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.
14
,
No
.
6
,
Dec
em
b
er
20
25
,
p
p
.
4
9
8
2
~
4
9
9
2
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
14
.i
6
.
p
p
4
9
8
2
-
4
9
9
2
4982
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Deep
lea
rni
ng
-
ba
sed
ev
a
lua
tion for
distribu
ted
denia
l of servi
ce
a
tt
a
cks det
ection
Nee
t
hu
S.
1
,
H
.
V.
Ra
v
is
h Ar
a
dh
y
a
2
,
Vis
wa
v
a
rdha
n Re
dd
y
K
a
rna
3
1
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
En
g
i
n
e
e
r
i
n
g
,
R
V
C
o
l
l
e
g
e
o
f
En
g
i
n
e
e
r
i
n
g
,
V
i
s
v
e
s
v
a
r
a
y
a
T
e
c
h
n
o
l
o
g
i
c
a
l
U
n
i
v
e
r
s
i
t
y
,
B
e
l
a
g
a
v
i
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
C
o
m
mu
n
i
c
a
t
i
o
n
En
g
i
n
e
e
r
i
n
g
,
R
V
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
,
V
i
s
v
e
s
v
a
r
a
y
a
Te
c
h
n
o
l
o
g
i
c
a
l
U
n
i
v
e
r
si
t
y
,
B
e
l
a
g
a
v
i
,
I
n
d
i
a
3
D
e
p
a
r
t
me
n
t
o
f
A
r
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
,
R
V
C
o
l
l
e
g
e
o
f
En
g
i
n
e
e
r
i
n
g
,
V
i
s
v
e
s
v
a
r
a
y
a
T
e
c
h
n
o
l
o
g
i
c
a
l
U
n
i
v
e
r
s
i
t
y
,
B
e
l
a
g
a
v
i
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
n
21
,
2
0
2
4
R
ev
is
ed
Au
g
4
,
2
0
2
5
Acc
ep
ted
Sep
7
,
2
0
2
5
S
o
ftwa
re
-
d
e
fin
e
d
n
e
tw
o
rk
(S
DN
)
in
tr
o
d
u
c
e
s
a
p
ro
g
ra
m
m
a
b
le
a
n
d
c
e
n
tralize
d
c
o
n
tr
o
l
m
e
c
h
a
n
ism
fo
r
m
a
n
a
g
i
n
g
n
e
two
r
k
i
n
fr
a
stru
c
tu
re
,
e
n
h
a
n
c
in
g
flex
i
b
il
i
ty
a
n
d
e
fficie
n
c
y
.
Ho
we
v
e
r,
th
is
a
rc
h
i
tec
tu
re
i
s
p
ro
n
e
t
o
se
c
u
rit
y
th
re
a
ts,
p
a
rti
c
u
larl
y
d
istri
b
u
ted
d
e
n
ial
o
f
se
rv
ice
(DD
o
S
)
a
t
tac
k
s
th
a
t
e
x
p
lo
it
c
e
n
tralize
d
c
o
n
tr
o
l.
T
h
is
stu
d
y
p
re
se
n
ts
a
c
o
m
p
a
ra
ti
v
e
a
n
a
ly
sis
o
f
se
v
e
ra
l
d
e
e
p
lea
rn
in
g
(DL)
m
o
d
e
ls
—
n
a
m
e
ly
,
m
u
l
ti
lay
e
r
p
e
rc
e
p
tro
n
(
M
LP
)
,
a
rti
ficia
l
n
e
u
ra
l
n
e
two
r
k
(AN
N
)
,
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
r
k
(CNN
)
,
re
c
u
rre
n
t
n
e
u
ra
l
n
e
two
r
k
(RNN
)
,
a
n
d
lo
n
g
sh
o
rt
-
term
m
e
m
o
ry
(LS
TM
)
—
fo
r
d
e
tec
ti
n
g
DD
o
S
t
h
re
a
ts
with
in
S
DN
e
n
v
ir
o
n
m
e
n
ts.
T
h
e
re
se
a
rc
h
in
c
o
rp
o
ra
tes
k
e
y
p
re
p
r
o
c
e
ss
in
g
tec
h
n
iq
u
e
s
su
c
h
a
s
fe
a
tu
re
se
le
c
ti
o
n
a
n
d
sy
n
th
e
ti
c
m
in
o
rit
y
o
v
e
rsa
m
p
li
n
g
tec
h
n
iq
u
e
(S
M
OTE)
t
o
h
a
n
d
le
c
las
s
imb
a
lan
c
e
.
Th
e
re
su
lt
s
i
n
d
ica
te
th
a
t
se
q
u
e
n
c
e
-
a
wa
re
m
o
d
e
ls
li
k
e
LS
TM
a
n
d
RNN
a
re
h
ig
h
l
y
e
ffe
c
ti
v
e
in
in
t
e
rp
re
ti
n
g
tem
p
o
ra
l
n
e
tw
o
rk
b
e
h
a
v
io
r,
with
LS
TM
a
c
h
iev
i
n
g
t
h
e
h
ig
h
e
st
p
e
r
fo
rm
a
n
c
e
(a
c
c
u
ra
c
y
:
9
1
%
,
p
re
c
is
io
n
:
8
6
%
,
re
c
a
ll
:
9
4
%
,
a
n
d
F
1
-
sc
o
re
:
9
0
%
)
.
Th
e
se
fin
d
i
n
g
s
u
n
d
e
rsc
o
re
th
e
p
o
ten
ti
a
l
o
f
a
d
v
a
n
c
e
d
DL
m
e
th
o
d
s
in
f
o
rti
f
y
in
g
S
DN
in
fra
str
u
c
tu
re
s
a
g
a
in
s
t
c
o
m
p
lex
c
y
b
e
r
th
re
a
ts.
K
ey
w
o
r
d
s
:
Dee
p
lear
n
in
g
D
i
s
t
r
i
b
u
t
e
d
d
e
n
i
a
l
o
f
s
e
r
v
i
c
e
a
t
t
a
c
k
Op
en
f
lo
w
p
r
o
to
co
l
SMOT
E
So
f
twar
e
-
d
ef
in
e
d
n
etwo
r
k
s
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
:
Nee
th
u
S
.
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
E
n
g
in
ee
r
in
g
,
R
V
C
o
lleg
e
o
f
E
n
g
i
n
ee
r
in
g
Vis
v
esv
ar
ay
a
T
ec
h
n
o
l
o
g
ical
Un
iv
er
s
ity
B
elag
av
i
,
Kar
n
atak
a,
I
n
d
ia
E
m
ail: n
ee
th
u
s
@
r
v
ce
.
ed
u
.
in
1.
I
NT
RO
D
UCT
I
O
N
So
f
twar
e
-
d
ef
in
e
d
n
etwo
r
k
(
SDN)
h
as
s
ig
n
if
ican
tly
tr
a
n
s
f
o
r
m
ed
th
e
la
n
d
s
ca
p
e
o
f
n
etwo
r
k
m
an
ag
em
en
t
b
y
d
ec
o
u
p
lin
g
th
e
co
n
tr
o
l
a
n
d
d
ata
p
lan
es,
e
n
a
b
lin
g
ce
n
tr
alize
d
an
d
p
r
o
g
r
am
m
ab
le
co
n
tr
o
l
o
v
e
r
n
etwo
r
k
in
f
r
astru
ctu
r
e
.
T
h
is
a
r
ch
itectu
r
al
ad
v
an
ce
m
e
n
t
f
o
s
t
er
s
d
y
n
am
ic
an
d
ef
f
icien
t
c
o
m
m
u
n
icatio
n
b
u
t
also
in
tr
o
d
u
ce
s
n
o
v
el
s
ec
u
r
ity
v
u
ln
er
ab
ilit
ies.
I
n
p
ar
ticu
lar
,
th
e
ce
n
tr
alize
d
co
n
tr
o
l
m
ec
h
a
n
is
m
b
ec
o
m
es
a
cr
itical
p
o
in
t
o
f
f
ailu
r
e
,
s
u
s
ce
p
tib
le
t
o
tar
g
eted
attac
k
s
s
u
ch
as
b
o
tn
ets
an
d
d
is
tr
ib
u
ted
d
e
n
ial
o
f
s
er
v
ice
(
DDo
S)
th
r
ea
ts
,
wh
ich
ca
n
c
o
m
p
r
o
m
i
s
e
b
o
th
c
o
n
tr
o
l
an
d
d
ata
p
lan
e
co
m
p
o
n
en
ts
.
DDo
S
attac
k
s
r
em
ain
a
p
er
s
is
ten
t
an
d
escalatin
g
ch
allen
g
e
in
S
DN
en
v
ir
o
n
m
e
n
ts
d
u
e
t
o
th
e
in
tr
icate
an
d
d
y
n
am
ic
n
atu
r
e
o
f
n
etwo
r
k
tr
af
f
ic.
B
y
f
lo
o
d
in
g
th
e
tar
g
et
s
y
s
tem
with
f
ak
e
r
eq
u
ests
,
attac
k
er
s
aim
to
ex
h
au
s
t
co
m
p
u
tatio
n
al
r
eso
u
r
ce
s
an
d
d
is
r
u
p
t
leg
itima
te
s
er
v
ices.
T
h
ese
attac
k
s
ar
e
n
o
t
o
n
ly
s
tealt
h
y
a
n
d
i
n
ex
p
e
n
s
iv
e
to
ex
ec
u
te
b
u
t
also
ca
p
a
b
le
o
f
in
f
lictin
g
s
u
b
s
tan
tial
d
am
ag
e
o
n
n
etwo
r
k
p
er
f
o
r
m
a
n
ce
an
d
av
ailab
ilit
y
[
1
]
,
[
2
]
.
T
o
ad
d
r
e
s
s
th
ese
co
n
ce
r
n
s
,
r
esear
ch
er
s
h
av
e
p
r
o
p
o
s
ed
l
ev
er
ag
in
g
d
ee
p
lear
n
i
n
g
(
D
L
)
tech
n
iq
u
es
f
o
r
r
ea
l
-
tim
e
tr
af
f
ic
an
aly
s
is
an
d
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
Dee
p
lea
r
n
in
g
-
b
a
s
ed
ev
a
lu
a
tio
n
fo
r
d
is
tr
ib
u
ted
d
en
i
a
l o
f ser
vice
a
tta
ck
s
d
etec
tio
n
(
N
ee
th
u
S
.
)
4983
an
o
m
aly
d
etec
tio
n
.
On
e
s
u
c
h
ap
p
r
o
ac
h
in
v
o
lv
es
th
e
u
s
e
o
f
d
ee
p
n
e
u
r
al
n
etwo
r
k
s
(
DNNs)
,
wh
ich
o
f
f
er
s
ca
lab
le
ar
ch
itectu
r
es
f
o
r
id
e
n
tify
in
g
c
o
m
p
lex
tr
a
f
f
ic
p
atte
r
n
s
ass
o
ciate
d
with
DDo
S
ac
tiv
ities
[
3
]
.
Sev
er
al
h
y
b
r
id
m
o
d
els
h
a
v
e
also
b
ee
n
d
ev
elo
p
ed
—
f
o
r
in
s
tan
ce
,
a
co
m
b
in
atio
n
o
f
c
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NNs)
an
d
b
id
ir
ec
tio
n
al
l
o
n
g
s
h
o
r
t
-
ter
m
m
e
m
o
r
y
(
B
iLST
M)
n
etwo
r
k
s
—
wh
ich
en
h
an
ce
d
etec
tio
n
ca
p
ab
ilit
ies
u
s
in
g
well
-
estab
lis
h
ed
d
atasets
s
u
ch
as
UNS
W
-
NB
1
5
an
d
NSL
-
KDD
f
o
r
b
o
th
b
in
ar
y
an
d
m
u
lticlas
s
cla
s
s
if
icatio
n
task
s
[
4
]
.
I
n
n
o
v
atio
n
s
in
an
o
m
al
y
d
etec
tio
n
co
n
tin
u
e
to
em
e
r
g
e.
Feed
f
o
r
war
d
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
etwo
r
k
s
(
FF
C
NNs)
h
av
e
b
ee
n
ap
p
lied
to
lo
w
-
r
ate
d
e
n
ial
o
f
s
er
v
ice
(
Do
S
)
s
ce
n
ar
io
s
in
I
o
T
-
SDN
s
ettin
g
s
,
o
f
te
n
p
air
e
d
with
f
ea
tu
r
e
s
elec
tio
n
tech
n
iq
u
es
lik
e
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM)
-
b
ased
wr
ap
p
er
s
.
T
h
ese
m
o
d
els
ar
e
b
en
ch
m
ar
k
e
d
ag
ai
n
s
t
class
ical
class
if
ier
s
,
in
clu
d
in
g
J
4
8
,
r
an
d
o
m
f
o
r
est
s
,
r
ed
u
ce
d
er
r
o
r
p
r
u
n
in
g
(
R
E
P)
t
r
ee
,
an
d
m
u
lti
-
lay
er
p
er
ce
p
t
r
o
n
(
ML
P)
[
5
]
.
C
lo
u
d
-
in
te
g
r
ated
DL
ar
ch
itectu
r
es
h
av
e
also
b
ee
n
p
r
o
p
o
s
ed
f
o
r
d
etec
tin
g
a
n
d
m
itig
atin
g
p
h
is
h
in
g
an
d
b
o
tn
et
attac
k
s
at
s
ca
le
[
6
]
.
Mo
r
e
o
v
er
,
s
tr
ateg
ies
s
u
ch
as
m
o
v
in
g
tar
g
et
d
ef
en
s
e
(
M
T
D)
h
av
e
b
ee
n
in
tr
o
d
u
ce
d
to
d
iv
er
t
m
alicio
u
s
tr
af
f
ic
to
war
d
d
ec
o
y
s
y
s
tem
s
,
ef
f
ec
tiv
ely
r
e
d
u
cin
g
th
e
im
p
a
ct
o
n
p
r
im
ar
y
s
er
v
er
s
[
7
]
.
A
v
ast
b
o
d
y
o
f
liter
atu
r
e
ex
p
l
o
r
es
v
ar
io
u
s
m
ac
h
in
e
lear
n
in
g
(
ML
)
a
n
d
DL
-
b
ased
s
o
lu
tio
n
s
to
d
etec
t
DDo
S
t
h
r
ea
ts
with
in
SDN
in
f
r
astru
c
tu
r
es
[
8
]
‒
[
1
2
]
.
Fo
r
in
s
tan
ce
,
a
two
-
tier
ed
ap
p
r
o
ac
h
em
p
lo
y
in
g
e
n
tr
o
p
y
-
b
ased
an
o
m
aly
d
etec
tio
n
f
o
llo
we
d
b
y
C
NN
-
b
ased
p
ac
k
et
class
if
icatio
n
h
as
s
h
o
wn
p
r
o
m
is
e
in
d
if
f
er
en
tiatin
g
leg
itima
t
e
an
d
s
u
s
p
icio
u
s
f
lo
ws
[
1
3
]
.
Oth
er
co
n
tr
ib
u
tio
n
s
in
clu
d
e
s
p
ar
s
e
a
u
to
en
co
d
er
s
c
o
m
b
in
ed
with
DNNs
f
o
r
f
ea
tu
r
e
r
ep
r
esen
tatio
n
an
d
class
if
icatio
n
[
1
4
]
,
as
well
as
in
teg
r
ated
in
tr
u
s
io
n
d
etec
tio
n
s
y
s
tem
s
(
I
DS)
an
d
d
ee
p
r
ein
f
o
r
ce
m
en
t
lear
n
in
g
(
DR
L
)
-
b
ased
in
tr
u
s
io
n
p
r
ev
en
tio
n
s
y
s
tem
s
(
I
PS
)
tailo
r
ed
f
o
r
lo
w
-
r
ate
DDo
S
attac
k
s
[
1
5
]
.
