s
se
lec
ted
1
8
o
p
ti
m
a
l
fe
a
tu
re
s
fro
m
th
e
in
it
ial
7
8
fe
a
tu
re
s
o
f
th
e
d
a
tas
e
t
a
n
d
a
c
h
ie
v
e
d
9
0
.
6
8
%
a
c
c
u
ra
c
y
,
9
1
.
7
4
%
p
re
c
isio
n
,
9
0
.
6
4
%
re
c
a
ll
,
9
1
.
1
9
%
F
1
-
sc
o
re
,
a
n
d
a
n
M
CC
o
f
0
.
9
1
4
8
.
T
h
e
p
ro
p
o
se
d
m
e
th
o
d
a
lso
a
c
h
iev
e
d
a
4
1
.
5
%
re
d
u
c
ti
o
n
in
tr
a
in
in
g
ti
m
e
c
o
m
p
a
re
d
with
th
e
f
u
ll
-
fe
a
tu
re
b
a
se
li
n
e
m
o
d
e
l
wh
il
e
imp
ro
v
in
g
th
e
o
v
e
ra
ll
c
las
sifica
ti
o
n
p
e
rf
o
rm
a
n
c
e
.
Th
e
re
su
lt
s
sh
o
w
th
a
t
Hy
b
F
S
-
NID
S
c
a
n
su
c
c
e
ss
fu
ll
y
imp
ro
v
e
t
h
e
e
fficie
n
c
y
o
f
in
tru
si
o
n
d
e
tec
ti
o
n
a
n
d
f
it
fo
r
r
e
a
l
-
ti
m
e
c
y
b
e
rse
c
u
rit
y
a
p
p
li
c
a
ti
o
n
s
with
a
larg
e
a
m
o
u
n
t
o
f
n
e
two
r
k
traffic d
a
ta.
K
ey
w
o
r
d
s
:
C
I
C
-
I
DS2
0
1
7
d
ataset
Gen
etic
alg
o
r
ith
m
Hy
b
r
id
f
ea
t
u
r
e
s
elec
tio
n
Mu
lti
-
lay
er
p
er
ce
p
tr
o
n
Netwo
r
k
in
tr
u
s
io
n
d
etec
tio
n
s
y
s
tem
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
:
J
o
s
ep
h
in
e
R
ath
in
am
Pre
s
id
en
cy
Sch
o
o
l o
f
C
o
m
p
u
t
er
Scien
ce
an
d
E
n
g
in
ee
r
i
n
g
,
P
r
esid
en
cy
Un
iv
er
s
ity
B
en
g
alu
r
u
,
I
n
d
ia
E
m
ail:
jo
s
ep
h
in
e.
2
0
2
2
3
cse0
0
2
6
@
p
r
esid
en
cy
u
n
iv
e
r
s
ity
.
in
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
g
r
o
wth
o
f
d
ig
ital
co
m
m
u
n
icatio
n
an
d
th
e
s
p
r
ea
d
o
f
in
t
er
n
et
-
co
n
n
ec
ted
d
ev
ices
h
a
v
e
led
to
an
in
cr
ea
s
e
in
th
e
p
r
e
v
alen
ce
o
f
cy
b
er
attac
k
s
o
n
c
o
m
p
u
tin
g
s
y
s
tem
s
[
1
]
.
Netwo
r
k
in
tr
u
s
io
n
d
etec
tio
n
s
y
s
tem
s
(
NI
DS)
ar
e
u
s
ed
to
m
o
n
ito
r
n
etwo
r
k
tr
af
f
ic
f
o
r
m
alicio
u
s
ac
tiv
ity
an
d
to
en
s
u
r
e
d
ata
s
ec
u
r
ity
,
in
clu
d
in
g
av
ailab
ilit
y
,
co
n
f
id
en
tiality
,
a
n
d
in
teg
r
ity
[
2
]
.
I
n
m
o
d
er
n
h
i
g
h
-
d
im
e
n
s
io
n
al
d
ata
s
u
ch
as
C
I
C
-
I
DS2
0
1
7
,
th
er
e
ex
is
ts
r
elev
an
t
d
ata
an
d
ir
r
ele
v
an
t
in
f
o
r
m
atio
n
[
3
]
.
T
h
e
r
ed
u
n
d
an
t
f
ea
tu
r
es
in
th
e
d
ataset
also
co
n
tr
ib
u
te
to
th
e
co
m
p
u
tin
g
tim
e,
co
m
p
lex
ity
,
an
d
o
v
e
r
f
itti
n
g
r
is
k
o
f
th
e
m
o
d
el
,
wh
ich
d
ec
r
ea
s
es
d
etec
tio
n
p
er
f
o
r
m
an
ce
[
4
]
.
R
ea
l
-
tim
e
in
tr
u
s
io
n
d
etec
tio
n
is
v
er
y
d
if
f
icu
lt
b
ec
a
u
s
e
o
f
h
ig
h
-
d
im
e
n
s
io
n
al
n
etwo
r
k
tr
af
f
ic
d
ata
[
5
]
.
Ma
n
y
tr
ad
itio
n
al
f
ea
tu
r
e
s
ele
ctio
n
m
eth
o
d
s
d
o
n
o
t
s
u
f
f
icie
n
tly
r
eso
lv
e
th
e
tr
a
d
e
-
o
f
f
b
et
wee
n
class
if
icatio
n
ac
cu
r
ac
y
an
d
th
e
co
m
p
u
tin
g
ec
o
n
o
m
y
[
6
]
.
T
h
is
m
ak
es
it
eith
er
h
av
e
lo
w
d
etec
tio
n
ac
cu
r
ac
y
o
r
a
h
ig
h
co
m
p
u
tin
g
co
s
t
[
7
]
.
An
ef
f
icien
t
f
ea
tu
r
e
s
elec
tio
n
m
eth
o
d
th
at
g
u
ar
an
tees
h
ig
h
ac
cu
r
ac
y
as
well
as
r
ed
u
ce
d
co
m
p
lex
ity
is
th
er
ef
o
r
e
n
ec
ess
ar
y
to
o
f
f
er
e
f
f
icien
t r
ea
l
-
tim
e
NI
DS f
u
n
ctio
n
ality
[
8
]
.
I
n
th
e
ca
s
e
o
f
in
tr
u
s
io
n
d
etec
tio
n
,
r
esear
ch
er
s
h
av
e
u
s
ed
m
o
s
t
f
ea
tu
r
e
s
elec
tio
n
m
eth
o
d
s
o
v
er
th
e
y
ea
r
s
[
9
]
,
[
1
0
]
.
B
ec
au
s
e
o
f
th
eir
lo
w
c
o
m
p
u
tatio
n
al
co
s
t,
f
ilter
-
b
ased
m
et
h
o
d
s
s
u
c
h
as
in
f
o
r
m
atio
n
g
ai
n
(
I
G)
,
ch
i
-
s
q
u
ar
e
test
,
a
n
d
m
u
tu
al
i
n
f
o
r
m
atio
n
ar
e
o
f
ten
u
s
ed
as
in
tr
u
s
io
n
d
etec
tio
n
m
eth
o
d
s
.
T
h
ey
h
a
v
e
also
b
ee
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Hyb
r
id
fea
tu
r
e
s
elec
tio
n
s
tr
a
te
g
ies fo
r
s
tr
o
n
g
n
etw
o
r
k
in
tr
u
s
i
o
n
d
etec
tio
n
(
Jo
s
ep
h
i
n
e
R
a
th
i
n
a
m
)
4367
u
s
ed
b
ec
au
s
e
o
f
th
e
a
b
ilit
y
o
f
wr
ap
p
er
tech
n
iq
u
es
s
u
ch
as
s
eq
u
en
tial
f
o
r
war
d
s
elec
tio
n
(
S
FS
)
an
d
s
eq
u
en
tial
b
ac
k
war
d
s
elec
tio
n
(
SB
S)
to
d
ir
ec
tly
o
p
tim
ize
f
ea
tu
r
e
s
u
b
s
ets,
d
ep
en
d
in
g
o
n
class
if
ier
p
er
f
o
r
m
a
n
ce
,
wh
ich
h
as
also
led
t
o
th
eir
u
tili
za
tio
n
.
I
n
o
r
d
er
t
o
s
tr
ik
e
a
b
ala
n
ce
b
etwe
en
p
r
ec
is
io
n
an
d
ef
f
ec
tiv
en
ess
,
s
o
m
e
h
y
b
r
i
d
s
y
s
tem
s
th
at
co
m
b
in
e
b
o
th
f
ilter
an
d
wr
ap
p
e
r
tech
n
iq
u
es
h
a
v
e
also
b
ee
n
d
ev
elo
p
e
d
.
Mo
r
e
o
v
er
,
NI
DS
f
ea
tu
r
e
s
elec
tio
n
h
as
also
m
ad
e
u
s
e
o
f
m
eta
-
h
eu
r
is
tic
o
p
tim
izatio
n
m
eth
o
d
s
s
u
ch
as
p
a
r
ticle
s
war
m
o
p
tim
izatio
n
(
PS
O)
,
an
t c
o
lo
n
y
o
p
tim
izatio
n
(
AC
O)
,
an
d
g
en
etic
alg
o
r
ith
m
(
GA)
.
T
h
ese
ex
is
tin
g
wo
r
k
s
h
a
v
e
th
e
f
o
llo
win
g
d
is
ad
v
an
tag
es:
f
il
ter
m
eth
o
d
s
ig
n
o
r
e
f
ea
tu
r
e
d
e
p
en
d
en
cies
an
d
g
en
e
r
ally
s
elec
t
ir
r
elev
a
n
t
f
ea
tu
r
es
alo
n
g
with
r
elev
a
n
t
o
n
es.
W
r
ap
p
er
m
eth
o
d
s
ar
e
co
m
p
u
tatio
n
ally
ex
p
en
s
iv
e
an
d
t
h
er
ef
o
r
e
n
o
t
s
u
itab
le
f
o
r
lar
g
e
-
s
ca
le
o
r
r
ea
l
-
tim
e
d
etec
tio
n
.
T
h
e
h
y
b
r
id
a
p
p
r
o
ac
h
es
u
s
ed
in
p
r
ev
io
u
s
wo
r
k
s
g
en
er
ally
co
m
b
in
e
b
asic
f
ilter
m
et
h
o
d
s
a
n
d
c
o
m
p
u
tatio
n
ally
e
x
p
en
s
iv
e
wr
ap
p
e
r
s
,
s
o
th
e
m
eth
o
d
s
d
ev
el
o
p
ed
a
r
e
n
o
t
s
ca
lab
le.
Ma
n
y
s
tu
d
ies
d
o
n
o
t
s
y
s
tem
atica
lly
co
m
b
in
e
th
e
s
tr
en
g
th
s
o
f
m
u
ltip
le
ad
v
a
n
ce
d
f
ilter
m
et
h
o
d
s
b
ef
o
r
e
a
p
p
ly
in
g
th
e
wr
ap
p
er
-
b
ased
o
p
tim
izatio
n
m
e
th
o
d
.
Op
tim
izatio
n
alg
o
r
ith
m
s
ar
e
ap
p
lied
with
o
u
t
f
ir
s
t
p
er
f
o
r
m
in
g
f
ea
tu
r
e
r
a
n
k
in
g
,
wh
ich
n
ee
d
less
ly
s
ea
r
ch
es
o
v
er
p
o
o
r
-
q
u
ality
f
ea
tu
r
es.
I
t
is
s
u
g
g
ested
in
th
is
r
esear
ch
th
at
a
n
o
v
el
tec
h
n
iq
u
e
ca
lled
h
y
b
r
id
f
ea
tu
r
e
s
el
ec
tio
n
f
o
r
n
etwo
r
k
in
tr
u
s
io
n
d
etec
tio
n
s
y
s
tem
s
(
Hy
b
FS
-
NI
DS)
is
in
tr
o
d
u
ce
d
to
ad
d
r
ess
th
e
af
o
r
e
m
en
tio
n
e
d
d
r
awb
ac
k
s
.
T
h
e
m
o
s
t
p
er
tin
en
t
ch
ar
ac
ter
is
tics
ar
e
o
p
tim
ized
th
r
o
u
g
h
th
e
in
teg
r
ati
o
n
o
f
m
u
ltip
le
ad
v
an
ce
d
f
ilter
tech
n
iq
u
es
with
a
m
eta
-
h
eu
r
is
tic
wr
ap
p
er
.
B
y
th
is
ap
p
r
o
ac
h
,
h
ig
h
d
etec
tio
n
ac
cu
r
ac
y
is
m
ain
tain
ed
,
a
n
d
c
o
m
p
u
tatio
n
al
co
m
p
lex
ity
is
r
ed
u
ce
d
.
