I
AE
S
I
nte
rna
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
8
,
No
.
4
,
Dec
em
b
er
201
9
,
p
p
.
3
60
~
3
66
I
SS
N:
2252
-
8938
,
DOI
: 1
0
.
1
1
5
9
1
/i
j
ai.
v
8
.i
4
.
p
p
3
60
-
3
66
360
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
es
co
r
e
.
co
m
Predic
ting
fata
lit
i
es a
m
o
n
g
sha
r
k
a
tt
a
c
k
s:
co
m
pa
riso
n of
cla
ss
ifiers
L
i
m
M
ei
Sh
i,
Aida
M
us
t
a
ph
a
,
Ya
na
M
a
zw
in
M
o
h
m
a
d H
a
s
s
i
m
F
a
c
u
lt
y
o
f
Co
m
p
u
ter S
c
ien
c
e
a
n
d
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
l
o
g
y
,
Un
iv
e
rs
it
i
T
u
n
Hu
ss
e
in
On
n
M
a
lay
sia
,
P
a
rit
Ra
ja,
8
6
4
0
0
Ba
t
u
P
a
h
a
r,
Jo
h
o
r,
M
a
lay
sia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
A
ug
2
,
2
0
1
9
R
ev
i
s
ed
O
ct
1
0
,
2
0
1
9
A
cc
ep
ted
O
ct
2
4
,
2
0
1
9
T
h
is
p
a
p
e
r
p
re
se
n
ts
th
e
c
o
m
p
a
riso
n
s
o
f
d
if
fe
re
n
t
c
las
si
f
iers
o
n
p
re
d
ictin
g
S
h
a
rk
a
tt
a
c
k
f
a
talit
ies
.
In
th
is
stu
d
y
,
w
e
a
re
c
o
m
p
a
rin
g
t
w
o
c
las
si
fiers
w
h
ich
a
re
S
u
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
s
(S
VMs)
a
n
d
Ba
y
e
s
P
o
in
t
M
a
c
h
i
n
e
s
(BP
M
s)
o
n
S
h
a
rk
a
tt
a
c
k
s
d
a
tas
e
t.
T
h
e
c
o
m
p
a
riso
n
o
f
th
e
c
las
sif
iers
w
e
r
e
b
a
se
d
o
n
t
h
e
a
c
c
u
ra
c
y
,
re
c
a
ll
,
p
re
c
isio
n
a
n
d
F
1
-
sc
o
re
a
s
th
e
p
e
rf
o
rm
a
n
c
e
m
e
a
su
re
m
e
n
t.
T
h
e
re
su
lt
s
o
b
tain
e
d
f
ro
m
th
is
stu
d
y
sh
o
we
d
th
a
t
BP
M
s
p
re
d
icte
d
t
h
e
f
a
talit
y
o
f
sh
a
c
k
a
tt
a
c
k
v
icti
m
e
x
p
e
ri
m
e
n
t
w
it
h
h
ig
h
e
r
a
c
c
u
ra
c
y
a
n
d
p
re
c
isio
n
t
h
a
n
th
e
S
V
M
s
b
e
c
a
u
se
BP
M
s
h
a
v
e
“
a
v
e
ra
g
e
”
id
e
n
ti
f
ier
w
h
ich
c
a
n
m
i
n
im
ize
th
e
p
ro
b
a
b
il
isti
c
e
rr
o
r
m
e
a
su
re
.
F
ro
m
th
is
e
x
p
e
rime
n
t,
it
is
c
o
n
c
lu
d
e
d
t
h
a
t
B
P
M
s
a
re
m
o
re
su
it
a
b
le
in
p
re
d
ictin
g
f
a
talit
y
o
f
sh
a
rk
a
tt
a
c
k
v
i
c
ti
m
a
s
B
P
M
s
is
a
n
im
p
ro
v
e
m
e
n
t
o
f
S
VMs.
K
ey
w
o
r
d
s
:
B
ay
e
s
p
o
in
t
m
ac
h
i
n
es
Data
m
i
n
i
n
g
Ma
ch
i
n
e
lear
n
i
n
g
P
r
ed
ictio
n
Su
p
p
o
r
t v
ec
to
r
m
ac
h
i
n
e
s
Co
p
y
rig
h
t
©
2
0
1
9
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Yan
a
Ma
z
w
i
n
Mo
h
m
ad
Hass
i
m
,
Facu
lt
y
o
f
C
o
m
p
u
ter
Scien
ce
an
d
I
n
f
o
r
m
atio
n
T
ec
h
n
o
lo
g
y
,
Un
i
v
er
s
iti T
u
n
H
u
s
s
ei
n
On
n
Ma
la
y
s
ia,
P
ar
it R
aj
a,
8
6
4
0
0
B
atu
P
ah
at,
J
o
h
o
r
,
Ma
lay
s
ia.
E
m
ail:
y
an
a
@
u
t
h
m
.
ed
u
.
m
y
1.
I
NT
RO
D
UCT
I
O
N
Sh
ar
k
attac
k
ca
n
b
e
ca
te
g
o
r
ized
in
to
t
w
o
t
y
p
e
s
w
h
ich
ar
e
p
r
o
v
o
k
ed
an
d
u
n
p
r
o
v
o
k
ed
.
A
cc
o
r
d
in
g
to
th
e
I
n
ter
n
atio
n
a
l
Sh
ar
k
A
ttac
k
Fil
e
(
I
S
A
F),
8
8
ca
s
es
w
er
e
co
n
f
ir
m
ed
to
b
e
u
n
p
r
o
v
o
k
ed
s
h
ar
k
attac
k
s
o
n
h
u
m
a
n
o
u
t
o
f
th
e
in
v
e
s
ti
g
ated
1
5
5
in
cid
en
ts
o
f
alleg
ed
s
h
ar
k
-
h
u
m
a
n
i
n
t
er
ac
tio
n
o
cc
u
r
r
in
g
w
o
r
ld
w
id
e
in
2
0
1
7
[
1
]
.
T
h
is
r
esu
lt
is
h
ig
h
er
th
a
n
th
e
m
o
s
t
r
e
ce
n
t
y
ea
r
s
(
2
0
1
2
-
2
0
1
6
)
,
w
it
h
av
er
a
g
e
o
f
8
3
in
cid
en
t
s
a
n
n
u
all
y
[
2
]
.
Gr
o
w
i
n
g
n
u
m
b
er
o
f
s
h
ar
k
a
tta
ck
s
h
av
e
ca
u
s
ed
h
u
m
a
n
to
f
ea
r
o
f
s
h
ar
k
[
3
]
.
Sh
ar
k
attac
k
co
u
ld
b
e
f
atal
an
d
n
o
n
-
f
ata
l.
No
n
e
th
ele
s
s
,
o
n
l
y
5
a
m
o
n
g
9
8
u
n
p
r
o
v
o
k
ed
attac
k
s
w
er
e
f
a
tal
w
o
r
l
d
w
id
e
in
2
0
1
5
[
4
]
,
w
h
ich
is
ar
o
u
n
d
5
%
.