T
o
s
tr
en
g
th
e
n
SC
ADA
s
y
s
tem
s
b
u
ilt
o
n
SDN,
ad
v
an
ce
d
ar
c
h
itectu
r
es
u
tili
zin
g
r
ec
u
r
r
e
n
t
n
eu
r
al
n
et
wo
r
k
s
(
R
NNs),
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
L
STM
)
,
an
d
g
ate
d
r
ec
u
r
r
e
n
t
u
n
its
(
GR
Us)
h
av
e
also
b
ee
n
p
r
o
p
o
s
ed
,
en
h
an
cin
g
d
etec
tio
n
ac
cu
r
ac
y
ac
r
o
s
s
v
ar
io
u
s
d
ep
lo
y
m
e
n
t scen
ar
io
s
[
1
6
]
‒
[
1
8
]
.
A
m
eth
o
d
lev
e
r
ag
in
g
s
p
atial
-
tem
p
o
r
al
g
r
ap
h
co
n
v
o
lu
tio
n
al
n
etwo
r
k
s
(
ST
-
GC
N)
h
as
b
ee
n
in
tr
o
d
u
ce
d
f
o
r
s
ec
u
r
in
g
th
e
d
ata
p
la
n
e
o
f
SDNs
.
T
h
i
s
m
o
d
el
u
tili
ze
s
in
-
b
an
d
n
etwo
r
k
telem
etr
y
(
I
NT
)
with
s
am
p
lin
g
to
m
o
n
ito
r
n
etwo
r
k
s
tate
an
d
p
in
p
o
in
t
th
e
s
witch
es
in
v
o
lv
e
d
in
f
o
r
war
d
in
g
DDo
S
tr
af
f
ic
f
lo
ws
[
1
9
]
.
I
n
an
o
t
h
e
r
ap
p
r
o
ac
h
f
o
cu
s
ed
o
n
th
e
ea
r
l
y
d
etec
tio
n
o
f
T
C
P
SYN
f
lo
o
d
attac
k
s
,
an
e
x
ten
d
e
d
ch
i
-
s
q
u
ar
e
g
o
o
d
n
ess
-
of
-
f
it
test
is
em
p
lo
y
ed
.
T
h
is
tech
n
iq
u
e
ev
alu
ates
th
e
d
is
tr
ib
u
tio
n
o
f
h
alf
-
o
p
en
co
n
n
ec
tio
n
s
b
y
ca
lcu
latin
g
th
e
p
-
v
alu
e,
wh
ich
h
elp
s
d
etec
t
a
n
o
m
alies
in
n
etwo
r
k
b
e
h
av
io
r
[
2
0
]
.
Fu
r
th
e
r
,
a
u
th
o
r
s
p
r
o
p
o
s
ed
a
DNN
-
b
ased
m
o
d
el
o
f
f
er
in
g
a
s
ca
lab
le
a
n
d
ef
f
ec
tiv
e
f
r
a
m
ewo
r
k
f
o
r
d
etec
tin
g
DDo
S
attac
k
s
in
SDN
en
v
ir
o
n
m
e
n
ts
,
d
em
o
n
s
tr
atin
g
s
u
p
e
r
io
r
ac
cu
r
ac
y
an
d
r
eliab
ilit
y
ac
r
o
s
s
d
i
v
er
s
e
d
atasets
an
d
r
ea
l
-
wo
r
ld
co
n
d
itio
n
s
[
2
1
]
.
Ad
d
itio
n
ally
,
L
STM
a
n
d
h
y
b
r
id
C
NN
-
L
STM
ar
ch
itectu
r
es
h
av
e
b
ee
n
em
p
lo
y
e
d
to
d
esig
n
I
DS
,
p
a
r
ticu
lar
ly
u
s
in
g
th
e
C
I
C
-
DDo
S2
0
1
9
d
ataset
f
o
r
tr
ain
in
g
an
d
v
alid
atio
n
[
2
2
]
.
A
n
o
v
el
en
s
em
b
le
m
eth
o
d
ca
lled
SE
-
I
DS
co
m
b
in
es
d
ec
is
io
n
b
o
u
n
d
ar
ies
f
r
o
m
f
iv
e
t
r
ee
-
b
ased
class
if
ier
s
,
with
a
ML
P
s
er
v
in
g
as
th
e
m
eta
-
lear
n
er
f
o
r
f
in
al
class
if
icatio
n
[
2
3
]
.
An
o
th
er
n
o
tab
le
d
e
v
elo
p
m
e
n
t
in
clu
d
es
th
e
ca
s
ca
d
e
f
o
r
war
d
b
ac
k
p
r
o
p
a
g
atio
n
n
eu
r
a
l
n
etwo
r
k
(
C
FB
PNN)
,
wh
ich
u
tili
ze
s
a
r
ef
in
ed
s
u
b
s
et
o
f
f
ea
tu
r
es
s
elec
ted
u
s
in
g
co
r
r
elatio
n
-
b
ased
f
ea
tu
r
e
s
elec
tio
n
(
C
FS
)
.
T
h
is
m
o
d
el
h
as
b
ee
n
v
alid
ated
ac
r
o
s
s
m
u
ltip
le
d
atasets
[
2
4
]
.
Mo
r
eo
v
er
,
a
n
o
m
aly
-
b
ased
I
DS
f
o
r
I
o
T
-
en
a
b
led
SDN
en
v
ir
o
n
m
en
ts
h
av
e
also
b
ee
n
p
r
o
p
o
s
ed
,
u
tili
zin
g
C
NN
m
o
d
els
to
ex
am
in
e
tr
af
f
ic
p
atter
n
s
an
d
d
etec
t su
s
p
icio
u
s
b
eh
av
io
r
ef
f
ec
tiv
ely
[
2
5
]
.
T
h
is
s
tu
d
y
aim
s
to
c
o
m
p
ar
at
iv
ely
ass
ess
a
v
ar
iety
o
f
DL
m
o
d
els
f
o
r
th
eir
ab
ilit
y
t
o
d
etec
t
an
d
m
itig
ate
DDo
S
attac
k
s
in
SD
N
s
ettin
g
s
.
A
cr
itical
co
m
p
o
n
en
t
o
f
th
is
r
esear
ch
in
v
o
lv
es
o
p
tim
izin
g
f
ea
tu
r
e
s
elec
tio
n
f
r
o
m
n
etwo
r
k
tr
af
f
i
c
d
ata,
wh
ich
ca
n
s
ig
n
if
ican
tly
en
h
an
ce
th
e
ac
cu
r
ac
y
an
d
ef
f
icien
cy
o
f
th
ese
m
o
d
els wh
ile
m
in
im
izin
g
c
o
m
p
u
tatio
n
al
lo
ad
o
n
t
h
e
SDN
co
n
tr
o
ller
.
Su
m
m
a
r
y
o
f
k
e
y
co
n
t
r
ib
u
tio
n
s
:
i)
E
x
p
lo
r
ato
r
y
d
ata
an
aly
s
is
(
E
DA)
:
a
th
o
r
o
u
g
h
an
aly
s
is
was
co
n
d
u
cted
to
ex
a
m
in
e
all
d
ataset
f
ea
tu
r
es,
in
clu
d
in
g
t
h
eir
ty
p
es,
r
a
n
g
es,
d
is
tr
ib
u
tio
n
s
,
an
d
a
n
y
m
is
s
in
g
o
r
an
o
m
alo
u
s
v
alu
es.
ii)
Data
clea
n
in
g
:
m
is
s
in
g
v
alu
e
s
wer
e
h
an
d
led
u
s
in
g
tec
h
n
i
q
u
es
s
u
ch
as
im
p
u
tatio
n
o
r
r
o
w
ex
clu
s
io
n
,
d
ep
en
d
i
n
g
o
n
th
eir
im
p
ac
t o
n
t
h
e
an
aly
s
is
.
iii)
Descr
ip
tiv
e
s
tati
s
tics
:
m
ea
s
u
r
es
s
u
ch
as
m
ea
n
,
m
ed
ian
,
s
tan
d
ar
d
d
ev
iatio
n
,
an
d
in
ter
q
u
ar
ti
le
r
an
g
e
wer
e
ca
lcu
lated
to
u
n
d
er
s
tan
d
d
is
tr
ib
u
tio
n
p
atter
n
s
.
Fo
r
in
s
tan
ce
,
a
s
ig
n
if
ican
t
g
ap
b
etwe
en
m
ea
n
an
d
m
e
d
ian
p
ac
k
et
co
u
n
ts
co
u
ld
in
d
icate
s
k
ewn
ess
d
u
e
to
o
u
tlier
s
.
iv
)
Data
v
is
u
aliza
tio
n
:
‒
B
o
x
p
lo
ts
r
ev
ea
led
o
u
tlier
s
th
at
co
u
ld
s
ig
n
if
y
attac
k
in
s
tan
ce
s
.
‒
Scatter
p
lo
ts
h
ig
h
lig
h
ted
r
elatio
n
s
h
ip
s
b
etwe
en
co
n
tin
u
o
u
s
v
ar
iab
les.
‒
B
ar
ch
ar
ts
d
is
p
lay
ed
ca
teg
o
r
ical
d
is
tr
ib
u
tio
n
s
,
s
u
ch
as p
r
o
to
c
o
l ty
p
es a
n
d
p
o
r
t
u
s
ag
e.
v)
Ad
d
r
ess
in
g
class
im
b
alan
ce
:
t
h
e
s
y
n
th
etic
m
in
o
r
ity
o
v
er
-
s
a
m
p
lin
g
tech
n
iq
u
e
(
SMOT
E
)
t
ec
h
n
iq
u
e
was
ap
p
lied
to
s
y
n
th
etica
lly
b
alan
ce
u
n
d
e
r
r
ep
r
esen
ted
class
es,
th
er
eb
y
r
ed
u
cin
g
m
o
d
el
b
ias
t
o
war
d
m
aj
o
r
ity
class
in
s
tan
ce
s
.
v
i)
Featu
r
e
s
elec
tio
n
an
d
im
p
o
r
tan
ce
:
co
r
r
elatio
n
m
atr
ices
an
d
th
e
r
an
d
o
m
f
o
r
est
alg
o
r
ith
m
wer
e
em
p
lo
y
e
d
to
ass
ess
an
d
s
elec
t th
e
m
o
s
t i
n
f
lu
en
tial f
ea
tu
r
es f
o
r
tr
ai
n
in
g
DL
m
o
d
els.
v
ii)
DL
m
o
d
els
u
s
ed
:
v
ar
io
u
s
D
L
m
o
d
els
wer
e
test
ed
,
in
clu
d
in
g
ML
P,
ar
tific
ial
n
eu
r
al
n
e
two
r
k
(
ANN)
,
C
NN,
R
N
N,
an
d
L
STM
,
ea
ch
ch
o
s
en
f
o
r
th
eir
a
b
ilit
y
to
m
o
d
el
co
m
p
lex
p
atter
n
s
in
n
etwo
r
k
tr
af
f
ic.
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.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
4
9
8
2
-
4
9
9
2
4984
v
iii)
Mo
d
el
o
p
tim
izatio
n
an
d
v
alid
atio
n
:
ea
ch
m
o
d
el
u
n
d
er
wen
t
h
y
p
er
p
ar
am
eter
t
u
n
in
g
an
d
cr
o
s
s
-
v
alid
atio
n
to
en
s
u
r
e
g
e
n
er
aliza
tio
n
a
n
d
p
r
ev
en
t o
v
er
f
itti
n
g
.
ix
)
Per
f
o
r
m
an
ce
ev
alu
atio
n
:
t
h
e
m
o
d
els
wer
e
e
v
alu
ated
u
s
in
g
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
,
o
f
f
er
in
g
in
s
ig
h
ts
in
to
d
etec
tio
n
ef
f
icac
y
a
n
d
c
o
m
p
u
tatio
n
al
ef
f
icien
cy
.
x)
DDo
S
m
itig
atio
n
v
ia
g
r
ap
h
th
eo
r
y
:
t
h
e
SDN
co
n
tr
o
ller
p
er
i
o
d
ically
co
llects
f
lo
w
s
tati
s
tic
s
(
e.
g
.
,
ev
er
y
5
s
ec
o
n
d
s
)
f
r
o
m
Op
en
Flo
w
s
witch
es.
T
h
ese
ar
e
an
aly
z
ed
u
s
in
g
a
tr
ain
ed
DL
m
o
d
el
to
id
en
tify
s
u
s
p
icio
u
s
f
lo
ws.
W
h
en
attac
k
s
ar
e
d
etec
ted
,
th
e
co
n
tr
o
lle
r
tr
ig
g
e
r
s
m
itig
atio
n
s
tep
s
b
a
s
ed
o
n
g
r
a
p
h
th
eo
r
y
to
m
in
im
ize
d
is
r
u
p
tio
n
wh
ile
m
ain
tain
in
g
s
er
v
ice
q
u
a
lity
.
T
h
e
r
em
ain
i
n
g
s
ec
tio
n
s
o
f
th
e
p
ap
er
c
o
v
er
th
e
f
o
llo
win
g
to
p
i
cs:
s
ec
tio
n
2
d
etails
th
e
d
atase
ts
u
s
ed
in
ex
p
er
im
en
tatio
n
,
th
e
ar
ch
itect
u
r
e
an
d
m
et
h
o
d
o
lo
g
y
o
f
th
e
p
r
o
p
o
s
ed
d
etec
tio
n
s
y
s
tem
.
Sectio
n
3
d
is
cu
s
s
es
r
esu
lts
,
m
etr
ics,
an
d
co
m
p
ar
i
s
o
n
s
with
co
n
tem
p
o
r
ar
y
s
o
lu
t
io
n
s
.
Fin
ally
,
se
ctio
n
4
co
n
cl
u
d
es
th
e
p
ap
e
r
a
n
d
s
u
g
g
ests
d
ir
ec
tio
n
s
f
o
r
f
u
tu
r
e
r
esear
ch
.
2.
M
E
T
H
O
D
2
.
1
.
E
x
perim
ent
a
l
s
et
up
T
h
e
ex
p
er
im
en
tal
f
r
a
m
ewo
r
k
was
co
n
s
tr
u
cted
u
s
in
g
th
e
Min
in
et
n
etwo
r
k
em
u
lato
r
in
co
n
ju
n
ctio
n
with
th
e
R
YU
SDN
co
n
tr
o
lle
r
,
as
s
h
o
wn
in
Fig
u
r
e
1
.
T
h
e
s
im
u
latio
n
s
wer
e
ex
ec
u
ted
u
s
i
n
g
Min
in
et
v
e
r
s
io
n
2
.
3
.
2
,
wh
ich
p
r
o
v
id
es
r
o
b
u
s
t
s
u
p
p
o
r
t
f
o
r
O
p
en
v
Switch
(
OVS)
,
a
wid
ely
ad
o
p
ted
v
ir
tu
al
s
witch
co
m
p
atib
le
with
Op
en
Flo
w
p
r
o
to
co
ls
.
T
h
e
ex
p
er
im
en
ts
wer
e
r
u
n
o
n
a
s
y
s
tem
eq
u
ip
p
ed
with
an
I
n
tel
C
o
r
e
i7
-
5
5
0
0
U
p
r
o
ce
s
s
o
r
,
8
GB
o
f
R
AM
,
an
d
W
in
d
o
ws
1
0
as
th
e
o
p
er
ati
n
g
s
y
s
tem
.
DL
m
o
d
els
wer
e
d
ev
elo
p
e
d
u
s
in
g
th
e
Ker
as
lib
r
ar
y
in
Py
th
o
n
.