T
o
p
r
o
d
u
ce
f
ea
tu
r
e
r
an
k
in
g
a
n
d
r
ed
u
n
d
an
c
y
in
f
o
r
m
atio
n
,
t
h
r
ee
ad
v
an
ce
d
f
ilter
tech
n
iq
u
es,
R
elief
F,
f
ast
co
r
r
elatio
n
-
b
ased
f
ilter
(
FC
B
F
)
,
an
d
m
in
im
u
m
r
ed
u
n
d
an
cy
m
ax
im
u
m
r
elev
a
n
ce
(
m
R
MR
)
,
ar
e
f
ir
s
t
ap
p
lied
in
th
e
p
r
o
p
o
s
ed
m
eth
o
d
.
T
h
e
o
b
tain
e
d
r
an
k
in
g
s
ar
e
th
en
co
m
b
in
ed
th
r
o
u
g
h
th
e
B
o
r
d
a
co
u
n
t
v
o
tin
g
m
eth
o
d
to
g
en
er
ate
a
to
p
-
r
a
n
k
ed
f
ea
tu
r
e
s
et
as
well.
T
h
is
r
e
d
u
ce
d
s
et
is
s
u
b
jecte
d
t
o
a
wr
ap
p
er
b
ased
o
n
GA.
Featu
r
e
s
u
b
s
ets
with
in
th
is
r
ed
u
ce
d
co
llectio
n
ar
e
e
v
alu
ate
d
u
s
in
g
a
m
u
lti
-
lay
er
p
e
r
ce
p
tr
o
n
(
MLP
)
class
if
ier
with
cr
o
s
s
-
v
alid
atio
n
.
T
h
e
wr
ap
p
er
-
s
elec
ted
an
d
f
ilter
-
r
an
k
e
d
f
ea
tu
r
es
ar
e
co
m
b
in
ed
b
y
an
ML
P
m
o
d
el
to
g
et
a
class
if
ier
-
awa
r
e
an
d
s
tatis
tically
im
p
o
r
tan
t
f
ea
tu
r
e
s
et.
T
h
is
s
tu
d
y
p
r
esen
t
Hy
b
FS
-
NI
DS,
a
h
y
b
r
id
f
ilter
-
wr
ap
p
er
m
eth
o
d
t
h
at
u
s
es
d
if
f
er
en
t
f
ilter
m
eth
o
d
s
,
t
h
e
B
o
r
d
a
co
u
n
t
r
an
k
in
g
m
et
h
o
d
,
a
n
d
GA
-
b
ased
im
p
r
o
v
em
e
n
t
in
th
is
p
ap
er
,
to
s
elec
t
s
m
aller
f
ea
tu
r
e
s
u
b
s
ets,
im
p
r
o
v
e
th
e
in
tr
u
s
io
n
d
etec
tio
n
ac
cu
r
ac
y
,
r
ed
u
ce
th
e
p
r
o
ce
s
s
in
g
co
s
t
an
d
tr
ain
i
n
g
tim
e
,
an
d
to
h
el
p
r
ea
l
-
tim
e
NI
DS
ap
p
licatio
n
s
.
Hy
b
FS
-
NI
DS
d
if
f
er
s
f
r
o
m
tr
ad
itio
n
al
m
eth
o
d
s
b
y
c
o
m
b
i
n
in
g
m
R
MR,
FC
B
F
,
an
d
R
elief
F
with
r
an
k
ag
g
r
eg
atio
n
a
n
d
s
et
in
ter
s
ec
tio
n
b
ef
o
r
e
GA
o
p
tim
izatio
n
,
th
u
s
im
p
r
o
v
in
g
th
e
d
etec
tio
n
p
er
f
o
r
m
an
ce
a
n
d
r
e
d
u
cin
g
th
e
co
m
p
u
tatio
n
al
co
m
p
lex
ity
.
T
h
er
ef
o
r
e,
th
e
p
r
o
p
o
s
ed
f
r
am
ewo
r
k
is
ap
p
licab
le
to
en
ter
p
r
is
e
an
d
g
o
v
er
n
m
e
n
t
n
etwo
r
k
s
,
clo
u
d
s
ec
u
r
ity
,
in
ter
n
et
o
f
th
in
g
s
(
I
o
T
)
an
d
i
n
d
u
s
tr
ial
s
y
s
tem
s
,
an
d
r
ea
l
-
tim
e
in
tr
u
s
io
n
d
etec
tio
n
.
Dif
f
er
en
t
f
ilter
,
wr
ap
p
er
,
en
s
e
m
b
le
,
an
d
o
p
tim
izatio
n
-
b
ased
tech
n
iq
u
es
h
av
e
b
ee
n
u
s
ed
t
o
im
p
r
o
v
e
f
ea
tu
r
e
s
elec
tio
n
in
NI
DS
,
l
ea
d
in
g
to
b
etter
d
etec
tio
n
a
cc
u
r
ac
y
an
d
r
ed
u
ce
d
d
im
e
n
s
io
n
ality
[
9
]
–
[
1
3
]
.
Pro
m
is
in
g
r
esu
lts
h
av
e
b
ee
n
r
ep
o
r
ted
f
o
r
ex
tr
em
e
g
r
ad
ien
t
b
o
o
s
tin
g
(
XGBo
o
s
t
)
with
f
ea
tu
r
e
s
elec
tio
n
[
1
1
]
,
im
p
r
o
v
e
d
g
r
ey
wo
lf
o
p
tim
iz
atio
n
(
I
GW
O
)
-
b
ased
f
ea
tu
r
e
s
elec
tio
n
[
1
2
]
,
n
e
u
r
al
n
etwo
r
k
-
ass
is
ted
f
ea
tu
r
e
s
elec
tio
n
[
1
3
]
,
a
n
d
d
ataset
s
tan
d
ar
d
izatio
n
m
eth
o
d
s
[
1
4
]
.
T
h
e
en
s
em
b
le
class
if
icatio
n
with
f
ea
tu
r
e
s
elec
tio
n
in
clo
u
d
e
n
v
ir
o
n
m
en
ts
[
1
5
]
,
m
ajo
r
ity
v
o
tin
g
[
1
6
]
,
ch
i
-
s
q
u
a
r
e
with
g
en
e
r
ativ
e
ad
v
er
s
ar
i
al
n
etwo
r
k
(
GAN
)
b
alan
cin
g
[
1
7
]
,
ch
a
o
tic
cr
o
w
s
ea
r
ch
[
1
8
]
,
a
n
d
b
io
-
in
s
p
ir
ed
alg
o
r
ith
m
s
s
u
ch
as
PS
O,
g
r
ey
wo
lf
o
p
tim
izatio
n
(
GW
O
)
,
f
ir
ef
ly
alg
o
r
ith
m
(
FFA
),
an
d
GA
[
1
9
]
wer
e
u
s
ed
to
o
b
tain
o
p
tim
al
f
ea
tu
r
e
s
u
b
s
ets.
Oth
er
s
tu
d
ies
f
o
cu
s
ed
o
n
b
io
-
in
s
p
ir
ed
I
o
T
in
tr
u
s
io
n
d
etec
tio
n
[
2
0
]
,
h
y
b
r
id
k
er
n
el
s
u
p
p
o
r
t
v
ec
to
r
m
a
ch
in
e
(
SVM
)
with
en
s
em
b
le
f
ea
tu
r
e
s
elec
tio
n
[
2
1
]
,
IG
on
C
I
C
-
I
DS2
0
1
7
[
2
2
]
,
weig
h
ted
f
ea
tu
r
e
s
elec
tio
n
[
2
3
]
,
o
p
tim
al
n
eu
r
al
n
etwo
r
k
s
[
2
4
]
,
an
d
m
ee
r
k
at
c
lan
alg
o
r
ith
m
[
2
5
]
.
Ho
wev
e
r
,
e
x
is
tin
g
m
eth
o
d
s
s
till
h
av
e
s
o
m
e
lim
itatio
n
s
,
s
u
ch
as
d
ep
en
d
en
cy
o
n
s
in
g
le
f
ea
tu
r
e
s
elec
tio
n
tech
n
iq
u
es,
lack
o
f
co
n
s
id
er
atio
n
o
f
f
ea
tu
r
e
d
ep
en
d
e
n
cies,
h
ig
h
co
m
p
u
tatio
n
al
c
o
s
t
in
wr
ap
p
e
r
-
b
ased
ap
p
r
o
ac
h
es,
lack
o
f
s
y
s
tem
atic
h
y
b
r
id
izatio
n
,
an
d
in
co
n
s
is
ten
cy
in
th
e
r
ep
r
o
d
u
cib
ilit
y
o
f
f
ea
tu
r
e
s
u
b
s
ets ac
r
o
s
s
d
if
f
er
en
t r
u
n
s
.
T
o
s
o
lv
e
th
ese
p
r
o
b
lem
s
,
th
e
p
r
o
p
o
s
ed
Hy
b
FS
-
NI
DS
u
s
es
th
e
o
p
tim
al
s
u
b
s
ets
f
r
o
m
th
e
f
ilter
an
d
wr
ap
p
er
s
tag
es
to
o
b
tain
th
e
f
in
al
f
ea
tu
r
e
s
u
b
s
et
.
Sin
ce
th
e
f
i
n
al
f
ea
tu
r
e
s
u
b
s
et
is
in
ter
s
ec
te
d
with
th
e
o
p
tim
al
s
u
b
s
ets
o
f
two
s
tag
es,
th
e
p
r
o
p
o
s
ed
m
et
h
o
d
ca
n
g
u
ar
an
t
ee
th
at
th
e
f
in
al
f
ea
tu
r
e
s
u
b
s
et
is
r
o
b
u
s
t
an
d
co
n
s
id
er
ed
as
im
p
o
r
tan
t
as
p
o
s
s
ib
le
b
y
d
if
f
er
en
t
s
tr
ateg
ie
s
.
I
n
co
m
p
ar
is
o
n
with
ex
is
ti
n
g
ap
p
r
o
ac
h
es,
t
h
e
p
r
o
p
o
s
ed
Hy
b
FS
-
NI
DS
d
em
o
n
s
tr
ates
im
p
r
o
v
e
d
d
etec
tio
n
a
cc
u
r
ac
y
,
g
r
ea
ter
s
tab
ilit
y
o
f
t
h
e
s
elec
ted
f
ea
t
u
r
e
s
u
b
s
et,
an
d
r
ed
u
ce
d
co
m
p
u
tatio
n
al
co
m
p
le
x
ity
.
T
h
e
m
eth
o
d
is
ap
p
r
o
p
r
iate
f
o
r
h
ig
h
-
d
im
en
s
io
n
al
d
atasets
s
u
ch
as
C
I
C
-
I
DS2
0
1
7
an
d
ca
n
b
e
ad
ap
ted
f
o
r
d
e
p
lo
y
m
e
n
t
in
r
ea
l
-
tim
e
in
tr
u
s
io
n
d
etec
tio
n
en
v
ir
o
n
m
en
ts
.
T
h
e
wea
k
n
ess
es
o
f
tr
ad
itio
n
al
f
e
atu
r
e
s
elec
tio
n
m
eth
o
d
s
in
NI
DS
p
o
in
ted
o
u
t
b
y
m
a
n
y
ea
r
lier
s
tu
d
ies.
Fo
r
ex
am
p
le,
f
ew
wr
ap
p
e
r
-
b
ased
t
ec
h
n
iq
u
es
test
ed
o
n
th
e
C
I
C
-
I
DS2
0
1
7
d
ataset
r
ep
o
r
ted
h
i
g
h
ac
cu
r
ac
y
.
B
u
t
th
ey
to
o
k
a
lo
n
g
tim
e
to
r
u
n
.
B
ec
au
s
e
th
e
m
o
d
el
h
a
d
to
b
e
t
r
ain
ed
ag
ain
an
d
ag
ain
wh
ile
c
h
ec
k
i
n
g
d
if
f
e
r
en
t
f
ea
tu
r
e
s
u
b
s
ets.
On
th
e
o
th
er
h
an
d
,
f
i
l
t
e
r
-
b
a
s
e
d
m
e
t
h
o
d
s
d
e
p
e
n
d
o
n
l
y
o
n
s
t
a
t
i
s
ti
c
a
l
m
e
as
u
r
e
s
an
d
o
f
t
e
n
k
e
p
t
e
x
t
r
a
f
e
a
t
u
r
e
s
.
As
a
r
es
u
l
t
,
t
h
e
i
m
p
r
o
v
e
m
e
n
t
i
n
a
c
c
u
r
a
c
y
w
as
s
m
a
l
l
.
A
l
s
o
,
t
h
e
n
u
m
b
e
r
o
f
f
e
a
t
u
r
es
w
a
s
n
o
t
el
i
m
i
n
a
te
d
e
f
f
e
c
t
i
v
e
l
y
.
