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
.
4
,
A
u
g
u
s
t
20
25
,
p
p
.
4
1
0
9
~
4
1
1
9
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
15
i
4
.
pp
4
1
0
9
-
4
1
1
9
4109
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
A comp
a
ra
tive st
udy
of deep
learn
ing
-
ba
sed net
wo
r
k int
rusio
n
detec
tion sy
stem
with
e
x
pla
ina
ble
a
rtif
icia
l in
tellige
nce
T
a
n J
ua
n K
a
i,
L
ee
-
Yeng
O
ng
,
M
eng
-
Chew
L
eo
w
F
a
c
u
l
t
y
o
f
I
n
f
o
r
ma
t
i
o
n
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
(
F
I
S
T)
,
M
u
l
t
i
me
d
i
a
U
n
i
v
e
r
si
t
y
,
M
e
l
a
k
a
,
M
a
l
a
y
si
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Oct
2
4
,
2
0
2
4
R
ev
is
ed
Ma
r
1
6
,
2
0
2
5
Acc
ep
ted
Ma
y
2
3
,
2
0
2
5
In
th
e
ra
p
i
d
l
y
e
v
o
l
v
i
n
g
la
n
d
sc
a
p
e
o
f
c
y
b
e
rse
c
u
rit
y
,
ro
b
u
st
n
e
two
r
k
in
tru
si
o
n
d
e
tec
ti
o
n
s
y
ste
m
s
(NID
S
)
a
re
c
ru
c
ial
to
c
o
u
n
teri
n
g
i
n
c
re
a
sin
g
ly
so
p
h
ist
ica
ted
c
y
b
e
r
t
h
re
a
ts,
in
c
lu
d
i
n
g
z
e
ro
-
d
a
y
a
tt
a
c
k
s.
De
e
p
l
e
a
rn
in
g
a
p
p
ro
a
c
h
e
s
i
n
NID
S
o
ffe
r
p
ro
m
isin
g
imp
r
o
v
e
m
e
n
ts
i
n
in
tr
u
sio
n
d
e
tec
ti
o
n
ra
tes
a
n
d
re
d
u
c
ti
o
n
o
f
fa
lse
p
o
siti
v
e
s.
Ho
we
v
e
r,
th
e
i
n
h
e
re
n
t
o
p
a
c
it
y
o
f
d
e
e
p
l
e
a
rn
in
g
m
o
d
e
ls
p
re
se
n
ts
sig
n
ifi
c
a
n
t
c
h
a
ll
e
n
g
e
s,
h
i
n
d
e
ri
n
g
th
e
u
n
d
e
rsta
n
d
in
g
a
n
d
tru
st
in
t
h
e
ir
d
e
c
isio
n
-
m
a
k
in
g
p
ro
c
e
ss
e
s.
Th
is stu
d
y
e
x
p
lo
re
s t
h
e
e
ffi
c
a
c
y
o
f
e
x
p
lain
a
b
le
a
rti
f
icia
l
in
tell
ig
e
n
c
e
(XA
I)
tec
h
n
iq
u
e
s,
sp
e
c
ifi
c
a
ll
y
S
h
a
p
ley
a
d
d
it
i
v
e
e
x
p
la
n
a
ti
o
n
s
(S
HA
P
)
a
n
d
l
o
c
a
l
in
ter
p
re
tab
le
m
o
d
e
l
-
a
g
n
o
stic
e
x
p
lan
a
ti
o
n
s
(LI
M
E),
i
n
e
n
h
a
n
c
i
n
g
t
h
e
tran
sp
a
re
n
c
y
a
n
d
tru
stwo
r
th
in
e
ss
o
f
NID
S
sy
ste
m
s.
Wi
t
h
t
h
e
imp
le
m
e
n
tatio
n
o
f
Tab
Ne
t
a
rc
h
it
e
c
tu
re
o
n
th
e
AWID3
d
a
tas
e
t,
it
is
a
b
le
to
a
c
h
iev
e
a
re
m
a
rk
a
b
le
a
c
c
u
ra
c
y
o
f
9
9
.
9
9
%
.
De
sp
it
e
th
is
h
i
g
h
p
e
rfo
rm
a
n
c
e
,
c
o
n
c
e
rn
s
re
g
a
r
d
in
g
th
e
i
n
terp
re
tab
il
it
y
o
f
th
e
Tab
Ne
t
m
o
d
e
l'
s
d
e
c
isio
n
s
p
e
rsi
st.
By
e
m
p
l
o
y
i
n
g
S
HA
P
a
n
d
L
IM
E,
t
h
is
stu
d
y
a
ims
to
e
lu
c
id
a
te
th
e
i
n
tri
c
a
c
ies
o
f
m
o
d
e
l
in
terp
re
tab
il
it
y
,
f
o
c
u
sin
g
o
n
b
o
t
h
g
l
o
b
a
l
a
n
d
l
o
c
a
l
a
sp
e
c
ts
o
f
th
e
Tab
Ne
t
m
o
d
e
l'
s
d
e
c
isio
n
-
m
a
k
in
g
p
ro
c
e
ss
e
s.
Ulti
m
a
tely
,
th
is
stu
d
y
u
n
d
e
rsc
o
re
s
th
e
p
i
v
o
tal
ro
le
o
f
XA
I
in
imp
ro
v
i
n
g
u
n
d
e
rsta
n
d
i
n
g
a
n
d
fo
s
terin
g
tr
u
st
i
n
d
e
e
p
lea
rn
i
n
g
-
b
a
s
e
d
NID
S
sy
ste
m
s.
Th
e
r
o
b
u
stn
e
ss
o
f
t
h
e
m
o
d
e
l
is
a
ls
o
b
e
in
g
tes
ted
b
y
a
d
d
i
n
g
th
e
sig
n
a
l
-
to
-
n
o
ise
ra
ti
o
(
S
NR) t
o
th
e
d
a
tas
e
ts.
K
ey
w
o
r
d
s
:
AW
I
D3
d
ataset
Dee
p
l
ea
r
n
in
g
E
x
p
lain
ab
le
ar
tific
ial
in
tellig
en
ce
L
o
ca
l
in
ter
p
r
eta
b
le
m
o
d
el
-
ag
n
o
s
tic
ex
p
lan
atio
n
N
etwo
r
k
in
tr
u
s
io
n
d
etec
tio
n
s
y
s
tem
Sh
ap
ley
ad
d
itiv
e
e
x
p
lan
atio
n
T
ab
Net
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
:
L
ee
-
Yen
g
On
g
Facu
lty
o
f
I
n
f
o
r
m
atio
n
Scien
c
e
an
d
T
ec
h
n
o
lo
g
y
,
M
u
ltime
d
ia
Un
iv
er
s
ity
J
alan
Ay
er
Ker
o
h
L
a
m
a
7
5
4
5
0
Me
lak
a,
Ma
lay
s
ia
E
m
ail: ly
o
n
g
@
m
m
u
.
e
d
u
.
m
y
1.
I
NT
RO
D
UCT
I
O
N
As
th
e
u
s
e
o
f
th
e
in
ter
n
et
co
n
tin
u
es
to
g
r
o
w,
m
ain
tai
n
in
g
r
o
b
u
s
t
s
ec
u
r
ity
m
ea
s
u
r
es
b
ec
o
m
es
in
cr
ea
s
in
g
ly
im
p
o
r
tan
t.
Mo
r
e
o
v
er
,
th
e
p
r
ev
alen
ce
o
f
ze
r
o
-
d
ay
attac
k
s
ad
d
s
a
lay
er
o
f
u
r
g
en
cy
o
n
d
ev
elo
p
i
n
g
an
d
im
p
lem
en
tin
g
s
u
ch
m
ea
s
u
r
es
[
1
]
.
As
s
u
ch
,
an
o
m
aly
-
b
as
ed
n
etwo
r
k
in
tr
u
s
io
n
d
etec
tio
n
s
y
s
tem
(
NI
DS)
is
in
tr
o
d
u
ce
d
to
e
f
f
ec
tiv
ely
d
etec
t
z
er
o
-
d
a
y
attac
k
s
th
r
o
u
g
h
o
u
t
co
m
p
ar
is
o
n
o
f
n
etwo
r
k
tr
af
f
ic
p
r
o
f
iles
,
u
tili
zin
g
p
o
wer
o
f
m
ac
h
in
e
lear
n
in
g
o
r
d
ee
p
lea
r
n
in
g
a
p
p
r
o
ac
h
es
[
2
]
.
Dee
p
lear
n
in
g
a
p
p
r
o
ac
h
es
in
NI
DS
m
o
d
el
d
ev
elo
p
m
e
n
t
h
a
d
p
r
o
v
e
n
to
b
e
m
o
r
e
ef
f
ec
tiv
e
as
it
o
f
ten
h
as
a
b
etter
p
e
r
f
o
r
m
an
ce
in
ter
m
s
o
f
p
r
o
d
u
cin
g
a
h
ig
h
d
etec
tio
n
r
ate
wh
ile
k
ee
p
in
g
t
h
e
lo
w
p
o
s
itiv
e
r
ate
[
3
]
.
Desp
ite
th
e
ef
f
ec
tiv
en
ess
o
f
d
ee
p
lea
r
n
in
g
-
b
ase
d
ap
p
r
o
ac
h
es
in
NI
DS,
th
eir
d
ec
is
io
n
-
m
ak
in
g
p
r
o
ce
s
s
es
o
f
ten
lack
tr
an
s
p
ar
en
c
y
an
d
clar
ity
[
4
]
.
T
h
e
ex
p
lain
ab
ilit
y
o
f
p
r
ed
ictio
n
a
n
d
class
if
icatio
n
m
o
d
els
is
t
y
p
ically
in
v
e
r
s
ely
p
r
o
p
o
r
tio
n
al
to
th
eir
lear
n
in
g
p
er
f
o
r
m
an
ce
,
esp
ec
ially
f
o
r
d
ee
p
lear
n
in
g
ap
p
r
o
ac
h
es
wh
ich
ar
e
o
f
ten
r
ef
er
r
ed
to
as
"b
l
ac
k
b
o
x
es"
d
u
e
to
th
eir
co
m
p
lex
s
tr
u
ctu
r
es a
n
d
o
p
aq
u
e
d
ec
is
io
n
-
m
ak
in
g
p
r
o
ce
s
s
es
[
5
]
,
[
6
]
.
T
h
is
lack
o
f
in
ter
p
r
etab
ilit
y
p
o
s
es
s
ig
n
if
ican
t
ch
allen
g
es
f
o
r
n
etwo
r
k
ad
m
in
is
tr
ato
r
s
wh
o
r
ely
o
n
th
ese
s
y
s
tem
s
to
id
en
tify
an
d
r
esp
o
n
d
to
a
b
n
o
r
m
al
n
etwo
r
k
b
eh
a
v
io
r
s
.
Mo
r
eo
v
e
r
,
it
is
c
r
u
cial
to
u
n
d
er
s
tan
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
.
4
,
Au
g
u
s
t
20
25
:
4
1
0
9
-
4119
4110
th
e
m
o
d
el’
s
b
e
h
av
io
r
,
as
t
h
is
u
n
d
er
s
tan
d
i
n
g
allo
ws
t
h
em
t
o
tr
u
s
t
th
e
s
y
s
tem
'
s
aler
ts
an
d
ta
k
e
in
f
o
r
m
ed
ac
tio
n
s
b
ased
o
n
th
e
m
o
d
el'
s
o
u
tp
u
t.
W
ith
o
u
t
th
is
k
n
o
wled
g
e,
ad
m
in
i
s
tr
ato
r
s
m
ay
s
tr
u
g
g
le
to
d
is
tin
g
u
is
h
b
etwe
en
tr
u
e
th
r
ea
ts
an
d
f
alse
p
o
s
itiv
es,
p
o
ten
tially
lead
in
g
to
eith
e
r
u
n
n
ec
ess
ar
y
d
is
r
u
p
tio
n
s
o
r
m
is
s
ed
attac
k
s
.
A
s
s
u
ch
,
th
e
em
er
g
e
n
ce
o
f
e
x
p
lain
ab
le
AI
(
XAI
)
h
as
b
ec
o
m
e
v
ital
in
th
e
r
ea
lm
o
f
n
etwo
r
k
i
n
tr
u
s
io
n
d
etec
t
io
n
,
en
h
an
cin
g
th
e
tr
a
n
s
p
ar
en
cy
an
d
in
ter
p
r
etab
ilit
y
o
f
AI
m
o
d
els.
T
h
is
s
tu
d
y
e
m
p
lo
y
s
Sh
ap
ley
a
d
d
itiv
e
ex
p
lan
atio
n
s
(
SHAP)
an
d
lo
c
al
in
ter
p
r
etab
le
m
o
d
el
-
ag
n
o
s
tic
ex
p
lan
atio
n
s
(
L
I
ME
)
,
wh
ic
h
ar
e
well
-
r
e
g
ar
d
e
d
XAI
tech
n
iq
u
es,
to
el
u
cid
ate
t
h
e
d
ec
is
io
n
-
m
a
k
in
g
p
r
o
ce
s
s
es o
f
th
ese
m
o
d
els.
T
h
e
co
n
tr
i
b
u
tio
n
s
o
f
th
is
p
a
p
e
r
ar
e
th
r
ee
f
o
ld
:
f
ir
s
t,
it
d
ev
el
o
p
s
a
NI
DS
m
o
d
el
u
s
in
g
a
d
ee
p
lear
n
in
g
ap
p
r
o
ac
h
th
at
ac
h
iev
es
a
h
ig
h
in
tr
u
s
io
n
d
etec
tio
n
r
ate
with
a
lo
w
f
alse
p
o
s
itiv
e
r
ate.
Seco
n
d
ly
,
it
e
v
alu
ates
th
e
r
o
b
u
s
tn
ess
o
f
th
e
NI
DS
m
o
d
e
l
in
n
o
is
y
e
n
v
ir
o
n
m
en
ts
.
L
astl
y
,
it
in
ter
p
r
ets
th
e
m
o
d
el'
s
d
ec
is
io
n
-
m
ak
in
g
s
tep
s
th
r
o
u
g
h
th
e
a
p
p
licatio
n
o
f
SHAP a
n
d
L
I
ME
.
T
h
e
r
em
ain
in
g
p
ar
t
o
f
th
e
p
a
p
er
is
s
tr
u
ctu
r
ed
as
f
o
llo
ws
.
Sectio
n
2
d
is
cu
s
s
es
r
elate
d
wo
r
k
o
n
d
ee
p
lear
n
in
g
-
b
ased
NI
DS
an
d
XAI
ap
p
r
o
ac
h
es
.
Sectio
n
3
c
o
n
ce
n
tr
ates
o
n
th
e
m
eth
o
d
o
l
o
g
y
o
f
th
e
e
x
p
er
im
e
n
t
p
ip
elin
e.
Sectio
n
4
s
h
o
ws
th
e
ex
p
er
im
e
n
tal
r
esu
lts
an
d
d
i
s
cu
s
s
io
n
,
an
d
f
in
all
y
s
ec
tio
n
5
co
n
clu
d
es th
e
co
n
tr
ib
u
tio
n
s
an
d
f
u
tu
r
e
wo
r
k
o
f
t
h
e
s
tu
d
y
.
2.
RE
L
AT
E
D
WO
RK
2
.