A
c
u
s
to
m
ized
SDN
to
p
o
lo
g
y
was
im
p
lem
en
ted
,
f
e
atu
r
in
g
a
h
ier
ar
ch
ical
tr
ee
-
b
ased
s
tr
u
ctu
r
e
m
a
n
ag
ed
b
y
a
ce
n
tr
a
lized
co
n
tr
o
ller
.
T
h
is
to
p
o
lo
g
y
in
teg
r
ated
s
ev
e
n
Op
e
n
Flo
w
s
witch
es,
with
o
n
e
s
witch
(
S1
)
co
n
n
ec
ted
to
s
ix
h
o
s
ts
,
an
d
th
e
r
e
m
ain
in
g
s
witch
es
co
n
n
ec
ted
t
o
th
r
ee
h
o
s
ts
ea
ch
.
T
h
e
n
etwo
r
k
ar
ch
itectu
r
e
co
n
s
is
ted
o
f
th
r
ee
lay
er
s
:
th
e
co
n
tr
o
l
p
lan
e
(
h
o
u
s
in
g
th
e
SDN
co
n
tr
o
ll
er
)
,
th
e
d
ata
p
lan
e
(
co
n
tain
in
g
h
o
s
ts
an
d
s
witch
e
s
)
,
an
d
th
e
ap
p
licatio
n
lay
er
,
wh
ich
in
clu
d
ed
f
o
u
r
co
m
p
o
n
e
n
ts
:
f
lo
w
s
tat
is
tic
s
co
llectio
n
,
f
ea
tu
r
e
ex
tr
ac
tio
n
,
a
DL
class
if
ier
,
an
d
an
attac
k
m
itig
atio
n
m
o
d
u
le.
Af
ter
d
ep
l
o
y
in
g
t
h
e
to
p
o
lo
g
y
,
co
n
n
ec
tiv
ity
was
v
alid
ated
u
s
in
g
th
e
p
i
n
g
c
o
m
m
an
d
ac
r
o
s
s
a
ll
h
o
s
ts
.
T
r
af
f
ic
f
lo
ws
wer
e
g
e
n
er
ated
u
s
in
g
T
C
P
with
p
ac
k
et
s
izes f
ix
ed
at
5
1
2
b
y
tes an
d
f
lo
w
d
u
r
atio
n
s
s
et
to
1
2
,
0
0
0
m
illi
s
ec
o
n
d
s
.
T
h
e
p
ac
k
et
co
u
n
t
f
o
r
f
lo
ws
r
an
g
ed
b
etwe
en
5
0
0
an
d
1
0
0
0
,
f
o
r
m
in
g
th
e
b
asis
f
o
r
th
e
d
ata
s
et
u
s
ed
in
tr
ain
in
g
a
n
d
ev
al
u
a
tio
n
.
Fig
u
r
e
1
.
A
s
ch
em
atic
d
iag
r
a
m
o
f
SDN
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
Dee
p
lea
r
n
in
g
-
b
a
s
ed
ev
a
lu
a
tio
n
fo
r
d
is
tr
ib
u
ted
d
en
i
a
l o
f ser
vice
a
tta
ck
s
d
etec
tio
n
(
N
ee
th
u
S
.
)
4985
T
h
e
d
ataset
em
p
lo
y
ed
in
th
i
s
s
tu
d
y
co
m
p
r
is
es
1
0
4
,
3
4
6
en
tr
ies
d
is
tr
ib
u
ted
ac
r
o
s
s
2
3
co
lu
m
n
s
,
en
co
m
p
ass
in
g
b
o
th
f
ea
tu
r
es a
n
d
tar
g
et
lab
els.
T
h
ese
co
lu
m
n
s
f
all
in
to
th
e
f
o
llo
win
g
ca
teg
o
r
ies:
‒
Featu
r
e
attr
ib
u
tes:
th
ese
r
ep
r
e
s
en
t
v
ar
io
u
s
n
etwo
r
k
tr
af
f
ic
in
d
icato
r
s
u
s
ed
to
d
if
f
er
en
tiate
b
etwe
en
b
en
ig
n
an
d
m
alicio
u
s
(
DDo
S)
ac
tiv
i
ties
.
No
tab
le
f
ea
tu
r
es
in
clu
d
e
p
ac
k
et
s
ize,
tr
an
s
m
is
s
io
n
r
ate,
b
y
te
co
u
n
t,
s
o
u
r
ce
an
d
d
esti
n
atio
n
I
P a
d
d
r
ess
es,
an
d
s
ev
er
al
o
th
er
s
tatis
ti
ca
l m
ea
s
u
r
es r
elev
an
t to
tr
a
f
f
i
c
f
lo
w
an
aly
s
is
.
‒
L
ab
el
co
lu
m
n
:
th
is
s
er
v
es
as
t
h
e
class
if
icatio
n
tar
g
et,
id
en
tify
in
g
wh
eth
er
ea
ch
d
ata
in
s
tan
ce
co
r
r
esp
o
n
d
s
to
leg
itima
te
tr
af
f
ic
o
r
a
DDo
S
attac
k
.
T
h
is
lab
el
is
cr
it
ical
f
o
r
s
u
p
er
v
is
ed
ML
,
allo
win
g
th
e
m
o
d
els
to
lear
n
f
r
o
m
h
is
to
r
ical
ex
am
p
les.
‒
Data
p
ar
titi
o
n
in
g
:
t
o
f
ac
ilit
ate
DL
,
th
e
d
ataset
is
s
p
lit
in
to
th
r
ee
s
u
b
s
ets.
Sev
en
ty
-
f
iv
e
p
er
ce
n
t
(
7
5
%)
o
f
th
e
r
ec
o
r
d
s
ar
e
allo
ca
ted
f
o
r
tr
ain
i
n
g
,
wh
ile
th
e
r
em
ain
in
g
2
5
%
ar
e
d
iv
id
e
d
b
etwe
en
v
alid
atio
n
an
d
test
in
g
to
ass
es
s
m
o
d
el
g
en
er
aliza
b
ilit
y
an
d
p
er
f
o
r
m
a
n
ce
.
I
t
is
wo
r
th
n
o
tin
g
th
at
m
an
y
o
f
th
e
tr
af
f
ic
-
r
elate
d
f
ea
t
u
r
es,
s
u
ch
as
p
ac
k
et
s
ize,
r
ate,
an
d
s
o
u
r
ce
/d
esti
n
atio
n
ad
d
r
ess
es,
ar
e
in
s
tr
u
m
en
tal
in
ca
p
tu
r
in
g
t
h
e
u
n
i
q
u
e
p
atter
n
s
ass
o
ciate
d
with
DDo
S b
eh
av
io
r
s
.
2
.
2
.
E
x
plo
ra
t
o
ry
da
t
a
a
na
l
y
s
is
T
h
e
p
r
elim
in
a
r
y
an
aly
s
is
o
f
th
e
d
ataset
in
v
o
l
v
es
b
o
th
s
tatis
tica
l
s
u
m
m
ar
izatio
n
a
n
d
f
ea
tu
r
e
ex
am
in
atio
n
.
T
h
e
s
u
m
m
ar
y
s
t
atis
tics
f
o
r
r
aw
d
ata
ar
e
p
r
esen
ted
in
T
ab
le
1
,
wh
ile
T
ab
le
2
lis
ts
th
e
s
elec
ted
f
ea
tu
r
es
u
s
ed
in
m
o
d
el
d
e
v
e
lo
p
m
en
t
af
ter
f
ea
tu
r
e
en
g
i
n
e
er
in
g
.
T
h
e
d
ataset
co
n
tain
s
8
,
7
1
4
r
ec
o
r
d
s
,
ea
ch
co
m
p
r
is
in
g
m
u
ltip
le
v
a
r
iab
les
th
at
p
r
o
v
id
e
a
d
etailed
v
iew
o
f
n
etwo
r
k
tr
af
f
ic
b
eh
av
i
o
r
.
Am
o
n
g
th
ese,
th
e
'
s
witch
'
attr
ib
u
te
lik
ely
d
en
o
tes
id
en
tifie
r
s
f
o
r
n
etwo
r
k
s
witch
es.
I
ts
m
ea
n
v
alu
e
o
f
1
1
,
0
8
9
.
8
3
m
ay
s
u
g
g
est
th
at
th
e
d
ata
is
eith
er
en
co
d
ed
o
r
s
ca
led
,
as
s
witch
id
en
tifie
r
s
ar
e
ty
p
ically
ca
teg
o
r
ic
al.
T
h
e
v
a
r
iab
les
'
p
k
tco
u
n
t'
an
d
'
b
y
teco
u
n
t'
r
ep
r
esen
t
th
e
n
u
m
b
er
o
f
p
ac
k
ets
an
d
to
tal
b
y
tes
p
e
r
ev
e
n
t,
r
esp
ec
tiv
ely
.
On
a
v
er
ag
e
,
ea
ch
ev
en
t
h
as
2
.
3
2
p
ac
k
ets,
s
u
g
g
esti
n
g
co
n
s
is
ten
t
p
ac
k
et
f
l
o
w
ac
r
o
s
s
r
ec
o
r
d
s
.
C
o
n
v
er
s
ely
,
th
e
av
e
r
ag
e
b
y
te
co
u
n
t
—
ap
p
r
o
x
im
ately
7
9
m
illi
o
n
b
y
tes
—
r
ef
lects
a
b
r
o
ad
er
v
ar
iatio
n
in
tr
af
f
ic
s
ize.
T
im
e
-
b
ased
f
ea
tu
r
es
s
u
ch
as
'
d
u
r
'
,
'
d
u
r
_
n
s
ec
'
,
an
d
'
to
t_
d
u
r
'
p
r
o
v
id
e
e
v
en
t
d
u
r
atio
n
s
in
s
ec
o
n
d
s
,
n
an
o
s
ec
o
n
d
s
,
a
n
d
as
an
ag
g
r
eg
ate
m
ea
s
u
r
e.
T
h
ese
in
d
icate
a
b
r
o
ad
s
p
ec
tr
u
m
o
f
tr
af
f
ic
d
u
r
atio
n
s
.
T
h
e
'
f
lo
ws
'
v
ar
iab
le,
av
er
ag
in
g
3
.
3
2
p
er
ev
e
n
t,
h
in
ts
at
m
u
ltip
le
co
n
c
u
r
r
en
t c
o
m
m
u
n
icatio
n
s
ess
io
n
s
.
T
ab
le
1
.
Data
s
et
p
k
t
c
o
u
n
t
b
y
t
e
c
o
u
n
t
dur
d
u
r
_
n
sec
t
o
t
_
d
u
r
p
a
c
k
e
t
i
n
s
p
k
t
p
e
r
f
l
o
w
c
o
u
n
t
1
4
.
0
0
8
,
7
1
4
.
0
0
8
,
7
1
4
.
0
0
8
,
7
1
4
.
0
0
8
,
7
1
4
.
0
0
8
,
7
1
4
.
0
0
8
,
7
1
4
.
0
0
mea
n
7
.
4
7
E+
0
4
7
.
9
1
E+
0
7
1
9
4
.
0
1
5
.
4
8
E+
0
8
1
.
9
5
E+
1
1
1
,
9
2
0
.
4
2
1
.
0
5
E+
0
4
st
d
4
.
0
7
E+
0
4
4
.
3
1
E+
0
7
1
1
7
.
3
5
2
.
3
2
E+
0
8
1
.
1
7
E+
1
1
2
5
4
.
1
3
4
.
0
5
E+
0
3
mi
n
2
8
4
.
0
0
3
.
0
3
E+
0
5
-
7
.
9
0
E+
0
7
8
.
3
7
E+
0
8
5
5
8
.
0
0
0
.
0
0
E+
0
0
2
5
%
3
.
7
3
E+
0
4
3
.
9
6
E+
0
7
1
0
0
.
0
0
3
.
9
1
E+
0
8
1
.
0
1
E+
1
1
1
,
9
3
1
.
0
0
8
.
6
4
E+
0
3
5
0
%
7
.
6
6
E+
0
4
8
.
1
7
E+
0
7
1
9
0
.
0
0
5
.
5
6
E+
0
8
1
.
9
1
E+
1
1
1
,
9
4
3
.
0
0
1
.
3
4
E+
0
4
7
5
%
1
.
1
3
E+
0
5
1
.
2
0
E+
0
8
2
8
0
.
0
0
7
.
2
6
E+
0
8
2
.
8
1
E+
1
1
1
,
9
4
3
.
0
0
1
.
3
5
E+
0
4
max
1
.
3
5
E+
0
5
1
.
4
4
E+
0
8
4
7
3
.
0
0
9
.
1
4
E+
0
8
4
.
7
3
E+
1
1
2
,
2
4
2
.
0
0
1
.
3
7
E+
0
4
T
ab
le
2
.
Featu
r
e
s
elec
tio
n
i
n
a
d
ataset
b
y
t
e
p
e
r
f
l
o
w
p
k
t
r
a
t
e
t
x
_
b
y
t
e
s
r
x
_
b
y
t
e
s
t
x
_
k
b
p
s
r
x
_
k
b
p
s
t
o
t
_
k
b
p
s
c
o
u
n
t
8
,
7
1
4
.
0
0
8
,
7
1
4
.
0
0
8
,
7
1
4
.
0
0
8
,
7
1
4
.
0
0
8
,
7
1
4
.
0
0
8
,
7
1
4
.
0
0
8
,
7
1
4
.
0
0
mea
n
1
.
1
1
E+
0
7
3
4
9
.
8
6
4
.
8
2
E+
0
7
4
.
8
1
E+
0
7
8
7
3
.
4
3
8
7
3
.
1
1
1
,
7
4
6
.
5
4
st
d
4
.
3
5
E+
0
6
1
3
4
.
9
1
1
.
5
1
E+
0
8
1
.
1
0
E+
0
8
2
,
8
4
8
.
9
7
2
,
2
6
9
.
0
7
3
,
4
2
6
.
3
6
mi
n
0
.
0
0
E+
0
0
-
2
.
8
5
E+
0
3
9
.
2
6
E+
0
2
-
-
2
5
%
9
.
2
1
E+
0
6
2
8
8
.
0
0
3
.
5
9
E+
0
3
1
.
4
7
E+
0
3
-
-
5
0
%
1
.
4
3
E+
0
7
4
4
6
.
0
0
3
.
8
4
E+
0
3
3
.
5
4
E+
0
3
-
-
7
5
%
1
.
4
4
E+
0
7
4
5
1
.
0
0
4
.
2
5
E+
0
3
6
.
1
6
E+
0
6
-
-
2
.
5
7
E+
0
3
max
1
.
4
6
E+
0
7
4
5
6
.
0
0
1
.
2
7
E+
0
9
9
.
9
1
E+
0
8
2
.
0
6
E+
0
4
1
.
6
6
E+
0
4
2
.
0
6
E+
0
4
T
h
e
'
p
ac
k
etin
s
'
f
ea
tu
r
e,
p
o
s
s
ib
ly
m
ea
s
u
r
in
g
in
co
m
in
g
p
a
ck
ets,
s
h
o
ws
an
av
er
ag
e
o
f
1
,
9
2
0
.
4
2
,
s
u
g
g
esti
n
g
s
ig
n
if
ican
t
v
ar
iab
il
ity
in
p
ac
k
et
r
ec
ep
tio
n
r
ates.
Me
tr
ics
lik
e
'
p
k
tp
er
f
lo
w'
an
d
'b
y
tep
er
f
l
o
w'
ca
p
tu
r
e
p
er
-
f
lo
w
tr
a
n
s
m
is
s
io
n
s
tatis
tic
s
,
r
ev
ea
lin
g
d
iv
er
s
e
p
atter
n
s
t
h
r
o
u
g
h
th
eir
h
ig
h
s
tan
d
ar
d
d
e
v
iatio
n
s
,
cr
itical
f
o
r
id
en
tify
in
g
an
o
m
alo
u
s
b
eh
av
i
o
r
s
.