Hu
m
a
n
h
as
b
ad
i
m
p
r
ess
io
n
b
ec
au
s
e
t
h
e
y
ten
d
to
th
i
n
k
th
a
t
s
h
ar
k
attac
k
as
th
e
ir
n
atu
r
e
[
5
]
ev
en
th
o
u
g
h
i
n
r
ea
lit
y
d
o
g
s
o
r
b
e
es
k
ill
m
o
r
e
p
eo
p
le
ev
er
y
y
e
ar
th
an
s
h
ar
k
s
[
6
]
.
Un
ited
State
s
is
th
e
lead
i
n
g
co
u
n
tr
y
th
a
t
h
a
s
t
h
e
m
o
s
t
s
h
ar
k
attac
k
,
w
it
h
6
0
p
er
ce
n
t
o
f
th
e
g
lo
b
e’
s
8
8
u
n
p
r
o
v
o
k
ed
s
h
ar
k
attac
k
s
i
n
2
0
1
7
[
7
]
.
Ho
w
e
v
er
,
th
e
U
n
ited
States
d
id
n
o
t
h
av
e
a
n
y
s
h
ar
k
attac
k
s
th
at
r
e
s
u
l
ted
in
f
atalit
y
.
Au
s
tr
alia,
o
n
t
h
e
o
th
er
h
a
n
d
,
h
ad
7
%
f
atalit
y
r
ate
in
s
h
ar
k
attac
k
s
,
w
h
ic
h
m
e
an
s
1
o
u
t
o
f
1
4
in
cid
en
t
s
r
ep
o
r
ted
in
Au
s
tr
alia
h
as
r
es
u
lted
i
n
a
f
atal
it
y
i
n
2
0
1
7
[
8
]
.
T
h
e
n
u
m
b
er
o
f
h
u
m
a
n
-
s
h
ar
k
i
n
ter
ac
tio
n
s
is
d
ir
ec
tl
y
co
r
r
elate
d
w
it
h
ti
m
e
s
p
en
t
b
y
h
u
m
an
s
i
n
th
e
s
ea
[
9
]
.
T
h
e
h
ig
h
er
th
e
n
u
m
b
er
o
f
h
u
m
an
-
s
h
ar
k
in
ter
ac
tio
n
,
t
h
e
h
i
g
h
er
t
h
e
r
is
k
o
f
b
ein
g
attac
k
ed
b
y
a
s
h
ar
k
.
Ho
w
e
v
er
,
o
n
l
y
ce
r
tain
s
p
ec
ies
o
f
s
h
ar
k
attac
k
ar
e
m
o
r
e
li
k
el
y
to
lead
to
f
atalit
y
.
W
h
ile
r
esear
ch
o
n
p
r
ed
ictin
g
s
p
ec
if
icall
y
s
h
ar
k
attac
k
i
s
li
m
ited
s
u
c
h
a
s
i
n
[
1
0
-
1
2
]
,
th
e
lit
er
atu
r
e
h
a
s
s
h
o
w
n
v
ar
io
u
s
d
ata
m
in
in
g
a
p
p
r
o
ac
h
es
u
s
ed
in
a
n
al
y
zi
n
g
f
atalit
ies
o
f
o
t
h
er
an
i
m
al
at
ta
ck
s
s
u
c
h
as
leo
p
ar
d
[
1
3
]
,
elep
h
an
t
[
1
4
]
,
an
d
s
n
a
k
e
[
1
5
]
.
T
h
e
m
a
in
m
o
t
iv
at
io
n
to
p
r
ed
ict
f
atalitie
s
a
m
o
n
g
s
h
ar
k
attac
k
s
is
atr
ib
u
ted
to
th
e
in
cr
ea
s
in
g
n
u
m
b
er
o
f
s
h
ar
k
attac
k
w
o
r
ld
w
id
e
o
v
er
th
e
p
ast
f
iv
e
y
ea
r
s
[
1
6
]
.
Ho
w
ev
er
,
s
o
m
e
s
t
u
d
ie
s
f
o
u
n
d
th
at
m
o
s
t
o
f
th
e
s
h
ar
k
attac
k
s
ar
e
n
o
t
f
a
tal.
I
n
u
n
p
r
o
v
o
k
ed
ca
s
es,
s
h
ar
k
attac
k
s
o
n
l
y
w
h
e
n
th
e
y
ar
e
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
P
r
ed
ictin
g
fa
ta
liti
es a
mo
n
g
s
h
a
r
k
a
tta
ck
s
:
co
mp
a
r
is
o
n
o
f c
l
a
s
s
ifie
r
s
... (
Ya
n
a
Ma
z
w
in
Mo
h
ma
d
Ha
s
s
im
)
361
co
n
f
u
s
ed
a
p
er
s
o
n
w
i
th
n
at
u
r
a
l p
r
e
y
d
u
e
to
it
s
p
o
o
r
v
is
io
n
.
T
h
u
s
,
t
h
e
s
h
ar
k
w
il
l a
p
p
r
o
ac
h
,
b
ite
an
d
s
w
i
m
a
w
a
y
af
ter
v
er
i
f
y
in
g
t
h
at
th
e
v
icti
m
is
n
o
t
p
ar
t
o
f
th
eir
d
iet.
I
n
f
ac
t
,
s
h
ar
k
s
h
a
v
e
n
o
p
ar
ticu
lar
li
k
i
n
g
f
o
r
h
u
m
a
n
f
les
h
as it c
o
n
tai
n
s
a
lo
w
er
lev
el
o
f
f
at
t
h
an
t
h
e
y
n
ee
d
.
B
esid
es,
th
e
f
al
s
e
i
m
p
r
es
s
io
n
o
f
h
u
m
a
n
o
n
s
h
ar
k
m
u
s
t
b
e
co
r
r
ec
ted
.
Sh
ar
k
d
o
es
n
o
t
s
im
p
l
y
at
tac
k
p
eo
p
le
as
w
h
at
m
o
v
ie
s
p
o
r
tr
ay
ed
[
1
7
]
.
Sh
ar
k
attac
k
s
ar
e
o
n
l
y
tr
ig
g
er
ed
w
h
e
n
h
u
m
a
n
d
i
d
ce
r
tain
ac
tio
n
s
o
r
d
o
in
g
s
w
h
ich
m
a
y
co
n
f
u
s
e
o
r
th
r
ea
ten
th
e
s
h
ar
k
.
T
h
e
ac
tiv
itie
s
co
n
d
u
cted
b
y
th
e
v
ict
i
m
w
ill
af
f
ec
t
t
h
e
ch
an
ce
s
o
f
b
ei
n
g
attac
k
ed
b
y
a
s
h
ar
k
.