1
.
Dee
p
le
a
rning
a
pp
ro
a
ches
in
NIDS
d
ev
elo
pm
ent
Dee
p
lear
n
in
g
ap
p
r
o
ac
h
es
in
NI
DS
co
n
s
is
t
o
f
d
ee
p
n
eu
r
al
n
etwo
r
k
(
DNN)
,
co
n
v
o
lu
t
io
n
n
eu
r
al
n
etwo
r
k
(
C
NN)
an
d
lo
n
g
-
s
h
o
r
t
ter
m
m
em
o
r
y
(
L
STM
)
.
I
n
ter
m
s
o
f
DNN,
T
an
g
et
a
l.
[
7
]
h
ad
d
ev
elo
p
ed
a
s
o
f
twar
e
-
d
ef
in
ed
n
etwo
r
k
-
b
ased
NI
DS
u
s
in
g
DNN
a
n
d
m
a
n
ag
e
t
o
h
it
an
ac
cu
r
ac
y
o
f
7
5
.
7
5
%
o
n
NSL
-
KDD
d
atasets
.
Simi
lar
ly
,
W
an
g
et
a
l.
[
8
]
f
o
u
n
d
o
u
t
th
at
DNN
em
er
g
es
in
ter
m
s
o
f
in
tr
u
s
io
n
d
etec
tio
n
f
o
r
th
e
C
E
S
-
C
I
C
-
I
DS
2
0
1
8
d
atasets
af
ter
co
m
p
ar
in
g
th
e
r
esu
lts
with
o
th
er
f
iv
e
d
ee
p
lear
n
in
g
m
o
d
els
an
d
m
an
ag
e
to
h
it
th
e
ac
cu
r
ac
y
o
f
9
8
.
7
9
% a
cc
u
r
ac
y
u
s
in
g
f
i
v
e
h
id
d
en
lay
er
s
w
ith
2
5
6
n
o
d
es.
I
n
ter
m
s
o
f
C
NN,
Ah
m
ad
e
t
a
l.
[
9
]
h
ad
p
r
o
p
o
s
ed
a
C
NN
m
o
d
el
u
s
in
g
AW
I
D3
d
a
tasets
af
ter
en
co
d
in
g
an
d
c
o
n
v
er
tin
g
th
e
t
ab
u
lar
d
ata
in
to
im
a
g
es
u
s
in
g
Gr
am
ian
an
g
u
la
r
f
ield
ap
p
r
o
ac
h
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
o
f
th
e
ar
ch
itectu
r
e
2
D
-
C
NN
-
1
l
ay
er
ac
h
ie
v
ed
t
h
e
b
es
t
p
er
f
o
r
m
an
ce
an
d
m
an
a
g
ed
to
h
it
a
n
ac
cu
r
ac
y
o
f
9
9
.
7
7
%,
with
a
p
r
ec
is
io
n
o
f
9
9
.
5
9
%,
r
ec
all
o
f
9
9
.
7
3
%
an
d
F1
-
s
co
r
e
o
f
9
9
.
6
6
%.
Mo
r
eo
v
e
r
,
an
L
STM
-
b
ased
m
o
d
el
f
o
r
i
n
tr
u
s
io
n
d
etec
tio
n
in
in
-
v
e
h
icle
C
AN
b
u
s
co
m
m
u
n
icatio
n
s
was
em
p
lo
y
ed
b
y
Ho
s
s
ain
et
a
l.
[
1
0
]
,
ac
h
iev
in
g
a
n
im
p
r
ess
iv
e
ac
cu
r
ac
y
o
f
9
9
.
9
9
5
% u
s
in
g
s
elf
-
c
o
llected
d
atasets
.
Hy
b
r
id
-
b
ased
a
p
p
r
o
ac
h
es
o
f
C
NN
an
d
L
STM
h
av
e
also
co
m
m
o
n
l
y
u
s
ed
in
th
e
d
ev
el
o
p
m
en
t
o
f
NI
DS.
Fo
r
in
s
tan
ce
,
Deo
r
e
an
d
B
h
o
s
ale
[
1
1
]
d
ev
elo
p
ed
C
NN
-
L
STM
m
o
d
el
b
y
u
s
in
g
th
e
C
NN
ar
ch
itectu
r
e
f
o
r
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
u
s
in
g
L
STM
as
it
s
clas
s
if
ier
,
t
h
r
o
u
g
h
in
teg
r
atio
n
with
ch
im
p
ch
ick
en
s
war
m
o
p
tim
izatio
n
ap
p
r
o
ac
h
.
T
h
e
C
NN
-
L
STM
m
o
d
el
m
an
ag
es
to
h
it
an
ac
cu
r
ac
y
o
f
9
3
.
9
7
%
f
o
r
n
o
n
-
attac
k
p
r
o
f
ile
an
d
9
8
.
8
8
%
f
o
r
t
h
e
in
tr
u
s
io
n
s
attem
p
t
in
NSL
-
KDD
d
atase
t,
wh
ile
h
itti
n
g
an
ac
cu
r
ac
y
o
f
9
8
.
8
8
%
f
o
r
n
o
n
-
attac
k
p
r
o
f
ile
a
n
d
9
0
.
5
8
%
a
cc
u
r
ac
y
o
f
attac
k
p
r
o
f
ile
in
th
e
B
o
T
-
I
o
T
d
ataset.
T
h
e
s
a
m
e
ap
p
r
o
ac
h
was
cu
s
to
m
ized
in
t
h
e
wo
r
k
o
f
[
1
2
]
,
wh
ich
m
an
a
g
ed
t
o
h
it
a
n
ac
cu
r
ac
y
o
f
9
9
.
8
4
%
f
o
r
b
in
a
r
y
class
if
icatio
n
an
d
9
9
.
8
0
ac
c
u
r
ac
y
f
o
r
m
u
lticlas
s
class
if
ica
tio
n
in
X
-
I
I
o
T
I
D
d
a
taset.
I
n
ad
d
itio
n
,
th
e
cu
s
to
m
ized
ar
ch
itectu
r
e
o
f
C
NN
-
L
STM
also
ac
h
iev
ed
an
ac
cu
r
ac
y
o
f
9
3
.
2
1
%
f
o
r
b
in
ar
y
class
if
icatio
n
an
d
9
2
.
9
%
f
o
r
m
u
lticlas
s
class
if
icatio
n
in
UNS
W
-
NB
1
5
d
ataset
.
2
.
2
.
E
x
pla
ina
ble A
I
a
pp
ro
a
ches
in N
I
DS
E
x
p
l
a
i
n
a
b
l
e
A
I
(
XA
I
)
a
p
p
r
o
a
ch
c
a
n
b
a
s
i
c
al
l
y
b
e
d
i
v
i
d
e
d
i
n
to
t
w
o
m
a
i
n
c
a
te
g
o
r
i
e
s
,
w
h
i
c
h
a
r
e
g
l
o
b
a
l
i
n
t
e
r
p
r
e
t
a
b
i
li
t
y
a
n
d
l
o
c
a
l
i
n
t
e
r
p
r
e
t
a
b
i
l
it
y
[
1
3
]
.
G
l
o
b
al
i
n
te
r
p
r
e
ta
b
i
l
i
t
y
r
e
f
e
r
s
t
o
u
n
d
e
r
s
t
a
n
d
i
n
g
th
e
o
v
e
r
a
l
l
b
e
h
a
v
i
o
r
a
n
d
d
e
c
i
s
i
o
n
-
m
a
k
i
n
g
p
r
o
c
e
s
s
o
f
t
h
e
e
n
t
i
r
e
m
o
d
e
l
,
p
r
o
v
i
d
i
n
g
i
n
s
i
g
h
t
s
i
n
t
o
h
o
w
t
h
e
m
o
d
e
l
m
a
k
e
s
p
r
e
d
i
c
t
i
o
n
s
a
c
r
o
s
s
a
ll
i
n
p
u
t
s
.
L
o
c
a
l
i
n
te
r
p
r
e
t
a
b
i
l
it
y
,
o
n
t
h
e
o
t
h
e
r
h
a
n
d
,
f
o
c
u
s
e
s
o
n
e
x
p
l
a
i
n
i
n
g
i
n
d
i
v
i
d
u
a
l
p
r
e
d
i
c
t
i
o
n
s
,
o
f
f
e
r
i
n
g
a
d
e
t
a
i
l
e
d
u
n
d
e
r
s
t
a
n
d
i
n
g
o
f
w
h
y
t
h
e
m
o
d
e
l
m
a
d
e
a
s
p
e
ci
f
i
c
d
ec
i
s
i
o
n
f
o
r
a
p
a
r
t
ic
u
l
a
r
i
n
p
u
t
i
n
s
ta
n
c
e
.
Fo
r
g
lo
b
al
in
ter
p
r
eta
b
ilit
y
,
SHAP
is
n
o
r
m
ally
u
s
ed
t
o
ac
c
ess
th
e
o
v
er
all
b
e
h
av
io
r
o
f
NI
DS
m
o
d
el
wh
ich
ar
e
r
ep
o
r
ted
i
n
v
ar
io
u
s
r
esear
ch
wo
r
k
s
[
1
4
]
–
[
1
8
]
u
s
i
n
g
d
if
f
e
r
en
t
ap
p
r
o
ac
h
es.
Fo
r
i
n
s
tan
ce
,
[
1
4
]
,
[
1
6
]
,
[
1
7
]
u
s
ed
th
e
s
u
m
m
ar
y
p
l
o
t
o
f
SHAP
to
v
iew
th
e
o
v
er
all
f
ea
tu
r
e
im
p
o
r
tan
ce
o
f
d
ata
an
d
s
h
o
w
th
e
f
ea
tu
r
es
co
n
tr
ib
u
tio
n
to
th
e
co
r
r
esp
o
n
d
in
g
lab
els
in
b
o
th
b
in
ar
y
class
if
icatio
n
an
d
m
u
lticlas
s
class
if
icatio
n
task
s
.
Me
an
wh
ile
,
s
tu
d
y
[
1
8
]
u
tili
ze
d
b
ee
s
war
m
p
l
o
t
to
in
ter
p
r
et
t
h
e
d
ec
is
io
n
-
m
a
k
in
g
s
tep
s
f
o
r
b
in
ar
y
class
th
r
o
u
g
h
d
if
f
er
en
t
class
if
ier
s
.
Oth
er
m
eth
o
d
s
th
at
co
u
ld
b
e
u
s
ed
to
ac
ce
s
s
th
e
g
lo
b
al
in
ter
p
r
etab
ilit
y
o
f
d
ee
p
lear
n
in
g
m
o
d
els s
u
ch
as,
p
er
m
u
tatio
n
i
m
p
o
r
tan
ce
(
PI)
,
c
o
n
tex
tu
al
im
p
o
r
tan
ce
a
n
d
u
tili
ty
(
C
I
U)
[
1
4
]
an
d
r
u
le
f
it
[
1
5
]
.
M
o
v
i
n
g
o
n
t
o
t
h
e
c
o
n
t
e
x
t
o
f
l
o
c
a
l
i
n
t
e
r
p
r
et
a
b
i
l
it
y
,
L
I
M
E
is
g
e
n
e
r
a
l
l
y
u
s
e
d
as
a
t
o
o
l
f
o
r
a
n
al
y
z
i
n
g
t
h
e
i
n
t
e
r
p
r
e
t
a
ti
o
n
o
f
i
n
d
i
v
i
d
u
a
l
p
r
e
d
i
c
t
i
o
n
.
C
o
m
m
o
n
u
t
i
li
z
a
ti
o
n
o
f
L
I
M
E
is
s
i
m
il
a
r
t
o
t
h
e
a
p
p
r
o
a
ch
d
e
s
c
r
i
b
e
d
i
n
[
1
7
]
,
w
h
e
r
e
l
o
c
a
l
p
r
o
b
a
b
i
l
it
y
p
r
e
d
i
c
t
i
o
n
s
a
r
e
d
i
s
p
la
y
e
d
a
l
o
n
g
s
id
e
w
i
t
h
t
h
e
f
e
a
t
u
r
es
t
h
at
c
o
n
t
r
i
b
u
t
e
d
t
o
t
h
o
s
e
p
r
e
d
i
c
t
i
o
n
s
.
M
ea
n
w
h
i
l
e
,
s
t
u
d
y
[
1
8
]
u
s
e
s
L
I
M
E
t
o
p
l
o
t
t
h
e
f
r
e
q
u
e
n
t
f
e
a
t
u
r
e
s
t
o
a
n
a
l
y
z
e
th
e
m
o
s
t
i
m
p
o
r
t
a
n
t
f
e
a
t
u
r
e
s
i
n
t
h
e
p
a
r
t
i
c
u
l
a
r
p
r
e
d
ic
t
i
o
n
.
O
n
t
h
e
o
t
h
e
r
h
a
n
d
,
s
t
u
d
y
[
1
5
]
h
i
g
h
l
i
g
h
t
e
d
t
h
e
f
e
a
t
u
r
e
s
t
h
a
t
o
f
t
e
n
l
e
a
d
t
o
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
A
co
mp
a
r
a
tive
s
tu
d
y
o
f d
ee
p
lea
r
n
in
g
-
b
a
s
ed
n
etw
o
r
k
in
tr
u
s
io
n
d
etec
tio
n
…
(
Ta
n
J
u
a
n
K
a
i
)
4111
c
o
r
r
e
c
t
o
r
w
r
o
n
g
p
r
e
d
i
ct
i
o
n
s
b
y
a
n
a
l
y
z
i
n
g
t
h
e
l
o
c
a
l
i
n
t
e
r
p
r
eta
b
i
l
i
t
y
o
f
p
o
s
i
ti
v
e
a
n
d
n
e
g
at
i
v
e
s
c
e
n
a
r
i
o
s
.
S
HA
P
c
a
n
a
ls
o
b
e
u
t
il
i
z
e
d
f
o
r
l
o
c
a
l
i
n
t
e
r
p
r
e
t
a
t
i
o
n
p
u
r
p
o
s
es
.
F
o
r
i
n
s
t
an
c
e
,
s
t
u
d
i
es
[
1
7
]
,
[
1
9
]
u
s
e
d
SHA
P
wa
t
e
r
f
a
ll
p
l
o
t
t
o
i
l
l
u
s
t
r
a
t
e
t
h
e
e
f
f
e
ct
o
f
f
e
a
t
u
r
es
o
n
a
p
a
r
t
i
c
u
l
a
r
cl
as
s
i
f
i
c
a
ti
o
n
m
a
d
e
b
a
s
e
d
o
n
t
h
e
s
e
le
c
t
e
d
f
ea
t
u
r
e
s
r
el
a
t
i
v
e
t
o
t
h
e
i
n
d
e
x
s
c
o
r
e
s
.
3.
M
E
T
H
O
D
I
n
th
e
p
r
o
p
o
s
ed
p
i
p
elin
e
f
o
r
d
ev
elo
p
in
g
d
ee
p
lear
n
in
g
-
b
as
ed
NI
DS
with
XAI
,
AW
I
D3
d
ataset
is
u
tili
ze
d
,
wh
er
eb
y
it
co
n
s
is
ts
o
f
1
3
ty
p
es
o
f
in
t
r
u
s
io
n
s
in
W
PA2
n
etwo
r
k
s
with
a
to
tal
s
am
p
le
o
f
3
0
,
3
8
7
,
0
9
9
n
o
r
m
al
tr
af
f
ic
an
d
6
,
5
2
6
,
4
0
4
m
alicio
u
s
tr
af
f
ic
[
2
0
]
.