Ad
d
itio
n
ally
,
'
p
k
tr
ate'
ex
h
ib
its
a
m
ea
n
v
alu
e
o
f
3
4
9
.
8
6
,
with
wid
e
v
a
r
iatio
n
,
p
o
ten
tially
r
ef
lectin
g
b
o
th
i
d
le
an
d
ac
tiv
e
co
m
m
u
n
icatio
n
p
er
io
d
s
in
th
e
tr
af
f
ic
lo
g
s
.
T
o
en
h
an
ce
m
o
d
el
p
er
f
o
r
m
an
ce
,
f
ea
tu
r
e
en
g
i
n
ee
r
in
g
was
em
p
l
o
y
ed
—
an
ess
en
tial
s
tep
in
v
o
lv
i
n
g
d
o
m
ain
-
d
r
iv
en
tr
an
s
f
o
r
m
atio
n
s
to
h
ig
h
li
g
h
t
m
ea
n
in
g
f
u
l
p
atte
r
n
s
in
th
e
d
ata.
L
in
ea
r
ass
o
ciatio
n
s
b
etwe
en
f
ea
tu
r
es
wer
e
u
n
co
v
er
ed
th
r
o
u
g
h
co
r
r
elatio
n
an
al
y
s
is
,
wh
ile
r
an
d
o
m
f
o
r
est
was
u
s
ed
to
co
m
p
u
te
f
ea
tu
r
e
im
p
o
r
ta
n
ce
,
f
ac
ilit
atin
g
d
im
en
s
io
n
ality
r
ed
u
ctio
n
a
n
d
d
ir
ec
tin
g
m
o
d
el
atten
tio
n
to
th
e
m
o
s
t in
f
o
r
m
ativ
e
v
ar
iab
les.
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.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
4
9
8
2
-
4
9
9
2
4986
Giv
en
th
e
ty
p
ical
im
b
alan
ce
in
DDo
S
-
r
elate
d
d
atasets
,
t
h
e
SMOT
E
was
ap
p
lied
.
T
h
is
m
eth
o
d
g
en
er
ates
s
y
n
th
etic
s
am
p
les
o
f
u
n
d
er
r
ep
r
esen
ted
attac
k
tr
af
f
ic,
th
u
s
en
s
u
r
i
n
g
th
at
th
e
class
if
ier
d
o
es
n
o
t
b
ec
o
m
e
b
iased
to
war
d
th
e
d
o
m
in
an
t
(
b
e
n
ig
n
)
class
.
A
d
iv
er
s
e
s
et
o
f
DL
m
o
d
els
wa
s
th
en
ch
o
s
en
f
o
r
ev
alu
atio
n
,
r
an
g
i
n
g
f
r
o
m
f
u
lly
co
n
n
ec
ted
n
etwo
r
k
s
to
ar
c
h
itectu
r
es
tailo
r
ed
f
o
r
s
eq
u
en
ti
al
d
ata.
T
h
is
v
ar
iety
was
k
ey
to
ca
p
tu
r
i
n
g
d
if
f
er
e
n
t
asp
ec
ts
o
f
th
e
tr
af
f
ic
p
atter
n
s
.
Ov
er
all,
th
e
en
tire
p
ip
elin
e
—
f
r
o
m
ex
p
lo
r
ato
r
y
an
aly
s
is
an
d
p
r
e
p
r
o
ce
s
s
in
g
to
m
o
d
el
s
elec
tio
n
—
was
d
esig
n
ed
to
en
s
u
r
e
a
r
o
b
u
s
t,
co
m
p
r
e
h
en
s
iv
e
ap
p
r
o
ac
h
to
DDo
S a
ttack
d
etec
tio
n
,
with
a
n
em
p
h
asis
o
n
r
eliab
ilit
y
an
d
p
r
ed
ictiv
e
ef
f
ec
ti
v
en
ess
.
2
.
3
.
M
o
del
dev
elo
pm
ent
A
r
an
g
e
o
f
well
-
estab
lis
h
ed
DL
m
o
d
els is
u
tili
ze
d
in
th
is
s
tu
d
y
,
ea
ch
tr
ain
e
d
o
n
th
e
d
ataset
g
en
er
ated
f
r
o
m
SDN
-
b
ased
s
im
u
latio
n
s
.
T
h
ese
m
o
d
els
ar
e
p
ar
ticu
lar
ly
ef
f
ec
tiv
e
in
id
en
tify
in
g
in
tr
icate
an
d
n
o
n
-
lin
ea
r
d
ata
p
atter
n
s
—
ca
p
ab
ilit
ies th
at
o
f
ten
s
u
r
p
ass
th
o
s
e
o
f
c
o
n
v
e
n
tio
n
al
ML
alg
o
r
ith
m
s
.
T
h
e
o
v
er
all
wo
r
k
f
l
o
w
an
d
m
eth
o
d
o
l
o
g
y
f
o
r
d
etec
tin
g
an
d
m
itig
atin
g
DDo
S
attac
k
s
u
s
in
g
th
ese
DL
m
o
d
els
ar
e
illu
s
tr
ated
in
Fig
u
r
e
2
.
Var
io
u
s
DL
m
o
d
els u
s
ed
f
o
r
c
lass
if
icatio
n
ar
e
d
escr
ib
ed
in
t
h
e
f
o
llo
win
g
s
u
b
-
s
ec
tio
n
.
Fig
u
r
e
2
.
Sch
em
atic
d
iag
r
am
o
f
DDo
S
attac
k
d
etec
tio
n
a
n
d
m
itig
atio
n
u
s
in
g
DL
m
o
d
els
2
.
3
.
1
.
T
he
m
ulti
-
la
y
er
perc
e
ptr
o
n c
la
s
s
if
ier
ML
P
c
lass
if
ier
is
a
ty
p
e
o
f
f
ee
d
-
f
o
r
war
d
ANN
c
o
m
p
o
s
ed
o
f
s
ev
er
al
lay
er
s
o
f
n
o
d
es,
ea
c
h
a
p
p
ly
in
g
a
n
o
n
lin
ea
r
ac
tiv
atio
n
f
u
n
ctio
n
.
T
h
e
s
p
ec
if
ic
ML
P
c
lass
if
ier
in
q
u
esti
o
n
is
s
et
u
p
with
t
wo
h
id
d
e
n
lay
er
s
co
n
tain
in
g
1
0
0
an
d
5
0
n
o
d
es,
r
esp
ec
tiv
ely
.
T
h
is
d
esig
n
aim
s
to
id
en
tify
co
m
p
lex
p
atter
n
s
in
th
e
d
ata
t
h
r
o
u
g
h
m
u
ltip
le
lev
els
o
f
ab
s
tr
ac
tio
n
.
T
h
e
p
ar
am
eter
m
ax
_
iter
=1
0
0
0
allo
ws
th
e
m
o
d
el
u
p
to
1
0
0
0
iter
atio
n
s
to
co
n
v
er
g
e
o
n
a
s
o
lu
tio
n
,
u
n
les
s
it
m
ee
ts
a
s
to
p
p
in
g
cr
iter
io
n
ea
r
lier
.
T
h
is
ex
ten
s
iv
e
n
u
m
b
er
o
f
iter
atio
n
s
is
ad
v
an
tag
e
o
u
s
f
o
r
i
n
tr
icate
d
a
tasets
wh
er
e
th
e
r
elatio
n
s
h
ip
s
b
etwe
en
in
p
u
ts
an
d
o
u
t
p
u
ts
ar
e
ch
allen
g
in
g
to
m
o
d
el.
Usi
n
g
a
r
a
n
d
o
m
_
s
tate
en
s
u
r
es
th
e
r
esu
lts
ar
e
r
ep
r
o
d
u
cib
le
b
y
f
i
x
in
g
th
e
s
ee
d
f
o
r
t
h
e
r
an
d
o
m
n
u
m
b
er
g
en
er
ato
r
u
s
ed
in
in
itializin
g
weig
h
ts
.
T
r
ain
in
g
th
e
ML
P
o
n
r
esam
p
led
d
ata,
lik
ely
ad
j
u
s
ted
to
co
r
r
ec
t
class
im
b
alan
ce
s
th
r
o
u
g
h
tec
h
n
iq
u
e
s
lik
e
SMOT
E
,
h
elp
s
th
e
class
if
ier
p
er
f
o
r
m
well
f
o
r
b
o
th
m
i
n
o
r
ity
a
n
d
m
aj
o
r
ity
class
es.
T
h
e
ML
P
'
s
ca
p
ab
ilit
y
to
ca
p
tu
r
e
n
o
n
-
lin
ea
r
r
elatio
n
s
h
ip
s
m
ak
es
it
e
s
p
ec
ially
ef
f
ec
tiv
e
f
o
r
task
s
s
u
ch
as
p
r
ed
ictin
g
DDo
S
attac
k
s
in
SDN
en
v
ir
o
n
m
en
ts
,
wh
e
r
e
attac
k
p
atter
n
s
m
ig
h
t
b
e
s
u
b
t
le
an
d
n
o
t
lin
ea
r
ly
s
ep
ar
ab
le.
2
.
3
.
2
.
Art
if
icia
l
neura
l net
wo
rk
mo
del
T
h
e
ANN
m
o
d
el,
i
m
p
lem
en
te
d
with
T
e
n
s
o
r
Flo
w'
s
Ker
as,
f
ea
tu
r
es
a
s
eq
u
e
n
tial
ar
c
h
itectu
r
e
with
two
lay
er
s
.
T
h
e
f
ir
s
t
lay
er
is
a
d
en
s
e
lay
er
with
6
4
n
eu
r
o
n
s
u
s
in
g
th
e
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
eL
U
)
ac
tiv
atio
n
f
u
n
ctio
n
,
aid
in
g
in
n
o
n
-
lin
ea
r
ity
an
d
m
itig
atin
g
th
e
v
an
is
h
in
g
g
r
a
d
ien
t
p
r
o
b
lem
.
T
h
e
o
u
tp
u
t
lay
er
h
as
o
n
e
n
eu
r
o
n
with
a
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
ctio
n
,
s
u
itab
le
f
o
r
b
in
ar
y
class
if
icatio
n
task
s
li
k
e
d
etec
tin
g
DDo
S
attac
k
s
.
T
h
e
m
o
d
el
em
p
lo
y
s
b
in
ar
y
_
cr
o
s
s
en
tr
o
p
y
as
th
e
lo
s
s
f
u
n
ctio
n
an
d
t
h
e
Ad
am
o
p
tim
izer
.
An
E
ar
ly
Sto
p
p
in
g
ca
llb
ac
k
with
a
p
atie
n
ce
o
f
1
0
e
p
o
ch
s
an
d
r
esto
r
e
_
b
est_
weig
h
ts
o
p
tio
n
h
elp
s
p
r
ev
e
n
t
o
v
er
f
itti
n
g
an
d
en
s
u
r
es
t
h
e
m
o
d
el
g
e
n
er
alize
s
well.
T
h
e
m
o
d
el
is
tr
ain
ed
o
n
r
esam
p
le
d
d
ata
to
a
d
d
r
ess
class
im
b
alan
ce
.
2
.
3
.
3
.
Co
nv
o
lutio
na
l
neura
l
net
wo
rk
mo
del
C
NN
is
tailo
r
ed
f
o
r
o
n
e
-
d
im
en
s
io
n
al
s
eq
u
en
ce
d
ata,
s
u
ch
as
tim
e
s
er
ies
o
r
n
etwo
r
k
tr
af
f
ic
f
lo
w
an
aly
s
is
.
T
h
e
ar
ch
itectu
r
e
in
cl
u
d
es
co
n
v
o
lu
tio
n
al
la
y
er
s
with
3
2
an
d
6
4
f
ilter
s
an
d
a
k
er
n
el
s
ize
o
f
3
,
w
h
ich
ex
tr
ac
t
h
ig
h
-
lev
el
f
ea
tu
r
es
b
y
ap
p
ly
i
n
g
f
ilter
s
ac
r
o
s
s
th
e
i
n
p
u
t
d
ata
to
ca
p
tu
r
e
lo
ca
l
d
e
p
en
d
en
cies
with
in
s
eq
u
en
ce
s
.
E
ac
h
co
n
v
o
lu
tio
n
al
lay
er
is
f
o
llo
wed
b
y
a
M
ax
Po
o
lin
g
lay
er
with
a
p
o
o
l
s
ize
o
f
2
,
wh
ich
d
o
wn
s
am
p
les
th
e
in
p
u
t
r
ep
r
esen
tatio
n
,
r
ed
u
ce
s
d
im
en
s
io
n
ality
,
an
d
en
h
a
n
ce
s
m
o
d
e
l
p
er
f
o
r
m
an
ce
b
y
in
tr
o
d
u
cin
g
tr
an
s
latio
n
al
in
v
a
r
ian
ce
.
Fo
llo
win
g
th
e
co
n
v
o
l
u
tio
n
al
an
d
p
o
o
lin
g
o
p
er
atio
n
s
,
th
e
f
ea
tu
r
e
m
ap
s
ar
e
f
latten
ed
i
n
to
a
o
n
e
-
d
im
e
n
s
io
n
al
ar
r
ay
,
en
a
b
lin
g
in
te
g
r
atio
n
with
f
u
lly
co
n
n
ec
ted
(
d
en
s
e)
lay
er
s
.
T
h
ese
th
ick
lay
er
s
p
er
f
o
r
m
ad
d
itio
n
al
tr
an
s
f
o
r
m
atio
n
s
b
ef
o
r
e
p
r
o
d
u
cin
g
th
e
f
in
al
class
if
icatio
n
o
u
tp
u
t.
A
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
ctio
n
is
ap
p
lied
in
th
e
o
u
tp
u
t
lay
er
to
s
u
p
p
o
r
t
b
in
ar
y
class
if
icatio
n
task
s
,
s
u
ch
as
d
is
tin
g
u
is
h
in
g
b
etwe
en
r
eg
u
lar
an
d
DDo
S
t
r
af
f
ic.
T
h
e
m
o
d
el
is
co
m
p
ile
d
u
s
in
g
th
e
b
in
ar
y
cr
o
s
s
-
en
tr
o
p
y
lo
s
s
f
u
n
ctio
n
,
p
air
ed
with
th
e
Ad
am
o
p
t
i
m
izer
to
en
s
u
r
e
e
f
f
icien
t
an
d
ad
ap
tiv
e
g
r
ad
ien
t
u
p
d
ates
d
u
r
in
g
tr
ain
in
g
.
T
o
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
Dee
p
lea
r
n
in
g
-
b
a
s
ed
ev
a
lu
a
tio
n
fo
r
d
is
tr
ib
u
ted
d
en
i
a
l o
f ser
vice
a
tta
ck
s
d
etec
tio
n
(
N
ee
th
u
S
.
)
4987
s
af
eg
u
ar
d
ag
ain
s
t
o
v
er
f
itti
n
g
,
an
ea
r
ly
s
to
p
p
in
g
m
ec
h
an
is
m
is
im
p
lem
en
ted
,
c
o
n
f
i
g
u
r
ed
w
ith
a
p
atien
ce
o
f
1
0
ep
o
ch
s
,
allo
win
g
tr
ai
n
in
g
to
h
alt
if
n
o
p
er
f
o
r
m
an
ce
im
p
r
o
v
em
e
n
t
is
o
b
s
er
v
ed
,
th
u
s
p
r
o
m
o
tin
g
b
etter
g
en
er
aliza
tio
n
to
u
n
s
ee
n
d
ata.
2
.
3
.
4
.