T
h
e
r
e
f
o
r
e
,
i
t
i
s
n
o
t
a
l
w
a
y
s
t
h
e
c
as
e
t
h
a
t
o
n
l
y
f
il
t
e
r
m
e
t
h
o
d
s
o
r
o
n
l
y
w
r
a
p
p
e
r
m
e
t
h
o
d
s
p
r
o
v
i
d
e
a
g
o
o
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
5
,
No
.
5
,
Octo
b
er
2
0
2
6
:
4
3
6
6
-
4
3
8
0
4368
t
r
a
d
e
-
o
f
f
b
e
t
w
e
e
n
d
e
t
e
ct
i
o
n
p
e
r
f
o
r
m
a
n
c
e
a
n
d
c
o
m
p
u
t
a
ti
o
n
a
l
c
o
s
t
.
T
h
is
i
s
t
h
e
r
e
a
s
o
n
w
h
y
t
h
e
H
y
b
F
S
-
N
I
D
S
f
r
a
m
e
w
o
r
k
i
s
r
e
q
u
i
r
e
d
.
T
h
e
r
est
o
f
th
is
p
ap
er
is
o
r
g
an
ized
as
f
o
llo
ws:
s
ec
tio
n
2
ex
p
lain
s
th
e
d
ataset
an
d
th
e
p
r
o
p
o
s
ed
Hy
b
FS
-
NI
DS
f
r
am
ewo
r
k
.
Sectio
n
3
p
r
esen
ts
th
e
r
esea
r
ch
m
eth
o
d
in
d
etail
.
Sectio
n
4
p
r
esen
ts
th
e
ex
p
er
im
en
tal
s
etu
p
an
d
d
is
cu
s
s
es
th
e
r
esu
lts
an
d
c
o
m
p
a
r
is
o
n
s
with
ex
is
tin
g
m
et
h
o
d
s
.
Sectio
n
6
is
th
e
s
u
m
m
ar
y
o
f
th
e
co
n
clu
s
io
n
s
a
n
d
p
r
o
p
o
s
als
f
o
r
f
u
tu
r
e
wo
r
k
.
2.
P
RO
P
O
SE
D
M
E
T
H
O
D
T
h
is
s
ec
tio
n
o
u
tlin
es
th
e
m
eth
o
d
o
lo
g
y
,
n
a
m
ely
Hy
b
FS
-
N
I
DS.
T
h
is
m
eth
o
d
is
u
s
ed
to
co
m
b
in
e
d
if
f
er
en
t
f
ilter
-
b
ased
r
an
k
in
g
m
eth
o
d
s
a
n
d
a
wr
ap
p
er
-
b
ased
m
eta
-
h
e
u
r
is
tic
o
p
tim
izatio
n
t
o
f
in
d
o
u
t
t
h
e
m
o
s
t
r
elev
an
t
an
d
n
o
n
-
r
ed
u
n
d
an
t
f
e
atu
r
es
to
b
e
u
s
ed
in
i
n
tr
u
s
io
n
d
etec
tio
n
.
T
h
e
last
o
p
tim
ized
s
u
b
s
et
is
test
ed
b
y
a
M
L
P
c
l
ass
i
f
i
e
r
.
T
h
e
p
e
r
f
o
r
m
a
n
c
e
i
s
t
h
e
n
a
s
s
e
s
s
e
d
b
as
e
d
o
n
v
a
r
i
o
u
s
m
e
a
s
u
r
es
t
o
e
n
s
u
r
e
i
ts
r
o
b
u
s
t
n
e
s
s
a
n
d
u
t
i
li
t
y
.
2
.
1
.
Da
t
a
s
et
des
cr
iptio
n
T
h
e
C
I
C
-
I
DS2
0
1
7
d
ataset
is
u
s
ed
as
th
e
ex
p
er
im
en
tal
f
r
am
ewo
r
k
,
co
n
s
is
tin
g
o
f
a
v
ast
in
tr
u
s
io
n
d
etec
tio
n
d
ataset
th
at
co
v
er
s
a
lar
g
e
v
ar
iety
o
f
n
o
r
m
al
tr
af
f
i
c
an
d
n
u
m
er
o
u
s
ty
p
es
o
f
attac
k
s
,
in
clu
d
i
n
g
d
en
ial
o
f
s
er
v
ice
(
Do
S
)
,
d
is
tr
ib
u
ted
d
en
ial
o
f
s
er
v
ice
(
DDo
S
)
,
in
tr
u
s
io
n
,
b
r
u
te
f
o
r
ce
,
o
n
lin
e,
an
d
b
o
tn
ets.
T
h
e
d
ataset
co
n
s
is
ts
o
f
ac
tu
al
tr
a
f
f
ic
c
o
n
d
itio
n
s
an
d
co
n
tai
n
s
o
v
e
r
8
0
n
etwo
r
k
p
a
r
am
eter
s
lik
e
f
lo
w
tim
e,
p
ac
k
et
s
ize
m
etr
ics,
h
ea
d
er
co
n
te
n
ts
,
an
d
in
ter
-
ar
r
iv
al
tim
e
(
I
AT
)
.
B
ef
o
r
e
f
ea
tu
r
e
s
elec
tio
n
was
d
o
n
e,
d
ata
p
r
ep
r
o
ce
s
s
in
g
was d
o
n
e
in
o
r
d
er
t
o
h
av
e
h
ig
h
d
ata
q
u
ality
:
−
Miss
in
g
v
alu
e
im
p
u
tatio
n
–
m
i
s
s
in
g
(
r
ec
o
r
d
s
)
o
f
s
o
m
e
f
lo
w
-
b
ased
f
ea
tu
r
es
wer
e
im
p
u
ted
b
ased
o
n
th
e
m
ea
n
(
n
u
m
er
ical)
an
d
m
o
d
e
(
c
ateg
o
r
ical)
attr
ib
u
tes.
−
Du
p
licate
r
em
o
v
al
–
d
u
p
licates
wer
e
r
em
o
v
e
d
s
o
as to
p
r
ev
en
t
b
ias in
th
e
f
ea
tu
r
e
s
elec
tio
n
p
r
o
ce
s
s
.
−
Ou
tlier
h
an
d
lin
g
–
th
e
in
ter
q
u
ar
tile
r
an
g
e
(
I
QR
)
was
u
s
ed
to
d
etec
t
ex
tr
em
e
v
alu
es,
wh
ich
w
er
e
ca
p
p
ed
t
o
d
ec
r
ea
s
e
s
k
ewn
ess
.
−
No
r
m
aliza
tio
n
–
all
o
f
th
e
n
u
m
er
ic
f
ea
tu
r
es
wer
e
n
o
r
m
aliz
ed
with
m
in
-
m
ax
n
o
r
m
aliza
tio
n
to
b
r
in
g
all
v
alu
es with
in
th
e
r
an
g
e
[
0
,
1
]
,
s
o
th
at
ea
ch
f
ea
tu
r
e
co
n
tr
i
b
u
te
s
th
e
s
am
e
d
is
tan
ce
-
b
ased
co
m
p
u
tatio
n
.
I
n
o
r
d
er
to
m
ea
s
u
r
e
th
e
e
f
f
ec
t
s
o
f
p
r
ep
r
o
ce
s
s
in
g
,
T
ab
le
1
p
r
o
v
id
es
p
er
ce
n
tag
es
o
f
r
ec
o
r
d
s
af
f
ec
ted
at
ev
er
y
s
tep
.
T
h
is
g
u
ar
an
tees
tr
an
s
p
ar
en
cy
in
ter
m
s
o
f
d
ata
q
u
ality
im
p
r
o
v
em
en
t
b
e
f
o
r
e
f
ea
tu
r
e
s
elec
tio
n
.
Af
ter
th
e
d
ataset
was
p
r
ep
r
o
ce
s
s
ed
,
th
e
r
esu
ltin
g
d
ataset
was
p
r
ep
ar
ed
to
r
u
n
th
e
h
y
b
r
id
f
ea
tu
r
e
s
elec
tio
n
p
ip
elin
e
to
p
r
o
d
u
ce
a
s
u
b
s
et
o
f
f
ea
tu
r
es o
p
tim
ized
f
o
r
in
tr
u
s
io
n
d
etec
tio
n
.
T
ab
le
1
.
I
m
p
ac
t o
f
d
ata
p
r
ep
r
o
ce
s
s
in
g
o
n
th
e
C
I
C
-
I
DS2
0
1
7
d
ataset
P
r
e
p
r
o
c
e
ss
i
n
g
s
t
e
p
A
f
f
e
c
t
e
d
r
e
c
o
r
d
s
P
e
r
c
e
n
t
a
g
e
(
%)
D
e
scri
p
t
i
o
n
o
f
a
c
t
i
o
n
M
i
ss
i
n
g
v
a
l
u
e
i
m
p
u
t
a
t
i
o
n
1
2
,
8
4
5
1
.
7
3
%
N
u
meri
c
a
l
f
e
a
t
u
r
e
s
i
m
p
u
t
e
d
u
si
n
g
m
e
a
n
;
c
a
t
e
g
o
r
i
c
a
l
u
si
n
g
m
o
d
e
D
u
p
l
i
c
a
t
e
r
e
mo
v
a
l
8
,
2
1
4
1
.
1
0
%
Ex
a
c
t
d
u
p
l
i
c
a
t
e
f
l
o
w
r
e
c
o
r
d
s
r
e
m
o
v
e
d
O
u
t
l
i
e
r
h
a
n
d
l
i
n
g
(
I
Q
R
c
a
p
p
i
n
g
)
2
4
,
5
0
6
3
.
2
9
%
Ex
t
r
e
me
v
a
l
u
e
s
c
a
p
p
e
d
t
o
I
Q
R
b
o
u
n
d
s
N
o
r
mal
i
z
a
t
i
o
n
1
0
0
%
o
f
n
u
mer
i
c
a
l
f
e
a
t
u
r
e
s
—
M
i
n
–
ma
x
s
c
a
l
i
n
g
t
o
r
a
n
g
e
[
0
,
1
]
3.
H
Y
B
F
S
-
N
I
D
S
A
L
G
O
R
I
T
H
M
A
N
D
I
M
P
L
E
M
E
N
T
A
T
I
O
N
T
h
e
Hy
b
FS
-
NI
DS
alg
o
r
ith
m
in
teg
r
ates
b
o
th
f
ilter
-
b
ased
an
d
wr
a
p
p
er
-
b
ased
s
elec
tio
n
s
tr
ateg
ies,
th
er
eb
y
co
m
b
in
in
g
th
e
ad
v
a
n
tag
es
o
f
th
e
two
ap
p
r
o
ac
h
es.
Fil
ter
m
eth
o
d
s
ar
e
f
ast
an
d
e
f
f
ec
tiv
e
s
in
ce
th
ey
r
ap
id
ly
s
cr
ee
n
t
h
e
f
ea
tu
r
es
in
ac
co
r
d
an
ce
with
th
eir
s
tatis
tical
s
ig
n
if
ican
ce
.
I
n
co
n
t
r
ast,
wr
ap
p
e
r
m
eth
o
d
s
co
n
s
id
er
s
u
b
s
ets
o
f
ch
ar
ac
te
r
is
tics
in
th
e
en
v
ir
o
n
m
en
t
o
f
a
p
ar
ticu
lar
class
if
ier
in
o
r
d
er
to
en
h
an
ce
th
e
p
r
ed
icted
ac
cu
r
ac
y
.
T
h
e
Hy
b
F
S
-
NI
DS
alg
o
r
ith
m
is
illu
s
tr
ated
in
Alg
o
r
ith
m
1
.
Alg
o
r
ith
m
1
.
Hy
b
FS
-
NI
DS
Input:
preprocessed
CIC
-
IDS2017
dataset (cleaned, normalized, and outlier
-
handled).
Output:
optimized
feature subset for improved NIDS accuracy and efficiency.
Steps:
1:
Dataset
loading
Load the preprocessed dataset into the feature selection environment.
2:
Filter
-
based sele
ction
ReliefF:
weigh
features by their ability to separate close but differently labeled
samples.
FCBF:
use symmetrical uncertainty (SU)
to retain relevant, non
-
redundant features.
mRMR:
select
features maximizing relevance to the target while minimizing redundancy.
3:
Filter
results aggregat
ion
Apply Borda
count
to merge rankings from all filter methods, selecting the top
-
N
features with highest aggregate scores.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Hyb
r
id
fea
tu
r
e
s
elec
tio
n
s
tr
a
te
g
ies fo
r
s
tr
o
n
g
n
etw
o
r
k
in
tr
u
s
i
o
n
d
etec
tio
n
(
Jo
s
ep
h
i
n
e
R
a
th
i
n
a
m
)
4369
4:
Wrapper
-
based selection (meta
-
heuristic
)
Use a GA:
Encode feature subsets as chromosomes. Evolve via selection, crossover, and mutation.