I
n
itially
,
d
ata
p
r
ep
r
o
ce
s
s
in
g
is
p
er
f
o
r
m
ed
to
clea
n
an
d
p
r
ep
ar
e
th
e
d
ata
f
o
r
t
h
e
latter
s
tag
e.
T
h
is
is
f
o
llo
wed
b
y
th
e
ap
p
licatio
n
o
f
a
f
ea
tu
r
e
s
elec
tio
n
alg
o
r
ith
m
to
id
en
tify
th
e
m
o
s
t
r
elev
an
t
f
ea
tu
r
es
f
o
r
th
e
m
o
d
el.
Su
b
s
eq
u
e
n
tly
,
m
o
d
el
d
ev
elo
p
m
en
t
is
c
o
n
d
u
cte
d
u
s
in
g
th
e
s
elec
ted
f
ea
tu
r
es
to
cr
e
ate
a
p
r
ed
ictiv
e
m
o
d
el.
T
h
e
p
e
r
f
o
r
m
an
ce
o
f
th
is
m
o
d
el
is
th
en
e
v
alu
ated
to
ass
ess
its
ef
f
ec
ti
v
en
ess
.
Ad
d
itio
n
ally
,
r
esu
lts
ar
e
in
ter
p
r
eted
u
s
in
g
XAI
tech
n
iq
u
e
to
s
h
o
w
th
e
tr
an
s
p
ar
en
cy
o
f
th
e
m
o
d
el'
s
d
ec
is
io
n
-
m
ak
in
g
p
r
o
c
ess
.
3
.
1
.
Da
t
a
p
re
pro
ce
s
s
ing
Am
o
n
g
th
e
1
3
ty
p
es
o
f
in
tr
u
s
i
o
n
s
av
ailab
le
in
th
e
AW
I
D3
d
atasets
,
7
s
p
ec
if
ic
in
tr
u
s
io
n
s
r
elev
an
t
t
o
th
e
n
etwo
r
k
ac
ce
s
s
lay
er
o
f
th
e
T
C
P/IP
m
o
d
el
h
av
e
b
ee
n
s
elec
ted
.
T
h
ese
in
tr
u
s
io
n
s
in
clu
d
e
d
ea
u
th
e
n
ticatio
n
attac
k
s
,
d
is
ass
o
ciatio
n
attac
k
s
,
(
r
e
)
ass
o
ciatio
n
attac
k
s
,
R
o
g
u
e
ac
ce
s
s
p
o
i
n
t
(
AP)
attac
k
s
,
E
v
il
T
win
attac
k
s
,
KR
A
C
K
attac
k
s
,
an
d
Kr
0
0
k
attac
k
s
.
T
h
e
attac
k
lab
els
ar
e
ca
teg
o
r
ized
i
n
to
th
r
ee
g
r
o
u
p
s
:
d
en
ial
-
of
-
s
er
v
ice
(
Do
S)
attac
k
s
,
m
an
-
in
-
th
e
-
m
i
d
d
le
(
MiT
M)
attac
k
s
,
an
d
tr
af
f
ic
d
ec
r
y
p
tio
n
attac
k
s
.
T
h
e
o
u
tco
m
e
o
f
lab
e
l
m
ap
p
in
g
is
illu
s
tr
ated
in
T
ab
le
1
.
Featu
r
es
with
m
o
r
e
th
an
8
0
%
m
is
s
in
g
v
alu
es
ar
e
ex
cl
u
d
ed
,
an
d
d
ata
im
p
u
tatio
n
tech
n
iq
u
es
ar
e
u
s
ed
to
ad
d
r
ess
th
e
r
em
ain
in
g
m
is
s
in
g
v
alu
es.
C
ateg
o
r
ical
d
ata
ar
e
p
r
e
-
p
r
o
ce
s
s
ed
u
s
in
g
o
r
d
i
n
al
en
co
d
in
g
,
wh
ile
n
u
m
er
ical
d
ata
a
r
e
p
r
o
ce
s
s
ed
u
s
in
g
m
in
-
m
a
x
s
ca
lin
g
.
T
ab
le
1
.
L
a
b
el
m
ap
p
i
n
g
o
f
A
W
I
D3
d
ataset
O
r
i
g
i
n
a
l
i
n
t
r
u
s
i
o
n
N
o
r
mal
t
r
a
f
f
i
c
M
a
l
i
c
i
o
u
s
t
r
a
f
f
i
c
La
b
e
l
m
a
p
p
i
n
g
D
e
a
u
t
h
e
n
t
i
c
a
t
i
o
n
1
,
5
8
7
,
5
2
7
3
8
,
9
4
2
D
e
n
i
a
l
-
of
-
serv
i
c
e
(
D
o
S
)
D
i
sass
o
c
i
a
t
i
o
n
1
,
9
3
8
,
5
8
5
7
5
,
1
3
1
(
R
e
)
a
ss
o
c
i
a
t
i
o
n
1
,
8
3
8
,
4
3
0
5
,
5
0
2
R
o
g
u
e
A
P
1
,
9
7
1
,
8
7
5
1
3
1
0
M
a
n
-
in
-
t
h
e
-
M
i
d
d
l
e
(
M
i
T
M
)
Ev
i
l
Tw
i
n
3
,
6
7
3
,
8
5
4
1
0
4
,
8
2
7
K
R
A
C
K
1
,
3
8
8
,
4
9
8
4
9
,
9
9
0
Tr
a
f
f
i
c
d
e
c
r
y
p
t
i
o
n
K
r
0
0
k
2
,
7
0
8
,
6
3
7
1
8
6
,
1
7
3
3
.
2
.
F
e
a
t
ure
s
elec
t
io
n
I
n
o
r
d
er
to
o
b
tain
th
e
o
p
tim
al
f
ea
tu
r
e
s
ets,
f
ea
tu
r
e
s
elec
tio
n
alg
o
r
ith
m
n
am
ed
p
h
i
-
K
is
b
ei
n
g
u
tili
ze
d
as
it
i
s
ab
le
to
co
m
p
u
te
th
e
co
r
r
elatio
n
b
etwe
en
ca
teg
o
r
ical
d
ata
an
d
n
u
m
er
ical
d
ata
[
2
1
]
.
T
h
e
p
h
i
-
K
m
atr
ix
s
co
r
es
an
d
th
eir
co
r
r
esp
o
n
d
in
g
s
ig
n
if
ican
ce
v
alu
es
ar
e
co
m
p
u
ted
.
T
h
e
to
p
1
5
f
ea
tu
r
es
with
th
e
h
ig
h
est
p
h
i
-
K
s
co
r
es
an
d
s
ig
n
if
ican
t
v
alu
es
ar
e
s
elec
ted
to
r
ed
u
ce
th
e
d
i
m
en
s
io
n
ality
o
f
d
ata.
T
h
ese
s
elec
ted
f
ea
tu
r
es
with
th
e
ass
o
ciate
d
v
alu
es a
n
d
d
es
c
r
ip
tio
n
ar
e
p
r
esen
ted
in
T
ab
le
2
.
3
.
3
.
M
o
del
d
ev
elo
pm
ent
T
ab
Net
is
em
p
lo
y
ed
in
th
e
d
e
v
elo
p
m
en
t
o
f
th
e
NI
DS
m
o
d
el
d
u
e
to
its
r
o
b
u
s
t
ca
p
ab
ilit
ies
in
h
an
d
lin
g
tab
u
lar
d
ata
[
2
2
]
.
T
ab
Net
is
a
d
ee
p
lear
n
in
g
a
r
ch
itectu
r
e
d
esig
n
ed
s
p
ec
if
ically
f
o
r
tab
u
lar
d
ata,
u
tili
zin
g
g
r
ad
ien
t
d
escen
t
-
b
ased
o
p
tim
izatio
n
to
en
ab
le
f
lex
i
b
le
en
d
-
to
-
en
d
lear
n
in
g
,
wh
ich
co
n
s
is
t
s
o
f
f
ea
tu
r
e
an
d
atten
tiv
e
tr
an
s
f
o
r
m
er
s
an
d
f
u
ll
y
co
n
n
ec
ted
lay
er
s
.
B
ef
o
r
e
f
itt
in
g
t
h
e
d
ata
in
to
th
e
m
o
d
els,
it
is
s
p
lit
in
to
th
r
ee
s
ets:
7
5
%
f
o
r
tr
ain
in
g
,
1
5
%
f
o
r
v
alid
atio
n
,
an
d
1
5
%
f
o
r
te
s
tin
g
.
T
h
e
p
ar
am
eter
s
an
d
m
o
d
el
ar
ch
itectu
r
e
o
f
T
ab
Net
ar
e
lis
ted
in
T
ab
le
3
.
No
te
th
at
th
e
p
ar
am
eter
w
eig
h
t
in
T
ab
Net
is
s
et
to
1
in
o
r
d
er
to
au
to
m
a
tically
d
is
tr
ib
u
te
th
e
weig
h
ts
am
o
n
g
t
h
e
class
es to
s
o
lv
e
th
e
class
i
m
b
alan
ce
d
is
s
u
e.
3
.
4
.
P
er
f
o
r
m
a
nce
a
nd
ro
bu
s
t
nes
s
ev
a
lua
t
io
n
T
h
e
p
er
f
o
r
m
an
ce
o
f
th
e
m
o
d
el
is
ev
alu
ated
u
s
in
g
a
co
n
f
u
s
io
n
m
atr
ix
,
ac
cu
r
ac
y
,
r
ec
all,
p
r
e
cisi
o
n
,
an
d
F1
-
s
co
r
e.
Su
b
s
eq
u
e
n
tly
,
th
e
m
o
d
el'
s
p
er
f
o
r
m
a
n
ce
is
co
m
p
ar
ed
with
f
o
u
r
s
tate
-
of
-
th
e
-
a
r
t
(
SOTA
)
m
o
d
els
t
o
b
en
ch
m
ar
k
its
ef
f
ec
tiv
en
ess
.
T
o
ass
ess
th
e
r
o
b
u
s
tn
ess
o
f
th
e
m
o
d
el,
th
e
s
am
e
p
er
f
o
r
m
an
c
e
ev
alu
atio
n
m
etr
ics
ar
e
ap
p
lied
to
th
e
AW
I
D3
d
at
aset
with
th
e
ad
d
itio
n
o
f
s
ig
n
a
l
-
to
-
n
o
is
e
r
atio
(
SNR
)
f
r
o
m
th
e
r
an
g
e
o
f
1
5
to
3
0
.
T
h
e
in
clu
s
io
n
o
f
SNR
is
in
ten
d
ed
to
s
im
u
late
th
e
lev
el
o
f
d
esire
d
s
ig
n
al
r
elativ
e
to
b
ac
k
g
r
o
u
n
d
n
o
is
e,
p
r
o
v
id
i
n
g
a
r
ea
lis
tic
s
ce
n
ar
io
to
test
th
e
m
o
d
el
'
s
ab
ili
ty
to
h
an
d
le
n
o
is
y
d
ata
in
a
r
ea
l
tim
e
en
v
ir
o
n
m
e
n
t.
T
h
e
SNR
v
alu
es a
r
e
co
m
p
u
ted
b
ased
o
n
(
1
)
as r
e
f
er
en
ce
d
in
s
o
u
r
ce
[
2
3
]
.
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
.
4
,
Au
g
u
s
t
20
25
:
4
1
0
9
-
4119
4112
=
10
l
og
10
(
)
(
1
)
T
ab
le
2
.
L
a
b
el
m
ap
p
i
n
g
o
f
A
W
I
D3
d
ataset
T
ab
le
3
.
Mo
d
el
ar
ch
itectu
r
e
a
n
d
p
ar
a
m
eter
s
o
f
T
a
b
Net
P
a
r
a
me
t
e
r
V
a
l
u
e
n
_
s
t
e
p
s
3
o
p
t
i
m
i
z
e
r
_
f
n
A
d
a
m
o
p
t
i
m
i
z
e
r
_
p
a
r
a
ms
d
i
c
t
(
l
r
=
0
.
0
0
5
)
n
_
d
14
n
_
a
14
sch
e
d
u
l
e
r
_
p
a
r
a
ms
{"
st
e
p
_
s
i
z
e
"
:
3
,
"
g
a
mm
a
"
:
0
.
7
}
sch
e
d
u
l
e
r
_
f
n
t
o
r
c
h
.
o
p
t
i
m.
l
t
_
sc
h
e
d
u
l
e
r
.
S
t
e
p
L
R
w
e
i
g
h
t
1
3
.
5
.
I
nte
rpre
t
ing
mo
del pre
dict
io
n wit
h e
x
pla
ina
ble
AI
T
o
ac
h
iev
e
a
co
m
p
r
eh
e
n
s
iv
e
u
n
d
e
r
s
tan
d
in
g
o
f
th
e
m
o
d
el's
p
r
ed
ictio
n
s
,
b
o
t
h
g
lo
b
al
an
d
lo
ca
l
in
ter
p
r
etatio
n
m
eth
o
d
s
a
r
e
em
p
lo
y
ed
.
Fo
r
g
lo
b
al
in
te
r
p
r
etati
o
n
,
th
e
SHAP
Ker
n
el
e
x
p
lain
e
r
is
u
tili
ze
d
.
SHAP
v
alu
es
p
r
o
v
id
e
a
c
o
n
s
is
ten
t
m
ea
s
u
r
e
o
f
f
ea
tu
r
e
im
p
o
r
tan
ce
b
y
q
u
a
n
tify
in
g
th
e
c
o
n
tr
ib
u
tio
n
o
f
ea
ch
f
ea
t
u
r
e
to
th
e
m
o
d
el'
s
p
r
ed
ictio
n
s
,
th
u
s
o
f
f
er
in
g
tr
an
s
p
a
r
en
cy
in
to
th
e
m
o
d
el'
s
o
v
er
all
b
eh
av
io
r
.
Fo
r
lo
ca
l
in
ter
p
r
etatio
n
,
L
I
ME
is
u
s
ed
to
co
n
s
tr
u
ct
i
n
ter
p
r
etab
le
lo
ca
l
m
o
d
els
ar
o
u
n
d
ea
ch
p
r
e
d
ictio
n
.
T
h
is
ap
p
r
o
ac
h
en
a
b
les
th
e
ex
p
lan
atio
n
o
f
in
d
i
v
id
u
al
p
r
ed
ictio
n
b
y
ap
p
r
o
x
im
atin
g
th
e
m
o
d
el
lo
ca
lly
with
a
s
im
p
ler
an
d
m
o
r
e
in
ter
p
r
etab
le
m
o
d
el.
4.
E
XP
E
R
I
M
E
N
T
R
E
SU
L
T
S AN
D
DIS
CUSS
I
O
N
T
o
co
m
p
r
eh
e
n
s
iv
ely
e
v
alu
ate
th
e
T
ab
Net
m
o
d
el
o
n
th
e
AW
I
D3
d
atasets
,
th
r
ee
e
x
p
er
i
m
en
ts
wer
e
co
n
d
u
cte
d
.