Fo
llo
w
in
g
r
ec
e
n
t
tr
en
d
s
,
s
u
r
f
er
s
an
d
th
o
s
e
p
ar
ticip
atin
g
i
n
b
o
ar
d
s
p
o
r
ts
ac
co
u
n
ted
f
o
r
m
o
s
t
in
cid
e
n
ts
as
th
is
g
r
o
u
p
s
p
en
d
s
t
h
e
m
o
s
t
ti
m
e
i
n
th
e
s
u
r
f
zo
n
e,
an
ar
ea
co
m
m
o
n
l
y
f
r
eq
u
en
ted
b
y
s
h
ar
k
s
an
d
m
a
y
u
n
i
n
ten
t
io
n
all
y
at
tr
ac
t
s
h
ar
k
s
b
y
s
p
las
h
i
n
g
o
r
p
ad
d
lin
g
.
S
w
i
m
m
er
s
a
n
d
w
ad
er
s
ac
co
u
n
ted
f
o
r
2
2
%
o
f
in
c
id
en
ts
,
9
%
f
o
r
s
n
o
r
k
eler
s
o
r
f
r
ee
d
iv
er
s
,
2
%
f
o
r
s
cu
b
a
d
i
v
er
s
,
3
%
f
o
r
b
o
d
y
-
s
u
r
f
er
s
an
d
5
%
f
o
r
th
o
s
e
p
ar
ticip
ati
n
g
in
o
t
h
er
s
h
allo
w
w
a
ter
ac
ti
v
iti
es.
T
h
e
lo
w
a
w
ar
en
e
s
s
o
f
h
u
m
an
o
n
s
h
ar
k
at
tack
is
th
e
n
e
x
t
m
o
t
iv
atio
n
f
o
r
th
i
s
r
esear
ch
.
Hu
m
an
i
s
n
o
t
a
w
ar
e
o
f
th
e
ef
f
ec
t
o
f
w
ea
r
in
g
b
r
ig
h
t
clo
th
es
w
h
ic
h
w
ill
ca
p
tu
r
e
s
h
ar
k
’
s
a
tten
tio
n
a
s
t
h
e
r
ef
lecte
d
li
g
h
t
f
r
o
m
t
h
e
c
lo
th
es
ca
n
b
e
co
n
f
u
s
ed
w
i
th
th
e
b
r
i
g
h
t
n
es
s
o
f
th
e
f
i
s
h
’
s
s
ca
le
s
.
A
d
d
itio
n
a
ll
y
,
h
u
m
a
n
w
h
o
co
n
d
u
cts
w
at
er
s
p
o
r
ts
ac
ti
v
itie
s
d
u
r
i
n
g
s
h
ar
k
f
ee
d
in
g
ti
m
e
is
ex
p
o
s
ed
to
h
ig
h
er
ch
a
n
ce
s
o
f
b
ein
g
attac
k
ed
as
s
h
ar
k
s
ar
e
m
o
r
e
s
en
s
iti
v
e
ea
r
l
y
in
t
h
e
m
o
r
n
in
g
o
r
late
at
n
ig
h
t.
T
h
u
s
,
t
h
is
s
h
ar
k
attac
k
d
ata
s
e
t
w
ill
also
d
eter
m
i
n
e
i
f
th
e
ti
m
e
o
f
co
n
d
u
cti
n
g
s
p
o
r
ts
ac
ti
v
ities
w
ill
af
f
ec
t
th
e
f
atalit
y
o
f
t
h
e
v
icti
m
o
f
s
h
ar
k
attac
k
.
Oth
er
th
a
n
t
h
at,
th
e
d
ec
r
ea
s
i
n
g
n
u
m
b
er
o
f
w
h
ite
s
h
ar
k
p
r
o
b
lem
h
as
m
o
ti
v
ated
u
s
to
c
h
o
o
s
e
th
i
s
d
ataset
[
1
8
]
.
T
h
e
n
eg
ati
v
e
i
m
ag
e
o
f
th
e
w
h
ite
s
h
ar
k
an
d
th
e
f
ea
r
it
p
r
o
j
ec
ted
o
n
h
u
m
a
n
s
o
f
te
n
r
esu
l
ted
in
u
n
w
a
r
r
an
ted
k
il
lin
g
o
f
th
e
s
p
ec
ies
[
1
9
]
.
T
h
ese
ac
tio
n
s
ar
e
m
ad
e
w
o
r
s
e
b
y
th
e
p
r
o
x
i
m
it
y
o
f
w
h
ite
s
h
ar
k
f
ee
d
in
g
an
d
b
r
ee
d
in
g
ar
ea
s
t
o
co
astal
h
u
m
a
n
p
o
p
u
latio
n
s
o
f
th
e
w
o
r
ld
’
s
s
h
ar
k
s
a
n
d
r
a
y
s
[
2
0
]
.
T
h
u
s
,
th
e
f
atalit
y
o
f
s
h
ar
k
attac
k
s
v
icti
m
s
is
d
eter
m
in
ed
to
u
n
d
er
s
ta
n
d
b
etter
if
s
h
ar
k
is
r
ea
ll
y
a
leth
al
an
i
m
al.
T
h
is
p
ap
er
p
r
esen
ts
t
h
e
co
m
p
ar
is
o
n
s
o
f
d
if
f
er
e
n
t
cla
s
s
i
f
ier
s
o
n
p
r
ed
ictin
g
S
h
ar
k
attac
k
f
atali
ties
.
.
I
n
th
i
s
s
t
u
d
y
,
we
ar
e
co
m
p
ar
i
n
g
t
w
o
clas
s
i
f
ier
s
w
h
ic
h
ar
e
S
u
p
p
o
r
t v
ec
to
r
m
ac
h
in
e
s
(
SV
Ms)
a
n
d
B
a
y
es P
o
in
t M
ac
h
i
n
es
(
B
P
Ms)
b
ased
o
n
f
o
u
r
s
tan
d
ar
d
p
er
f
o
r
m
an
ce
m
ea
s
u
r
e
m
e
n
t
:
ac
c
u
r
ac
y
,
r
ec
all,
p
r
ec
is
io
n
a
n
d
F1
-
s
co
r
e.
W
e
aim
ed
to
s
ee
k
a
b
etter
class
i
f
er
t
h
at
ca
n
b
e
u
s
e
to
p
r
ed
ict
th
e
s
h
ar
k
att
ac
k
f
ata
liti
es
t
h
at
ca
n
h
elp
to
av
o
id
th
is
u
n
w
an
ted
in
cid
en
t i
n
t
h
e
f
u
t
u
r
e.
2.
RE
S
E
ARCH
M
E
T
H
O
D
T
h
is
s
tu
d
y
i
s
ca
r
r
ied
o
u
t
b
as
ed
o
n
SEM
MA
m
e
th
o
d
o
lo
g
y
w
h
ich
co
n
s
is
t
o
f
f
i
v
e
p
h
ase
s
;
s
a
m
p
le
,
ex
p
lo
r
e,
m
o
d
if
y
,
m
o
d
el
an
d
a
s
s
ess
[
2
1
]
.