Rec
urre
nt
neura
l net
wo
rk
m
o
del
R
NN
m
o
d
el
with
Simp
leR
NN
lay
er
s
ex
ce
ls
at
h
an
d
lin
g
s
eq
u
en
ce
s
wh
er
e
cu
r
r
en
t
o
u
t
p
u
ts
d
ep
en
d
o
n
p
r
ev
io
u
s
c
o
m
p
u
tatio
n
s
,
m
ak
i
n
g
it
s
u
itab
le
f
o
r
an
aly
zin
g
s
eq
u
en
tial
d
ata
lik
e
n
etwo
r
k
tr
af
f
ic.
T
h
e
m
o
d
el
f
ea
tu
r
es
a
Simp
leR
NN
lay
er
with
5
0
u
n
its
,
ca
p
ab
le
o
f
ca
p
t
u
r
in
g
tem
p
o
r
al
d
y
n
am
ics
b
u
t
p
o
ten
tially
s
tr
u
g
g
lin
g
with
lo
n
g
-
ter
m
d
ep
e
n
d
en
cies
d
u
e
to
th
e
v
an
is
h
in
g
g
r
a
d
ien
t
p
r
o
b
lem
in
h
er
en
t
in
b
asic
R
N
Ns.
T
h
e
R
N
N
lay
er
u
tili
ze
s
th
e
R
eL
U
ac
tiv
atio
n
f
u
n
ctio
n
,
e
n
ab
lin
g
t
h
e
n
etwo
r
k
to
ca
p
tu
r
e
n
o
n
lin
ea
r
r
elatio
n
s
h
ip
s
an
d
ex
tr
ac
t
co
m
p
lex
te
m
p
o
r
al
f
ea
tu
r
es
f
r
o
m
s
eq
u
en
tial
d
ata.
Fo
r
b
in
ar
y
class
if
icatio
n
p
u
r
p
o
s
es,
th
e
o
u
tp
u
t
lay
er
em
p
lo
y
s
a
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
ctio
n
,
ef
f
ec
tiv
ely
m
ap
p
i
n
g
t
h
e
o
u
tp
u
t
to
a
p
r
o
b
ab
ilit
y
s
co
r
e
.
T
h
e
m
o
d
el
is
co
m
p
iled
u
s
in
g
th
e
Ad
am
o
p
tim
izer
,
k
n
o
wn
f
o
r
its
ad
ap
tiv
e
lear
n
i
n
g
ca
p
ab
ilit
ies,
alo
n
g
with
th
e
b
in
ar
y
cr
o
s
s
-
en
tr
o
p
y
lo
s
s
f
u
n
ctio
n
,
wh
ich
is
a
s
tan
d
ar
d
ch
o
ice
f
o
r
h
an
d
lin
g
b
in
a
r
y
class
if
icatio
n
p
r
o
b
lem
s
.
E
a
r
ly
s
to
p
p
in
g
is
u
s
e
d
d
u
r
in
g
tr
ai
n
in
g
,
m
o
n
ito
r
in
g
v
alid
atio
n
ac
cu
r
ac
y
,
an
d
h
alt
in
g
th
e
p
r
o
ce
s
s
if
n
o
im
p
r
o
v
em
en
t
o
cc
u
r
s
o
v
er
s
ev
er
al
ep
o
ch
s
,
p
r
e
v
en
tin
g
o
v
er
f
itti
n
g
an
d
en
s
u
r
in
g
th
e
m
o
d
el
g
en
e
r
alize
s
well.
T
h
e
R
NN
ar
ch
itectu
r
e
is
well
-
s
u
ited
f
o
r
r
ea
l
-
tim
e
s
tr
ea
m
in
g
d
ata,
wh
e
r
e
r
ec
en
t
d
ata
p
o
in
ts
ar
e
c
r
itical
f
o
r
p
r
ed
ict
io
n
s
.
Ho
wev
er
,
f
o
r
v
er
y
lo
n
g
s
eq
u
en
ce
s
o
r
d
is
p
er
s
ed
im
p
o
r
tan
t
in
f
o
r
m
atio
n
,
L
STM
o
r
GR
U
m
o
d
els
m
ay
b
e
m
o
r
e
ef
f
ec
tiv
e
d
u
e
to
th
eir
ad
v
a
n
ce
d
g
atin
g
m
ec
h
an
is
m
s
.
2
.
3
.
5
.
L
o
ng
s
ho
rt
-
t
er
m m
e
mo
ry
mo
del
T
h
e
L
STM
m
o
d
el,
a
n
ad
v
a
n
ce
d
R
NN
ar
ch
itectu
r
e,
ex
c
els
at
lear
n
in
g
l
o
n
g
-
te
r
m
d
e
p
en
d
en
cies,
cr
u
cial
f
o
r
s
eq
u
en
tial
d
ata
w
ith
im
p
o
r
tan
t
tem
p
o
r
al
f
ea
t
u
r
es.
I
t
in
clu
d
es
an
L
STM
lay
er
with
5
0
u
n
its
,
allo
win
g
it
to
r
etain
i
n
f
o
r
m
atio
n
o
v
er
ex
ten
d
e
d
p
er
io
d
s
,
wh
i
ch
is
ess
en
tial
f
o
r
n
etwo
r
k
tr
a
f
f
ic
s
eq
u
en
ce
s
.
T
h
e
in
p
u
t
s
h
ap
e
m
atch
es
th
e
r
esh
a
p
ed
tr
ain
i
n
g
d
ata,
p
r
esen
tin
g
n
etwo
r
k
tr
a
f
f
ic
as
a
s
eq
u
en
ce
.
L
STM
av
o
i
d
s
th
e
v
an
is
h
in
g
g
r
a
d
ien
t p
r
o
b
lem
,
m
ak
in
g
it id
ea
l f
o
r
c
o
m
p
lex
s
e
q
u
en
ce
s
lik
e
n
etwo
r
k
tr
a
f
f
ic
d
ata.
T
h
e
o
u
tp
u
t la
y
er
co
m
p
r
is
es
a
s
in
g
le
n
eu
r
o
n
ac
t
iv
ated
b
y
a
s
ig
m
o
id
f
u
n
ctio
n
,
m
ak
in
g
it
well
-
s
u
ited
f
o
r
b
in
a
r
y
class
if
icatio
n
b
y
p
r
o
d
u
cin
g
a
p
r
o
b
ab
ilit
y
s
co
r
e
b
etwe
en
0
an
d
1
.
T
h
e
m
o
d
e
l
is
co
m
p
iled
u
s
in
g
th
e
A
d
a
m
o
p
tim
izer
,
wh
ic
h
en
s
u
r
es
ef
f
icien
t
tr
ain
i
n
g
th
r
o
u
g
h
ad
ap
tiv
e
lear
n
in
g
r
ates,
an
d
t
h
e
b
in
a
r
y
c
r
o
s
s
-
en
tr
o
p
y
lo
s
s
f
u
n
ctio
n
,
a
co
m
m
o
n
l
y
u
s
ed
cr
iter
io
n
f
o
r
ev
alu
atin
g
p
er
f
o
r
m
a
n
ce
in
b
i
n
ar
y
class
if
icatio
n
task
s
.
E
ar
ly
s
to
p
p
in
g
m
o
n
ito
r
s
v
alid
atio
n
lo
s
s
,
h
altin
g
tr
ain
in
g
wh
en
th
er
e'
s
n
o
im
p
r
o
v
e
m
en
t,
an
d
r
ev
er
tin
g
to
th
e
b
est
m
o
d
el
weig
h
ts
,
p
r
ev
en
tin
g
o
v
e
r
f
itti
n
g
an
d
en
s
u
r
in
g
g
en
er
aliza
tio
n
t
o
u
n
s
ee
n
d
ata.
2
.
4
.
P
er
f
o
r
m
a
nce
m
e
t
rics
T
o
ass
ess
th
e
ef
f
ec
tiv
en
ess
o
f
ea
c
h
DL
m
o
d
el,
a
v
ar
i
ety
o
f
ev
alu
atio
n
m
etr
ics
s
u
itab
le
f
o
r
class
if
icatio
n
task
s
wer
e
em
p
lo
y
ed
.
T
h
ese
in
clu
d
e
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all
,
a
n
d
F
1
-
s
co
r
e,
o
f
f
er
in
g
a
co
m
p
r
eh
e
n
s
iv
e
p
er
s
p
ec
tiv
e
o
n
ea
ch
m
o
d
el’
s
p
r
ed
ictiv
e
ca
p
a
b
ilit
ies.
‒
Acc
u
r
ac
y
r
e
f
lects
th
e
o
v
er
all
co
r
r
ec
tn
ess
o
f
th
e
m
o
d
el
an
d
is
ca
lcu
lated
as
th
e
p
r
o
p
o
r
t
io
n
o
f
co
r
r
ec
tly
p
r
ed
icted
in
s
tan
ce
s
to
th
e
t
o
tal
n
u
m
b
e
r
o
f
p
r
ed
ictio
n
s
m
ad
e.
=
(
+
)
(
+
+
+
)
(
1
)
Pre
cisi
o
n
q
u
an
tifie
s
th
e
r
atio
o
f
ac
c
u
r
ate
p
o
s
itiv
e
d
etec
tio
n
s
to
th
e
to
tal
in
s
tan
ce
s
th
at
wer
e
p
r
ed
icted
as
p
o
s
itiv
e.
I
t m
ea
s
u
r
es th
e
m
o
d
e
l’
s
ab
ilit
y
to
av
o
id
f
alse a
lar
m
s
wh
en
id
en
tify
in
g
attac
k
tr
af
f
ic.
=
(
+
)
(
2
)
‒
R
ec
all
(
also
k
n
o
w
n
as
s
en
s
itiv
ity
o
r
th
e
ac
tu
al
p
o
s
itiv
e
r
ate
)
g
a
u
g
es
h
o
w
well
t
h
e
m
o
d
el
id
en
tifie
s
ac
tu
al
attac
k
ca
s
es.
I
t i
s
co
m
p
u
ted
b
y
d
iv
id
in
g
th
e
n
u
m
b
er
o
f
tr
u
e
p
o
s
itiv
es b
y
th
e
s
u
m
o
f
tr
u
e
p
o
s
itiv
es a
n
d
f
alse
n
eg
ativ
es.
=
(
+
)
(
3
)
F1
-
s
co
r
e
p
r
o
v
id
es
a
b
alan
ce
d
m
ea
s
u
r
e
b
y
c
o
m
p
u
tin
g
th
e
h
a
r
m
o
n
ic
m
ea
n
o
f
p
r
ec
is
io
n
an
d
r
ec
all.
T
h
is
m
etr
ic
is
h
an
d
y
in
ca
s
es
o
f
class
im
b
alan
ce
,
e
n
s
u
r
in
g
b
o
t
h
f
als
e
p
o
s
itiv
es
an
d
f
alse
n
eg
ativ
es
ar
e
ad
eq
u
ately
co
n
s
id
er
ed
.
1
=
2
×
(
×
)
(
+
)
(
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.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
4
9
8
2
-
4
9
9
2
4988
2
.
5
.
DDo
S
m
it
ig
a
t
io
n str
a
t
e
g
y
us
ing
g
ra
ph
-
ba
s
ed
dy
na
m
ic
f
lo
w
co
ntr
o
l
I
n
th
e
SDN
en
v
ir
o
n
m
en
t,
th
e
co
n
tr
o
ller
p
lay
s
a
p
iv
o
tal
r
o
le
in
m
o
n
ito
r
in
g
o
n
g
o
in
g
t
r
af
f
ic
an
d
en
s
u
r
in
g
p
r
o
tectio
n
a
g
ain
s
t
m
alicio
u
s
in
tr
u
s
io
n
s
.
Up
o
n
i
d
en
tify
in
g
s
u
s
p
icio
u
s
b
eh
av
i
o
r
,
it
m
u
s
t
s
wif
tly
ac
tiv
ate
a
d
ef
e
n
s
e
m
ec
h
an
is
m
to
lim
it
th
e
im
p
ac
t
a
n
d
m
ain
tain
s
ea
m
less
n
etwo
r
k
p
er
f
o
r
m
an
ce
.
W
h
ile
co
n
v
en
tio
n
al
s
o
lu
tio
n
s
o
f
ten
in
v
o
lv
e
f
ilter
in
g
o
r
b
lo
ck
i
n
g
m
alicio
u
s
tr
af
f
ic,
th
ey
ty
p
ically
leav
e
b
eh
in
d
r
esid
u
al
f
lo
w
en
t
r
ies
with
in
t
h
e
s
witch
es.
T
h
ese
lef
to
v
er
e
n
tr
ies
ca
n
h
in
d
er
p
ac
k
et
f
o
r
w
ar
d
in
g
a
n
d
i
m
p
o
s
e
u
n
n
ec
ess
ar
y
p
r
o
ce
s
s
in
g
o
v
e
r
h
ea
d
o
n
b
o
th
t
h
e
co
n
tr
o
ller
an
d
th
e
s
witch
es.
T
o
co
u
n
ter
th
is
lim
itatio
n
,
we
p
r
o
p
o
s
e
a
g
r
a
p
h
-
t
h
eo
r
y
-
b
ased
m
itig
atio
n
m
ec
h
a
n
is
m
th
at
in
c
o
r
p
o
r
ates
d
y
n
am
ic
f
lo
w
d
eletio
n
.
I
n
o
u
r
ap
p
r
o
ac
h
,
th
e
co
n
tr
o
ller
p
er
io
d
ically
(
e.
g
.
,
e
v
er
y
5
s
ec
o
n
d
s
)
co
llects
f
lo
w
s
tatis
t
ics
f
r
o
m
ass
o
ciate
d
Op
e
n
Flo
w
s
witch
es.
T
h
ese
s
tatis
ti
cs,
co
n
tain
in
g
v
ital
tr
af
f
ic
ch
a
r
ac
ter
is
tics
,
ar
e
f
ed
in
to
a
p
r
e
-
tr
ain
ed
DL
class
if
ier
to
d
eter
m
i
n
e
wh
eth
er
t
h
e
f
lo
w
is
b
en
ig
n
o
r
in
d
icativ
e
o
f
an
attac
k
.
Su
p
p
o
s
e
th
e
class
if
ier
f
lag
s
a
f
lo
w
as
m
alicio
u
s
.
I
n
th
at
ca
s
e,
th
e
co
n
tr
o
ller
lo
g
s
th
is
in
a
"g
r
ay
lis
t"
(
S<su
b
>g
</su
b
>)
,
is
o
latin
g
th
e
s
u
s
p
ec
t
f
lo
ws
f
r
o
m
th
o
s
e
o
r
ig
in
atin
g
in
s
witch
es
id
en
tifie
d
as
ca
r
r
y
in
g
attac
k
tr
af
f
ic.
T
h
is
m
ec
h
an
is
m
allo
ws
f
o
r
d
ee
p
er
an
aly
s
is
an
d
r
ed
u
ce
s
th
e
r
is
k
o
f
p
r
em
atu
r
ely
d
r
o
p
p
in
g
leg
itima
te
p
ac
k
ets.
T
h
e
c
o
n
tr
o
ller
co
n
ti
n
u
es
to
r
e
-
r
o
u
te
f
lo
ws
in
th
e
g
r
ay
lis
t
th
r
o
u
g
h
th
e
class
if
ier
f
o
r
ad
d
itio
n
a
l
v
er
if
icatio
n
.
A
co
u
n
ter
m
ain
tain
s
a
tally
o
f
d
etec
ted
m
alicio
u
s
f
lo
ws,
an
d
a
p
r
ed
ef
i
n
ed
th
r
esh
o
l
d
h
elp
s
d
eter
m
in
e
wh
en
f
u
r
t
h
er
ac
tio
n
is
n
ec
ess
ar
y
.
On
ce
th
is
th
r
esh
o
ld
is
m
et,
th
e
co
n
tr
o
ller
cr
ea
tes two
ad
d
i
tio
n
al
lis
ts
:
‒
Dele
te
lis
t (
S<
s
u
b
>d
</su
b
>)
:
co
n
tain
s
f
lo
w
en
tr
ies s
ch
ed
u
le
d
f
o
r
r
em
o
v
al.