Fitness: MLP classifier performance via cross
-
validation.
5:
Hybrid
selection
Intersect top filter
-
selected features with GA
-
optimized wrapper results to retain only
consistently
.
3
.
1
.
Ste
p 1
:
f
ilte
r
-
ba
s
ed
s
elec
t
io
n
T
h
e
p
r
o
ce
s
s
o
f
f
ilter
s
elec
tio
n
th
at
in
v
o
lv
es
th
e
u
s
e
o
f
f
il
ter
s
b
eg
in
s
b
y
ap
p
l
y
in
g
th
r
ee
d
if
f
er
en
t
alg
o
r
ith
m
s
,
wh
ich
ar
e
R
elief
F,
FC
B
F,
an
d
m
R
MR
,
to
ass
e
s
s
an
d
r
an
k
t
h
e
s
ig
n
if
ican
ce
o
f
in
d
iv
id
u
al
f
ea
t
u
r
es.
T
h
ese
d
if
f
er
en
t
r
an
k
i
n
g
s
ar
e
th
en
co
m
b
in
e
d
b
y
th
e
u
s
e
o
f
th
e
B
o
r
d
a
co
u
n
t
tech
n
iq
u
e.
T
h
is
s
tr
ateg
y
s
u
m
s
u
p
th
e
s
co
r
es
to
d
eter
m
in
e
th
e
attr
ib
u
tes
th
at
h
av
e
a
h
ig
h
s
co
r
e
ac
r
o
s
s
all
ap
p
r
o
ac
h
es.
T
h
e
N
f
ea
tu
r
es
with
th
e
h
ig
h
est cu
m
u
lativ
e
s
co
r
es a
r
e
ch
o
s
en
,
th
u
s
en
d
in
g
t
h
e
f
ilter
-
b
ased
s
elec
tio
n
s
tag
e.
B
etwe
en
Hy
b
FS
-
NI
DS,
R
elief
F,
FC
B
F,
an
d
m
R
MR
,
th
e
latt
er
is
s
elec
ted
b
ec
au
s
e
o
f
th
e
n
ec
ess
ity
to
p
r
o
ce
s
s
h
ig
h
-
d
im
en
s
io
n
al
c
y
b
er
s
ec
u
r
ity
d
ata
m
o
r
e
ef
f
icien
tly
.
Mo
s
t
s
tan
d
ar
d
f
ilter
tech
n
iq
u
es,
s
u
ch
as
IG
an
d
ch
i
-
s
q
u
ar
e
,
ty
p
ically
ex
a
m
in
e
th
e
r
elatio
n
s
h
ip
o
f
ea
ch
f
ea
tu
r
e
to
th
e
tar
g
et
class
.
Ho
w
ev
er
,
th
ey
d
o
n
o
t
ad
d
r
ess
th
e
in
ter
r
elatio
n
s
h
ip
s
o
f
f
ea
tu
r
es
in
t
h
e
r
i
g
h
t
way
.
C
o
m
p
ar
ativ
ely
,
m
R
MR
p
ick
s
f
e
atu
r
es
b
y
r
etain
in
g
th
o
s
e
h
av
in
g
s
tr
o
n
g
r
elate
d
n
e
s
s
to
th
e
tar
g
et
as
well
as
elim
in
atin
g
th
o
s
e
t
h
at
h
av
e
s
im
ilar
in
f
o
r
m
atio
n
.
I
t
p
r
ev
en
ts
r
ep
etitio
n
.
Mo
r
e
o
v
er
,
FC
B
F a
s
s
is
t
s
in
th
is
p
r
o
ce
s
s
b
y
r
em
o
v
in
g
ir
r
elev
an
t f
ea
t
u
r
es with
h
elp
o
f
SU
.
T
h
is
SU
wo
r
k
s
well
f
o
r
lar
g
e
d
atasets
s
u
ch
as
C
I
C
-
I
DS2
0
1
7
.
R
elief
F
ad
d
s
o
n
e
m
o
r
e
b
en
ef
it
b
y
ev
alu
atin
g
f
ea
tu
r
es
u
s
in
g
in
d
i
v
id
u
al
d
ata
in
s
tan
ce
s
an
d
ca
p
tu
r
in
g
lo
ca
l
p
atter
n
s
in
th
e
d
ata.
T
h
is
m
ak
es
it
u
s
ef
u
l f
o
r
i
d
en
tify
in
g
s
m
all
d
if
f
er
en
ce
s
b
etwe
en
n
o
r
m
al
an
d
attac
k
tr
af
f
ic.
B
y
co
m
b
in
in
g
t
h
ese
th
r
ee
m
eth
o
d
s
,
th
e
Hy
b
FS
-
NI
DS
f
r
am
ewo
r
k
cr
ea
tes
a
m
o
r
e
b
alan
ce
d
an
d
r
eliab
le
f
ea
tu
r
e
r
an
k
in
g
p
r
o
ce
s
s
th
an
o
n
ly
r
ely
in
g
o
n
b
asic statis
tical
f
ilter
s
.
3
.
1
.
1
.
Relief
F
a
lg
o
rit
hm
R
elief
F
ass
e
s
s
e
s
th
e
s
ig
n
if
ican
ce
o
f
a
f
ea
tu
r
e
b
y
r
ep
ea
ted
l
y
s
elec
tin
g
s
am
p
les
an
d
co
m
p
ar
in
g
th
e
d
if
f
er
en
ce
s
b
etwe
en
t
h
e
clo
s
est
n
eig
h
b
o
r
s
b
elo
n
g
in
g
to
t
h
e
s
am
e
class
,
ca
lled
th
e
n
e
ar
est
h
it,
an
d
t
h
o
s
e
b
elo
n
g
in
g
to
o
th
er
class
es,
ca
lled
th
e
n
ea
r
est
m
is
s
.
Gr
e
ater
weig
h
t
is
ass
ig
n
ed
to
f
ea
tu
r
es
th
at
m
o
r
e
ef
f
ec
tiv
ely
d
is
tin
g
u
is
h
b
etwe
en
class
es.
T
h
e
co
r
r
esp
o
n
d
in
g
weig
h
t
u
p
d
ate
eq
u
atio
n
is
g
iv
en
as
(
1
)
,
wh
er
e
[
]
r
ep
r
esen
ts
th
e
weig
h
t
ass
ig
n
ed
to
f
ea
tu
r
e
A,
d
en
o
tes
th
e
to
tal
n
u
m
b
er
o
f
s
am
p
les,
an
d
(
.
)
in
d
icate
s
th
e
n
o
r
m
alize
d
d
if
f
e
r
en
ce
ca
lcu
lated
b
etwe
en
t
h
e
c
o
r
r
esp
o
n
d
in
g
f
ea
tu
r
e
v
alu
es.
[
]
=
[
]
−
(
ℎ
)
+
(
)
(
1
)
3
.
1
.
2
.
F
a
s
t
co
rr
el
a
t
io
n
-
ba
s
ed
f
ilte
r
I
n
FC
B
F,
SU
is
em
p
lo
y
ed
to
id
en
tify
f
ea
tu
r
es
th
at
ar
e
s
tr
o
n
g
ly
r
elate
d
to
th
e
tar
g
et
v
ar
i
ab
le
wh
ile
r
em
ain
in
g
non
-
r
ed
u
n
d
an
t
with
r
esp
ec
t to
ea
ch
o
th
e
r
.
T
h
e
SU
m
etr
ic
is
d
ef
in
ed
as (
2
)
.
W
h
er
e
(
|
)
d
en
o
tes
th
e
IG
an
d
(
.
)
r
ep
r
esen
ts
th
e
e
n
tr
o
p
y
f
u
n
ctio
n
.
(
,
)
=
2
(
|
)
(
)
+
(
)
(
2
)
3
.
1
.
3
.
M
ini
m
um
re
du
nd
a
ncy
m
a
x
i
m
um
re
lev
a
nc
e
T
h
e
m
R
MR
ap
p
r
o
ac
h
ch
o
o
s
es
f
ea
tu
r
es
b
y
m
ax
im
izin
g
th
ei
r
r
elev
an
ce
to
th
e
tar
g
et
v
ar
ia
b
le
wh
ile
s
im
u
ltan
eo
u
s
ly
m
in
im
izin
g
r
e
d
u
n
d
a
n
cy
am
o
n
g
th
e
s
elec
ted
f
ea
tu
r
es.
T
h
e
s
co
r
in
g
f
u
n
ctio
n
is
d
ef
i
n
ed
as
(
3
)
,
wh
er
e
(
.
;.)
d
en
o
tes
th
e
m
u
t
u
al
in
f
o
r
m
atio
n
,
in
d
icate
s
th
e
f
ea
tu
r
e
s
u
b
s
et,
an
d
c
r
ef
er
s
to
th
e
class
lab
el.
T
h
e
f
ilter
-
b
ased
f
ea
tu
r
e
s
elec
tio
n
p
r
o
ce
s
s
is
illu
s
tr
ated
in
Fig
u
r
e
1
.
[
1
|
|
∑
(
;
)
∈
−
1
|
|
2
∑
(
;
)
,
∈
]
(
3
)
3
.
2
.
Ste
p 2
:
ra
nk
a
g
g
re
g
a
t
io
n v
ia
B
o
rda
co
un
t
T
h
e
r
an
k
i
n
g
s
p
r
o
d
u
ce
d
b
y
th
e
th
r
ee
f
ilter
tech
n
iq
u
es
ar
e
c
o
m
b
in
ed
b
y
u
s
in
g
th
e
B
o
r
d
a
c
o
u
n
t
m
eth
o
d
as
s
h
o
wn
in
(
4
)
.
Her
e
d
en
o
tes
th
e
to
tal
n
u
m
b
e
r
o
f
f
e
atu
r
es,
r
ep
r
esen
ts
th
e
n
u
m
b
er
o
f
r
an
k
i
n
g
ap
p
r
o
ac
h
es,
an
d
ₖ
(
)
is
th
e
r
an
k
in
g
p
o
s
itio
n
ass
ig
n
ed
to
f
ea
tu
r
e
b
y
m
eth
o
d
.
T
h
e
f
ea
tu
r
es
th
at
h
av
e
o
b
tain
ed
th
e
h
ig
h
est co
m
b
in
ed
s
co
r
es a
r
e
s
elec
ted
f
o
r
th
e
wr
ap
p
er
s
tag
e.
(
)
=
∑
(
−
(
)
)
=
1
(
4
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
5
,
No
.
5
,
Octo
b
er
2
0
2
6
:
4
3
6
6
-
4
3
8
0
4370
3
.
3
.
Ste
p 3
:
wra
pp
er
-
ba
s
ed
s
elec
t
io
n wit
h g
enet
ic
a
lg
o
rit
hm
T
h
e
wr
ap
p
e
r
-
b
ased
f
ea
tu
r
e
s
e
lectio
n
p
r
o
ce
s
s
is
in
itiated
b
y
en
co
d
in
g
p
r
o
s
p
ec
tiv
e
f
ea
tu
r
e
s
u
b
s
ets
as
ch
r
o
m
o
s
o
m
es,
wh
ich
ar
e
t
h
e
n
ev
o
lv
ed
b
y
th
e
GA
o
p
er
a
tio
n
s
,
wh
ich
i
n
clu
d
e
s
elec
tio
n
,
cr
o
s
s
o
v
er
,
a
n
d
m
u
tatio
n
.
T
h
e
ef
f
icac
y
o
f
ea
c
h
s
u
b
s
et
is
ev
alu
ated
b
y
th
e
p
e
r
f
o
r
m
a
n
ce
o
f
an
ML
P
class
if
ier
.
T
h
e
p
r
o
ce
d
u
r
e
is
r
ep
ea
ted
u
n
til
co
n
v
er
g
en
ce
is
ac
h
iev
ed
o
r
a
s
to
p
p
in
g
cr
iter
io
n
is
m
et.
T
h
is
r
esu
lts
in
th
e
i
d
en
tific
atio
n
o
f
th
e
o
p
tim
al
f
ea
tu
r
e
s
u
b
s
et
th
at
m
ax
im
ally
en
h
an
ce
s
class
if
icatio
n
p
er
f
o
r
m
an
ce
.
W
ith
in
th
e
GA
f
r
am
ewo
r
k
,
a
ch
r
o
m
o
s
o
m
e
is
en
co
d
ed
as
a
b
in
ar
y
s
tr
in
g
in
wh
ich
a
v
alu
e
o
f
1
d
en
o
tes
th
at
a
f
ea
tu
r
e
is
s
elec
ted
,
wh
er
ea
s
0
in
d
icate
s
th
at
it i
s
n
o
t in
clu
d
ed
.
T
h
e
f
itn
ess
f
u
n
ctio
n
is
d
ef
in
e
d
as (
5
)
.