Firstl
y
,
th
e
m
o
d
el's
p
er
f
o
r
m
an
ce
m
etr
ics,
in
clu
d
in
g
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
F1
-
s
co
r
e,
u
n
d
etec
ted
in
tr
u
s
io
n
s
,
an
d
f
al
s
e
alar
m
r
ates,
wer
e
ass
e
s
s
ed
an
d
co
m
p
a
r
ed
to
SOTA
m
o
d
e
ls
to
b
en
ch
m
ar
k
its
ef
f
ec
tiv
en
ess
in
d
etec
tin
g
n
et
wo
r
k
in
t
r
u
s
io
n
s
.
Seco
n
d
ly
,
th
e
m
o
d
el'
s
r
o
b
u
s
tn
ess
was
tes
ted
u
n
d
er
v
ar
y
i
n
g
SNR
co
n
d
itio
n
s
to
en
s
u
r
e
h
ig
h
d
etec
tio
n
ac
cu
r
ac
y
an
d
lo
w
f
alse
p
o
s
itiv
es
in
n
o
is
y
en
v
ir
o
n
m
e
n
ts
.
L
astl
y
,
in
ter
p
r
etab
ilit
y
was
ex
a
m
in
ed
u
s
in
g
SHAP
an
d
L
I
ME
tech
n
iq
u
es.
SHAP
p
r
o
v
id
ed
in
s
ig
h
ts
in
to
th
e
g
lo
b
al
f
ea
tu
r
e
im
p
o
r
tan
ce
f
o
r
d
if
f
e
r
en
t
attac
k
ty
p
es,
wh
ile
L
I
ME
o
f
f
er
e
d
lo
ca
l
i
n
ter
p
r
etat
io
n
s
o
f
i
n
d
iv
id
u
al
p
r
ed
ictio
n
s
,
h
i
g
h
lig
h
tin
g
f
ea
tu
r
e
co
n
tr
ib
u
tio
n
s
to
co
r
r
ec
t a
n
d
i
n
co
r
r
ec
t c
lass
if
icatio
n
s
.
F
e
a
t
u
r
e
s
P
h
i
k
m
a
t
r
i
x
P
h
i
k
s
i
g
n
i
f
i
c
a
n
c
e
D
e
scri
p
t
i
o
n
r
a
d
i
o
t
a
p
.
l
e
n
g
t
h
0
.
1
3
0
9
6
2
7
1
6
.
1
2
9
3
7
8
Th
e
l
e
n
g
t
h
o
f
t
h
e
r
a
d
i
o
t
a
p
h
e
a
d
e
r
,
w
h
i
c
h
p
r
o
v
i
d
e
s m
e
t
a
d
a
t
a
a
b
o
u
t
t
h
e
w
i
r
e
l
e
s
s
p
a
c
k
e
t
.
f
r
a
me.
l
e
n
0
.
1
4
0
2
6
2
5
9
7
.
8
2
3
2
3
3
Th
e
t
o
t
a
l
l
e
n
g
t
h
o
f
t
h
e
f
r
a
me
,
i
n
c
l
u
d
i
n
g
h
e
a
d
e
r
s a
n
d
p
a
y
l
o
a
d
.
r
a
d
i
o
t
a
p
.
c
h
a
n
n
e
l
.
f
l
a
g
s
.
c
c
k
0
.
1
6
1
9
5
1
2
1
1
.
3
9
1
5
8
6
I
n
d
i
c
a
t
e
s
w
h
e
t
h
e
r
t
h
e
c
o
m
p
l
e
me
n
t
a
r
y
c
o
d
e
k
e
y
i
n
g
(
C
C
K
)
m
o
d
u
l
a
t
i
o
n
i
s
u
se
d
i
n
t
h
e
c
h
a
n
n
e
l
.
w
l
a
n
_
r
a
d
i
o
.
p
h
y
0
.
1
8
0
6
5
1
7
8
5
.
5
0
3
0
5
9
S
p
e
c
i
f
i
e
s t
h
e
p
h
y
si
c
a
l
l
a
y
e
r
t
y
p
e
u
s
e
d
f
o
r
t
h
e
w
i
r
e
l
e
ss
t
r
a
n
smiss
i
o
n
,
su
c
h
a
s
8
0
2
.
1
1
a
,
b
,
g
,
n
,
o
r
a
c
.
r
a
d
i
o
t
a
p
.
p
r
e
se
n
t
.
t
sf
t
0
.
2
4
1
7
5
8
6
2
9
.
8
6
9
9
9
0
A
f
l
a
g
i
n
d
i
c
a
t
i
n
g
t
h
e
p
r
e
s
e
n
c
e
o
f
t
h
e
t
i
me
s
y
n
c
h
r
o
n
i
z
a
t
i
o
n
f
u
n
c
t
i
o
n
t
i
m
e
r
(
TSF
T)
f
i
e
l
d
i
n
t
h
e
r
a
d
i
o
t
a
p
h
e
a
d
e
r
.
w
l
a
n
.
f
c
.
d
s
0
.
2
4
1
8
8
6
6
9
3
.
6
5
8
1
6
4
Th
e
d
i
st
r
i
b
u
t
i
o
n
sy
s
t
e
m
(
D
S
)
st
a
t
u
s fi
e
l
d
i
n
t
h
e
8
0
2
.
1
1
f
r
a
m
e
c
o
n
t
r
o
l
f
i
e
l
d
,
i
n
d
i
c
a
t
i
n
g
t
h
e
d
i
r
e
c
t
i
o
n
o
f
t
h
e
f
r
a
me
r
e
l
a
t
i
v
e
t
o
t
h
e
d
i
s
t
r
i
b
u
t
i
o
n
sy
st
e
m.
w
l
a
n
.
f
c
.
p
r
o
t
e
c
t
e
d
0
.
2
5
3
6
8
6
6
7
0
.
3
7
4
6
5
3
I
n
d
i
c
a
t
e
s
w
h
e
t
h
e
r
t
h
e
f
r
a
m
e
i
s
p
r
o
t
e
c
t
e
d
b
y
e
n
c
r
y
p
t
i
o
n
.
r
a
d
i
o
t
a
p
.
t
i
mes
t
a
m
p
.
t
s
0
.
2
6
9
6
0
6
9
5
0
.
6
8
2
0
3
9
Th
e
t
i
mes
t
a
m
p
o
f
w
h
e
n
t
h
e
f
r
a
me
w
a
s
c
a
p
t
u
r
e
d
,
p
r
o
v
i
d
e
d
b
y
t
h
e
r
a
d
i
o
t
a
p
h
e
a
d
e
r
.
f
r
a
me.
t
i
me
_
r
e
l
a
t
i
v
e
0
.
2
9
8
1
5
4
1
,
0
4
1
.
1
9
6
1
0
4
Th
e
r
e
l
a
t
i
v
e
t
i
m
e
f
r
o
m
t
h
e
b
e
g
i
n
n
i
n
g
o
f
t
h
e
c
a
p
t
u
r
e
t
o
w
h
e
n
t
h
e
f
r
a
m
e
w
a
s c
a
p
t
u
r
e
d
,
t
y
p
i
c
a
l
l
y
me
a
s
u
r
e
d
i
n
s
e
c
o
n
d
s.
w
l
a
n
_
r
a
d
i
o
.
c
h
a
n
n
e
l
0
.
3
2
3
9
2
7
6
8
0
.
2
7
0
1
7
2
Th
e
r
a
d
i
o
c
h
a
n
n
e
l
o
n
w
h
i
c
h
t
h
e
f
r
a
m
e
w
a
s t
r
a
n
s
mi
t
t
e
d
.
w
l
a
n
.
f
c
.
t
y
p
e
0
.
4
3
7
6
7
3
1
,
3
7
5
.
6
5
6
4
3
3
Th
e
t
y
p
e
o
f
f
r
a
m
e
,
s
u
c
h
a
s m
a
n
a
g
e
me
n
t
,
c
o
n
t
r
o
l
,
o
r
d
a
t
a
f
r
a
me
.
w
l
a
n
_
r
a
d
i
o
.
d
a
t
a
_
r
a
t
e
0
.
4
7
2
0
9
1
1
,
3
7
4
.
4
8
8
7
4
6
Th
e
d
a
t
a
r
a
t
e
a
t
w
h
i
c
h
t
h
e
f
r
a
me
w
a
s
t
r
a
n
sm
i
t
t
e
d
.
w
l
a
n
_
r
a
d
i
o
.
s
i
g
n
a
l
_
d
b
m
0
.
5
8
0
6
3
8
1
,
0
7
0
.
5
7
3
3
4
1
Th
e
si
g
n
a
l
st
r
e
n
g
t
h
o
f
t
h
e
f
r
a
m
e
i
n
d
e
c
i
b
e
l
s
-
m
i
l
l
i
w
a
t
t
s (d
B
m)
.
r
a
d
i
o
t
a
p
.
d
b
m
_
a
n
t
si
g
n
a
l
0
.
7
3
7
4
0
1
1
,
4
6
3
.
1
8
1
8
9
2
T
h
e
s
i
g
n
a
l
s
t
r
e
n
g
t
h
r
e
c
e
i
v
e
d
b
y
t
h
e
a
n
t
e
n
n
a
i
n
d
e
c
i
b
e
l
s
-
m
i
l
l
i
w
a
t
t
s
(
d
B
m
)
.
w
l
a
n
.
f
c
.
s
u
b
t
y
p
e
0
.
7
5
6
2
8
5
1
,
5
2
9
.
5
7
6
7
4
2
Th
e
su
b
t
y
p
e
o
f
t
h
e
f
r
a
m
e
,
p
r
o
v
i
d
i
n
g
mo
r
e
s
p
e
c
i
f
i
c
i
n
f
o
r
ma
t
i
o
n
a
b
o
u
t
t
h
e
f
r
a
me's
p
u
r
p
o
s
e
,
s
u
c
h
a
s
a
sso
c
i
a
t
i
o
n
r
e
q
u
e
st
,
d
a
t
a
a
n
d
a
c
k
n
o
w
l
e
d
g
me
n
t
.
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
A
co
mp
a
r
a
tive
s
tu
d
y
o
f d
ee
p
lea
r
n
in
g
-
b
a
s
ed
n
etw
o
r
k
in
tr
u
s
io
n
d
etec
tio
n
…
(
Ta
n
J
u
a
n
K
a
i
)
4113
4
.
1
.
P
er
f
o
r
m
a
nce
e
v
a
lua
t
io
n o
f
T
a
bNet
m
o
del
T
h
e
co
n
f
u
s
io
n
m
etr
ics
o
f
th
e
T
ab
Net
m
o
d
el
ar
e
b
ein
g
s
h
o
wn
in
Fig
u
r
e
1
.
No
tab
ly
,
th
e
m
o
d
el
ex
h
ib
its
co
m
m
en
d
ab
le
ef
f
ec
tiv
en
ess
in
d
etec
tin
g
n
etwo
r
k
in
tr
u
s
io
n
,
with
o
n
ly
4
in
tr
u
s
io
n
attem
p
ts
g
o
in
g
u
n
d
etec
ted
ac
r
o
s
s
th
e
en
tire
test
in
g
d
atasets
.
Fu
r
th
er
m
o
r
e,
th
e
T
a
b
Net
m
o
d
el
d
em
o
n
s
tr
ates
p
r
o
f
icien
c
y
i
n
ad
d
r
ess
in
g
th
e
p
r
ev
alen
t
is
s
u
e
o
f
h
ig
h
f
alse
alar
m
r
ates,
as
s
h
o
wn
b
y
th
e
g
en
e
r
atio
n
o
f
o
n
l
y
2
1
3
f
alse
alar
m
s
o
u
t o
f
2
,
3
3
6
,
2
3
7
test
in
g
s
am
p
l
es.
T
h
e
p
er
f
o
r
m
an
ce
m
et
r
ics
lis
te
d
in
T
ab
le
4
ar
e
u
s
ed
to
ev
alu
ate
th
e
co
m
p
ar
is
o
n
b
etwe
en
T
ab
Net
with
o
th
er
SOTA
m
o
d
els.
C
o
m
p
a
r
ed
with
o
th
er
SOTA
r
esu
lts
,
it
is
ev
id
en
t
th
at
T
ab
Net
o
u
tp
er
f
o
r
m
s
o
th
er
m
eth
o
d
s
wh
ile
u
tili
zin
g
m
o
r
e
co
m
p
r
e
h
en
s
iv
e
d
atasets
.
I
n
t
er
m
s
o
f
in
tr
u
s
io
n
d
etec
tio
n
,
T
ab
Net
ac
h
iev
ed
a
r
em
ar
k
ab
le
in
d
icato
r
,
d
e
m
o
n
s
tr
ated
b
y
its
9
9
.
9
9
%
p
r
ec
is
io
n
.
No
tab
ly
,
th
e
lo
w
f
alse
alar
m
r
ate,
in
d
icate
d
b
y
its
r
ec
all,
m
atch
es
th
e
h
ig
h
p
r
ec
is
io
n
,
s
h
o
wca
s
in
g
th
e
m
o
d
el'
s
r
eliab
ilit
y
an
d
ac
cu
r
ac
y
.
T
h
is
p
er
f
o
r
m
an
ce
s
u
g
g
ests
th
at
T
ab
Net
is
ex
c
ep
tio
n
ally
ef
f
ec
tiv
e
in
i
d
en
ti
f
y
in
g
an
d
m
itig
atin
g
v
a
r
io
u
s
ty
p
es
o
f
attac
k
s
,
in
clu
d
in
g
d
ea
u
th
en
ticatio
n
,
d
is
a
s
s
o
ciat
io
n
,
r
ea
s
s
o
ciatio
n
,
R
o
g
u
e
AP,
Kr
ac
k
,
Kr
0
0
k
,
a
n
d
E
v
il
T
win
.
T
h
e
b
alan
ce
d
h
ig
h
s
co
r
es
ac
r
o
s
s
all
m
etr
ics
h
ig
h
lig
h
t
T
ab
Net'
s
s
u
p
er
io
r
ity
i
n
m
ain
tain
in
g
s
ec
u
r
ity
an
d
ac
cu
r
ately
d
etec
tin
g
in
tr
u
s
io
n
s
,
m
a
k
in
g
it
a
r
o
b
u
s
t c
h
o
ice
f
o
r
in
tr
u
s
io
n
d
etec
tio
n
s
y
s
t
em
s
.
Fig
u
r
e
1
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
T
ab
Net
o
n
AW
I
D3
d
atasets
T
ab
le
4
.
Per
f
o
r
m
an
ce
e
v
alu
ati
o
n
with
o
th
e
r
SOTA
o
n
AW
I
D3
d
atasets
M
o
d
e
l
u
se
d
U
t
i
l
i
z
a
t
i
o
n
o
f
d
a
t
a
se
t
s
P
e
r
f
o
r
ma
n
c
e
m
e
t
r
i
c
s (%)
B
a
g
g
i
n
g
c
l
a
ssi
f
i
e
r
[
2
4
]
B
o
t
n
e
t
,
M
a
l
w
a
r
e
,
S
S
H
,
S
Q
L
i
n
j
e
c
t
i
o
n
,
S
S
D
P
a
mp
l
i
f
i
c
a
t
i
o
n
,
a
n
d
W
e
b
s
i
t
e
s
p
o
o
f
i
n
g
A
c
c
u
r
a
c
y
:
9
6
.