Fig
u
r
e
1
s
h
o
w
s
th
e
p
r
o
ce
s
s
f
lo
w
o
f
ea
c
h
p
h
a
s
es
i
n
SEM
M
A
m
et
h
o
d
.
Sa
m
p
le
p
h
ase
w
i
ll
s
elec
t
t
h
e
d
ata
an
d
d
eter
m
i
n
e
t
h
e
s
o
u
r
ce
o
f
th
e
d
ataset
w
h
er
ea
s
ex
p
lo
r
i
n
g
t
h
e
in
p
u
t
d
ata
is
p
h
ase
2
.
Sh
ar
k
attac
k
d
ataset
i
s
th
e
i
n
p
u
t n
ee
d
ed
to
d
o
th
e
an
al
y
s
i
s
o
n
f
atalit
y
o
f
s
h
ar
k
attac
k
v
ic
ti
m
.
1
0
o
u
t o
f
2
0
attr
ib
u
tes
w
il
l
b
e
u
s
ed
as
th
e
in
p
u
t
o
f
t
h
is
a
n
al
y
s
is
.
T
h
e
f
o
l
lo
w
i
n
g
m
o
d
i
f
y
p
h
a
s
e
in
c
lu
d
e
th
e
p
r
o
ce
s
s
o
f
p
r
ep
ar
in
g
,
r
ep
air
in
g
an
d
tr
an
s
f
o
r
m
in
g
th
e
d
ata.
T
h
is
d
ataset
h
as
to
b
e
p
r
ep
r
o
ce
s
s
ed
b
ef
o
r
e
ap
p
ly
i
n
g
to
alg
o
r
ith
m
.
Ot
h
er
w
is
e,
t
h
e
r
esu
lt
o
f
an
al
y
s
i
s
w
ill
b
e
af
f
ec
ted
an
d
n
o
t
ac
cu
r
ate.
O
n
th
e
o
t
h
er
h
an
d
,
m
o
d
el
p
h
as
e
w
il
l
u
n
d
er
g
o
t
h
e
p
r
o
ce
s
s
o
f
ap
p
ly
in
g
t
h
e
a
lg
o
r
it
h
m
o
r
tech
n
i
q
u
es to
cr
ea
te
t
h
e
m
o
d
el
t
h
at
p
o
s
s
ib
l
y
p
r
o
v
id
e
t
h
e
d
esire
d
o
u
tco
m
e.
C
lass
if
ica
tio
n
alg
o
r
it
h
m
i
s
u
s
ed
in
t
h
i
s
an
al
y
s
i
s
as
th
e
r
esp
o
n
s
e
class
c
o
n
tain
s
b
in
ar
y
d
ata
w
h
ic
h
is
eit
h
er
f
ata
l
o
r
n
o
n
-
f
atal
.
C
lass
if
ica
tio
n
i
s
w
ell
k
n
o
w
n
s
u
p
er
v
i
s
ed
lear
n
i
n
g
t
ask
t
h
at
h
as
b
ee
n
p
r
ev
io
u
s
l
y
u
s
ed
in
o
th
er
d
o
m
a
in
s
s
u
c
h
as
in
m
u
s
ic
[
2
2
]
,
h
ea
r
t
d
is
ea
s
e
[
2
3
]
,
an
d
tr
af
f
ics
[
2
4
]
.
T
h
e
f
in
al
ass
e
s
s
p
h
ase
w
ill e
v
al
u
ate
th
e
p
er
f
o
r
m
an
ce
o
f
ea
ch
al
g
o
r
ith
m
b
y
u
s
in
g
s
o
m
e
s
ta
n
d
a
r
d
m
etr
ics.
Fig
u
r
e
1
.
SEM
MA
m
et
h
o
d
o
lo
g
y
[
2
1
]
2
.
1
.
Da
t
a
s
et
Sh
ar
k
attac
k
d
ata
is
u
s
ed
to
co
n
d
u
ct
th
is
e
x
p
er
i
m
en
t.
I
n
th
i
s
e
x
p
er
i
m
e
n
t,
6
,
0
9
5
in
s
t
an
ce
s
,
2
2
attr
ib
u
tes
in
cl
u
d
ed
1
class
at
tr
ib
u
te
ar
e
i
n
v
o
l
v
ed
.
T
h
e
2
2
attr
ib
u
tes
ar
e
ca
s
e
n
u
m
b
er
1
,
d
ate,
y
ea
r
,
t
y
p
e,
co
u
n
tr
y
,
ar
ea
,
lo
ca
tio
n
,
ac
tiv
it
y
,
n
a
m
e,
s
e
x
,
a
g
e,
i
n
j
u
r
y
,
f
ata
l,
ti
m
e,
s
p
ec
ies,
in
v
es
tig
a
to
r
o
r
s
o
u
r
ce
,
p
d
f
,
h
r
ef
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
8
,
No
.
4
,
Dec
em
b
er
20
1
9
:
3
6
0
–
3
66
362
f
o
r
m
u
la,
h
r
ef
,
ca
s
e
n
u
m
b
er
2
,
ca
s
e
n
u
m
b
er
3
,
o
r
ig
in
al
o
r
d
er
.
T
h
e
class
at
tr
ib
u
te
n
a
m
el
y
f
a
tal.
Ho
w
e
v
er
,
o
n
l
y
1
0
attr
ib
u
tes
w
h
ic
h
ar
e
t
y
p
e,
c
o
u
n
tr
y
,
ar
ea
,
lo
ca
tio
n
,
ac
ti
v
it
y
,
s
ex
,
ag
e,
f
atal,
ti
m
e
a
n
d
s
p
ec
ies
w
ill
b
e
u
s
ed
to
test
th
e
f
a
talit
y
o
f
v
icti
m
s
i
n
th
e
ca
s
es
o
f
s
h
ar
k
attac
k
.
Oth
er
s
attr
ib
u
tes
s
u
ch
as
ca
s
e
n
u
m
b
er
1
,
d
ate,
y
ea
r
,
n
a
m
e,
i
n
v
esti
g
ato
r
o
r
s
o
u
r
ce
,
p
d
f
,
h
r
ef
f
o
r
m
u
la,
h
r
e
f
,
ca
s
e
n
u
m
b
er
2
,
ca
s
e
n
u
m
b
er
3
an
d
o
r
ig
in
a
l
n
u
m
b
er
w
il
l
n
o
t b
e
test
ed
in
t
h
e
ex
p
er
i
m
e
n
t.
T
h
e
d
ata
d
ictio
n
ar
y
f
o
r
th
is
d
ataset
is
lis
ted
in
T
ab
le
1
.
T
ab
le
1
.