W
h
er
e
an
d
r
ep
r
esen
t w
eig
h
ts
(
h
er
e,
ea
ch
weig
h
t
is
0
.
5
to
b
e
eq
u
ally
im
p
o
r
tan
t)
.
T
h
e
GA
is
u
s
ed
to
ev
o
lv
e
th
e
p
o
p
u
latio
n
b
y
m
ak
i
n
g
u
s
e
o
f
s
elec
tio
n
,
cr
o
s
s
o
v
er
,
an
d
m
u
t
atio
n
to
m
ax
im
ize
class
if
icati
o
n
.
T
h
is
wr
ap
p
er
-
b
ased
s
elec
tio
n
o
f
f
ea
tu
r
es
is
illu
s
tr
ated
in
Fig
u
r
e
2
.
=
⋅
+
⋅
1
−
(
5
)
Fig
u
r
e
1
.
Fil
ter
-
b
ased
f
ea
tu
r
e
s
elec
tio
n
Fig
u
r
e
2
.
W
r
ap
p
er
-
b
ased
f
ea
tu
r
e
s
elec
tio
n
3
.
4
.
Ste
p 4
:
hy
brid s
elec
t
io
n (
inte
rsect
io
n)
T
h
e
h
y
b
r
id
in
ter
s
ec
tio
n
p
r
o
ce
s
s
b
eg
in
s
with
th
e
f
ea
tu
r
e
s
u
b
s
ets
th
at
wer
e
s
elec
ted
in
th
e
f
ilter
-
b
ased
an
d
th
e
GA
-
b
ased
p
r
o
ce
s
s
es.
T
h
en
,
it
r
u
n
s
a
s
et
in
ter
s
ec
tio
n
to
r
etain
o
n
l
y
f
ea
tu
r
es
th
at
ap
p
ea
r
im
p
o
r
tan
t
to
b
o
th
m
et
h
o
d
s
.
T
h
is
r
esu
lts
in
th
e
o
p
tim
ized
f
ea
tu
r
e
s
et,
wh
i
ch
ter
m
in
ates
th
e
h
y
b
r
id
s
elec
tio
n
p
r
o
ce
s
s
.
T
h
is
g
u
ar
an
tees
th
at
o
n
l
y
f
ea
t
u
r
es
th
at
ar
e
im
p
o
r
ta
n
t
in
all
s
tag
e
s
ar
e
r
etain
e
d
,
a
n
d
a
b
ala
n
ce
b
etwe
en
s
tatis
tical
s
ig
n
if
ican
ce
an
d
p
r
ed
ictio
n
is
ac
h
iev
ed
.
T
h
e
d
ef
in
itio
n
s
o
f
th
e
f
ea
tu
r
e
s
ets
ar
e
p
r
o
v
id
ed
to
ex
p
lain
t
h
e
h
y
b
r
id
in
ter
s
ec
tio
n
s
tep
clea
r
ly
.
First,
th
e
t
op
-
N
f
ea
tu
r
es
th
at
h
ad
b
ee
n
ch
o
s
en
b
y
t
h
e
th
r
ee
f
ilter
m
eth
o
d
s
with
t
h
e
ap
p
licatio
n
o
f
th
e
B
o
r
d
a
r
an
k
ag
g
r
eg
atio
n
ar
e
FR
elief
,
FF
C
B
F,
FF
C
B
F,
an
d
Fm
R
MR
.
A
s
a
wh
o
le,
in
d
iv
id
u
al
f
ilter
s
p
r
o
v
id
e
th
eir
lis
ts
o
f
r
an
k
in
g
s
,
a
n
d
th
ese
r
a
n
k
in
g
s
ar
e
m
er
g
e
d
to
cr
ea
te
a
co
m
m
o
n
r
an
k
in
g
.
Nex
t,
f
u
zz
y
g
en
etic
alg
o
r
ith
m
(
FGA)
is
th
e
o
p
tim
ized
f
ea
tu
r
e
s
u
b
s
et
u
s
ed
b
y
th
e
GA
d
u
r
in
g
th
e
wr
ap
p
er
s
tep
.
T
h
e
p
h
ase
is
u
s
ed
to
d
eter
m
in
e
a
g
o
o
d
c
o
m
b
in
atio
n
o
f
f
ea
tu
r
es
th
r
o
u
g
h
test
in
g
d
if
f
er
e
n
t
s
u
b
s
ets
.
No
w
co
m
es
th
e
m
ai
n
id
ea
.
I
n
o
r
d
er
t
o
b
e
in
clu
d
ed
in
t
h
e
f
i
n
al
s
u
b
s
et,
a
f
ea
t
u
r
e
m
u
s
t
m
ee
t
two
co
n
d
itio
n
s
.
I
t
s
h
o
u
ld
b
e
in
cl
u
d
ed
in
th
e
co
m
p
o
s
ite
f
ilter
s
et,
an
d
it
s
h
o
u
ld
also
b
e
ch
o
s
en
b
y
th
e
GA.
So
,
b
o
t
h
s
tag
es m
u
s
t a
g
r
ee
.
T
h
is
m
ea
n
s
th
at
th
e
f
ea
tu
r
e
is
s
tatis
tica
lly
s
ig
n
if
ican
t a
n
d
also
u
s
ef
u
l to
t
h
e
class
if
ier
.
T
h
e
ag
g
r
e
g
ated
f
ilter
s
et
is
d
ef
in
ed
as
=
{
|
ap
p
ea
r
s
in
at
least
two
o
f
th
e
th
r
ee
f
ilter
r
an
k
in
g
s
}
.
I
n
p
r
ac
tice,
th
is
m
ea
n
s
a
f
ea
tu
r
e
s
h
o
u
ld
b
e
s
u
p
p
o
r
t
ed
b
y
at
least
two
f
ilter
s
b
ef
o
r
e
m
o
v
in
g
f
o
r
war
d
.
On
e
r
ea
s
o
n
f
o
r
th
is
r
u
le
m
ay
b
e
to
av
o
id
d
ep
e
n
d
in
g
to
o
m
u
ch
o
n
a
s
in
g
le
m
eth
o
d
.
So
m
et
im
es
o
n
e
f
ilter
m
ay
f
av
o
r
a
f
ea
tu
r
e
th
at
o
t
h
er
s
d
o
n
o
t.
T
h
e
p
r
o
ce
s
s
s
ee
m
s
m
o
r
e
s
tab
le
b
y
h
av
in
g
to
ag
r
ee
.
M
ea
n
wh
ile,
th
e
f
in
al
d
ec
is
io
n
is
d
eter
m
in
e
d
b
y
th
e
GA,
wh
ich
v
er
if
ies
th
e
r
esu
lt
o
f
t
h
e
f
ea
t
u
r
es
o
p
er
atin
g
to
g
eth
er
in
th
e
m
o
d
el.
T
h
e
f
in
al
o
p
tim
ized
f
ea
tu
r
e
s
u
b
s
et
is
th
u
s
co
m
p
u
ted
as
in
(
6
)
.
=
∩
(
6
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Hyb
r
id
fea
tu
r
e
s
elec
tio
n
s
tr
a
te
g
ies fo
r
s
tr
o
n
g
n
etw
o
r
k
in
tr
u
s
i
o
n
d
etec
tio
n
(
Jo
s
ep
h
i
n
e
R
a
th
i
n
a
m
)
4371
T
h
is
m
ajo
r
ity
-
b
ased
co
n
s
en
s
u
s
(
at
least
two
f
ilter
s
)
en
s
u
r
es
r
o
b
u
s
tn
ess
b
y
r
ed
u
cin
g
t
h
e
in
f
l
u
en
ce
o
f
an
y
s
in
g
le
r
an
k
in
g
m
eth
o
d
,
wh
ile
th
e
s
u
b
s
eq
u
en
t
in
ter
s
ec
tio
n
with
th
e
GA
-
s
elec
ted
s
u
b
s
et
g
u
ar
an
tees
p
r
ed
ictiv
e
ef
f
ec
tiv
en
ess
th
r
o
u
g
h
class
if
ier
-
b
ased
v
alid
atio
n
.
T
h
e
h
y
b
r
id
in
ter
s
ec
tio
n
is
in
d
icate
d
in
Fi
g
u
r
e
3
.
Fig
u
r
e
3
.
Hy
b
r
id
in
te
r
s
ec
tio
n
3
.
5
.
Ste
p 5
:
mo
del e
v
a
lua
t
io
n m
et
rics
T
h
e
f
in
a
l
s
u
b
s
e
t
i
s
v
a
l
i
d
a
te
d
w
i
t
h
t
h
e
h
e
l
p
o
f
a
n
M
L
P
c
la
s
s
i
f
i
e
r
a
n
d
a
s
s
e
s
s
e
d
w
i
t
h
t
h
e
h
e
l
p
o
f
v
a
r
io
u
s
m
e
t
r
i
c
s
:
i)
Acc
u
r
ac
y
:
in
d
icate
s
th
e
o
v
er
all
class
if
icatio
n
co
r
r
ec
tn
ess
b
y
ca
lcu
latin
g
th
e
p
r
o
p
o
r
tio
n
o
f
co
r
r
ec
tl
y
p
r
ed
icted
in
s
tan
ce
s
am
o
n
g
a
ll
s
am
p
les
(
7
)
.
W
h
er
e
tr
u
e
p
o
s
itiv
es
(
T
P),
tr
u
e
n
eg
ativ
e
s
(
T
N)
,
f
alse
p
o
s
itiv
es
(
FP
)
,
an
d
f
alse
n
eg
a
tiv
es
(
FN)
.
T
h
is
m
etr
ic
was
u
s
ed
to
m
ea
s
u
r
e
th
e
p
e
r
ce
n
tag
e
o
f
ac
cu
r
at
e
p
r
ed
ictio
n
s
o
f
all
th
e
class
es.
=
+
+
+
+
(
7
)
ii)
Pre
cisi
o
n
:
ev
alu
ates
th
e
ac
cu
r
ac
y
o
f
p
o
s
itiv
e
p
r
e
d
ictio
n
s
b
y
d
eter
m
in
in
g
th
e
p
r
o
p
o
r
tio
n
o
f
co
r
r
ec
tly
p
r
ed
icted
p
o
s
itiv
e
in
s
tan
ce
s
am
o
n
g
all
p
r
ed
icted
p
o
s
itiv
es
in
(
8
).
=
+
(
8
)
Pre
cisi
o
n
is
co
n
s
id
er
ed
im
p
o
r
tan
t
f
o
r
e
v
alu
atin
g
h
o
w
ef
f
e
ctiv
ely
f
alse
alar
m
s
ar
e
m
in
i
m
ized
b
y
th
e
NI
DS.
R
ec
all
m
ea
s
u
r
es
th
e
p
r
o
p
o
r
ti
o
n
o
f
ac
t
u
al
p
o
s
itiv
e
in
s
tan
ce
s
th
at
ar
e
c
o
r
r
ec
tly
id
en
tifie
d
b
y
th
e
m
o
d
el
in
(
9
).
=
+
(
9
)
Hig
h
r
ec
all
is
en
s
u
r
ed
s
o
th
at
m
o
s
t a
ttack
in
s
tan
ce
s
ar
e
co
r
r
e
ctly
d
etec
ted
.
iii)
F1
-
s
co
r
e
:
r
ep
r
esen
ts
th
e
h
ar
m
o
n
ic
av
er
ag
e
o
f
p
r
ec
is
io
n
an
d
r
ec
all,
o
f
f
er
in
g
a
c
o
m
b
in
e
d
ev
alu
atio
n
o
f
b
o
th
m
ea
s
u
r
es
as d
ef
i
n
ed
in
(
10
).
1
-
=
2
⋅
⋅
+
(
10
)
B
y
th
is
b
alan
ce
d
ass
ess
m
en
t,
th
e
d
etec
tio
n
p
er
f
o
r
m
an
ce
a
n
d
th
e
f
alse a
lar
m
s
ar
e
m
ain
tain
e
d
.
iv
)
Ma
tth
ew’
s
co
r
r
elatio
n
co
e
f
f
i
cien
t
(
MCC
)
:
s
er
v
es
as
a
b
alan
ce
d
p
e
r
f
o
r
m
an
ce
m
etr
i
c
f
o
r
b
in
ar
y
class
if
icatio
n
task
s
b
y
tak
in
g
in
to
ac
co
u
n
t a
ll p
r
ed
ictio
n
o
u
tco
m
es
as d
escr
ib
ed
in
(1
1
).
=
.
−
.
√
(
+
)
(
+
)
(
+
)
(
+
)
(1
1
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
5
,
No
.
5
,
Octo
b
er
2
0
2
6
:
4
3
6
6
-
4
3
8
0
4372
MCC
is
s
e
lecte
d
as
it
ac
co
u
n
ts
f
o
r
all
f
o
u
r
co
n
f
u
s
io
n
m
atr
ix
ca
teg
o
r
ies
a
n
d
r
em
ain
s
d
e
p
en
d
ab
le
ev
e
n
wh
en
th
e
d
ataset
is
im
b
alan
ce
d
.