7
0
R
e
c
a
l
l
:
9
5
.
0
3
P
r
e
c
i
s
i
o
n
:
9
6
.
8
4
F1
-
sc
o
r
e
:
8
8
.
0
7
K
-
n
e
a
r
e
s
t
n
e
i
g
h
b
o
r
s
K
-
NN
[
2
5
]
A
r
o
u
n
d
1
mi
l
l
i
o
n
s
u
b
se
t
s
t
a
k
e
n
f
r
o
m
t
h
e
d
a
t
a
s
e
t
(
1
5
%
o
f
d
a
t
a
se
t
s)
A
c
c
u
r
a
c
y
:
9
9
.
0
0
R
e
c
a
l
l
:
N
/
A
P
r
e
c
i
s
i
o
n
:
N
/
A
F1
-
sc
o
r
e
:
N
/
A
2D
-
C
N
N
-
1
l
a
y
e
r
[
9
]
2
0
%
o
f
D
e
a
u
t
h
e
n
t
i
c
a
t
i
o
n
,
D
i
sass
o
c
i
a
t
i
o
n
,
R
e
a
ss
o
c
i
a
t
i
o
n
,
R
o
g
u
e
A
P
,
K
r
a
c
k
,
K
r
0
0
k
,
a
n
d
E
v
i
l
Tw
i
n
d
a
t
a
A
c
c
u
r
a
c
y
:
9
9
.
7
7
R
e
c
a
l
l
:
9
9
.
7
3
P
r
e
c
i
s
i
o
n
:
9
9
.
5
9
F1
-
sc
o
r
e
:
9
9
.
6
6
Ex
t
r
a
t
r
e
e
[
2
6
]
D
e
a
u
t
h
e
n
t
i
c
a
t
i
o
n
,
D
i
sass
o
c
i
a
t
i
o
n
,
R
e
a
sso
c
i
a
t
i
o
n
,
R
o
g
u
e
A
P
,
K
r
a
c
k
,
K
r
0
0
k
,
a
n
d
E
v
i
l
T
w
i
n
A
c
c
u
r
a
c
y
:
9
9
.
9
6
R
e
c
a
l
l
:
9
9
.
2
8
P
r
e
c
i
s
i
o
n
:
9
9
.
7
5
F1
-
sc
o
r
e
:
9
9
.
5
2
Ta
b
N
e
t
D
e
a
u
t
h
e
n
t
i
c
a
t
i
o
n
,
D
i
sass
o
c
i
a
t
i
o
n
,
R
e
a
sso
c
i
a
t
i
o
n
,
R
o
g
u
e
A
P
,
K
r
a
c
k
,
K
r
0
0
k
,
a
n
d
E
v
i
l
T
w
i
n
A
c
c
u
r
a
c
y
:
9
9
.
9
9
R
e
c
a
l
l
:
9
9
.
9
9
P
r
e
c
i
s
i
o
n
:
9
9
.
9
9
F1
-
sc
o
r
e
:
9
9
.
9
9
4
.
2
.
Ro
bu
s
t
nes
s
e
v
a
lua
t
io
n
SNR
s
er
v
es
as
a
m
etr
ic
to
ass
es
s
th
e
ef
f
icac
y
o
f
t
h
e
NI
DS
in
m
an
ag
in
g
n
o
is
e
with
in
r
ea
l
-
tim
e
wir
eless
n
etwo
r
k
s
.
B
y
q
u
an
tify
in
g
th
e
r
atio
o
f
s
ig
n
al
p
o
wer
to
b
ac
k
g
r
o
u
n
d
n
o
is
e
p
o
wer
,
SNR
f
ac
ili
tates
an
u
n
d
er
s
tan
d
i
n
g
o
f
th
e
s
y
s
tem
'
s
ca
p
ab
ilit
y
to
d
etec
t
in
tr
u
s
io
n
s
am
id
s
t
v
ar
y
in
g
lev
els
o
f
in
ter
f
er
en
ce
.
T
h
e
lo
wer
th
e
SNR
v
alu
es
b
ein
g
u
tili
ze
d
,
th
e
m
o
r
e
th
e
n
o
is
e
o
v
er
wh
el
m
s
th
e
s
ig
n
al.
SNR
v
alu
es
r
an
g
in
g
f
r
o
m
1
5
t
o
3
0
ar
e
em
p
lo
y
ed
th
r
o
u
g
h
o
u
t
th
e
s
ec
o
n
d
ex
p
er
im
en
t
to
en
co
m
p
ass
wea
k
to
s
tr
o
n
g
s
ig
n
al
co
n
d
itio
n
s
.
Fig
u
r
e
2
p
r
esen
ts
a
v
is
u
al
d
ep
ictio
n
o
f
T
ab
Net
p
er
f
o
r
m
a
n
ce
ac
r
o
s
s
d
if
f
er
en
t
SNR
lev
els,
s
h
ed
d
in
g
l
ig
h
t
o
n
its
b
eh
a
v
io
r
u
n
d
er
v
ar
y
i
n
g
n
o
is
e
in
ten
s
ities
.
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
.
4
,
Au
g
u
s
t
20
25
:
4
1
0
9
-
4119
4114
Acr
o
s
s
th
e
r
an
g
e
o
f
SNR
,
th
e
r
ec
all
m
etr
ic,
wh
ich
is
r
ep
r
esen
ted
b
y
th
e
g
r
ay
b
ar
s
,
c
o
n
s
is
ten
tly
m
ain
tain
s
a
s
tab
le
p
atter
n
.
T
h
is
co
n
s
is
ten
cy
p
r
o
v
ed
th
e
ca
p
ab
ilit
y
o
f
T
ab
Net
to
ac
cu
r
ately
id
en
tify
g
e
n
u
in
e
in
tr
u
s
io
n
s
r
em
ain
s
u
n
af
f
ec
ted
b
y
t
h
e
f
lu
ctu
ati
o
n
s
o
f
s
ig
n
al
q
u
ality
.
T
h
is
r
esil
ien
ce
in
r
ec
all
u
n
d
er
s
co
r
es
t
h
e
m
o
d
el'
s
ef
f
ec
tiv
en
ess
in
d
etec
tin
g
in
tr
u
s
io
n
s
,
u
s
in
g
tr
u
e
p
o
s
itiv
es a
s
in
d
icato
r
s
,
ir
r
esp
ec
tiv
e
o
f
n
o
is
e
lev
els.
C
o
n
v
er
s
ely
,
p
r
ec
is
io
n
,
in
d
icate
d
b
y
th
e
o
r
an
g
e
b
ar
s
,
ex
h
ib
i
ts
n
o
ticea
b
le
v
ar
ia
b
ilit
y
ac
r
o
s
s
d
if
f
er
en
t
SNR
v
alu
es.
Par
ticu
lar
ly
ev
id
en
t
at
lo
wer
SNR
lev
els,
s
u
ch
as
SNR
1
0
,
p
r
ec
is
io
n
ten
d
s
t
o
b
e
lo
wer
r
elativ
e
to
h
ig
h
er
SNR
v
alu
es.
T
h
is
in
d
ic
ates
th
at
th
er
e
is
a
g
r
ea
ter
lik
e
lih
o
o
d
o
f
af
f
ec
ted
n
etwo
r
k
p
a
ck
et
co
n
ten
t
b
ein
g
m
is
class
if
ied
as
an
o
m
alies
w
h
en
th
e
en
v
ir
o
n
m
e
n
t
co
n
s
is
ts
o
f
a
h
ig
h
e
r
lev
el
o
f
n
o
is
e.
As
a
r
esu
lt,
th
e
m
o
d
el
ten
d
s
to
p
r
o
d
u
ce
m
o
r
e
f
alse
p
o
s
itiv
es
in
t
h
e
s
itu
atio
n
o
f
p
o
o
r
er
s
ig
n
al
q
u
ality
,
lead
i
n
g
to
a
d
ec
r
ea
s
e
in
p
r
ec
is
io
n
.
Fig
u
r
e
2
.
Per
f
o
r
m
an
c
e
v
is
u
aliza
tio
n
in
AW
I
D3
with
d
if
f
e
r
e
n
t SNR
v
alu
es
4
.
3
.
G
l
o
ba
l
i
nte
rpre
t
a
t
io
n o
f
T
a
bNet
m
o
del us
ing
SH
AP
SHAP
s
u
m
m
ar
y
p
lo
ts
in
Fig
u
r
e
3
to
5
p
r
o
v
id
e
a
c
o
m
p
r
e
h
e
n
s
iv
e
v
is
u
aliza
tio
n
o
f
th
e
im
p
ac
t
o
f
ea
c
h
f
ea
tu
r
e
o
n
t
h
e
m
o
d
el'
s
o
u
tp
u
t.
T
h
e
f
ea
tu
r
es
ar
e
lis
ted
alo
n
g
th
e
y
-
ax
is
,
o
r
d
e
r
ed
b
y
th
eir
o
v
er
all
in
f
lu
en
ce
o
n
th
e
p
r
ed
ictio
n
.
E
ac
h
p
o
in
t
r
e
p
r
esen
ts
a
SHAP
v
alu
e
f
o
r
a
f
ea
tu
r
e
in
a
p
ar
ticu
lar
in
s
ta
n
ce
,
with
th
e
co
lo
r
in
d
icatin
g
th
e
f
ea
tu
r
e
v
al
u
e,
wh
er
e
b
lu
e
s
ig
n
if
ies
lo
w
a
n
d
r
ed
s
ig
n
if
ies
h
i
g
h
.
T
h
is
o
r
d
er
i
n
g
h
el
p
s
to
q
u
ick
ly
id
en
tify
wh
ich
f
ea
tu
r
es a
r
e
th
e
m
o
s
t in
f
lu
en
tial in
d
eter
m
in
in
g
th
e
m
o
d
el'
s
p
r
ed
ictio
n
s
.
SHAP
f
ea
tu
r
e
v
alu
e
d
is
tr
ib
u
t
io
n
in
Fig
u
r
e
3
s
h
ed
s
lig
h
t
o
n
MiT
M
attac
k
s
.
No
tab
ly
,
f
e
atu
r
es
lik
e
w
la
n
.
fc.
typ
e
,
w
la
n
.
fc.
s
u
b
typ
e
,
an
d
fr
a
me.
len
e
x
h
ib
it
h
ig
h
er
v
alu
es,
wh
ich
ar
e
c
o
n
s
is
ten
tly
s
h
o
wn
in
r
ed
p
lo
ts
.
Wla
n
.
fc.
typ
e
s
ig
n
if
ies
th
e
g
en
er
al
ca
teg
o
r
y
o
f
tr
an
s
m
itted
f
r
am
es,
wh
ile
w
la
n
.
fc.
s
u
b
typ
e
p
r
o
v
i
d
es
m
o
r
e
s
p
ec
if
ic
d
etails
with
in
th
at
ca
t
eg
o
r
y
.
R
o
g
u
e
APs
an
d
E
v
il
T
win
s
,
aim
in
g
to
im
p
er
s
o
n
ate
l
eg
itima
te
AP,
o
f
ten
u
s
e
b
ea
co
n
f
r
am
es
to
lu
r
e
u
s
er
s
.
T
h
ese
b
ea
co
n
f
r
a
m
es
g
en
er
ate
s
u
b
ty
p
e
8
p
ac
k
ets,
ca
teg
o
r
ized
a
s
d
ata
f
r
am
es,
with
w
la
n
.
fc.
typ
e
n
u
m
b
e
r
s
co
r
r
esp
o
n
d
i
n
g
to
2
.
Ad
d
itio
n
all
y
,
f
o
r
d
if
f
er
e
n
tiatio
n
,
th
e
a
u
th
o
r
f
ilter
s
R
o
g
u
e
AP
attac
k
s
b
ased
o
n
fr
a
me.
len
b
ein
g
less
th
an
2
6
4
an
d
E
v
il
T
win
attac
k
s
with
fr
a
me.
len
less
th
an
2
4
2
.
Mo
r
e
o
v
er
,
an
ex
tr
a
f
ilter
is
ap
p
l
ied
to
E
v
il
T
win
attac
k
s
,
in
v
o
lv
in
g
d
ea
u
th
en
ticatio
n
f
r
am
es
to
d
is
co
n
n
ec
t
d
ev
ices
f
r
o
m
th
e
o
r
ig
in
al
AP,
f
ac
ilit
atin
g
th
eir
co
n
n
ec
tio
n
to
th
e
m
alicio
u
s
o
n
e.
Mo
v
in
g
o
n
t
o
th
e
SHAP
s
u
b
p
lo
t
in
Do
S
attac
k
s
illu
s
tr
ated
in
Fig
u
r
e
4
,
s
h
o
ws
t
h
at
t
h
e
f
ea
tu
r
e
w
la
n
.
fc.
s
u
b
typ
e
h
as
s
ev
er
al
r
e
d
p
o
in
ts
.
T
h
is
p
o
s
itio
n
in
g
s
u
g
g
ests
th
at
h
ig
h
er
v
alu
es
o
f
th
i
s
f
ea
tu
r
e
ar
e
clo
s
ely
lin
k
ed
to
a
n
in
cr
ea
s
ed
lik
elih
o
o
d
o
f
a
Do
S
attac
k
o
cc
u
r
r
en
ce
.
C
o
n
s
eq
u
en
tly
,
it
im
p
lies
a
s
tr
o
n
g
ass
o
ciatio
n
b
etwe
en
s
p
ec
if
ic
f
r
am
e
ty
p
es
an
d
h
eig
h
ten
ed
r
is
k
s
o
f
Do
S
a
ttack
s
.
Fo
r
in
s
tan
ce
,
th
e
NI
DS
m
o
d
el
s
cr
u
tin
izes
n
etwo
r
k
p
ac
k
ets
to
d
etec
t
p
o
ten
tial
f
lo
o
d
in
g
o
f
ce
r
tai
n
f
r
am
e
ty
p
es.
Go
in
g
d
ee
p
in
to
th
e
f
ea
tu
r
es,
d
ea
u
th
en
ticatio
n
attac
k
s
co
r
r
e
s
p
o
n
d
to
s
u
b
ty
p
e
1
0
,
d
is
ass
o
ciatio
n
attac
k
s
to
s
u
b
ty
p
e
1
2
,
an
d
r
ea
s
s
o
ciatio
n
attac
k
s
to
s
u
b
ty
p
es
0
,
2
,
an
d
8
as
p
e
r
f
ilter
a
p
p
lied
b
y
t
h
e
au
th
o
r
s
[
2
0
]
.
As
a
r
esu
lt,
th
is
im
p
lies
th
at
th
e
T
ab
Net
m
o
d
el
class
if
ies
Do
S
attac
k
s
in
a
m
a
n
n
er
th
at
clo
s
el
y
r
esem
b
les
h
o
w
n
etwo
r
k
a
d
m
in
is
tr
ato
r
s
ev
alu
ate
s
u
ch
attac
k
s
in
r
ea
l
-
wo
r
ld
en
v
i
r
o
n
m
en
ts
.
Mo
r
eo
v
er
,
Fig
u
r
e
5
s
h
o
ws
th
e
SHAP
s
u
m
m
ar
y
s
u
b
p
lo
ts
o
n
tr
af
f
ic
d
ec
r
y
p
tio
n
attac
k
s
.