Data
d
ictio
n
ar
y
N
u
mb
e
r
A
t
t
r
i
b
u
t
e
s
D
e
scri
p
t
i
o
n
1
c
a
se
_
n
u
m
b
e
r
D
a
t
e
w
h
e
n
t
h
e
i
n
c
i
d
e
n
t
w
a
s re
p
o
r
t
e
d
2
D
a
t
e
D
a
t
e
w
h
e
n
t
h
e
i
n
c
i
d
e
n
t
w
a
s h
a
p
p
e
n
e
d
3
Y
e
a
r
Y
e
a
r
w
h
e
n
t
h
e
i
n
c
i
d
e
n
t
w
a
s h
a
p
p
e
n
e
d
4
Ty
p
e
Ty
p
e
o
f
sh
a
r
k
a
t
t
a
c
k
5
c
o
u
n
t
r
y
C
o
u
n
t
r
y
w
h
e
r
e
sh
a
r
k
a
t
t
a
c
k
t
o
o
k
p
l
a
c
e
6
A
r
e
a
A
r
e
a
w
h
e
r
e
sh
a
r
k
a
t
t
a
c
k
t
o
o
k
p
l
a
c
e
7
l
o
c
a
t
i
o
n
L
o
c
a
t
i
o
n
w
h
e
r
e
sh
a
r
k
a
t
t
a
c
k
t
o
o
k
p
l
a
c
e
8
a
c
t
i
v
i
t
y
A
c
t
i
v
i
t
y
c
o
n
d
u
c
t
e
d
b
y
t
h
e
v
i
c
t
i
m
9
N
a
me
N
a
me
o
f
v
i
c
t
i
m
10
S
e
x
S
e
x
o
f
v
i
c
t
i
m
11
A
g
e
A
g
e
o
f
v
i
c
t
i
m
12
I
n
j
u
r
y
Ty
p
e
s o
f
i
n
j
u
r
y
h
a
p
p
e
n
e
d
t
o
t
h
e
v
i
c
t
i
m
13
f
a
t
a
l
_
y
_
n
F
a
t
a
l
i
t
y
o
f
v
i
c
t
i
m
14
T
i
me
T
i
me
o
f
i
n
c
i
d
e
n
t
15
sp
e
c
i
e
s
S
p
e
c
i
e
s o
f
sh
a
r
k
16
i
n
v
e
st
i
g
a
t
o
r
_
o
r
_
so
u
r
c
e
I
n
v
e
st
i
g
a
t
o
r
i
n
v
o
l
v
e
d
i
n
t
h
e
c
a
se
17
P
d
f
C
a
se
n
a
me
18
h
r
e
f
_
f
o
r
mu
l
a
R
e
f
e
r
e
n
c
e
h
y
p
e
r
l
i
n
k
19
H
r
e
f
R
e
f
e
r
e
n
c
e
h
y
p
e
r
l
i
n
k
2
20
c
a
se
_
n
u
m
b
e
r
_
2
R
e
l
a
t
e
d
c
a
se
n
u
mb
e
r
2
21
c
a
se
_
n
u
m
b
e
r
_
3
R
e
l
a
t
e
d
c
a
se
n
u
mb
e
r
3
22
o
r
i
g
i
n
a
l
_
o
r
d
e
r
O
r
d
e
r
o
f
f
i
l
e
T
h
e
s
h
ar
k
attac
k
d
atase
t
w
as
s
o
u
r
ce
d
f
r
o
m
a
w
eb
s
ite
ca
lled
d
ata.
w
o
r
ld
.
Data
.
w
o
r
ld
is
a
web
s
ite
th
a
t
co
n
tain
s
n
u
m
er
o
u
s
d
i
f
f
er
e
n
t
t
y
p
es
o
f
r
a
w
d
atasets
,
p
u
b
l
is
h
ed
b
y
n
u
m
er
o
u
s
co
n
tr
ib
u
t
o
r
s
f
r
o
m
d
if
f
er
en
t
co
u
n
tr
ies.
I
t
is
a
p
latf
o
r
m
f
o
r
p
eo
p
le
t
o
co
llab
o
r
ate,
c
o
n
tr
ib
u
te
an
d
s
o
lv
e
p
r
o
b
lem
s
r
elati
n
g
to
d
atasets
.
Ne
w
d
ata
r
an
g
i
n
g
f
r
o
m
f
in
a
n
ce
to
h
ea
lth
to
s
p
o
r
ts
an
d
p
o
liti
cs
ca
n
b
e
d
is
co
v
er
ed
f
r
o
m
v
ar
io
u
s
s
o
u
r
ce
s
th
r
o
u
g
h
th
i
s
w
eb
s
i
te.
User
w
o
u
ld
o
n
l
y
n
ee
d
to
f
ilter
t
h
e
s
ea
r
c
h
es
ac
co
r
d
in
g
to
th
eir
p
r
e
f
er
en
ce
s
.
T
h
is
d
ataset
is
co
llected
f
r
o
m
Glo
b
al
Sh
ar
k
Attack
Fil
e
[
2
5
]
.
T
h
e
ai
m
o
f
t
h
i
s
w
eb
s
i
t
e
is
to
p
r
o
v
id
e
c
u
r
r
en
t
an
d
h
i
s
t
o
r
ical
d
ata
o
n
s
h
ar
k
o
r
h
u
m
an
in
ter
ac
tio
n
s
f
o
r
th
o
s
e
w
h
o
s
ee
k
ac
cu
r
ate
a
n
d
m
ea
n
in
g
f
u
l
i
n
f
o
r
m
atio
n
a
n
d
v
er
i
f
i
ab
le
r
ef
er
en
ce
s
.
O
n
th
e
o
t
h
er
h
an
d
,
t
h
i
s
w
o
r
k
s
p
ac
e
is
co
n
tr
ib
u
ted
b
y
Sh
r
u
ti
J
a
y
ar
a
m
P
r
ab
h
u
o
n
2
2
J
u
n
e
2
0
1
7
.
T
h
is
d
ataset
is
p
u
b
licl
y
s
h
ar
ed
an
d
co
n
tai
n
s
s
h
ar
k
attac
k
r
ep
o
r
tin
g
s
f
r
o
m
o
v
er
a
ce
n
tu
r
y
.
2
.
2
.
P
re
-
P
ro
ce
s
s
ing
B
ef
o
r
e
b
u
ild
in
g
th
e
clas
s
i
f
ica
tio
n
m
o
d
el
u
s
i
n
g
t
h
e
d
ata
s
et,
th
e
d
ataset
w
as
f
ir
s
t
p
r
e
-
p
r
o
ce
s
s
ed
to
ca
ter
is
s
u
es
s
u
c
h
as
m
i
s
s
i
n
g
an
d
co
n
ti
n
u
o
u
s
v
al
u
es.
Data
t
o
b
e
an
al
y
ze
d
b
y
d
ata
m
in
in
g
tech
n
iq
u
es
ca
n
b
e
in
co
m
p
lete,
n
o
is
y
a
n
d
i
n
co
n
s
is
ten
t.