T
ab
le
2
p
r
esen
ts
th
e
p
a
r
am
e
ter
co
n
f
ig
u
r
atio
n
f
o
r
H
y
b
FS
-
NI
DS.
T
ab
le
2
.
E
x
p
er
im
en
tal
p
ar
am
e
ter
s
ettin
g
s
f
o
r
Hy
b
FS
-
NI
DS
C
o
m
p
o
n
e
n
t
P
a
r
a
me
t
e
r
V
a
l
u
e
F
i
l
t
e
r
st
a
g
e
I
n
i
t
i
a
l
n
u
mb
e
r
o
f
f
e
a
t
u
r
e
s
78
To
p
-
N
f
e
a
t
u
r
e
s se
l
e
c
t
e
d
a
f
t
e
r
B
o
r
d
a
a
g
g
r
e
g
a
t
i
o
n
30
GA
(
w
r
a
p
p
e
r
st
a
g
e
)
P
o
p
u
l
a
t
i
o
n
s
i
z
e
50
N
u
mb
e
r
o
f
g
e
n
e
r
a
t
i
o
n
s
1
0
0
S
e
l
e
c
t
i
o
n
m
e
t
h
o
d
To
u
r
n
a
me
n
t
sel
e
c
t
i
o
n
(
k
=
3)
C
r
o
ss
o
v
e
r
r
a
t
e
0
.
8
M
u
t
a
t
i
o
n
r
a
t
e
0
.
0
2
C
h
r
o
mo
so
m
e
r
e
p
r
e
s
e
n
t
a
t
i
o
n
B
i
n
a
r
y
e
n
c
o
d
i
n
g
F
i
t
n
e
ss f
u
n
c
t
i
o
n
w
e
i
g
h
t
s (α
,
β)
0
.
5
,
0
.
5
M
LP
c
l
a
ssi
f
i
e
r
I
n
p
u
t
l
a
y
e
r
si
z
e
D
e
p
e
n
d
s
o
n
se
l
e
c
t
e
d
f
e
a
t
u
r
e
s
H
i
d
d
e
n
l
a
y
e
r
s
2
N
e
u
r
o
n
s (
h
i
d
d
e
n
l
a
y
e
r
1)
64
N
e
u
r
o
n
s (
h
i
d
d
e
n
l
a
y
e
r
2)
32
A
c
t
i
v
a
t
i
o
n
f
u
n
c
t
i
o
n
R
e
LU
(
h
i
d
d
e
n
)
,
s
o
f
t
m
a
x
(
o
u
t
p
u
t
)
O
p
t
i
mi
z
e
r
A
d
a
m
Le
a
r
n
i
n
g
r
a
t
e
0
.
0
0
1
B
a
t
c
h
si
z
e
1
2
8
Ep
o
c
h
s
50
Lo
ss f
u
n
c
t
i
o
n
C
a
t
e
g
o
r
i
c
a
l
c
r
o
ss
-
e
n
t
r
o
p
y
Ev
a
l
u
a
t
i
o
n
C
r
o
ss
-
v
a
l
i
d
a
t
i
o
n
10
-
f
o
l
d
T
h
e
Hy
b
FS
-
NI
DS
m
eth
o
d
co
n
s
is
ts
o
f
s
tatis
tical
f
ea
tu
r
e
r
an
k
in
g
an
d
e
v
o
lu
tio
n
ar
y
s
ea
r
ch
to
d
eter
m
in
e
th
e
m
o
s
t
ap
p
r
o
p
r
i
ate
s
et
o
f
f
ea
tu
r
es
o
f
t
h
e
C
I
C
-
I
DS2
0
1
7
d
ataset.
T
h
e
m
eth
o
d
en
s
u
r
es
th
at
co
m
p
u
tatio
n
al
an
d
p
r
e
d
ic
t
i
o
n
p
o
wer
ar
e
h
ig
h
th
r
o
u
g
h
R
elief
F,
FC
B
F,
an
d
m
R
MR,
as
well
as
B
o
r
d
a
co
u
n
t
ag
g
r
eg
atio
n
an
d
GA
-
b
ased
w
r
ap
p
er
s
elec
tio
n
.
N
o
t
o
n
ly
will
th
e
f
in
al
m
o
d
el
b
e
ca
p
a
b
le
o
f
d
eter
m
i
n
in
g
th
e
in
tr
u
s
io
n
s
with
g
r
ea
t
ac
cu
r
ac
y
,
b
u
t
it
will
also
b
e
ca
p
ab
l
e
o
f
ad
d
r
ess
in
g
r
ea
l
-
wo
r
ld
i
n
tr
u
s
io
n
d
etec
tio
n
s
itu
atio
n
s
q
u
ite
well,
as
th
e
v
ar
io
u
s
e
v
alu
atio
n
m
ea
s
u
r
es
ar
e
u
s
ed
,
in
clu
d
in
g
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all
,
F1
-
s
co
r
e,
an
d
MCC
.
T
h
e
o
p
t
im
ized
f
ea
tu
r
e
s
u
b
s
et
d
ev
elo
p
ed
u
s
in
g
th
is
p
r
o
ce
s
s
is
an
ti
cip
ated
to
d
ec
r
ea
s
e
co
m
p
u
tatio
n
al
o
v
er
h
ea
d
wh
ile
im
p
r
o
v
i
n
g
th
e
d
etec
tio
n
ca
p
a
b
ilit
ies o
f
NI
DS sy
s
tem
s
.
–
C
o
m
p
u
tatio
n
al
co
m
p
lex
ity
an
aly
s
is
:
l
et
,
is
n
u
m
b
er
o
f
s
am
p
les
,
is
n
u
m
b
er
o
f
f
ea
tu
r
es
,
is
n
u
m
b
er
o
f
GA
g
en
er
atio
n
s
,
is
GA
p
o
p
u
latio
n
s
ize
,
an
d
is
tim
e
co
m
p
lex
ity
o
f
ML
P tr
ain
in
g
.
–
Fil
ter
s
tag
e:
R
elief
F
is
r
eq
u
ir
e
s
n
ea
r
est
n
eig
h
b
o
r
s
ea
r
ch
f
o
r
ea
ch
s
am
p
le
→
co
m
p
lex
ity
:
(
2
)
.
FC
B
F
is
c
o
m
p
u
tes
s
y
m
m
etr
ical
u
n
ce
r
tain
ty
b
etwe
en
f
ea
tu
r
es
→
c
o
m
p
lex
ity
:
(
2
)
.
m
R
MR
is
m
u
tu
al
in
f
o
r
m
atio
n
p
air
wis
e
co
m
p
u
tatio
n
→
co
m
p
lex
ity
:
(
2
)
.
B
o
r
d
a
ag
g
r
eg
atio
n
is
s
o
r
tin
g
a
n
d
s
u
m
m
atio
n
→
co
m
p
lex
ity
:
(
)
.
Ov
er
all
f
ilter
s
tag
e
co
m
p
lex
ity
:
(
2
+
2
)
.
–
W
r
ap
p
er
s
tag
e
(
GA
)
:
f
o
r
ea
ch
g
en
er
atio
n
is
f
itn
ess
ev
alu
atio
n
:
P
ch
r
o
m
o
s
o
m
es
×
ML
P
tr
ain
in
g
→
(
×
)
.
Fo
r
G
g
en
e
r
atio
n
s
:
(
×
×
)
.
–
Hy
b
r
id
in
ter
s
ec
tio
n
:
s
et
in
ter
s
ec
tio
n
o
p
er
atio
n
is
(
)
.
Ov
er
all
co
m
p
lex
ity
,
th
e
to
tal
c
o
m
p
u
tatio
n
al
co
m
p
lex
ity
o
f
Hy
b
FS
-
NI
DS is
(
2
+
2
+
×
×
)
.
T
h
e
c
o
m
p
u
t
a
t
i
o
n
a
l
c
o
m
p
l
e
x
i
t
y
a
n
a
l
y
s
i
s
s
h
o
w
s
t
h
a
t
H
y
b
F
S
-
N
I
D
S
i
s
c
o
m
p
u
t
a
t
i
o
n
a
l
l
y
c
o
s
t
l
y
d
u
r
i
n
g
t
h
e
t
r
a
i
n
i
n
g
a
n
d
f
e
a
t
u
r
e
-
s
e
l
e
c
t
i
o
n
s
t
a
g
e
,
e
s
p
e
c
i
a
l
l
y
w
h
e
n
t
h
e
n
u
m
b
e
r
o
f
s
a
m
p
l
e
s
a
n
d
f
e
a
t
u
r
e
s
i
s
h
i
g
h
.
T
h
e
m
a
j
o
r
c
o
m
p
u
t
a
t
i
o
n
a
l
b
u
r
d
e
n
i
s
f
r
o
m
t
h
e
q
u
a
d
r
a
t
i
c
o
p
e
r
a
t
i
o
n
s
i
n
t
h
e
f
i
l
t
e
r
s
t
e
p
,
e
s
p
e
c
i
a
l
l
y
t
h
e
n
e
a
r
e
s
t
-
n
e
i
g
h
b
o
r
c
o
m
p
u
t
a
t
i
o
n
o
f
R
e
l
i
e
f
F
,
a
n
d
t
h
e
G
A
-
b
a
s
e
d
w
r
a
p
p
e
r
s
t
e
p
t
h
a
t
r
e
q
u
i
r
e
s
t
r
a
i
n
i
n
g
t
h
e
M
L
P
c
l
a
s
s
i
f
i
e
r
f
o
r
a
n
u
m
b
e
r
o
f
c
h
r
o
m
o
s
o
m
e
s
o
v
e
r
m
a
n
y
g
e
n
e
r
a
t
i
o
n
s
.
T
h
i
s
l
e
a
d
s
t
o
r
e
l
a
t
i
v
e
l
y
h
i
g
h
c
o
m
p
u
t
a
t
i
o
n
a
l
c
o
s
t
s
i
n
t
h
e
d
e
v
e
l
o
p
m
e
n
t
o
f
t
h
e
o
f
f
l
i
n
e
m
o
d
e
l
f
o
r
t
h
e
p
r
o
p
o
s
e
d
m
e
t
h
o
d
.
H
o
w
e
v
e
r
,
t
h
e
c
o
m
p
u
t
a
t
i
o
n
a
l
c
o
s
t
i
s
m
a
i
n
l
y
i
n
t
h
e
f
e
a
t
u
r
e
s
e
l
e
c
t
i
o
n
a
n
d
t
r
a
i
n
i
n
g
s
t
a
g
e
,
b
u
t
n
o
t
i
n
t
h
e
f
i
n
a
l
i
n
t
r
u
s
i
o
n
d
e
t
e
c
t
i
o
n
p
r
o
c
e
s
s
.
A
f
t
e
r
o
p
t
i
m
i
z
a
t
i
o
n
,
H
y
b
F
S
-
N
I
D
S
r
e
d
u
c
e
s
t
h
e
o
r
i
g
i
n
a
l
f
e
a
t
u
r
e
s
p
a
c
e
t
o
a
c
o
m
p
a
c
t
s
u
b
s
e
t
o
f
1
8
f
e
a
t
u
r
e
s
.
I
t
r
e
d
u
c
e
s
t
h
e
a
m
o
u
n
t
o
f
d
a
t
a
p
r
o
c
e
s
s
e
d
i
n
s
u
b
s
e
q
u
e
n
t
i
n
t
r
u
s
i
o
n
d
e
t
e
c
t
i
o
n
.
T
h
u
s
,
t
h
e
p
r
o
p
o
s
e
d
m
e
t
h
o
d
o
f
f
e
r
s
a
t
r
a
d
e
-
o
f
f
b
e
t
w
e
e
n
h
i
g
h
e
r
o
f
f
l
i
n
e
c
o
m
p
u
t
a
t
i
o
n
a
l
c
o
s
t
a
n
d
r
e
d
u
c
e
d
f
e
a
t
u
r
e
p
r
o
c
e
s
s
i
n
g
r
e
q
u
i
r
e
m
e
n
t
s
d
u
r
i
n
g
d
e
p
l
o
y
m
e
n
t
.
Sin
ce
wr
ap
p
er
ev
al
u
atio
n
d
o
m
in
ates
r
u
n
tim
e,
th
e
GA
s
tag
e
i
s
th
e
p
r
im
ar
y
c
o
m
p
u
tatio
n
al
b
o
ttlen
ec
k
,
an
d
th
e
f
ilter
s
tag
e
g
r
ea
tly
r
ed
u
ce
s
d
im
en
s
io
n
ality
p
r
i
o
r
t
o
wr
ap
p
er
o
p
tim
izatio
n
,
th
u
s
d
e
cr
ea
s
in
g
th
e
o
v
e
r
all
co
s
t
o
f
p
r
o
ce
s
s
in
g
.