B
ased
o
n
th
e
SHAP
d
i
s
tr
ib
u
tio
n
,
it
ca
n
b
e
o
b
s
er
v
ed
th
at
th
e
f
ea
tu
r
e
w
la
n
.
f
c.
s
u
b
typ
e
h
as
th
e
h
ig
h
est
im
p
ac
t
v
alu
e,
f
o
llo
wed
by
w
la
n
_
r
a
d
io
.
ch
a
n
n
el
.
T
h
is
s
ce
n
ar
io
m
ay
h
a
p
p
en
d
u
e
to
th
e
m
eth
o
d
o
lo
g
y
o
f
t
h
e
au
t
h
o
r
in
co
llectin
g
th
e
AW
I
D3
d
ataset
s
o
n
th
e
K
R
AC
K
an
d
Kr
0
0
k
attac
k
s
.
Sp
ec
if
ically
,
th
e
s
ig
n
if
ica
n
t
im
p
ac
t
o
f
th
e
w
la
n
.
fc.
s
u
b
typ
e
f
ea
tu
r
e
alig
n
s
with
th
e
d
ataset
au
th
o
r
s
'
m
eth
o
d
o
f
f
ilter
in
g
an
d
lab
elin
g
n
etwo
r
k
p
ac
k
ets.
T
h
ey
lab
eled
p
ac
k
ets
wh
er
e
th
e
f
ea
tu
r
e
w
la
n
.
fc.
typ
e_
s
u
b
ty
p
e
is
eq
u
iv
alen
t
to
1
0
as
Kr
0
0
k
attac
k
s
.
Nex
t,
th
e
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
A
co
mp
a
r
a
tive
s
tu
d
y
o
f d
ee
p
lea
r
n
in
g
-
b
a
s
ed
n
etw
o
r
k
in
tr
u
s
io
n
d
etec
tio
n
…
(
Ta
n
J
u
a
n
K
a
i
)
4115
f
ea
tu
r
e
o
f
w
la
n
_
r
a
d
io
.
ch
a
n
n
el
wh
ich
in
d
icate
s
th
e
n
etw
o
r
k
c
h
an
n
el
o
f
th
e
n
etwo
r
k
p
ac
k
et
is
lo
ca
te
d
.
KR
A
C
K
attac
k
s
p
r
im
ar
ily
c
o
n
d
u
cte
d
o
n
ch
a
n
n
els
2
an
d
1
3
ac
co
r
d
in
g
to
th
e
a
u
th
o
r
s
[
2
0
]
,
wh
ic
h
is
co
n
tr
ad
icted
to
t
h
e
n
o
r
m
al
p
a
ck
et
p
r
o
f
ile
an
d
o
th
e
r
in
tr
u
s
io
n
s
th
at
ar
e
in
itially
co
llected
o
n
ch
an
n
el
3
6
.
As
a
r
esu
lt,
th
e
m
o
d
el
te
n
d
s
to
class
if
y
t
h
e
n
etwo
r
k
p
ac
k
et
allo
ca
ted
to
c
h
an
n
el
o
u
ts
id
e
o
f
ch
an
n
el
3
6
,
s
p
ec
if
ically
ch
an
n
el
2
a
n
d
c
h
an
n
el
1
3
as K
R
AC
K
attac
k
attem
p
ts
.
Mo
r
eo
v
er
,
w
la
n
.
fc.
p
r
o
tecte
d
is
th
e
o
n
ly
f
ea
tu
r
e
t
h
at
h
as
a
h
ig
h
f
ea
tu
r
e
v
alu
e
(
m
ix
ed
with
r
ed
co
lo
r
)
as
co
m
p
ar
e
d
to
Fig
u
r
es
3
an
d
4
wh
ich
h
a
v
e
o
n
ly
lo
w
f
ea
tu
r
e
v
alu
e
(
en
tire
ly
b
l
u
e
co
lo
r
)
.
I
t
is
d
u
e
to
th
e
n
atu
r
e
o
f
t
r
af
f
ic
d
ec
r
y
p
tio
n
attac
k
s
wh
ich
ca
u
s
es
th
e
en
cr
y
p
tio
n
k
ey
o
f
t
h
e
n
etwo
r
k
p
ac
k
et
c
o
n
ten
t
to
b
e
r
eset
to
a
n
all
-
ze
r
o
v
alu
e
wh
ich
m
ea
n
n
o
en
c
r
y
p
tio
n
p
r
o
tectio
n
is
av
ailab
le
.
T
h
is
ass
er
tio
n
is
d
em
o
n
s
tr
ated
b
y
th
e
a
u
th
o
r
s
u
s
in
g
a
W
ir
esh
ar
k
f
ilter
,
s
p
e
cif
ically
b
y
s
ettin
g
w
la
n
.
fc.
p
r
o
tecte
d
to
ze
r
o
,
to
id
en
tify
an
d
lab
el
th
e
Kr
0
0
k
atta
ck
s
.
I
n
co
n
clu
s
io
n
,
t
h
e
g
lo
b
al
i
n
ter
p
r
etatio
n
o
f
SHAP
v
alu
es
p
r
o
v
id
es
v
alu
ab
le
in
s
ig
h
ts
in
to
t
h
e
alig
n
m
en
t
b
etwe
en
th
e
NI
D
S
m
o
d
el'
s
co
m
p
r
eh
en
s
io
n
o
f
o
v
er
all
r
esu
lts
an
d
th
e
a
u
th
o
r
'
s
d
ata
f
ilter
in
g
m
eth
o
d
o
l
o
g
y
alo
n
g
s
id
e
th
e
in
tr
in
s
ic
ch
ar
ac
ter
is
tics
o
f
th
e
a
ttack
s
.
T
h
e
d
is
ce
r
n
ib
le
co
r
r
esp
o
n
d
e
n
ce
b
etwe
en
th
e
SHAP
v
alu
es
an
d
th
e
ap
p
lied
d
ata
f
ilter
in
g
ap
p
r
o
ac
h
h
ig
h
lig
h
ts
n
o
t
o
n
ly
th
e
ef
f
i
ca
cy
o
f
th
e
f
ea
tu
r
e
s
elec
tio
n
p
r
o
ce
s
s
b
u
t
also
co
n
tr
ib
u
tes
to
a
d
ee
p
er
u
n
d
er
s
tan
d
in
g
o
f
th
e
d
ec
is
io
n
-
m
ak
in
g
f
r
a
m
ewo
r
k
em
p
lo
y
ed
b
y
th
e
m
o
d
el
an
d
h
en
ce
in
cr
ea
s
es
th
e
tr
u
s
two
r
th
in
ess
o
f
th
e
in
tr
u
s
io
n
d
etec
tio
n
s
m
ad
e
am
o
n
g
th
e
n
etwo
r
k
ad
m
in
is
tr
ato
r
s
.
T
h
is
co
n
c
o
r
d
a
n
ce
b
etwe
en
th
e
m
o
d
el'
s
in
ter
p
r
etatio
n
an
d
th
e
o
b
s
er
v
ed
atta
ck
p
atter
n
s
s
er
v
es.
Fig
u
r
e
3
.
SHAP
s
u
b
p
lo
ts
f
o
r
MiT
M
a
ttack
s
Fig
u
r
e
4
.
SHAP
s
u
b
p
lo
ts
f
o
r
Do
S
a
ttack
s
Fig
u
r
e
5
.
SHAP
s
u
b
p
lo
ts
f
o
r
tr
af
f
ic
d
ec
r
y
p
tio
n
attac
k
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
.
4
,
Au
g
u
s
t
20
25
:
4
1
0
9
-
4119
4116
4
.
4
.
L
o
ca
l
i
nte
rpre
t
a
t
io
n o
f
T
a
bNet
m
o
del us
ing
L
I
M
E
T
h
e
L
I
ME
i
n
ter
p
r
etatio
n
f
ig
u
r
es
p
r
o
v
i
d
ed
in
th
is
s
tu
d
y
co
n
s
is
t
o
f
th
r
ee
p
a
r
ts
.
T
h
e
lef
t
tab
le
s
h
o
ws
th
e
p
r
e
d
ictio
n
p
r
o
b
ab
ilit
ies
o
f
th
e
m
o
d
el
in
ter
p
r
etatio
n
wh
ic
h
also
i
n
d
icate
s
th
e
co
n
f
id
en
c
e
s
co
r
e
o
f
th
e
m
o
d
el
m
ak
in
g
th
e
p
r
e
d
ictio
n
s
.
T
h
e
r
ig
h
t
two
-
s
id
ed
b
ar
ch
ar
t
s
h
o
ws
th
e
d
etailed
b
r
ea
k
d
o
wn
o
f
th
e
co
n
tr
ib
u
tio
n
o
f
v
ar
io
u
s
f
ea
tu
r
es
to
p
r
ed
ictio
n
r
esu
lts
,
wh
er
eb
y
th
e
r
e
d
b
a
r
a
t
th
e
lef
t
s
id
e
s
h
o
ws
th
e
n
eg
at
iv
e
in
d
icato
r
to
th
e
p
r
ed
ictio
n
s
wh
ile
g
r
ee
n
b
ar
at
th
e
r
ig
h
t
s
id
e
s
h
o
ws
th
e
p
o
s
itiv
e
in
d
icato
r
to
t
h
e
p
r
e
d
ictio
n
s
.
T
h
ese
m
ag
n
itu
d
e
lev
els
s
h
o
wn
in
th
e
r
ig
h
t
two
-
si
d
ed
b
ar
ar
e
th
e
i
n
d
icato
r
o
f
co
n
tr
ib
u
tio
n
s
o
n
h
o
w
th
e
T
ab
Net
m
o
d
el
m
ak
e
th
e
class
if
icatio
n
r
esu
lts
,
wh
er
eb
y
a
p
o
s
itiv
e
m
ag
n
itu
d
e
le
v
e
l
co
n
tr
ib
u
tes
to
th
e
class
if
icatio
n
m
ad
e,
an
d
a
n
eg
ativ
e
m
ag
n
itu
d
e
lev
el
is
o
p
p
o
s
in
g
th
e
class
if
icatio
n
r
esu
lt
s
m
ad
e.
Fig
u
r
e
6
illu
s
tr
ates
a
lo
ca
l
i
n
te
r
p
r
etatio
n
u
s
in
g
L
I
ME
f
o
r
a
m
o
d
el
p
r
e
d
ictio
n
m
ar
k
ed
as
a
D
o
S
attac
k
.
T
h
e
in
ter
p
r
etatio
n
h
ig
h
lig
h
ts
ac
cu
r
ately
class
if
ied
in
tr
u
s
io
n
attem
p
ts
o
f
Do
S
attac
k
s
in
th
e
lef
t
tab
le.
T
h
e
p
r
ed
ictio
n
p
r
o
b
a
b
ilit
ies
in
d
icate
1
0
0
%
co
n
f
id
en
ce
th
at
t
h
e
n
etwo
r
k
p
ac
k
et
is
a
Do
S
attac
k
,
with
ze
r
o
p
r
o
b
a
b
ilit
ies f
o
r
o
th
er
class
es,
in
clu
d
in
g
tr
a
f
f
ic
d
ec
r
y
p
tio
n
attac
k
s
,
MiT
M,
an
d
n
o
r
m
al.
T
h
e
r
ig
h
t
two
-
s
id
ed
b
ar
ch
ar
t
o
f
a
Do
S
attac
k
u
s
es
th
e
len
g
th
o
f
th
e
b
ar
s
to
r
ep
r
esen
t
th
e
m
ag
n
itu
d
e
o
f
ea
ch
f
ea
tu
r
e
'
s
co
n
tr
ib
u
tio
n
to
th
e
p
r
ed
ictio
n
,
with
lo
n
g
er
b
ar
s
in
d
icatin
g
a
s
tr
o
n
g
er
in
f
lu
en
ce
.
Gr
ee
n
b
ar
s
r
ep
r
esen
t
f
ea
tu
r
es
th
at
s
u
p
p
o
r
t
th
e
Do
S
clas
s
if
icatio
n
,
in
clu
d
in
g
fr
a
me.
time_
r
ela
tive
with
a
m
ag
n
itu
d
e
lev
el
m
o
r
e
th
an
0
.
0
0
8
,
w
la
n
.
fc.
s
u
b
typ
e
with
a
m
ag
n
itu
d
e
lev
el
ar
o
u
n
d
0
.
0
0
6
,
w
la
n
.
f
c.
p
r
o
tecte
d
an
d
r
a
d
io
ta
p
.
timest
a
mp
.
ts
with
m
ag
n
itu
d
e
lev
el
o
f
s
lig
h
tl
y
less
th
an
0
.
0
0
6
,
r
a
d
io
t
a
p
.
d
b
m
_
a
n
ts
ig
n
a
l
,
r
a
d
io
ta
p
.
ch
a
n
n
el.
fla
g
s
.
ck
k
with
a
v
alu
e
o
f
0
.
0
0
3
5
an
d
last
ly
w
la
n
_
r
a
d
io
.
d
a
ta
_
r
a
te
with
m
ag
n
itu
d
e
le
v
el
o
f
0
.
0
0
2
.
C
o
n
v
er
s
ely
,
r
ed
b
ar
s
in
d
icate
f
ea
tu
r
es
th
at
o
p
p
o
s
e
to
th
e
p
r
ed
ictio
n
,
s
u
ch
as
r
a
d
io
t
a
p
.
p
r
esen
t.tsf
t
with
m
ag
n
itru
d
e
lev
el
o
f
-
0
.
0
0
6
an
d
w
la
n
_
r
a
d
io
.
p
h
y
with
m
ag
n
itu
d
e
lev
el
ar
o
u
n
d
-
0
.
0
0
3
5
.
B
ased
o
n
t
h
e
co
n
tr
ib
u
tio
n
s
o
f
th
e
m
ag
n
it
u
d
e
lev
el
as
p
er
in
d
icate
d
in
t
h
e
r
ig
h
t
b
ar
ch
ar
t,
it
c
o
u
ld
b
e
o
b
s
er
v
ed
th
at
m
o
s
t
o
f
th
e
m
ag
n
itu
d
e
v
o
tin
g
s
is
to
w
ar
d
s
th
e
p
o
s
itiv
e
s
id
e
in
Do
S
class
if
icatio
n
an
d
h
en
ce
,
T
a
b
Net
m
o
d
el
is
ab
le
co
r
r
ec
tly
class
if
y
th
e
p
a
r
ticu
lar
n
etwo
r
k
p
ac
k
et
as a
Do
S a
tte
m
p
t.
L
o
o
k
in
g
in
to
th
e
s
p
ec
if
ic
c
o
n
tr
ib
u
tio
n
s
o
f
ea
ch
f
ea
t
u
r
e
,
th
e
lo
ca
l
in
ter
p
r
etatio
n
ali
g
n
s
with
estab
lis
h
ed
p
r
in
cip
les
in
n
etw
o
r
k
s
ec
u
r
ity
,
as
well
as
th
e
g
lo
b
al
in
ter
p
r
etatio
n
d
e
r
iv
ed
f
r
o
m
SHAP
v
alu
es
f
o
r
Do
S
attac
k
clas
s
if
icatio
n
.