T
h
e
p
u
r
p
o
s
e
o
f
d
ata
cl
ea
n
in
g
is
to
clea
n
t
h
e
d
ata
to
b
e
an
al
y
ze
d
[
2
6
]
.
T
h
er
e
ar
e
m
an
y
d
i
f
f
er
en
t
clea
n
in
g
m
o
d
es
av
a
ilab
le
f
o
r
u
s
er
to
s
elec
t
s
u
c
h
a
s
r
e
m
o
v
i
n
g
t
h
e
en
tire
co
lu
m
n
o
r
r
e
m
o
v
i
n
g
t
h
e
e
n
tire
r
o
w
.
I
n
s
h
ar
k
attac
k
r
a
w
d
ataset,
m
a
n
y
m
is
s
in
g
v
a
lu
e
s
w
er
e
f
o
u
n
d
an
d
th
ese
m
i
s
s
i
n
g
v
alu
e
s
w
o
u
ld
b
e
r
e
m
o
v
ed
i
n
o
r
d
er
to
p
r
o
v
id
e
a
b
etter
an
d
ac
cu
r
ate
r
e
s
u
lt
d
u
r
i
n
g
m
i
n
in
g
p
r
o
ce
s
s
.
Fig
u
r
e
2
s
h
o
w
s
o
n
e
o
f
th
e
at
tr
ib
u
te
w
it
h
m
i
s
s
in
g
v
al
u
e
i
n
s
h
ar
k
attac
k
d
ataset.
I
n
[
2
7
]
,
SMOT
E
m
o
d
u
le
w
as
u
s
ed
to
tr
ea
t
th
e
i
m
b
ala
n
ce
d
d
ataset.
T
h
is
is
b
ec
a
u
s
e
t
h
is
s
h
ar
k
attac
k
d
ataset
is
b
ias
to
n
o
n
-
f
atal
o
f
s
h
ar
k
attac
k
v
ict
i
m
.
T
h
u
s
,
t
h
is
i
m
b
ala
n
ce
d
d
ataset
n
ee
d
s
to
b
e
co
r
r
ec
te
d
to
p
r
o
d
u
ce
ac
cu
r
ate
r
esu
lt
d
u
r
i
n
g
an
al
y
s
is
.
Da
ta
tr
an
s
f
o
r
m
atio
n
is
u
s
ed
to
tr
an
s
f
o
r
m
o
r
co
n
s
o
lid
ate
th
e
d
ata
in
to
f
o
r
m
s
ap
p
r
o
p
r
iate
f
o
r
m
in
in
g
[
2
8
]
.
T
h
e
t
y
p
e,
ac
tiv
i
t
y
,
s
ex
,
s
p
ec
ies
a
n
d
ag
e
a
ttri
b
u
te
s
i
n
s
h
ar
k
at
tack
d
ataset
w
o
u
ld
b
e
tr
an
s
f
o
r
m
ed
to
ca
teg
o
r
ical
f
ea
tu
r
e
t
y
p
e.
T
h
is
is
b
ec
au
s
e
t
h
e
v
al
u
es
o
f
d
ata
in
t
h
o
s
e
attr
ib
u
tes
ca
n
b
e
s
o
r
ted
in
to
g
r
o
u
p
s
o
r
ca
te
g
o
r
ies.
C
ate
g
o
r
ical
d
ata
m
u
s
t
b
e
c
ast
ca
te
g
o
r
ies
s
o
th
at
t
h
e
co
m
p
u
ter
ca
n
tr
ea
t
t
h
e
m
co
r
r
ec
tly
w
h
e
n
u
s
i
n
g
clas
s
i
f
ic
atio
n
al
g
o
r
ith
m
.
Fi
g
u
r
e
3
s
h
o
w
s
o
n
e
o
f
t
h
e
attr
ib
u
te
t
h
at
i
s
tr
an
s
f
o
r
m
ed
f
r
o
m
s
tr
in
g
f
ea
tu
r
e
t
y
p
e
to
ca
te
g
o
r
ical
f
ea
t
u
r
e
t
y
p
e
in
o
r
d
er
to
u
s
e
t
h
e
class
if
icatio
n
lear
n
in
g
.
Nex
t,
d
ata
r
ed
u
ct
io
n
i
s
u
s
ed
to
r
ed
u
ce
th
e
n
u
m
b
er
o
f
at
tr
i
b
u
t
es
w
h
ich
ar
e
n
o
t
r
ele
v
a
n
t
to
t
h
e
an
al
y
s
i
s
w
it
h
o
u
t
co
m
p
r
o
m
i
s
in
g
t
h
e
in
t
eg
r
it
y
o
f
th
e
o
r
ig
in
al
d
ata
a
n
d
y
et
p
r
o
d
u
cin
g
th
e
q
u
a
lit
y
k
n
o
w
led
g
e
[
2
9
]
.
T
h
is
is
to
r
ed
u
ce
th
e
co
m
p
lex
it
y
o
f
d
ata,
m
a
k
i
n
g
th
e
a
n
al
y
s
i
s
p
r
o
ce
s
s
b
ec
o
m
e
q
u
ic
k
er
an
d
i
n
cr
ea
s
e
its
e
f
f
icie
n
c
y
.
On
l
y
1
0
attr
ib
u
te
s
o
u
t
o
f
2
2
attr
ib
u
tes
w
o
u
ld
b
e
u
s
ed
i
n
t
h
is
a
n
al
y
s
is
.
T
h
e
i
n
j
u
r
y
,
ca
s
e
n
u
m
b
er
,
d
ate,
y
ea
r
,
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
P
r
ed
ictin
g
fa
ta
liti
es a
mo
n
g
s
h
a
r
k
a
tta
ck
s
:
co
mp
a
r
is
o
n
o
f c
l
a
s
s
ifie
r
s
... (
Ya
n
a
Ma
z
w
in
Mo
h
ma
d
Ha
s
s
im
)
363
co
u
n
tr
y
,
ar
ea
,
lo
ca
tio
n
,
n
a
m
e,
in
v
e
s
ti
g
ato
r
o
r
s
o
u
r
ce
p
d
f
,
h
r
e
f
f
o
r
m
u
la,
h
r
e
f
,
ca
s
e
n
u
m
b
er
(
2
)
,
ca
s
e
n
u
m
b
er
(
3
)
an
d
o
r
ig
in
a
l
o
r
d
er
ar
e
ex
clu
d
e
d
in
th
is
a
n
al
y
s
i
s
.
T
h
is
is
b
e
ca
u
s
e
t
h
ese
a
ttrib
u
te
s
ar
e
n
o
t
s
ig
n
i
f
ica
n
t
an
d
n
o
t
u
s
e
f
u
l
in
th
i
s
an
a
l
y
s
is
.