T
h
e
co
m
p
lete
ex
p
er
im
en
tal
f
lo
w
o
f
th
e
Hy
b
FS
-
NI
DS
f
r
am
ewo
r
k
is
illu
s
tr
ated
in
Fig
u
r
e
4
.
T
h
e
p
r
o
ce
d
u
r
es
wer
e
p
er
f
o
r
m
ed
in
an
id
en
tifia
b
le
s
eq
u
en
ce
to
en
s
u
r
e
th
at
o
th
er
in
d
iv
id
u
als
ca
n
p
er
f
o
r
m
th
e
s
am
e
p
r
o
ce
d
u
r
e
with
o
u
t c
o
n
f
u
s
io
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Hyb
r
id
fea
tu
r
e
s
elec
tio
n
s
tr
a
te
g
ies fo
r
s
tr
o
n
g
n
etw
o
r
k
in
tr
u
s
i
o
n
d
etec
tio
n
(
Jo
s
ep
h
i
n
e
R
a
th
i
n
a
m
)
4373
First,
r
aw
d
ata
o
f
th
e
t
r
af
f
i
c
f
lo
w
p
r
o
v
i
d
ed
in
th
e
C
I
C
-
I
DS2
0
1
7
d
ataset
wer
e
co
l
lecte
d
an
d
p
r
e
-
p
r
o
ce
s
s
ed
.
Me
an
im
p
u
tatio
n
was
u
s
ed
to
r
em
o
v
e
d
u
p
licate
r
ec
o
r
d
s
an
d
m
is
s
in
g
v
alu
es.
Pra
ctica
lly
,
th
is
s
tag
e
o
f
clea
n
in
g
th
e
d
ata
is
ai
m
ed
at
p
r
e
v
en
tin
g
u
n
ex
p
ec
ted
er
r
o
r
s
in
th
e
f
u
tu
r
e.
Nex
t,
th
e
en
c
o
d
in
g
o
f
ca
teg
o
r
i
ca
l
f
ea
tu
r
es
in
to
n
u
m
er
ical
o
n
es
was
d
o
n
e.
Min
-
m
ax
s
ca
lin
g
was
u
s
ed
to
n
o
r
m
alize
th
e
n
u
m
er
ical
a
ttrib
u
tes.
T
h
is
s
tep
m
ain
tain
s
all
v
alu
es
at
a
s
im
ilar
r
a
n
g
e,
wh
ich
ty
p
ically
f
ac
ilit
ates e
asier
lear
n
in
g
b
y
th
e
class
if
ier
.
T
h
e
f
ea
tu
r
es
wer
e
r
an
k
ed
b
y
th
r
ee
d
is
tin
ct
f
ilter
m
eth
o
d
s
a
f
ter
p
r
e
p
r
o
ce
s
s
in
g
:
R
elief
F,
FC
B
F,
an
d
m
R
MR.
B
o
th
ap
p
r
o
ac
h
es
r
esu
lted
in
th
eir
r
an
k
in
g
lis
ts
.
T
h
e
r
esu
lts
o
f
th
ese
r
an
k
in
g
s
wer
e
th
en
co
m
b
in
ed
with
th
e
B
o
r
d
a
co
u
n
t
to
s
ele
ct
th
e
N
ca
n
d
id
ate
f
ea
t
u
r
es.
T
h
is
s
u
g
g
ests
th
at
f
ea
tu
r
es
s
u
p
p
o
r
ted
b
y
m
u
ltip
le
f
ilter
s
ar
e
g
iv
en
m
o
r
e
im
p
o
r
ta
n
ce
.
T
h
e
n
ex
t
s
tep
was
th
e
GA
w
r
ap
p
er
.
I
t
was
r
u
n
u
n
d
e
r
f
ix
e
d
p
ar
am
eter
s
,
s
u
c
h
as
p
o
p
u
la
tio
n
s
ize,
m
u
tatio
n
r
ate,
cr
o
s
s
o
v
er
r
ate,
an
d
n
u
m
b
er
o
f
g
e
n
er
atio
n
s
.
An
ML
P
class
if
ier
was
u
s
e
d
to
ev
alu
ate
ea
ch
f
ea
tu
r
e
co
m
b
in
atio
n
d
u
r
i
n
g
th
e
s
ea
r
ch
p
r
o
ce
s
s
.
I
n
s
im
p
le
te
r
m
s
,
th
e
GA
ex
p
er
im
e
n
ted
wi
th
v
ar
io
u
s
s
u
b
s
ets
an
d
k
e
p
t th
e
b
etter
o
n
es b
ased
o
n
class
if
ier
p
er
f
o
r
m
an
ce
.
Fin
ally
,
1
0
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
was
u
s
ed
to
v
alid
ate
th
e
s
elec
ted
f
ea
tu
r
e
s
u
b
s
et.
T
h
is
r
ep
etitio
n
o
f
test
in
g
h
elp
s
to
en
h
an
ce
th
e
s
tatis
tical
s
tab
ilit
y
an
d
r
ed
u
ce
s
th
e
ef
f
ec
t
o
f
r
an
d
o
m
d
ata
s
p
lits
.
All
th
e
ex
p
er
im
en
ts
wer
e
p
er
f
o
r
m
e
d
i
n
a
co
n
tr
o
lle
d
co
m
p
u
tatio
n
al
en
v
ir
o
n
m
en
t.
So
,
th
e
r
esu
lts
s
h
o
u
ld
b
e
co
n
s
is
ten
t
an
d
r
ea
s
o
n
a
b
ly
ea
s
y
to
r
ep
lica
te
u
n
d
er
s
im
ilar
co
n
d
itio
n
s
.
Fig
u
r
e
4
.
Flo
w
d
ia
g
r
am
o
f
Hy
b
FS
-
NI
DS m
eth
o
d
o
lo
g
y
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
4
.
1
.
E
x
perim
ent
a
l set
up
T
h
is
s
tu
d
y
was
ca
r
r
ied
o
u
t
in
th
e
J
av
a
p
r
o
g
r
am
m
in
g
la
n
g
u
a
g
e
an
d
in
th
e
Ap
ac
h
e
NetBean
s
I
DE
1
5
,
o
p
er
atin
g
o
n
W
in
d
o
ws
1
1
.
J
av
a
(
J
DK
1
.
8
)
an
d
NetBean
s
wer
e
ad
v
a
n
tag
eo
u
s
s
elec
tio
n
s
d
u
e
to
t
h
eir
lar
g
e
lib
r
ar
ies
an
d
d
eb
u
g
g
in
g
to
o
ls
th
at
h
elp
ed
to
c
r
ea
te
an
d
test
th
e
o
f
f
e
r
ed
h
y
b
r
id
f
ea
t
u
r
e
s
elec
tio
n
alg
o
r
ith
m
.
W
ek
a
is
an
o
p
en
-
s
o
u
r
ce
s
o
f
t
war
e
th
at
was
u
s
ed
to
p
er
f
o
r
m
m
ac
h
in
e
lear
n
in
g
class
if
icatio
n
an
d
ev
alu
atio
n
.
W
ek
a's
co
m
p
r
eh
en
s
iv
e
s
elec
tio
n
o
f
class
if
ier
s
,
wh
ich
in
clu
d
es
th
e
ML
P,
allo
wed
W
ek
a
to
ef
f
ec
tiv
ely
tr
ain
m
o
d
els,
v
alid
ate
th
em
,
an
d
ev
alu
ate
th
eir
p
er
f
o
r
m
an
ce
.
Her
e
,
W
ek
a
3
.
9
.
5
is
u
s
ed
.
T
h
is
c
o
n
f
ig
u
r
atio
n
h
el
p
ed
to
m
ak
e
s
u
r
e
t
h
at
th
e
Hy
b
FS
-
NI
DS a
lg
o
r
ith
m
was r
ig
o
r
o
u
s
ly
ev
alu
ated
i
n
a
s
tab
le
an
d
r
ep
r
o
d
u
cib
le
s
ettin
g
.
4
.
2
.
Co
m
pa
riso
n r
esu
lt
s
T
a
b
l
e
3
c
o
m
p
a
r
e
s
th
e
p
r
o
p
o
s
e
d
a
lg
o
r
i
t
h
m
o
f
H
y
b
F
S
-
N
I
DS
w
i
t
h
o
th
e
r
s
o
p
h
i
s
t
i
c
a
t
ed
m
e
t
h
o
d
s
th
a
t
w
e
r
e
p
r
e
s
en
t
e
d
i
n
t
h
e
l
i
t
er
a
tu
r
e
.
T
h
e
y
a
r
e
ev
a
lu
a
t
e
d
o
n
ac
c
u
r
ac
y
,
p
r
e
c
i
s
i
o
n
,
r
ec
a
l
l
,
F
1
-
s
c
o
r
e
,
an
d
M
C
C
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
5
,
No
.
5
,
Octo
b
er
2
0
2
6
:
4
3
6
6
-
4
3
8
0
4374
T
h
e
s
e
m
e
t
r
i
c
s
to
g
et
h
er
p
r
o
v
id
e
a
b
r
o
ad
p
ic
t
u
r
e
o
f
th
e
c
la
s
s
i
f
i
c
a
t
i
o
n
e
f
f
i
c
i
en
c
y
o
f
t
h
e
s
y
s
t
e
m
a
n
d
p
o
in
t
o
u
t
i
t
s
k
e
y
s
t
r
e
n
g
t
h
s
.
T
ab
le
3.
Per
f
o
r
m
an
ce
m
etr
ics
co
m
p
ar
is
o
n
of
NI
DS
mod
e
l
s
M
o
d
e
l
A
c
c
u
r
a
c
y
(
%
)
P
r
e
c
i
s
i
o
n
(
%
)
R
e
c
a
l
l
(
%
)
F1
-
s
c
o
r
e
(
%
)
M
C
C
(
%
)
B
u
l
a
v
a
s
et
a
l
.
[
2
6
]
81
.
0
0
77
.
0
0
83
.
0
0
76
.
0
0
6
1
.
0
6
R
a
mz
a
n
et
a
l
.
[
2
7
]
73
.
0
0
72
.
0
0
79
.
0
0
68
.
0
0
4
9
.
2
2
O
l
i
v
e
i
r
a
et
a
l
.
[
2
8
]
7
6
.
2
2
7
8
.
6
8
7
3
.
7
5
7
5
.
7
9
5
4
.
1
3
Al
F
a
r
s
i
et
a
l
.
[
2
9
]
82
.
0
0
84
.
0
0
80
.
0
0
80
.
0
0
6
4
.
8
7
H
y
b
F
S
-
N
I
D
S
9
0
.
6
8
9
1
.
7
4
9
0
.
6
4
9
1
.
1
9
9
1
.
4
8
Fro
m
T
ab
le
3
,
th
e
Hy
b
FS
-
NI
DS
alg
o
r
ith
m
p
er
f
o
r
m
s
b
etter
th
an
all
th
e
o
th
er
e
x
is
tin
g
m
o
d
els
in
ter
m
s
o
f
all
p
er
f
o
r
m
a
n
ce
r
esu
lts
.
T
h
e
m
o
d
el
is
p
er
f
o
r
m
in
g
b
etter
in
th
e
o
v
er
all
p
r
e
d
ic
tio
n
with
a
h
ig
h
ac
cu
r
ac
y
o
f
9
0
.
6
8
%.
T
h
e
ac
c
u
r
ac
y
(
9
1
.
7
4
%)
an
d
r
ec
all
(
9
0
.
6
4
%)
s
co
r
es
s
h
o
w
th
at
it
is
u
s
ef
u
l
to
co
r
r
ec
tl
y
id
en
tify
attac
k
s
with
lo
w
f
als
e
p
o
s
itiv
es.
I
t
h
as
a
h
ig
h
F1
-
s
co
r
e
(
9
1
.
1
9
%),
wh
ich
g
iv
es
a
g
o
o
d
b
alan
ce
o
f
p
r
ec
is
io
n
an
d
r
ec
all,
wh
ich
ar
e
ess
en
tial
to
in
tr
u
s
io
n
d
etec
tio
n
as
b
o
th
f
alse
p
o
s
itiv
es
an
d
f
alse
n
eg
ativ
es
ca
n
h
av
e
s
er
io
u
s
r
ep
e
r
cu
s
s
io
n
s
.
T
h
e
MCC
o
f
9
1
.
4
8
%
s
u
g
g
e
s
ts
th
at
th
e
class
if
ier
is
r
eliab
l
e
an
d
wo
r
k
s
well
in
d
ea
lin
g
with
in
tr
u
s
io
n
d
atas
ets
co
n
tain
in
g
u
n
e
v
en
n
u
m
b
e
r
s
o
f
attac
k
ty
p
es.