I
n
n
etwo
r
k
s
ec
u
r
ity
,
ce
r
tain
f
ea
t
u
r
es
s
u
ch
as
w
la
n
.
fc.
p
r
o
tecte
d
b
ein
g
0
,
in
d
icatin
g
u
n
p
r
o
tecte
d
f
r
am
es,
an
d
w
la
n
_
r
a
d
io
.
p
h
y
b
ein
g
1
,
in
d
icatin
g
th
e
u
tili
za
tio
n
o
f
p
h
y
s
ical
r
ad
io
s
ettin
g
s
,
ar
e
cr
u
cial
f
ea
tu
r
es
in
d
icato
r
s
o
f
p
o
ten
tial
Do
S
ac
tiv
ity
.
Ad
d
itio
n
ally
,
as
p
r
ev
io
u
s
ly
d
is
cu
s
s
ed
in
th
e
g
lo
b
al
in
ter
p
r
etatio
n
o
f
Do
S
attac
k
s
,
th
e
ac
cu
r
ate
class
if
icatio
n
o
f
Do
S
attac
k
s
in
v
o
l
v
es
r
ec
o
g
n
iz
in
g
w
la
n
.
fc.
s
u
b
ty
p
e
as
a
p
iv
o
tal
in
d
icato
r
.
Fu
r
th
e
r
m
o
r
e,
th
e
p
o
s
itiv
e
d
ir
ec
tio
n
o
n
th
e
b
a
r
o
f
r
a
d
io
ta
p
.
d
b
m_
a
n
ts
ig
n
a
l
s
h
o
wn
in
Fig
u
r
e
6
r
ein
f
o
r
ce
s
th
is
class
if
ica
tio
n
,
as
th
is
f
ea
tu
r
e
s
h
o
ws
s
ig
n
al
s
tr
en
g
th
c
o
n
d
itio
n
in
r
ea
l
-
tim
e
en
v
ir
o
n
m
en
t,
wh
ich
m
ea
n
s
th
a
t th
e
m
o
d
el
is
ca
p
ab
le
to
d
etec
t th
e
ab
n
o
r
m
al
s
ig
n
al
s
tr
en
g
th
o
cc
u
r
r
e
d
.
I
n
co
n
tr
ast,
Fig
u
r
e
7
f
o
r
a
f
alse
alar
m
s
ce
n
ar
io
wh
er
e
a
n
o
r
m
al
n
etwo
r
k
p
ac
k
et
is
in
co
r
r
ec
tly
class
if
ied
as
a
Do
S
at
tack
.
T
h
e
p
r
ed
ictio
n
p
r
o
b
a
b
ilit
ies
s
h
o
w
a
9
8
%
lik
elih
o
o
d
f
o
r
th
e
D
o
S
class
,
with
v
er
y
lo
w
p
r
o
b
a
b
ilit
ies
f
o
r
o
th
er
cla
s
s
es,
d
esp
ite
th
e
tr
u
e
lab
el
b
ein
g
'
No
r
m
al'
.
T
h
is
m
is
c
lass
if
icatio
n
h
ig
h
lig
h
ts
th
e
m
o
d
el'
s
er
r
o
r
.
T
h
e
r
ig
h
t
-
s
id
e
two
-
s
id
ed
b
ar
ch
ar
t
s
h
o
w
s
th
at
ce
r
tain
f
ea
tu
r
es
n
e
g
a
tiv
ely
im
p
ac
t
th
e
class
if
icatio
n
o
f
th
e
p
ac
k
et
as
a
Do
S
attac
k
,
s
u
g
g
esti
n
g
it
s
h
o
u
ld
b
e
co
r
r
ec
tly
class
if
ied
as
a
n
o
r
m
al
p
ac
k
et.
Sp
ec
if
ically
,
th
e
f
ea
tu
r
es
w
la
n
.
fc.
s
u
b
typ
e
with
a
m
ag
n
itu
d
e
s
lig
h
lty
lo
we
r
th
an
-
0
.
0
0
6
,
w
la
n
_
r
a
d
io
.
p
h
y
with
m
ag
n
itu
d
e
o
f
ap
p
r
o
x
im
ately
-
0
.
0
0
6
,
r
a
d
io
ta
p
.
p
r
esen
t.tsf
t
with
m
ag
n
itu
d
e
ar
o
u
n
d
-
0
.
0
0
4
,
an
d
r
a
d
io
ta
p
.
d
b
m_
a
n
ts
ig
n
a
l
with
m
ag
n
itu
d
e
ar
o
u
n
d
-
0
.
0
0
2
c
o
n
tr
ib
u
te
n
e
g
ativ
ely
to
th
e
Do
S
class
if
icatio
n
.
Ho
wev
er
,
th
e
m
ajo
r
ity
v
o
tin
g
o
f
th
e
r
em
ain
i
n
g
f
ea
tu
r
es
an
d
m
ag
n
itu
d
e
ar
e
m
o
r
e
to
w
ar
d
s
to
th
e
p
o
s
itiv
e
d
ir
ec
tio
n
,
ca
u
s
in
g
a
f
alse a
lar
m
s
ce
n
ar
io
,
wh
er
e
b
y
th
e
n
o
r
m
al
p
ac
k
et
is
b
ein
g
m
is
class
if
ied
as a
Do
S a
ttem
p
t.
Mo
v
in
g
o
n
to
th
e
p
e
r
s
p
ec
tiv
e
o
f
th
e
n
etwo
r
k
s
ec
u
r
ity
f
ie
ld
,
th
e
NI
DS
m
o
d
e
l
h
as
id
e
n
tifie
d
k
ey
f
ac
to
r
s
f
o
r
c
o
r
r
ec
tly
class
if
y
in
g
th
e
n
etwo
r
k
p
ac
k
et
as
a
n
o
r
m
al
p
ac
k
et.
T
h
is
s
ce
n
ar
io
c
o
u
ld
b
e
f
o
u
n
d
i
n
th
e
f
ea
tu
r
e
w
la
n
.
fc.
s
u
b
typ
e
an
d
r
a
d
io
ta
p
.
d
b
m.
a
n
t
_
s
ig
n
a
l
wh
e
r
eb
y
th
e
n
e
g
ativ
e
s
id
e
in
th
e
r
ig
h
t
b
ar
c
h
ar
t
in
d
icate
s
th
at
th
e
m
o
d
el
r
ea
lizes
th
at
th
ese
f
ea
tu
r
es
o
p
p
o
s
e
th
e
class
if
icatio
n
o
f
th
e
p
a
r
ticu
lar
n
etwo
r
k
p
ac
k
et
as
a
Do
S
attac
k
.
T
o
b
e
m
o
r
e
s
p
ec
if
ic,
th
e
f
ea
tu
r
e
wlan
.
f
c.
s
u
b
ty
p
e
s
h
o
ws
th
at
a
n
o
r
m
al
ty
p
e
o
f
n
etwo
r
k
p
ac
k
e
t
is
b
ein
g
tr
an
s
m
itted
,
wh
ile
th
e
f
ea
tu
r
e
r
a
d
io
ta
p
.
d
b
m.
a
n
t_
s
i
gnal
in
d
icate
s
th
at
th
e
s
ig
n
al
s
tr
en
g
th
o
f
th
e
n
etwo
r
k
is
ac
tu
ally
n
o
r
m
al.
H
o
wev
er
,
th
e
NI
DS
m
o
d
el
g
ets
co
n
f
u
s
ed
w
h
en
ce
r
tai
n
f
ea
tu
r
e
s
cr
ea
te
am
b
ig
u
ity
,
s
u
ch
as
w
la
n
.
fc.
p
r
o
tecte
d
an
d
r
a
d
io
ta
p
.
ch
a
n
n
el.
fla
g
s
.
cc
k
.
I
n
th
e
f
ea
tu
r
e
w
la
n
.
fc.
p
r
o
tect
ed
,
th
is
co
n
f
u
s
io
n
ar
is
es
wh
en
u
n
en
cr
y
p
ted
f
r
am
es
ar
e
tr
an
s
m
itted
,
wh
ich
d
o
es
n
o
t
n
ec
ess
ar
ily
i
n
d
icate
an
in
tr
u
s
io
n
attem
p
t.
Fo
r
i
n
s
tan
ce
,
d
u
r
in
g
th
e
tr
a
n
s
m
is
s
io
n
o
f
p
r
o
b
e
r
eq
u
ests
,
th
e
f
r
a
m
es
ar
e
u
n
en
cr
y
p
ted
as
p
ar
t
o
f
n
o
r
m
al
n
etwo
r
k
o
p
er
atio
n
s
.
Mo
r
eo
v
er
,
t
h
e
f
ea
tu
r
e
ra
d
io
ta
p
.
ch
a
n
n
el.
fla
g
s
.
cc
k
in
d
icate
s
th
e
p
r
esen
ce
o
f
ce
r
tain
m
o
d
u
latio
n
s
ch
em
es,
wh
ich
a
r
e
co
m
m
o
n
in
b
o
th
n
o
r
m
al
a
n
d
attac
k
s
ce
n
ar
io
s
.
T
h
is
o
v
e
r
lap
ca
n
lead
th
e
m
o
d
el
to
m
is
in
ter
p
r
et
n
o
r
m
a
l
m
o
d
u
latio
n
as
a
p
o
ten
tial
th
r
ea
t.
C
o
n
s
eq
u
en
tly
,
th
e
m
o
d
el'
s
ch
allen
g
e
lies
in
d
is
en
tan
g
lin
g
th
ese
am
b
ig
u
o
u
s
s
ig
n
als
to
m
ak
e
a
co
r
r
ec
t c
lass
if
icatio
n
.
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
A
co
mp
a
r
a
tive
s
tu
d
y
o
f d
ee
p
lea
r
n
in
g
-
b
a
s
ed
n
etw
o
r
k
in
tr
u
s
io
n
d
etec
tio
n
…
(
Ta
n
J
u
a
n
K
a
i
)
4117
Fig
u
r
e
6
.
L
I
ME
in
ter
p
r
etatio
n
f
o
r
co
r
r
ec
t p
r
e
d
ictio
n
in
D
o
S
a
ttack
s
Fig
u
r
e
7
.
L
I
ME
in
ter
p
r
etatio
n
f
o
r
f
alse a
lar
m
in
Do
S
a
ttack
s
5.
CO
NCLU
SI
O
N
T
h
is
p
ap
er
p
r
esen
ts
a
co
m
p
r
e
h
en
s
iv
e
s
tu
d
y
an
aly
zin
g
th
e
p
er
f
o
r
m
an
ce
o
f
T
a
b
Net
in
NI
DS
m
o
d
el
d
ev
elo
p
m
e
n
t.
Usi
n
g
f
ea
tu
r
e
s
ets
o
b
tain
ed
th
r
o
u
g
h
th
e
p
h
i
-
K
m
eth
o
d
,
th
e
m
o
d
el
a
ch
iev
ed
im
p
r
ess
iv
e
ac
cu
r
ac
y
,
r
ec
all,
a
n
d
p
r
ec
is
io
n
o
f
9
9
.
9
9
%,
s
u
r
p
ass
in
g
th
e
p
er
f
o
r
m
a
n
ce
o
f
f
o
u
r
e
x
is
tin
g
wo
r
k
s
.
T
h
ese
r
esu
lts
d
em
o
n
s
tr
ate
th
at
th
e
T
ab
Net
m
o
d
el
is
h
ig
h
l
y
ef
f
ec
tiv
e
i
n
i
n
tr
u
s
io
n
d
etec
tio
n
.
Ad
d
itio
n
ally
,
it
s
u
cc
ess
f
u
lly
m
itig
ates th
e
r
is
k
o
f
alar
m
f
atig
u
e,
wh
ich
is
o
f
ten
ca
u
s
ed
b
y
a
lar
g
e
v
o
l
u
m
e
o
f
f
alse a
lar
m
s
.
T
h
e
r
o
b
u
s
tn
es
s
o
f
t
h
e
m
o
d
el
is
ev
alu
ated
b
y
in
tr
o
d
u
cin
g
S
NR
v
alu
es
r
an
g
in
g
f
r
o
m
1
5
t
o
3
0
.
T
h
e
m
o
d
el'
s
co
n
s
is
ten
t
p
er
f
o
r
m
an
ce
in
te
r
m
s
o
f
p
r
ec
is
io
n
h
as
d
em
o
n
s
tr
ated
th
at
th
e
T
a
b
Net
m
o
d
el
ca
n
d
etec
t
in
tr
u
s
io
n
s
r
eg
ar
d
less
o
f
th
e
n
o
is
e
lev
el
in
th
e
wir
eless
ch
an
n
el.
Ho
wev
e
r
,
it
is
n
o
tab
le
th
at
th
e
n
o
is
e
lev
el
ca
n
ca
u
s
e
n
o
r
m
al
p
r
o
f
iles
to
b
e
m
is
class
if
ied
as
in
tr
u
s
io
n
atte
m
p
ts
.
T
h
is
is
ev
id
en
t
f
r
o
m
th
e
d
ec
r
ea
s
e
in
r
ec
all
wh
en
th
e
SNR
v
alu
e
is
lo
wer
,
in
d
icatin
g
th
at
h
i
g
h
n
o
is
e
lev
e
l is p
r
esen
t in
th
e
n
etwo
r
k
p
r
o
f
ile.
Fin
ally
,
th
e
d
ec
is
io
n
-
m
a
k
in
g
s
tep
s
o
f
th
e
NI
DS
m
o
d
el
ar
e
in
ter
p
r
etate
d
th
r
o
u
g
h
th
e
XAI
a
p
p
r
o
ac
h
es,
p
ar
ticu
lar
ly
u
s
in
g
SHAP
an
d
L
I
ME
.
B
y
ex
am
in
i
n
g
b
o
th
t
h
e
g
lo
b
al
a
n
d
lo
ca
l
i
n
ter
p
r
etatio
n
s
o
f
th
e
m
o
d
el,
k
ey
f
ea
tu
r
es
th
at
s
ig
n
if
ican
tly
in
f
lu
en
ce
th
e
m
o
d
el'
s
p
r
ed
icti
o
n
s
ar
e
id
en
tifie
d
.
T
h
ese
in
s
ig
h
ts
h
elp
in
u
n
d
er
s
tan
d
i
n
g
h
o
w
th
e
m
o
d
el
d
if
f
er
en
tiates
b
etwe
en
n
o
r
m
a
l
an
d
m
alicio
u
s
ac
tiv
ities
,
th
er
eb
y
en
h
an
cin
g
th
e
tr
an
s
p
ar
en
cy
a
n
d
tr
u
s
two
r
th
i
n
ess
o
f
th
e
in
tr
u
s
io
n
d
etec
t
io
n
p
r
o
ce
s
s
.
Fro
m
th
e
p
e
r
s
p
ec
tiv
e
o
f
NI
DS
in
ter
p
r
et
atio
n
,
its
m
eth
o
d
o
f
d
eter
m
in
in
g
ab
n
o
r
m
al
tr
af
f
i
c
clo
s
ely
alig
n
s
with
th
e
way
s
ec
u
r
ity
ex
p
e
r
ts
id
en
tify
a
b
n
o
r
m
al
tr
a
f
f
ic
i
n
r
e
al
life
.
T
h
is
d
em
o
n
s
tr
ates
th
at
it
is
r
eliab
le
an
d
e
f
f
ec
tiv
e
in
a
cc
u
r
ately
d
etec
tin
g
s
ec
u
r
ity
th
r
ea
ts
.