T
y
p
e,
co
u
n
tr
y
,
ar
ea
,
lo
ca
tio
n
,
ac
ti
v
it
y
,
s
e
x
,
ag
e,
ti
m
e,
s
p
ec
ies
a
n
d
f
atalit
y
o
f
v
icti
m
s
attr
ib
u
tes
ar
e
u
s
e
f
u
l
in
t
h
is
a
n
al
y
s
i
s
an
d
th
u
s
t
h
e
y
ar
e
in
cl
u
d
ed
in
th
is
an
al
y
s
is
.
Fi
g
u
r
e
4
s
h
o
w
s
t
h
e
attr
ib
u
te
u
s
ed
in
a
n
al
y
s
i
s
.
Fig
u
r
e
2
.
Miss
i
n
g
v
al
u
es o
f
f
at
al
attr
ib
u
te
Fig
u
r
e
3
.
C
ateg
o
r
ical
f
ea
tu
r
e
t
y
p
e
o
f
f
atal
attr
ib
u
te
Fig
u
r
e
4
.
A
ttrib
u
te
u
s
ed
in
an
al
y
s
is
Dif
f
er
en
t
a
lg
o
r
it
h
m
w
o
u
ld
r
e
q
u
ir
e
d
if
f
er
en
t
s
p
ec
if
ic
co
n
ten
t
t
y
p
es
i
n
o
r
d
er
to
f
u
n
ctio
n
co
r
r
ec
tly
.
I
n
d
ata
d
is
cr
etiza
tio
n
,
v
alu
e
s
ar
e
p
u
t
in
to
b
u
c
k
ets
s
o
th
a
t
th
er
e
ar
e
a
lim
i
ted
n
u
m
b
er
o
f
p
o
s
s
i
b
le
s
tates
[
3
0
]
.
Data
in
t
h
e
co
l
u
m
n
s
ca
n
b
e
d
is
cr
eti
ze
d
to
en
a
b
le
t
h
e
u
s
e
o
f
t
h
e
al
g
o
r
ith
m
s
to
p
r
o
d
u
ce
m
i
n
i
n
g
m
o
d
el.
Ag
e
attr
ib
u
t
e
in
s
h
ar
k
attac
k
d
ataset
w
o
u
ld
b
e
ca
teg
o
r
ized
an
d
co
n
v
er
ted
to
h
ig
h
er
co
n
ce
p
tu
al
lev
e
l
th
r
o
u
g
h
t
h
e
lev
el
o
f
h
ier
ar
ch
ie
s
.
T
h
e
v
alu
es
f
o
r
ag
e
attr
ib
u
te
w
o
u
ld
b
e
d
iv
id
ed
i
n
to
s
ev
er
al
ca
teg
o
r
i
es
w
it
h
f
i
x
ed
s
ize
o
f
in
ter
v
al.
W
ith
th
i
s
b
ein
g
d
o
n
e,
th
e
d
ep
en
d
en
c
y
b
et
w
ee
n
t
h
e
class
a
n
d
th
e
in
ter
v
a
l
ar
e
in
cr
ea
s
ed
an
d
p
r
o
v
id
e
a
m
o
r
e
ac
cu
r
ate
r
esu
l
t.
2
.
3
.
Alg
o
rit
h
m
s
T
h
is
p
ap
er
ad
o
p
ted
th
e
cl
ass
i
f
icatio
n
tech
n
iq
u
e
f
o
r
p
r
ed
ictin
g
s
h
ar
k
attac
k
f
atali
ti
es.
T
h
e
ex
p
er
i
m
e
n
ts
w
er
e
ca
r
r
ied
o
u
t
u
s
i
n
g
t
h
e
A
z
u
r
e
M
L
to
o
l
[
3
1
]
w
it
h
1
0
-
f
o
ld
v
a
lid
atio
n
m
et
h
o
d
to
ev
alu
a
te
t
h
e
SVM
an
d
B
P
M
class
i
f
ier
s
p
e
r
f
o
r
m
an
ce
.
C
r
o
s
s
-
v
a
lid
atio
n
m
o
d
el
m
o
d
u
le
w
a
s
u
s
ed
in
A
zu
r
e
ML
to
p
er
f
o
r
m
th
is
v
alid
atio
n
p
r
o
ce
s
s
.
C
r
o
s
s
-
v
al
id
atio
n
p
a
r
a
m
eter
p
ar
titi
o
n
ed
th
e
d
ata
in
to
1
0
f
o
ld
s
to
esti
m
ate
f
o
r
class
i
f
icatio
n
m
o
d
el.
9
s
et
s
wer
e
u
s
ed
to
tr
ain
th
e
cla
s
s
i
f
ier
w
h
ile
t
h
e
p
er
f
o
r
m
an
ce
o
f
cla
s
s
i
f
ier
w
as
a
s
s
e
s
s
ed
o
n
t
h
e
1
le
f
t
s
u
b
s
e
t.
T
h
is
w
a
s
th
e
n
iter
ated
ten
ti
m
es
a
s
s
u
b
s
ets
w
er
e
i
n
cl
u
d
ed
in
tr
ain
in
g
a
n
d
test
s
et
s
.
T
h
e
av
er
ag
e
p
er
f
o
r
m
a
n
ce
i
s
co
n
s
id
er
ed
as
th
e
f
in
a
l
p
er
f
o
r
m
a
n
ce
o
f
a
c
lass
if
ier
.
Qu
al
it
y
o
f
d
ata
s
et
ca
n
b
e
d
eter
m
in
ed
b
y
co
m
p
ar
in
g
t
h
e
ac
cu
r
ac
y
s
tati
s
tics
f
o
r
all
t
h
e
f
o
ld
s
.
Sin
ce
t
h
e
s
h
ar
k
d
ataset
u
s
ed
o
n
l
y
h
a
v
e
t
w
o
class
es
;
f
atal
o
r
n
o
t
f
a
tal,
t
wo
-
class
t
y
p
e
o
f
al
g
o
r
ith
m
w
a
s
b
e
s
elec
ted
f
o
r
th
e
class
if
i
ca
tio
n
ex
p
er
i
m
e
n
t.
Su
p
p
o
r
t
v
ec
to
r
m
ac
h
i
n
es
(
SV
Ms)
an
d
B
a
y
es
P
o
in
t
Ma
ch
i
n
es
(
B
P
Ms)
ca
n
b
e
u
s
ed
i
n
th
e
class
i
f
icatio
n
o
f
s
u
p
er
v
i
s
ed
lear
n
i
n
g
d
atase
t.
B
P
Ms
ar
e
a
t
y
p
e
o
f
l
in
ea
r
cla
s
s
i
f
ica
tio
n
al
g
o
r
ith
m
w
h
ic
h
w
a
s
in
tr
o
d
u
ce
d
b
y
R
al
f
Her
b
r
ix
h
,
T
h
eo
r
e
Gr
ap
el,
an
d
C
o
lin
C
a
m
p
b
ell
i
n
2
0
0
1
.