‒
B
lo
ck
lis
t
(
S<su
b
>b
</su
b
>)
:
in
clu
d
es
h
o
s
ts
id
en
tifie
d
as
m
alicio
u
s
,
s
to
r
in
g
attr
ib
u
tes
s
u
c
h
as
MA
C
/I
P
ad
d
r
ess
es,
p
o
r
t n
u
m
b
er
s
,
a
n
d
i
n
g
r
ess
d
etails f
o
r
f
u
t
u
r
e
r
e
f
er
e
n
ce
.
Usi
n
g
its
h
o
s
t tr
ac
k
in
g
ca
p
ab
ilit
ies,
th
e
co
n
tr
o
ller
g
ath
er
s
id
en
tify
in
g
in
f
o
r
m
atio
n
a
b
o
u
t th
e
attac
k
in
g
s
o
u
r
ce
s
.
Up
o
n
r
ea
c
h
in
g
t
h
e
attac
k
f
lo
w
th
r
esh
o
ld
,
a
g
r
ap
h
-
th
e
o
r
etic
tr
ac
in
g
alg
o
r
it
h
m
is
in
v
o
k
ed
to
r
ec
o
n
s
tr
u
ct
th
e
attac
k
p
ath
.
T
h
is
in
v
o
lv
es
id
en
tify
i
n
g
th
e
s
eq
u
en
ce
o
f
s
witch
es
(
h
o
p
s
)
th
r
o
u
g
h
wh
ich
th
e
m
alicio
u
s
tr
af
f
ic
tr
av
er
s
ed
.
T
h
e
f
o
llo
win
g
e
x
p
r
ess
io
n
r
e
p
r
esen
ts
th
e
attac
k
p
ath
:
,
=
∑
(
,
)
→
(
,
)
,
ℎ
,
(
5
)
Her
e,
E
(i,
j
)
d
en
o
tes
an
ed
g
e
i
n
th
e
attac
k
g
r
a
p
h
,
an
d
S
attack
is
th
e
s
et
o
f
s
witch
es
in
v
o
lv
ed
in
r
o
u
tin
g
th
e
DDo
S
tr
af
f
ic.
I
f
tr
af
f
ic
f
l
o
ws
th
r
o
u
g
h
b
o
th
s
witch
es
s
i
an
d
s
j
with
v
alid
f
o
r
war
d
in
g
r
u
les,
a
c
o
n
n
ec
tio
n
(
ed
g
e)
b
etwe
en
th
em
is
estab
lis
h
ed
.
T
h
e
ce
n
tr
al
o
b
jectiv
e
o
f
th
is
ap
p
r
o
ac
h
is
to
p
in
p
o
in
t th
e
ex
ac
t a
ttack
p
ath
,
th
er
eb
y
allo
win
g
tar
g
eted
d
r
o
p
p
in
g
o
f
m
alicio
u
s
tr
af
f
ic.
W
e
ass
u
m
e
th
at
s
witch
es
clo
s
er
to
th
e
s
o
u
r
ce
o
f
th
e
attac
k
ca
r
r
y
a
h
ig
h
er
co
n
ce
n
t
r
atio
n
o
f
m
alicio
u
s
p
ac
k
ets.
As
a
r
esu
lt,
ed
g
e
s
witch
es
(
wh
er
e
attac
k
tr
af
f
ic
en
ter
s
)
ar
e
ass
ig
n
ed
h
ig
h
er
d
r
o
p
p
in
g
r
ates,
wh
er
ea
s
in
te
r
m
ed
iate
s
witch
es
r
ec
eiv
e
lo
wer
r
ates
to
a
v
o
i
d
co
llater
al
d
am
ag
e
to
leg
itima
te
tr
af
f
ic.
A
d
r
o
p
p
in
g
r
ate
f
o
r
ea
ch
s
witch
is
co
m
p
u
ted
u
s
in
g
tr
af
f
ic
-
b
ased
in
d
icato
r
s
.
I
f
a
s
witch
is
o
n
ly
h
an
d
lin
g
clea
n
tr
a
f
f
ic,
n
o
d
r
o
p
p
in
g
is
en
f
o
r
ce
d
.
Fo
r
s
witch
es
u
n
d
er
s
u
s
p
icio
n
,
th
e
d
r
o
p
r
ate
is
d
eter
m
in
e
d
u
s
i
n
g
th
e
f
o
llo
win
g
f
o
r
m
u
la:
=
(
∆
,
∆
)
(
6
)
W
h
er
e
Δ
H
i
s
th
e
ch
an
g
e
in
e
n
tr
o
p
y
o
f
s
o
u
r
ce
I
P
ad
d
r
ess
es
o
v
er
tim
e
,
an
d
Δ
N
is
th
e
ch
an
g
e
in
p
ac
k
et
c
o
u
n
t
o
v
er
tim
e
at
th
e
s
witch
.
Fin
ally
,
th
e
co
n
t
r
o
ller
s
en
d
s
a
n
OFPF
C
_
ADD
m
es
s
ag
e
to
th
e
af
f
ec
ted
s
witch
es,
in
s
er
tin
g
n
ew
f
lo
w
r
u
les
th
at
d
r
o
p
tr
af
f
ic
as
p
er
t
h
e
d
elete
lis
t
(
S<su
b
>d
</su
b
>
)
an
d
ca
lc
u
lated
d
r
o
p
r
ates.
I
f
a
h
o
s
t’
s
d
r
o
p
p
in
g
r
ate
r
ea
ch
es
1
0
0
%,
it
is
a
d
d
ed
to
th
e
b
lo
c
k
lis
t
.
A
ll
f
u
t
u
r
e
tr
a
f
f
ic
f
r
o
m
th
at
h
o
s
t
is
b
lo
ck
e
d
,
th
er
eb
y
n
eu
tr
alizin
g
th
e
attac
k
at
its
s
o
u
r
ce
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
ev
alu
atio
n
o
f
v
ar
io
u
s
DL
m
o
d
els
ap
p
lied
to
DDo
S
attac
k
d
etec
tio
n
with
in
a
n
SDN
f
r
am
ewo
r
k
h
ig
h
lig
h
ts
d
is
tin
ct
p
er
f
o
r
m
an
ce
tr
en
d
s
ac
r
o
s
s
d
if
f
er
en
t
al
g
o
r
ith
m
s
.
A
d
etailed
co
m
p
ar
is
o
n
is
p
r
o
v
id
ed
in
T
ab
le
3
,
wh
ich
s
u
m
m
ar
izes
th
e
class
if
icatio
n
ef
f
ec
tiv
en
ess
o
f
ea
ch
m
o
d
el
b
ased
o
n
k
ey
p
e
r
f
o
r
m
a
n
ce
m
etr
ics.
T
h
is
co
m
p
r
e
h
en
s
iv
e
a
n
aly
s
is
o
f
f
e
r
s
in
s
ig
h
ts
in
to
h
o
w
w
ell
ea
ch
DL
ap
p
r
o
ac
h
id
en
ti
f
ies
an
d
m
itig
ates
m
alicio
u
s
tr
af
f
ic.
T
h
e
ML
P
m
o
d
el
h
as
a
lo
w
ac
cu
r
ac
y
o
f
0
.
4
8
b
u
t
an
ex
ce
p
tio
n
ally
h
ig
h
r
ec
all
o
f
0
.
9
9
,
in
d
icatin
g
it
is
s
u
s
ce
p
tib
le
an
d
co
r
r
ec
tly
i
d
en
tifie
s
n
ea
r
ly
a
ll
p
o
s
itiv
e
ca
s
es.
Ho
wev
e
r
,
t
h
e
p
r
ec
is
io
n
is
o
n
ly
0
.
4
5
,
m
ea
n
i
n
g
m
an
y
f
alse
p
o
s
itiv
es
ar
e
lik
ely
t
o
b
e
p
r
ed
ic
ted
.
T
h
e
F1
s
co
r
e,
at
0
.
6
2
,
s
u
g
g
ests
a
m
o
d
er
ate
b
alan
ce
b
etwe
en
p
r
ec
is
io
n
an
d
r
ec
all.
Sti
ll,
th
e
m
o
d
el'
s
u
tili
ty
in
a
r
ea
l
-
wo
r
ld
s
ettin
g
m
ig
h
t
b
e
lim
ited
d
u
e
to
its
ten
d
en
cy
to
o
v
er
-
p
r
ed
ict
th
e
p
o
s
itiv
e
class
.
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
Dee
p
lea
r
n
in
g
-
b
a
s
ed
ev
a
lu
a
tio
n
fo
r
d
is
tr
ib
u
ted
d
en
i
a
l o
f ser
vice
a
tta
ck
s
d
etec
tio
n
(
N
ee
th
u
S
.
)
4989
T
ab
le
3
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
o
f
DL
alg
o
r
ith
m
s
in
th
e
d
e
tectio
n
o
f
DDo
S
attac
k
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
LP
0
.
4
8
0
.
9
9
0
.
4
5
0
.
6
2
ANN
0
.
7
3
0
.
6
3
0
.
8
8
0
.
7
4
C
N
N
0
.
4
8
0
.
4
5
0
.
8
1
0
.
5
8
LSTM
0
.
9
1
0
.
8
6
0
.
9
4
0
.
9
0
R
N
N
0
.
6
6
0
.
5
8
0
.
7
4
0
.
6
6
T
h
e
ANN
d
em
o
n
s
tr
ates
a
s
o
li
d
p
r
ed
ictiv
e
ca
p
a
b
ilit
y
,
ac
h
iev
in
g
an
ac
cu
r
ac
y
o
f
0
.
7
3
.
I
ts
p
r
ec
is
io
n
o
f
0
.
6
3
an
d
r
ec
all
o
f
0
.
8
8
in
d
ica
te
th
at
wh
ile
th
e
m
o
d
el
is
ef
f
ec
tiv
e
in
d
etec
tin
g
ac
tu
al
attac
k
in
s
tan
ce
s
,
it
d
o
es
allo
w
f
o
r
a
m
o
d
er
ate
lev
el
o
f
f
alse
p
o
s
itiv
es.
An
F1
-
s
co
r
e
o
f
0
.
7
4
r
ef
lects
a
f
air
tr
ad
e
-
o
f
f
b
etwe
en
p
r
ec
is
io
n
an
d
r
ec
all,
s
u
g
g
esti
n
g
th
at
wit
h
f
u
r
t
h
er
o
p
tim
izatio
n
,
p
ar
ticu
lar
ly
to
en
h
an
ce
p
r
ec
is
io
n
,
A
NN
co
u
ld
s
er
v
e
as
a
v
iab
le
s
o
lu
tio
n
f
o
r
DDo
S
d
etec
tio
n
.
T
h
e
C
NN
m
o
d
el
h
as
an
ac
c
u
r
ac
y
o
n
p
a
r
with
th
e
ML
P
at
0
.
4
8
,
s
u
g
g
esti
n
g
ch
allen
g
es in
co
r
r
e
ctly
class
if
y
in
g
ca
s
es.
T
h
e
p
r
ec
is
io
n
is
lo
w
at
0
.
4
5
,
an
d
th
e
r
ec
all
is
0
.
8
1
,
wh
ich
is
lo
wer
th
an
ML
P's
b
u
t
s
till
in
d
icate
s
a
ten
d
en
cy
to
id
e
n
tify
m
o
s
t
p
o
s
itiv
e
in
s
tan
ce
s
.
T
h
e
F1
s
co
r
e
o
f
0
.
5
8
r
ef
lects
a
m
o
d
el
th
at
is
b
etter
at
en
s
u
r
in
g
attac
k
s
ar
e
n
o
t
m
is
s
ed
r
ath
er
th
an
p
r
ec
is
ely
id
en
tify
in
g
o
n
ly
th
e
ac
tu
al
attac
k
s
.
Fo
r
n
etwo
r
k
s
ec
u
r
ity
,
th
is
co
u
l
d
m
ea
n
a
h
ig
h
er
o
p
er
atio
n
al
lo
ad
d
u
e
to
f
alse a
lar
m
s
.
T
h
e
R
N
N
y
i
e
l
d
s
a
m
o
d
e
r
a
t
e
a
c
c
u
r
a
c
y
o
f
0
.
6
6
,
w
i
t
h
p
r
e
c
i
s
i
o
n
a
n
d
r
e
c
a
l
l
v
a
l
u
e
s
o
f
0
.
5
8
a
n
d
0
.
7
4
,
r
e
s
p
e
c
t
i
v
e
l
y
.
I
t
s
F
1
-
s
c
o
r
e
o
f
0
.
6
6
s
h
o
w
s
t
h
a
t
t
h
e
m
o
d
e
l
s
t
r
i
k
e
s
a
r
e
a
s
o
n
a
b
l
e
b
a
l
a
n
c
e
b
e
t
w
e
e
n
d
e
t
e
c
t
i
n
g
a
t
t
a
c
k
s
a
n
d
m
i
n
i
m
i
z
i
n
g
i
n
c
o
r
r
e
c
t
c
l
a
s
s
i
f
i
c
a
t
i
o
n
s
.
H
o
w
e
v
e
r
,
i
t
s
p
e
r
f
o
r
m
a
n
c
e
t
r
a
i
l
s
t
h
a
t
o
f
L
S
T
M
,
l
i
k
e
l
y
d
u
e
t
o
R
N
N
’
s
l
i
m
i
t
a
t
i
o
n
s
i
n
r
e
t
a
i
n
i
n
g
l
o
n
g
-
t
e
r
m
d
e
p
e
n
d
e
n
c
i
e
s
—
a
n
e
s
s
e
n
t
i
a
l
f
a
c
t
o
r
w
h
e
n
D
D
o
S
a
t
t
a
c
k
p
a
t
t
e
r
n
s
s
p
a
n
a
c
r
o
s
s
e
x
t
e
n
d
e
d
t
r
a
f
f
i
c
s
e
q
u
e
n
c
e
s
.
I
n
c
o
n
t
r
a
s
t
,
t
h
e
L
S
T
M
m
o
d
e
l
e
x
c
e
l
s
i
n
p
e
r
f
o
r
m
a
n
c
e
w
i
t
h
a
n
a
c
c
u
r
a
c
y
o
f
0
.
9
1
.
I
t
s
p
r
e
c
i
s
i
o
n
o
f
0
.
8
6
a
n
d
r
e
c
a
l
l
o
f
0
.
9
4
h
i
g
h
l
i
g
h
t
i
t
s
a
b
i
l
i
t
y
t
o
n
o
t
o
n
l
y
d
e
t
e
c
t
a
t
t
a
c
k
f
l
o
w
s
w
i
t
h
h
i
g
h
s
e
n
s
i
t
i
v
i
t
y
b
u
t
a
l
s
o
t
o
m
i
n
i
m
i
z
e
f
a
l
s
e
a
l
a
r
m
s
.
F
1
-
s
c
o
r
e
o
f
0
.
9
0
r
e
i
n
f
o
r
c
e
s
i
t
s
e
f
f
e
c
t
i
v
e
n
e
s
s
i
n
h
a
n
d
l
i
n
g
s
e
q
u
e
n
t
i
a
l
n
e
t
w
o
r
k
d
a
t
a
,
m
a
k
i
n
g
L
S
T
M
p
a
r
t
i
c
u
l
a
r
l
y
w
e
l
l
-
s
u
i
t
e
d
f
o
r
i
d
e
n
t
i
f
y
i
n
g
s
o
p
h
i
s
t
i
c
a
t
e
d
D
D
o
S
p
a
t
t
e
r
n
s
i
n
S
D
N
e
n
v
i
r
o
n
m
e
n
t
s
.