As
s
u
ch
,
th
is
s
tu
d
y
h
as
in
v
esti
g
ated
th
e
r
eliab
ilit
y
o
f
NI
DS
th
r
o
u
g
h
t
h
e
co
m
p
r
eh
e
n
s
iv
e
m
o
d
el
an
d
r
o
b
u
s
tn
ess
ev
alu
atio
n
,
ef
f
ec
ti
v
e
f
ea
tu
r
e
s
elec
tio
n
,
an
d
th
e
in
teg
r
atio
n
o
f
in
ter
p
r
etab
ilit
y
ap
p
r
o
ac
h
es.
T
h
ese
co
n
tr
ib
u
tio
n
s
p
r
o
v
e
n
th
e
ca
p
ab
ilit
y
o
f
th
e
T
ab
Net
m
o
d
el
in
en
h
an
cin
g
n
etwo
r
k
s
ec
u
r
ity
b
y
ac
cu
r
ately
id
en
tify
in
g
an
d
m
itig
atin
g
v
ar
io
u
s
th
r
ea
ts
.
Fu
tu
r
e
wo
r
k
is
s
u
g
g
ested
to
f
o
c
u
s
o
n
m
in
im
iz
in
g
f
alse
alar
m
s
in
n
o
is
y
en
v
ir
o
n
m
en
ts
b
y
u
s
in
g
h
y
b
r
id
a
p
p
r
o
ac
h
es.
Sp
ec
if
ically
,
r
u
le
-
b
ased
m
eth
o
d
s
s
h
o
u
ld
b
e
in
teg
r
ated
in
to
th
e
m
o
d
el
to
r
ed
u
ce
f
alse a
lar
m
s
ac
r
o
s
s
d
if
f
er
en
t n
o
is
e
lev
els.
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
.
4
,
Au
g
u
s
t
20
25
:
4
1
0
9
-
4119
4118
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
wo
r
k
was
s
u
p
p
o
r
ted
b
y
th
e
T
ele
k
o
m
Ma
lay
s
ia
R
e
s
ea
r
ch
an
d
Dev
elo
p
m
e
n
t
u
n
d
er
Gr
an
t
R
DT
C
/2
4
1
1
2
5
(
MM
UE
/2
4
0
0
6
6
)
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
J
u
an
-
Kai
T
an
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
L
ee
-
Yen
g
On
g
✓
✓
✓
✓
✓
✓
✓
✓
Me
n
g
-
C
h
ew
L
eo
w
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
in
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
o
p
en
ly
av
ailab
le
in
A
W
I
D3
at
1
0
.
1
1
0
9
/AC
C
E
SS
.
2
0
2
1
.
3
0
6
1
6
0
9
,
r
ef
er
en
ce
n
u
m
b
er
[
2
6
]
.
RE
F
E
R
E
NC
E
S
[
1
]
R
.
A
h
m
a
d
,
I
.
A
l
sma
d
i
,
W
.
A
l
h
a
m
d
a
n
i
,
a
n
d
L.
T
a
w
a
l
b
e
h
,
“
Z
e
r
o
-
d
a
y
a
t
t
a
c
k
d
e
t
e
c
t
i
o
n
:
a
sy
s
t
e
m
a
t
i
c
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
,
”
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
Re
v
i
e
w
,
v
o
l
.
5
6
,
n
o
.
1
0
,
p
p
.
1
0
7
3
3
–
1
0
8
1
1
,
O
c
t
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
0
4
6
2
-
023
-
1
0
4
3
7
-
z.
[
2
]
B
.
M
.
S
e
r
i
n
e
l
l
i
,
A
.
C
o
l
l
e
n
,
a
n
d
N
.
A
.
N
i
j
d
a
m,
“
Tr
a
i
n
i
n
g
g
u
i
d
a
n
c
e
w
i
t
h
K
D
D
c
u
p
1
9
9
9
a
n
d
N
S
L
-
K
D
D
d
a
t
a
se
t
s
o
f
A
N
I
D
I
N
R
:
a
n
o
ma
l
y
-
b
a
s
e
d
n
e
t
w
o
r
k
i
n
t
r
u
si
o
n
d
e
t
e
c
t
i
o
n
sy
s
t
e
m
,
”
Pr
o
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
1
7
5
,
p
p
.
5
6
0
–
5
6
5
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s
.
2
0
2
0
.
0
7
.
0
8
0
.
[
3
]
A
.
N
i
s
i
o
t
i
,
A
.
M
y
l
o
n
a
s,
P
.
D
.
Y
o
o
,
a
n
d
V
.
K
a
t
o
s
,
“
F
r
o
m
i
n
t
r
u
s
i
o
n
d
e
t
e
c
t
i
o
n
t
o
a
t
t
a
c
k
e
r
a
t
t
r
i
b
u
t
i
o
n
:
a
c
o
m
p
r
e
h
e
n
s
i
v
e
s
u
r
v
e
y
o
f
u
n
s
u
p
e
r
v
i
s
e
d
m
e
t
h
o
d
s,
”
I
EEE
C
o
m
m
u
n
i
c
a
t
i
o
n
s
S
u
r
v
e
y
s
&
T
u
t
o
r
i
a
l
s
,
v
o
l
.
2
0
,
n
o
.
4
,
p
p
.
3
3
6
9
–
3
3
8
8
,
2
0
1
8
,
d
o
i
:
1
0
.
1
1
0
9
/
C
O
M
S
T.
2
0
1
8
.
2
8
5
4
7
2
4
.
[
4
]
S
.
R
o
y
,
J.
L
i
,
V
.
P
a
n
d
e
y
,
a
n
d
Y
.
B
a
i
,
“
A
n
e
x
p
l
a
i
n
a
b
l
e
d
e
e
p
n
e
u
r
a
l
f
r
a
m
e
w
o
r
k
f
o
r
t
r
u
st
w
o
r
t
h
y
n
e
t
w
o
r
k
i
n
t
r
u
si
o
n
d
e
t
e
c
t
i
o
n
,
”
i
n
20
2
2
1
0
t
h
I
E
EE
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
ren
c
e
o
n
M
o
b
i
l
e
C
l
o
u
d
C
o
m
p
u
t
i
n
g
,
S
e
r
v
i
c
e
s,
a
n
d
En
g
i
n
e
e
ri
n
g
(
M
o
b
i
l
e
C
l
o
u
d
)
,
A
u
g
.
2
0
2
2
,
p
p
.
2
5
–
3
0
,
d
o
i
:
1
0
.
1
1
0
9
/
M
o
b
i
l
e
C
l
o
u
d
5
5
3
3
3
.
2
0
2
2
.
0
0
0
1
1
.
[
5
]
D
.
G
a
sp
a
r
,
P
.
S
i
l
v
a
,
a
n
d
C
.
S
i
l
v
a
,
“
E
x
p
l
a
i
n
a
b
l
e
A
I
f
o
r
i
n
t
r
u
si
o
n
d
e
t
e
c
t
i
o
n
sy
st
e
m
s:
LI
M
E
a
n
d
S
H
A
P
a
p
p
l
i
c
a
b
i
l
i
t
y
o
n
m
u
l
t
i
-
l
a
y
e
r
p
e
r
c
e
p
t
r
o
n
,
”
I
EE
E
A
c
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
3
0
1
6
4
–
3
0
1
7
5
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
3
6
8
3
7
7
.
[
6
]
V
.
H
a
ss
i
j
a
e
t
a
l
.
,
“
I
n
t
e
r
p
r
e
t
i
n
g
b
l
a
c
k
-
b
o
x
m
o
d
e
l
s
:
a
r
e
v
i
e
w
o
n
e
x
p
l
a
i
n
a
b
l
e
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
,
”
C
o
g
n
i
t
i
v
e
C
o
m
p
u
t
a
t
i
o
n
,
v
o
l
.
1
6
,
n
o
.
1
,
p
p
.
4
5
–
7
4
,
J
a
n
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
2
5
5
9
-
023
-
1
0
1
7
9
-
8.
[
7
]
T.
A
.
Ta
n
g
,
L
.
M
h
a
md
i
,
D
.
M
c
L
e
r
n
o
n
,
S
.
A
.
R
.
Za
i
d
i
,
a
n
d
M
.
G
h
o
g
h
o
,
“
D
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
f
o
r
n
e
t
w
o
r
k
i
n
t
r
u
si
o
n
d
e
t
e
c
t
i
o
n
i
n
s
o
f
t
w
a
r
e
d
e
f
i
n
e
d
n
e
t
w
o
r
k
i
n
g
,
”
i
n
2
0
1
6
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
W
i
re
l
e
ss
N
e
t
w
o
rks
a
n
d
Mo
b
i
l
e
C
o
m
m
u
n
i
c
a
t
i
o
n
s
(
WINC
O
M)
,
O
c
t
.
2
0
1
6
,
p
p
.
2
5
8
–
2
6
3
,
d
o
i
:
1
0
.
1
1
0
9
/
W
I
N
C
O
M
.
2
0
1
6
.
7
7
7
7
2
2
4
.
[
8
]
Y.
-
C
.
W
a
n
g
,
Y
.
-
C
.
H
o
u
n
g
,
H
.
-
X
.
C
h
e
n
,
a
n
d
S
.
-
M
.
Ts
e
n
g
,
“
N
e
t
w
o
r
k
a
n
o
ma
l
y
i
n
t
r
u
si
o
n
d
e
t
e
c
t
i
o
n
b
a
se
d
o
n
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
,
”
S
e
n
so
rs
,
v
o
l
.
2
3
,
n
o
.
4
,
p
.
2
1
7
1
,
F
e
b
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
s2
3
0
4
2
1
7
1
.
[
9
]
R
.
S
.
A
h
m
a
d
,
A
.
H
.
A
l
i
,
S
.
M
.
K
a
z
i
m,
a
n
d
Q
.
N
i
y
a
z
,
“
A
G
A
F
a
n
d
C
N
N
b
a
s
e
d
W
i
-
F
i
n
e
t
w
o
r
k
i
n
t
r
u
si
o
n
d
e
t
e
c
t
i
o
n
s
y
st
e
m,
”
i
n
I
EEE
I
N
FO
C
O
M
2
0
2
3
-
I
EEE
C
o
n
f
e
r
e
n
c
e
o
n
C
o
m
p
u
t
e
r C
o
m
m
u
n
i
c
a
t
i
o
n
s
Wo
r
k
sh
o
p
s (I
N
FO
C
O
M
W
K
S
H
PS
)
,
M
a
y
2
0
2
3
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
N
F
O
C
O
M
W
K
S
H
P
S
5
7
4
5
3
.
2
0
2
3
.
1
0
2
2
6
0
3
6
.
[
1
0
]
M
.
D
.
H
o
ssa
i
n
,
H
.
I
n
o
u
e
,
H
.
O
c
h
i
a
i
,
D
.
F
a
l
l
,
a
n
d
Y
.
K
a
d
o
b
a
y
a
sh
i
,
“
LSTM
-
b
a
se
d
i
n
t
r
u
si
o
n
d
e
t
e
c
t
i
o
n
s
y
s
t
e
m
f
o
r
i
n
-
v
e
h
i
c
l
e
c
a
n
b
u
s
c
o
mm
u
n
i
c
a
t
i
o
n
s
,
”
I
E
EE
Ac
c
e
ss
,
v
o
l
.
8
,
p
p
.
1
8
5
4
8
9
–
1
8
5
5
0
2
,
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
0
.
3
0
2
9
3
0
7
.
[
1
1
]
B
.
D
e
o
r
e
a
n
d
S
.
B
h
o
s
a
l
e
,
“
H
y
b
r
i
d
o
p
t
i
mi
z
a
t
i
o
n
e
n
a
b
l
e
d
r
o
b
u
s
t
C
N
N
-
LSTM
t
e
c
h
n
i
q
u
e
f
o
r
n
e
t
w
o
r
k
i
n
t
r
u
s
i
o
n
d
e
t
e
c
t
i
o
n
,
”
I
EE
E
Ac
c
e
ss
,
v
o
l
.
1
0
,
p
p
.
6
5
6
1
1
–
6
5
6
2
2
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
2
.
3
1
8
3
2
1
3
.
[
1
2
]
H
.
C
.
A
l
t
u
n
a
y
a
n
d
Z.
A
l
b
a
y
r
a
k
,
“
A
h
y
b
r
i
d
C
N
N
+
LST
M
-
b
a
se
d
i
n
t
r
u
s
i
o
n
d
e
t
e
c
t
i
o
n
sy
st
e
m
f
o
r
i
n
d
u
st
r
i
a
l
I
o
T
n
e
t
w
o
r
k
s,”
En
g
i
n
e
e
ri
n
g
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
a
n
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
,
v
o
l
.
3
8
,
p
.
1
0
1
3
2
2
,
F
e
b
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
e
st
c
h
.
2
0
2
2
.
1
0
1
3
2
2
.
[
1
3
]
M
.
G
r
a
z
i
a
n
i
e
t
a
l
.
,
“
A
g
l
o
b
a
l
t
a
x
o
n
o
my
o
f
i
n
t
e
r
p
r
e
t
a
b
l
e
A
I
:
u
n
i
f
y
i
n
g
t
h
e
t
e
r
mi
n
o
l
o
g
y
f
o
r
t
h
e
t
e
c
h
n
i
c
a
l
a
n
d
so
c
i
a
l
sc
i
e
n
c
e
s,”
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
Re
v
i
e
w
,
v
o
l
.
5
6
,
n
o
.
4
,
p
p
.
3
4
7
3
–
3
5
0
4
,
A
p
r
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
0
4
6
2
-
022
-
1
0
2
5
6
-
8.
[
1
4
]
S
.
H
a
r
i
h
a
r
a
n
,
R
.
R
.
R
e
j
i
m
o
l
R
o
b
i
n
s
o
n
,
R
.
R
.
P
r
a
s
a
d
,
C
.
T
h
o
mas
,
a
n
d
N
.
B
a
l
a
k
r
i
s
h
n
a
n
,
“
X
A
I
f
o
r
i
n
t
r
u
si
o
n
d
e
t
e
c
t
i
o
n
sy
s
t
e
m
:
c
o
m
p
a
r
i
n
g
e
x
p
l
a
n
a
t
i
o
n
s
b
a
s
e
d
o
n
g
l
o
b
a
l
a
n
d
l
o
c
a
l
sc
o
p
e
,
”
J
o
u
r
n
a
l
o
f
C
o
m
p
u
t
e
r
V
i
ro
l
o
g
y
a
n
d
H
a
c
k
i
n
g
T
e
c
h
n
i
q
u
e
s
,
v
o
l
.
1
9
,
n
o
.
2
,
p
p
.
2
1
7
–
2
3
9
,
J
u
l
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
1
4
1
6
-
0
2
2
-
0
0
4
4
1
-
2.
[
1
5
]
Z.
A
.
El
H
o
u
d
a
,
B
.
B
r
i
k
,
a
n
d
L
.
K
h
o
u
k
h
i
,
“
W
h
y
s
h
o
u
l
d
i
t
r
u
s
t
y
o
u
r
I
D
S
?
’
:
A
n
e
x
p
l
a
i
n
a
b
l
e
d
e
e
p
l
e
a
r
n
i
n
g
f
r
a
m
e
w
o
r
k
f
o
r
i
n
t
r
u
si
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.