T
h
e
ad
v
an
ta
g
es
ar
is
e
d
u
e
to
a
h
y
b
r
id
f
ea
tu
r
e
s
elec
tio
n
s
tr
a
teg
y
,
wh
ich
elim
in
ates
r
ed
u
n
d
an
t
f
ea
tu
r
es
an
d
o
p
tim
izes
th
e
s
u
b
s
et
to
en
h
an
ce
th
e
class
if
ier
.
T
h
is
m
in
im
izes
n
o
is
e
an
d
in
cr
ea
s
e
s
th
e
g
en
er
aliza
b
ilit
y
o
f
t
h
e
m
o
d
el.
T
h
e
ML
P
class
if
ier
was
s
elec
ted
b
ec
a
u
s
e
s
ev
er
al
r
ec
en
t
in
tr
u
s
io
n
d
etec
tio
n
s
tu
d
ies,
in
clu
d
in
g
[
2
6
]
‒
[
2
9
]
,
u
s
e
d
MLP
-
b
ased
m
o
d
els
f
o
r
ev
al
u
atio
n
.
Usi
n
g
th
e
s
am
e
clas
s
if
ier
en
s
u
r
es
a
f
air
co
m
p
ar
is
o
n
u
n
d
er
s
im
ilar
co
n
d
itio
n
s
.
M
o
r
eo
v
er
,
ML
P
i
s
ef
f
ec
tiv
e
in
h
a
n
d
lin
g
co
m
p
lex
n
o
n
lin
ea
r
r
elatio
n
s
h
ip
s
i
n
h
ig
h
-
d
im
e
n
s
io
n
al
cy
b
er
s
ec
u
r
ity
d
ata.
T
h
e
m
ain
o
b
jectiv
e
o
f
th
is
s
tu
d
y
is
to
e
v
alu
ate
th
e
ef
f
ec
tiv
en
ess
o
f
t
h
e
p
r
o
p
o
s
ed
h
y
b
r
id
f
ea
tu
r
e
s
elec
tio
n
m
eth
o
d
r
ath
e
r
th
an
co
m
p
ar
e
m
u
ltip
le
class
if
ier
s
.
T
h
er
e
f
o
r
e,
m
ain
tain
i
n
g
a
co
n
s
is
ten
t
ML
P
ar
ch
itectu
r
e
h
elp
s
en
s
u
r
e
th
at
p
er
f
o
r
m
a
n
ce
im
p
r
o
v
em
en
ts
ar
e
m
ain
ly
d
u
e
to
th
e
p
r
o
p
o
s
ed
Hy
b
FS
-
NI
DS
f
r
am
ewo
r
k
.
T
h
e
ac
h
ie
v
ed
r
e
s
u
lts
,
in
clu
d
in
g
9
0
.
6
8
%
ac
c
u
r
ac
y
,
9
1
.
1
9
%
F1
-
s
co
r
e,
a
n
d
9
1
.
4
8
%
MCC
,
co
n
f
ir
m
t
h
e
ef
f
ec
tiv
e
n
ess
o
f
th
e
p
r
o
p
o
s
ed
f
ea
tu
r
e
s
elec
tio
n
a
p
p
r
o
ac
h
.
4
.
3
.
Dis
cus
s
io
ns
T
h
e
ex
p
er
im
e
n
tal
r
esu
lts
in
Fi
g
u
r
es
5
to
9
s
h
o
w
th
e
e
f
f
ec
tiv
en
ess
o
f
Hy
b
FS
-
NI
DS
o
v
er
th
e
cu
r
r
en
t
ap
p
r
o
ac
h
es.
As
s
h
o
wn
in
Fi
g
u
r
e
5
,
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
y
ield
s
b
etter
ac
c
u
r
ac
y
.
T
h
is
is
attr
ib
u
ted
to
th
e
in
teg
r
atio
n
o
f
s
tatis
tical
an
d
wr
ap
p
er
-
b
ased
f
ea
tu
r
e
s
elec
tio
n
m
eth
o
d
s
,
wh
ich
s
elec
t
th
e
m
o
s
t
s
ig
n
if
ican
t
f
ea
tu
r
es
an
d
im
p
r
o
v
e
th
e
class
if
icatio
n
p
er
f
o
r
m
a
n
ce
.
Fig
u
r
e
6
in
d
icate
s
th
e
ac
cu
r
ac
y
,
an
d
Hy
b
FS
-
NI
DS
h
as
s
h
o
wn
a
co
n
s
id
er
a
b
le
im
p
r
o
v
em
en
t.
T
h
is
is
d
ir
ec
tly
d
u
e
to
th
e
r
em
o
v
al
o
f
e
x
tr
an
eo
u
s
an
d
u
n
n
ec
ess
ar
y
ch
ar
ac
ter
is
tics
,
an
d
th
u
s
m
in
im
izes
th
e
am
o
u
n
t
o
f
f
alse
p
o
s
itiv
es,
an
d
r
ep
o
r
te
d
in
cu
r
s
io
n
s
ar
e
h
ig
h
ly
r
eliab
le
.
Fig
u
r
e
7
r
ep
r
esen
ts
th
e
r
ec
all
co
m
p
ar
is
o
n
.
T
h
e
h
ig
h
r
ec
all
o
f
Hy
b
FS
-
NI
DS
d
ep
icts
th
at
it
h
as
a
h
ig
h
ab
ilit
y
to
id
en
tify
r
ea
l
in
tr
u
s
io
n
s
with
a
lo
w
f
alse
n
e
g
ativ
e
r
ate
.
T
h
is
h
as
a
s
ig
n
if
ican
t
r
o
le
in
n
etwo
r
k
s
ec
u
r
ity
,
in
wh
ich
f
alse d
etec
tio
n
s
ca
n
ca
u
s
e
s
er
io
u
s
v
u
ln
er
a
b
ilit
ies.
Fig
u
r
e
5.
Acc
u
r
ac
y
c
o
m
p
a
r
i
s
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Hyb
r
id
fea
tu
r
e
s
elec
tio
n
s
tr
a
te
g
ies fo
r
s
tr
o
n
g
n
etw
o
r
k
in
tr
u
s
i
o
n
d
etec
tio
n
(
Jo
s
ep
h
i
n
e
R
a
th
i
n
a
m
)
4375
Fig
u
r
e
6
.
Pre
cisi
o
n
c
o
m
p
a
r
i
s
o
n
Fig
u
r
e
7
.
R
ec
all
co
m
p
a
r
is
o
n
T
h
e
F1
-
s
co
r
es
ar
e
s
h
o
wn
in
Fig
u
r
e
8
,
an
d
th
is
r
esu
lt
in
d
icate
s
th
at
Hy
b
FS
-
NI
DS
is
th
e
lead
in
g
p
er
f
o
r
m
er
,
wh
ic
h
af
f
ir
m
s
th
at
th
e
m
o
d
el
m
ain
tain
s
a
f
air
an
d
s
tr
o
n
g
t
r
ad
e
-
o
f
f
b
etwe
en
p
r
ec
is
io
n
an
d
r
ec
al
l
an
d
en
h
an
ce
s
th
e
o
v
e
r
all
ef
f
ec
tiv
en
ess
o
f
th
e
d
etec
tio
n
s
y
s
tem
.
T
h
e
co
m
p
ar
is
o
n
o
f
MCC
is
g
iv
en
in
Fig
u
r
e
9
.
T
h
e
Hy
b
FS
-
NI
DS
r
ec
eiv
ed
th
e
b
est
s
co
r
e.
T
h
e
h
i
g
h
MCC
v
alu
e
in
d
icate
s
th
at
th
e
class
if
icatio
n
p
er
f
o
r
m
an
ce
is
ef
f
ec
tiv
e
ev
e
n
wh
en
th
e
d
ataset
is
clas
s
-
im
b
alan
ce
d
.
T
h
is
r
ef
lects
th
e
r
o
b
u
s
tn
ess
an
d
d
ep
en
d
a
b
ilit
y
o
f
th
e
alg
o
r
ith
m
.
T
h
e
ex
p
er
im
e
n
tal
r
esu
lts
an
d
an
aly
s
is
s
h
o
w
th
at
th
e
Hy
b
FS
-
NI
DS
alg
o
r
ith
m
ca
n
b
e
u
s
ed
f
o
r
im
p
r
o
v
in
g
th
e
p
er
f
o
r
m
an
ce
o
f
in
tr
u
s
io
n
d
etec
tio
n
.
I
t
c
o
m
b
in
es
s
ev
er
al
f
ilter
in
g
m
eth
o
d
s
with
a
wr
ap
p
er
ap
p
r
o
ac
h
b
ased
o
n
GA
.
I
t
s
u
cc
ess
f
u
lly
id
en
tifie
s
th
e
m
o
s
t
r
elev
an
t
in
f
o
r
m
atio
n
,
im
p
r
o
v
es
class
if
icatio
n
ac
cu
r
ac
y
,
an
d
o
b
tain
s
b
etter
p
r
ec
is
io
n
,
r
ec
all,
F1
-
s
co
r
e,
a
n
d
MCC
m
etr
ics
th
an
p
r
ev
io
u
s
m
o
d
els.
T
h
e
r
esu
lts
in
d
icate
th
at
Hy
b
FS
-
NI
DS
ca
n
b
e
u
tili
ze
d
to
o
f
f
er
a
f
ea
s
ib
le
s
o
lu
tio
n
f
o
r
ef
f
ec
tiv
e
an
d
s
o
u
n
d
n
etwo
r
k
in
tr
u
s
io
n
d
etec
tio
n
f
o
r
r
ea
l
-
w
o
r
ld
ap
p
licatio
n
s
.
I
n
o
r
d
er
to
f
u
r
th
er
test
th
e
ef
f
ec
tiv
en
ess
o
f
Hy
b
FS
-
NI
DS,
it
was
co
m
p
ar
ed
with
o
th
e
r
p
o
p
u
lar
tech
n
i
q
u
es
o
f
f
ea
tu
r
e
s
elec
tio
n
lik
e
I
G,
SF
S,
an
d
PS
O.
All
th
e
m
et
h
o
d
s
wer
e
test
ed
in
t
h
e
s
am
e
s
etu
p
.
T
h
e
y
wer
e
tr
ain
ed
o
n
th
e
s
am
e
ML
P
m
o
d
el
an
d
th
e
s
am
e
C
I
C
-
I
DS2
0
1
7
d
ataset; th
er
ef
o
r
e,
th
e
co
m
p
a
r
i
s
o
n
wo
u
ld
b
e
f
air
.
I
n
p
r
ac
tice,
I
G
r
ed
u
ce
s
f
ea
tu
r
es
q
u
ick
ly
.
T
h
at
is
h
elp
f
u
l.
B
u
t
it
d
o
es
n
o
t
alwa
y
s
d
r
o
p
f
ea
tu
r
es
th
at
s
to
r
e
s
im
ilar
in
f
o
r
m
atio
n
.
T
h
is
ca
n
lead
to
m
in
o
r
v
ar
iatio
n
s
in
p
er
f
o
r
m
an
ce
wh
e
n
th
e
m
o
d
e
l is tr
ain
ed
ag
ain
.
On
e
p
o
ten
tial
ca
u
s
e
is
th
at
I
G
will
ex
am
in
e
in
d
iv
id
u
al
f
ea
tu
r
es
s
ep
ar
ately
an
d
n
o
t
co
m
p
r
eh
e
n
s
iv
ely
ex
am
in
e
th
e
in
ter
ac
tio
n
o
f
f
ea
tu
r
es.
I
n
s
tead
,
SF
S
co
n
s
tr
u
cts
th
e
f
ea
tu
r
e
s
et
in
s
tag
es.
T
h
is
ten
d
s
to
in
cr
ea
s
e
th
e
q
u
ality
o
f
th
e
ch
o
s
en
s
u
b
s
et.
Ho
wev
er
,
th
e
p
r
o
ce
s
s
i
s
m
o
r
e
tim
e
-
co
n
s
u
m
i
n
g
.
I
t
m
ak
es
n
u
m
er
o
u
s
co
m
b
in
atio
n
s
;
h
en
ce
th
e
tr
ain
in
g
is
ca
r
r
ied
o
u
t
r
ep
ea
ted
ly
.
I
n
a
r
ea
l
-
life
ca
s
e,
if
a
s
ec
u
r
ity
team
n
ee
d
s
to
u
p
d
ate
a
m
o
d
el
q
u
ick
l
y
af
ter
a
n
ew
ty
p
e
o
f
attac
k
ap
p
ea
r
s
,
th
is
ex
tr
a
tim
e
m
ay
b
ec
o
m
e
a
p
r
o
b
lem
.
PS
O
also
Evaluation Warning : The document was created with Spire.PDF for Python.