T
h
e
s
e
r
e
s
u
l
t
s
s
u
g
g
e
s
t
t
h
a
t
L
S
T
M
i
s
a
s
t
r
o
n
g
c
a
n
d
i
d
a
t
e
f
o
r
r
e
a
l
-
t
i
m
e
d
e
p
l
o
y
m
e
n
t
i
n
i
n
t
e
l
l
i
g
e
n
t
n
e
t
w
o
r
k
d
e
f
e
n
s
e
s
y
s
t
e
m
s
.
Fig
u
r
e
3
r
e
p
r
esen
ts
a
p
er
f
o
r
m
an
ce
co
m
p
ar
is
o
n
o
f
DL
alg
o
r
ith
m
s
in
ter
m
s
o
f
ac
c
u
r
ac
y
an
d
r
ec
all.
Fig
u
r
e
4
d
ep
icts
th
e
e
v
alu
atio
n
o
f
th
e
m
o
d
els
in
ter
m
s
o
f
p
r
ec
is
io
n
an
d
F1
s
co
r
e.
I
t
ca
n
b
e
co
n
clu
d
ed
th
at
DL
m
o
d
els,
esp
ec
ially
th
o
s
e
tail
o
r
ed
f
o
r
s
eq
u
en
ce
d
ata
s
u
ch
as
L
STM
s
an
d
R
NNs,
h
av
e
p
r
o
v
e
n
s
u
p
er
io
r
in
ac
cu
r
ac
y
co
m
p
ar
ed
to
o
th
er
m
o
d
els
wh
en
h
an
d
lin
g
task
s
in
v
o
lv
in
g
s
eq
u
en
tial
o
r
tim
e
-
s
er
ies
d
ata.
T
h
is
s
u
p
er
io
r
ity
s
tem
s
f
r
o
m
th
eir
ar
ch
itectu
r
e'
s
ad
ep
tn
ess
at
c
ap
tu
r
in
g
tem
p
o
r
al
d
ep
en
d
en
c
ies
an
d
c
o
n
tex
tu
al
n
u
an
ce
s
f
r
o
m
th
e
d
ata,
cr
u
cial
f
o
r
id
en
tify
i
n
g
p
atter
n
s
in
d
icativ
e
o
f
DDo
S
attac
k
s
.
Desp
ite
th
eir
a
b
ilit
y
t
o
d
eliv
er
h
ig
h
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
,
DL
m
o
d
els
in
tr
o
d
u
ce
ch
allen
g
es
s
u
ch
as
h
eig
h
ten
ed
co
m
p
lex
ity
,
in
cr
ea
s
ed
co
m
p
u
tatio
n
al
o
v
e
r
h
ea
d
,
an
d
o
f
ten
a
n
o
p
a
q
u
e
n
at
u
r
e
th
at
co
m
p
licates in
ter
p
r
etati
o
n
an
d
i
n
teg
r
atio
n
.
I
n
r
ea
l
-
tim
e
DDo
S
d
etec
tio
n
s
y
s
tem
s
,
wh
er
e
er
r
o
r
s
lik
e
f
a
ls
e
p
o
s
itiv
es
o
r
f
alse
n
eg
ativ
es
ca
r
r
y
s
ig
n
if
ican
t
co
n
s
eq
u
en
ce
s
,
th
e
s
u
p
er
i
o
r
p
r
ed
ictiv
e
ac
cu
r
ac
y
o
f
DL
m
o
d
els,
p
ar
ticu
lar
ly
L
STM
s
,
r
en
d
er
s
th
em
m
o
r
e
s
u
itab
le
d
esp
ite
th
eir
g
r
ea
ter
r
eso
u
r
ce
d
em
an
d
s
.
T
h
er
ef
o
r
e,
d
esp
ite
b
ein
g
m
o
r
e
r
eso
u
r
ce
-
in
ten
s
iv
e
an
d
less
in
ter
p
r
etab
le,
DL
m
o
d
els
o
f
f
er
en
h
an
ce
d
p
r
ed
ictiv
e
ca
p
a
b
ilit
ies
an
d
ar
e
b
etter
s
u
ited
f
o
r
en
v
ir
o
n
m
e
n
ts
p
r
io
r
itizin
g
to
p
-
tier
p
er
f
o
r
m
an
ce
,
n
o
twith
s
tan
d
in
g
th
e
ir
h
eig
h
te
n
ed
c
o
m
p
lex
ity
an
d
co
m
p
u
tatio
n
al
r
eq
u
ir
em
e
n
ts
.
Fig
u
r
e
3
.
Per
f
o
r
m
an
c
e
co
m
p
ar
is
o
n
o
f
v
a
r
io
u
s
m
o
d
els co
n
ce
r
n
in
g
th
e
ac
c
u
r
ac
y
a
n
d
r
ec
all
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.
14
,
No
.
6
,
Dec
em
b
er
20
25
:
4
9
8
2
-
4
9
9
2
4990
Fig
u
r
e
4
.
Per
f
o
r
m
an
c
e
co
m
p
ar
is
o
n
o
f
v
a
r
io
u
s
m
o
d
els co
n
ce
r
n
in
g
th
e
p
r
ec
is
io
n
an
d
F1
s
co
r
e
4.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
ex
p
lo
r
ed
th
e
u
s
e
o
f
m
u
ltip
le
p
r
o
m
in
en
t
DL
m
o
d
els
f
o
r
th
e
d
etec
tio
n
o
f
DDo
S
attac
k
s
in
SDN
en
v
ir
o
n
m
en
ts
.
Un
lik
e
m
an
y
ex
is
tin
g
w
o
r
k
s
th
at
e
x
am
in
e
o
n
ly
a
lim
ited
s
elec
tio
n
o
f
tec
h
n
iq
u
es,
o
u
r
ap
p
r
o
ac
h
o
f
f
e
r
ed
a
co
m
p
r
eh
e
n
s
iv
e
ev
alu
atio
n
ac
r
o
s
s
a
d
iv
er
s
e
r
an
g
e
o
f
DL
ar
c
h
itectu
r
e
s
.
W
h
ile
th
e
r
esu
lts
h
ig
h
lig
h
t
s
tr
o
n
g
p
r
ed
ictiv
e
ca
p
ab
ilit
ies
f
o
r
s
o
m
e
m
o
d
els,
p
r
ac
tical
d
ep
lo
y
m
e
n
t
r
e
q
u
ir
es
m
o
r
e
th
a
n
ju
s
t
h
ig
h
ac
cu
r
ac
y
.
Ov
er
all,
th
e
ar
ch
ite
ctu
r
e
o
f
L
STM
s
em
er
g
ed
as
p
ar
ticu
lar
ly
well
-
s
u
ited
f
o
r
h
an
d
lin
g
th
e
s
eq
u
en
tial
an
d
tem
p
o
r
al
co
m
p
lex
ities
i
n
h
er
en
t
in
DDo
S
attac
k
p
r
ed
ictio
n
.
R
NN
p
r
o
v
id
ed
a
s
im
p
ler
y
et
ef
f
ec
tiv
e
alter
n
ativ
e.
ANN
an
d
ML
P
co
u
ld
b
e
v
iab
le
f
o
r
less
in
tr
icate
o
r
tem
p
o
r
ally
d
ep
e
n
d
en
t
d
ata
.
At
th
e
s
am
e
tim
e,
C
NN
m
ig
h
t
n
ec
ess
ita
te
s
ig
n
i
f
ican
t
ad
ju
s
tm
en
ts
o
r
b
e
m
o
r
e
s
u
itab
le
f
o
r
p
r
o
b
lem
s
alig
n
ed
with
its
s
p
atia
l
p
r
o
ce
s
s
in
g
s
tr
en
g
th
s
.
L
STM
e
x
h
ib
ited
a
h
i
g
h
r
ec
all
r
ate
with
o
u
t sig
n
if
ican
tly
co
m
p
r
o
m
is
in
g
p
r
ec
is
io
n
,
c
r
u
cial
in
DDo
S
d
etec
tio
n
,
wh
er
e
m
is
s
in
g
an
attac
k
ca
n
h
av
e
s
ev
er
e
co
n
s
eq
u
en
ce
s
.
T
h
e
h
ig
h
F1
s
co
r
es
attain
ed
b
y
L
STM
s
u
g
g
est
its
p
r
o
f
icien
c
y
in
ac
cu
r
ately
class
if
y
in
g
b
o
th
attac
k
a
n
d
n
o
n
-
attac
k
in
s
ta
n
ce
s
in
a
b
alan
ce
d
m
an
n
er
.
T
o
en
s
u
r
e
r
o
b
u
s
tn
ess
an
d
r
ea
l
-
wo
r
ld
ap
p
licab
ilit
y
,
m
eth
o
d
s
s
u
ch
as
cr
o
s
s
-
v
al
id
atio
n
an
d
test
in
g
ag
ain
s
t
p
r
ev
io
u
s
ly
u
n
s
ee
n
d
ata
m
u
s
t
b
e
co
n
s
is
ten
tly
ap
p
lied
;
m
o
r
eo
v
er
,
m
o
d
els
d
e
m
o
n
s
tr
atin
g
lo
wer
ef
f
ec
tiv
en
ess
p
r
esen
t
o
p
p
o
r
tu
n
ities
f
o
r
im
p
r
o
v
e
m
en
t
th
r
o
u
g
h
r
ef
in
e
d
f
ea
tu
r
e
s
elec
tio
n
s
tr
ateg
ies
an
d
h
y
p
er
p
ar
am
eter
tu
n
i
n
g
.
Giv
en
th
e
ev
o
lv
in
g
th
r
ea
t
lan
d
s
ca
p
e
,
in
teg
r
atin
g
DL
ap
p
r
o
ac
h
es
in
to
DDo
S
d
etec
tio
n
an
d
m
itig
atio
n
s
tr
ateg
ies is
n
o
lo
n
g
er
o
p
tio
n
al
—
it is
ess
en
tial
f
o
r
m
o
d
er
n
,
in
tellig
en
t n
etwo
r
k
d
ef
e
n
s
e.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
r
esear
ch
r
ec
ei
v
ed
n
o
s
p
e
cif
ic
g
r
an
t
f
r
o
m
an
y
f
u
n
d
in
g
a
g
en
cy
i
n
th
e
p
u
b
lic,
co
m
m
er
ci
al,
o
r
n
o
t
-
f
o
r
-
p
r
o
f
it secto
r
s
.
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
Nee
th
u
S.
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
H.
V.
R
av
is
h
Ar
ad
h
y
a
✓
✓
✓
✓
✓
✓
✓
Vis
wav
ar
d
h
an
R
ed
d
y
Kar
n
a
✓
✓
✓
✓
✓
✓
✓
✓
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
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
Dee
p
lea
r
n
in
g
-
b
a
s
ed
ev
a
lu
a
tio
n
fo
r
d
is
tr
ib
u
ted
d
en
i
a
l o
f ser
vice
a
tta
ck
s
d
etec
tio
n
(
N
ee
th
u
S
.
)
4991
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
T
h
e
au
t
h
o
r
s
d
ec
lar
e
th
at
th
e
y
h
av
e
n
o
k
n
o
wn
c
o
m
p
etin
g
f
in
an
cial
in
ter
ests
o
r
p
er
s
o
n
al
r
el
atio
n
s
h
ip
s
th
at
co
u
ld
h
av
e
ap
p
ea
r
ed
t
o
in
f
lu
en
ce
th
e
wo
r
k
r
e
p
o
r
te
d
in
t
h
is
p
ap
er
.
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
ter
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
av
ailab
l
e
f
r
o
m
th
e
co
r
r
esp
o
n
d
in
g
au
t
h
o
r
,
[
NS]
,
u
p
o
n
r
ea
s
o
n
ab
le
r
eq
u
est.
RE
F
E
R
E
NC
E
S
[
1
]
B
.
A
l
h
i
j
a
w
i
,
S
.
A
l
m
a
j
a
l
i
,
H
.
E
l
g
a
l
a
,
H
.
B
.
S
a
l
a
m
e
h
,
a
n
d
M
.
A
y
y
a
sh
,
“
A
su
r
v
e
y
o
n
D
o
S
/
D
D
o
S
mi
t
i
g
a
t
i
o
n
t
e
c
h
n
i
q
u
e
s
i
n
S
D
N
s:
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
c
o
mp
a
r
i
s
o
n
,
s
o
l
u
t
i
o
n
s
,
t
e
s
t
i
n
g
t
o
o
l
s
a
n
d
d
a
t
a
set
s,”
C
o
m
p
u
t
e
rs
a
n
d
El
e
c
t
r
i
c
a
l
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
9
9
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
e
l
e
c
e
n
g
.
2
0
2
2
.
1
0
7
7
0
6
.
[
2
]
J.
S
i
n
g
h
a
n
d
S
.
B
e
h
a
l
,
“
D
e
t
e
c
t
i
o
n
a
n
d
mi
t
i
g
a
t
i
o
n
o
f
D
D
o
S
a
t
t
a
c
k
s
i
n
S
D
N
:
a
c
o
m
p
r
e
h
e
n
si
v
e
r
e
v
i
e
w
,
r
e
s
e
a
r
c
h
c
h
a
l
l
e
n
g
e
s
a
n
d
f
u
t
u
r
e
d
i
r
e
c
t
i
o
n
s
,
”
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
Re
v
i
e
w
,
v
o
l
.
3
7
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
sr
e
v
.
2
0
2
0
.
1
0
0
2
7
9
.
[
3
]
V
.
H
n
a
mt
e
,
A
.
A
.
N
a
j
a
r
,
H
.
N
.
-
N
g
u
y
e
n
,
J.
H
u
ss
a
i
n
,
a
n
d
M
.
N
.
S
u
g
a
l
i
,
“
D
D
o
S
a
t
t
a
c
k
d
e
t
e
c
t
i
o
n
a
n
d
m
i
t
i
g
a
t
i
o
n
u
si
n
g
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
i
n
S
D
N
e
n
v
i
r
o
n
me
n
t
,
”
C
o
m
p
u
t
e
rs &
S
e
c
u
ri
t
y
,
v
o
l
.
1
3
8
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
se.
2
0
2
3
.
1
0
3
6
6
1
.
[
4
]
R
.
B
.
S
a
i
d
,
Z.
S
a
b
i
r
,
a
n
d
I
.
A
sk
e
r
z
a
d
e
,
“
C
N
N
-
B
i
LST
M
:
A
h
y
b
r
i
d
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
s
i
o
n
d
e
t
e
c
t
i
o
n
s
y
st
e
m
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
w
i
t
h
h
y
b
r
i
d
f
e
a
t
u
r
e
se
l
e
c
t
i
o
n
,
”
I
E
EE
Ac
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
1
3
8
7
3
2
–
1
3
8
7
4
7
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
3
4
0
1
4
2
.
[
5
]
H
.
S
.
I
l
a
n
g
o
,
M
.
M
a
,
a
n
d
R
.
S
u
,
“
A
f
e
e
d
f
o
r
w
a
r
d
–
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
t
o
d
e
t
e
c
t
l
o
w
-
r
a
t
e
D
o
S
i
n
I
o
T,
”
E
n
g
i
n
e
e
ri
n
g
Ap
p
l
i
c
a
t
i
o
n
s
o
f
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
1
1
4
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
n
g
a
p
p
a
i
.
2
0
2
2
.
1
0
5
0
5
9
.
[
6
]
G
.
D
.
L
.
T
.
P
a
r
r
a
,
P
.
R
a
d
,
K
.
-
K
.
R
.
C
h
o
o
,
a
n
d
N
.
B
e
e
b
e
,
“
D
e
t
e
c
t
i
n
g
i
n
t
e
r
n
e
t
o
f
t
h
i
n
g
s
a
t
t
a
c
k
s
u
si
n
g
d
i
s
t
r
i
b
u
t
e
d
d
e
e
p
l
e
a
r
n
i
n
g
,
”
J
o
u
rn
a
l
o
f
N
e
t
w
o
rk
a
n
d
C
o
m
p
u
t
e
r
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
1
6
3
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
n
c
a
.
2
0
2
0
.
1
0
2
6
6
2
.
[
7
]
M
.
A
.
R
i
b
e
i
r
o
,
M
.
S
.
P
.
F
o
n
s
e
c
a
,
a
n
d
J.
d
e
S
a
n
t
i
,
“
D
e
t
e
c
t
i
n
g
a
n
d
mi
t
i
g
a
t
i
n
g
D
D
o
S
a
t
t
a
c
k
s
w
i
t
h
m
o
v
i
n
g
t
a
r
g
e
t
d
e
f
e
n
s
e
a
p
p
r
o
a
c
h
b
a
s
e
d
o
n
a
u
t
o
ma
t
e
d
f
l
o
w
c
l
a
ssi
f
i
c
a
t
i
o
n
i
n
S
D
N
n
e
t
w
o
r
k
s,
”
C
o
m
p
u
t
e
rs
&
S
e
c
u
r
i
t
y
,
v
o
l
.
1
3
4
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
s
e
.
2
0
2
3
.
1
0
3
4
6
2
.
[
8
]
O
.
H
a
b
i
b
i
,
M
.
C
h
e
mm
a
k
h
a
,
a
n
d
M
.
La
z
a
a
r
,
“
I
mb
a
l
a
n
c
e
d
t
a
b
u
l
a
r
d
a
t
a
mo
d
e
l
i
z
a
t
i
o
n
u
s
i
n
g
C
TG
A
N
a
n
d
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
t
o
i
mp
r
o
v
e
I
o
T
b
o
t
-
n
e
t
a
t
t
a
c
k
s
d
e
t
e
c
t
i
o
n
,
”
En
g
i
n
e
e
r
i
n
g
A
p
p
l
i
c
a
t
i
o
n
s
o
f
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
1
1
8
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
n
g
a
p
p
a
i
.
2
0
2
2
.
1
0
5
6
6
9
.
[
9
]
N
.
S
.
a
n
d
H
.
V
.
R
.
A
r
a
d
h
y
a
,
“
D
e
t
e
c
t
i
o
n
o
f
(
D
D
o
S
)
a
t
t
a
c
k
s
i
n
S
D
N
,
”
EC
S
T
ra
n
s
a
c
t
i
o
n
s
,
v
o
l
.
1
0
7
,
n
o
.
1
,
p
p
.
1
8
3
0
5
–
1
8
3
1
3
,
A
p
r
.
2
0
2
2
,
d
o
i
:
1
0
.
1
1
4
9
/
1
0
7
0
1
.
1
2
1
8
9
e
c
st
.
[
1
0
]
M
.
A
.
A
l
so
u
f
i
e
t
a
l
.
,
“
A
n
o
ma
l
y
-
b
a
s
e
d
i
n
t
r
u
s
i
o
n
d
e
t
e
c
t
i
o
n
mo
d
e
l
u
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
f
o
r
I
o
T
n
e
t
w
o
r
k
s,
”
C
o
m
p
u
t
e
r
Mo
d
e
l
i
n
g
i
n
En
g
i
n
e
e
ri
n
g
&
S
c
i
e
n
c
e
s
,
v
o
l
.
1
4
1
,
n
o
.
1
,
p
p
.
1
–
1
0
,
A
u
g
.
2
0
2
4
,
d
o
i
:
1
0
.
3
2
6
0
4
/
c
mes.
2
0
2
4
.
0
5
2
1
1
2
.
[
1
1
]
E.
-
B
.
D
o
n
k
o
l
,
A
.
G
.
H
a
f
e
z
,
A
.
I
.
H
u
s
sei
n
,
a
n
d
M
.
M
.
M
a
b
r
o
o
k
,
“
O
p
t
i
mi
z
a
t
i
o
n
o
f
i
n
t
r
u
s
i
o
n
d
e
t
e
c
t
i
o
n
u
si
n
g
l
i
k
e
l
y
p
o
i
n
t
P
S
O
a
n
d
e
n
h
a
n
c
e
d
LSTM
-
R
N
N
h
y
b
r
i
d
t
e
c
h
n
i
q
u
e
i
n
c
o
mm
u
n
i
c
a
t
i
o
n
n
e
t
w
o
r
k
s,
”
I
EEE
A
c
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
9
4
6
9
–
9
4
8
2
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
2
4
0
1
0
9
.
[
1
2
]
H
.
Z
h
o
u
,
Y
.
Z
h
e
n
g
,
X
.
J
i
a
,
a
n
d
J
.
S
h
u
,
“
C
o
l
l
a
b
o
r
a
t
i
v
e
p
r
e
d
i
c
t
i
o
n
a
n
d
d
e
t
e
c
t
i
o
n
o
f
D
D
o
S
a
t
t
a
c
k
s
i
n
e
d
g
e
c
o
mp
u
t
i
n
g
:
a
d
e
e
p
l
e
a
r
n
i
n
g
-
b
a
s
e
d
a
p
p
r
o
a
c
h
w
i
t
h
d
i
st
r
i
b
u
t
e
d
S
D
N
,
”
C
o
m
p
u
t
e
r
N
e
t
w
o
rks
,
v
o
l
.
2
2
5
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
mn
e
t
.
2
0
2
3
.
1
0
9
6
4
2
.
[
1
3
]
Y
.
Li
u
,
T
.
Z
h
i
,
M
.
S
h
e
n
,
L
.
W
a
n
g
,
Y
.
Li
,
a
n
d
M
.
W
a
n
,
“
S
o
f
t
w
a
r
e
-
d
e
f
i
n
e
d
D
D
o
S
d
e
t
e
c
t
i
o
n
w
i
t
h
i
n
f
o
r
m
a
t
i
o
n
e
n
t
r
o
p
y
a
n
a
l
y
si
s
a
n
d
o
p
t
i
m
i
z
e
d
d
e
e
p
l
e
a
r
n
i
n
g
,
”
Fu
t
u
re
G
e
n
e
r
a
t
i
o
n
C
o
m
p
u
t
e
r
S
y
st
e
m
s
,
v
o
l
.
1
2
9
,
p
p
.
9
9
–
1
1
4
,
A
p
r
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
f
u
t
u
r
e
.
2
0
2
1
.
1
1
.
0
0
9
.
[
1
4
]
K
.
N
.
R
a
o
,
K
.
V
.
R
a
o
,
a
n
d
P
.
V
.
G
.
D
.
P
.
R
e
d
d
y
,
“
A
h
y
b
r
i
d
i
n
t
r
u
s
i
o
n
d
e
t
e
c
t
i
o
n
sy
st
e
m
b
a
s
e
d
o
n
s
p
a
r
se
a
u
t
o
e
n
c
o
d
e
r
a
n
d
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
C
o
m
p
u
t
e
r
C
o
m
m
u
n
i
c
a
t
i
o
n
s
,
v
o
l
.
1
8
0
,
p
p
.
7
7
–
8
8
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
c
o
m
.
2
0
2
1
.
0
8
.
0
2
6
.
[
1
5
]
N
.
M
.
Y
.
-
N
a
u
l
a
,
C
.
V
.
-
R
o
s
a
l
e
s,
J.
A
.
P
.
-
D
í
a
z
,
a
n
d
D
.
F
.
C
a
r
r
e
r
a
,
“
A
f
l
e
x
i
b
l
e
S
D
N
-
b
a
se
d
f
r
a
m
e
w
o
r
k
f
o
r
sl
o
w
-
r
a
t
e
D
D
o
S
a
t
t
a
c
k
mi
t
i
g
a
t
i
o
n
b
y
u
si
n
g
d
e
e
p
r
e
i
n
f
o
r
c
e
me
n
t
l
e
a
r
n
i
n
g
,
”
J
o
u
rn
a
l
o
f
N
e
t
w
o
rk
a
n
d
C
o
m
p
u
t
e
r
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
2
0
5
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
n
c
a
.
2
0
2
2
.
1
0
3
4
4
4
.
[
1
6
]
H
.
P
o
l
a
t
,
M
.
T
ü
r
k
o
ğ
l
u
,
O
.
P
o
l
a
t
,
a
n
d
A
.
Ş
e
n
g
ü
r
,
“
A
n
o
v
e
l
a
p
p
r
o
a
c
h
f
o
r
a
c
c
u
r
a
t
e
d
e
t
e
c
t
i
o
n
o
f
t
h
e
D
D
o
S
a
t
t
a
c
k
s
i
n
S
D
N
-
b
a
se
d
S
C
A
D
A
sy
s
t
e
ms
b
a
se
d
o
n
d
e
e
p
r
e
c
u
r
r
e
n
t
n
e
u
r
a
l
n
e
t
w
o
r
k
s
,
”
Ex
p
e
r
t
S
y
s
t
e
m
s
w
i
t
h
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
1
9
7
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
sw
a
.
2
0
2
2
.
1
1
6
7
4
8
.
[
1
7
]
V
.
R
.
K
a
r
n
a
a
n
d
K
.
V
.
V
.
R
e
d
d
y
,
“
1
-
d
i
m
e
n
s
i
o
n
a
l
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s
f
o
r
p
r
e
d
i
c
t
i
n
g
s
u
d
d
e
n
c
a
r
d
i
a
c
,
”
I
AE
S
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
(
I
J
-
A
I
)
,
v
o
l
.
1
3
,
n
o
.
1
,
p
p
.
9
8
4
–
9
9
3
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
a
i
.
v
1
3
.
i
1
.
p
p
9
8
4
-
993.
[
1
8
]
Z.
X
u
,
“
D
e
e
p
l
e
a
r
n
i
n
g
b
a
se
d
D
D
o
S
a
t
t
a
c
k
d
e
t
e
c
t
i
o
n
,
”
I
T
M
W
e
b
o
f
C
o
n
f
e
r
e
n
c
e
s
,
v
o
l
.
7
0
,
2
0
2
5
,
d
o
i
:
1
0
.
1
0
5
1
/
i
t
m
c
o
n
f
/
2
0
2
5
7
0
0
3
0
0
5
.
[
1
9
]
Y
.
C
a
o
e
t
a
l
.
,
“
D
e
t
e
c
t
i
n
g
a
n
d
m
i
t
i
g
a
t
i
n
g
D
D
o
S
a
t
t
a
c
k
s
i
n
S
D
N
u
si
n
g
s
p
a
t
i
a
l
-
t
e
mp
o
r
a
l
g
r
a
p
h
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
,
”
I
E
EE
T
ra
n
s
a
c
t
i
o
n
s
o
n
D
e
p
e
n
d
a
b
l
e
a
n
d
S
e
c
u
re
C
o
m
p
u
t
i
n
g
,
v
o
l
.
1
9
,
n
o
.
6
,
p
p
.
3
8
5
5
–
3
8
7
2
,
2
0
2
2
,
d
o
i
:
1
0
.
1
1
0
9
/
TD
S
C
.
2
0
2
1
.
3
1
0
8
7
8
2
.
[
2
0
]
P
.
V
.
S
h
a
l
i
n
i
,
V
.
R
a
d
h
a
,
a
n
d
S
.
G
.
S
a
n
j
e
e
v
i
,
“
E
a
r
l
y
d
e
t
e
c
t
i
o
n
a
n
d
mi
t
i
g
a
t
i
o
n
o
f
TC
P
S
Y
N
f
l
o
o
d
a
t
t
a
c
k
s
i
n
S
D
N
u
si
n
g
c
h
i
-
s
q
u
a
r
e
t
e
st
,
”
J
o
u
r
n
a
l
o
f
S
u
p
e
r
c
o
m
p
u
t
i
n
g
,
v
o
l
.
7
9
,
p
p
.
1
0
3
5
3
–
1
0
3
8
5
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
1
2
2
7
-
023
-
0
5
0
5
7
-
x.
[
2
1
]
V
.
H
n
a
m
t
e
,
A
.
A
.
N
a
j
a
r
,
H
.
N
.
-
N
g
u
y
e
n
,
J.
H
u
ssa
i
n
,
a
n
d
M
.
N
a
i
k
,
“
D
D
o
S
a
t
t
a
c
k
d
e
t
e
c
t
i
o
n
a
n
d
m
i
t
i
g
a
t
i
o
n
u
si
n
g
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
i
n
S
D
N
e
n
v
i
r
o
n
me
n
t
,”
C
o
m
p
u
t
e
rs &
S
e
c
u
ri
t
y
,
v
o
l
.
1
3
8
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
se.
2
0
2
3
.
1
0
3
6
6
1
.
[
2
2
]
T.
H
.
H
.
A
l
d
h
y
a
n
i
a
n
d
H
.
A
l
k
a
h
t
a
n
i
,
“
C
y
b
e
r
sec
u
r
i
t
y
f
o
r
d
e
t
e
c
t
i
n
g
d
i
s
t
r
i
b
u
t
e
d
d
e
n
i
a
l
o
f
ser
v
i
c
e
a
t
t
a
c
k
s
i
n
a
g
r
i
c
u
l
t
u
r
e
4
.
0
:
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
,
”
Ma
t
h
e
m
a
t
i
c
s
,
v
o
l
.
1
1
,
n
o
.
1
,
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
mat
h
1
1
0
1
0
2
3
3
.
[
2
3
]
E.
M
u
sh
t
a
q
,
A
.
Za
m
e
e
r
,
a
n
d
A
.
K
h
a
n
,
“
A
t
w
o
-
s
t
a
g
e
st
a
c
k
e
d
e
n
sem
b
l
e
i
n
t
r
u
si
o
n
d
e
t
e
c
t
i
o
n
s
y
st
e
m
u
s
i
n
g
f
i
v
e
b
a
se
c
l
a
ss
i
f
i
e
r
s
a
n
d
M
LP w
i
t
h
o
p
t
i
m
a
l
f
e
a
t
u
r
e
se
l
e
c
t
i
o
n
,
”
Mi
c
r
o
p
ro
c
e
ss
o
rs
a
n
d
M
i
c
r
o
sys
t
e
m
s
,
v
o
l
.
9
4
,
O
c
t
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
mi
c
p
r
o
.
2
0
2
2
.
1
0
4
6
6
0
.
[
2
4
]
P
.
L.
S
.
Ja
y
a
l
a
x
mi
e
t
a
l
.
,
“
D
e
B
o
t
:
A
d
e
e
p
l
e
a
r
n
i
n
g
-
b
a
s
e
d
mo
d
e
l
f
o
r
b
o
t
d
e
t
e
c
t
i
o
n
i
n
i
n
d
u
st
r
i
a
l
i
n
t
e
r
n
e
t
-
of
-
t
h
i
n
g
s
,”
C
o
m
p
u
t
e
rs
a
n
d
El
e
c
t
r
i
c
a
l
En
g
i
n
e
e
r
i
n
g
,
v
o
l
.
1
0
2
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
mp
e
l
e
c
e
n
g
.
2
0
2
2
.
1
0
8
2
1
4
.
[
2
5
]
T.
S
a
b
a
,
A
.
R
e
h
ma
n
,
T.
S
a
d
a
d
,
H
.
K
o
l
i
v
a
n
d
,
a
n
d
S
.
A
.
B
a
h
a
j
,
“
A
n
o
ma
l
y
-
b
a
se
d
i
n
t
r
u
s
i
o
n
d
e
t
e
c
t
i
o
n
sy
s
t
e
m
f
o
r
I
o
T
n
e
t
w
o
r
k
s
t
h
r
o
u
g
h
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
,
”
C
o
m
p
u
t
e
rs
a
n
d
El
e
c
t
r
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
9
9
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
e
l
e
c
e
n
g
.
2
0
2
2
.
1
0
7
8
1
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.