B
PMs
ar
e
k
n
o
w
n
as
a
n
“
a
v
er
ag
e
”
class
i
f
ier
t
h
at
ca
n
ef
f
icien
tl
y
ap
p
r
o
x
i
m
ate
t
h
e
th
eo
r
etica
l
o
p
tim
al
B
a
y
esia
n
av
er
ag
e
o
f
s
e
v
er
al
li
n
ea
r
class
i
f
ier
s
b
ased
o
n
th
ei
r
ab
ilit
y
to
g
e
n
er
alize
[
1
0
]
.
T
h
is
class
if
ier
i
s
u
s
ed
to
m
i
n
i
m
iz
e
th
e
p
r
o
b
ab
ilis
tic
er
r
o
r
m
ea
s
u
r
e.
T
h
e
“
av
er
a
g
e”
class
i
f
ier
is
k
n
o
w
n
as
B
a
y
es
p
o
in
t.
B
P
Ms
alg
o
r
ith
m
u
s
ed
in
A
z
u
r
e
Ma
ch
i
n
e
L
ea
r
n
i
n
g
ar
e
b
ased
o
n
I
n
f
er
t.N
e
t
an
d
ca
n
p
er
f
o
r
m
b
etter
th
a
n
t
h
e
o
th
er
B
ay
e
s
ian
al
g
o
r
ith
m
.
Nu
m
b
er
o
f
iter
atio
n
o
f
test
c
an
b
e
s
et
i
n
A
zu
r
e
M
L
.
T
h
e
h
ig
h
er
n
u
m
b
er
o
f
iter
atio
n
s
,
t
h
e
h
i
g
h
er
th
e
ac
c
u
r
ac
y
o
f
th
e
r
es
u
lt.
B
P
Ms
ar
e
m
o
r
e
r
o
b
u
s
t
a
n
d
less
p
r
o
n
e
to
o
v
er
-
f
itti
n
g
o
f
th
e
d
ata
w
h
er
eb
y
th
e
p
r
o
d
u
ctio
n
o
f
a
n
al
y
s
i
s
ar
e
to
o
s
i
m
ilar
to
t
h
e
d
ata
an
d
ca
u
s
i
n
g
it
to
f
ail
to
f
it
ad
d
iti
o
n
al
d
ata
o
r
p
r
ed
ict
f
u
tu
r
e
o
b
s
er
v
atio
n
s
r
eliab
l
y
.
I
t
ca
n
also
r
ed
u
ce
th
e
n
ee
d
to
p
er
f
o
r
m
p
er
f
o
r
m
an
ce
t
u
n
in
g
s
an
d
th
er
ef
o
r
e
ti
m
e
n
ee
d
ed
to
r
u
n
t
h
e
e
x
p
er
i
m
e
n
t
ca
n
b
e
d
ec
r
ea
s
ed
.
E
x
p
ec
tatio
n
p
r
o
p
ag
atio
n
i
s
u
s
ed
i
n
B
P
Ms
as
t
h
e
m
es
s
ag
e
-
p
ass
in
g
a
lg
o
r
it
h
m
w
h
ich
p
ass
es
t
h
e
m
es
s
ag
e
to
o
th
er
n
o
d
es
ac
r
o
s
s
th
e
ed
g
es
o
f
m
o
d
el
a
n
d
t
h
u
s
p
r
o
d
u
ce
s
a
f
ast
a
n
d
ac
c
u
r
ate
r
esu
lt
[
3
2
]
.
SVM
al
g
o
r
ith
m
w
a
s
i
n
tr
o
d
u
ce
d
[
3
3
]
.
T
h
is
alg
o
r
ith
m
w
i
ll
ass
i
g
n
d
ata
to
o
n
e
class
o
r
th
e
o
th
e
r
b
y
d
is
co
v
er
i
n
g
h
y
p
er
p
lan
e
s
w
h
ich
clea
n
l
y
s
eg
r
e
g
ate
d
ata
in
to
class
e
s
[
3
4
]
.
Ne
w
d
ata
p
o
in
ts
ca
n
b
e
ea
s
il
y
cla
s
s
i
f
ied
o
n
ce
i
d
ea
l h
y
p
er
p
lan
es a
r
e
d
is
co
v
er
ed
.
T
w
o
-
C
la
s
s
B
a
y
e
s
P
o
in
t M
ac
h
i
n
e.
T
h
e
f
o
r
m
u
la
is
s
h
o
w
n
i
n
(
1
)
,
w
h
er
e
A
an
d
B
ar
e
ev
en
t
s
an
d
P(
B
)
≠
0
.
P
(
A
|
)
=
(
|
)
(
)
(
)
(
1
)
T
w
o
-
C
la
s
s
S
u
p
p
o
r
t V
ec
to
r
M
ac
h
in
e.
T
h
e
f
o
r
m
u
la
i
s
s
h
o
w
n
in
(
2
)
.
|
|
|
|
.
(
x
2
−
1
)
=
ℎ
=
2
|
|
|
|
(
2
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
8
,
No
.
4
,
Dec
em
b
er
20
1
9
:
3
6
0
–
3
66
364
2
.
4
.
E
v
a
lua
t
i
o
n
M
et
rics
T
h
e
ev
al
u
atio
n
m
e
tr
ics
u
s
ed
in
t
h
e
e
x
p
er
i
m
e
n
ts
ar
e
ac
c
u
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all
an
d
F1
s
co
r
e.
A
cc
u
r
ac
y
p
er
f
o
r
m
s
b
e
s
t
i
f
f
als
e
p
o
s
itiv
e
s
an
d
f
al
s
e
n
e
g
ati
v
es
h
a
v
e
s
i
m
ilar
co
s
t
[
3
5
]
.
P
r
ec
is
i
o
n
an
d
R
ec
all
w
i
l
l
b
e
u
s
ed
if
th
e
co
s
t
o
f
f
alse
p
o
s
itiv
es
a
n
d
f
alse
n
eg
a
tiv
e
s
ar
e
v
er
y
d
i
f
f
er
e
n
t.
T
r
u
e
p
o
s
itiv
e
h
ap
p
en
s
w
h
e
n
p
r
ed
icted
class
an
d
ac
tu
a
l
clas
s
ar
e
t
r
u
e
w
h
er
ea
s
tr
u
e
n
e
g
ati
v
e
h
ap
p
en
s
w
h
en
p
r
ed
icted
class
a
n
d
ac
tu
al
c
las
s
ar
e
b
o
th
f
alse.
Fals
e
p
o
s
iti
v
e
h
ap
p
en
s
w
h
e
n
ac
tu
a
l
class
i
s
f
alse
b
u
t
p
r
ed
icted
class
is
tr
u
e
w
h
er
ea
s
f
a
ls
e
n
eg
at
iv
e
h
ap
p
en
s
w
h
e
n
p
r
ed
icted
class
is
f
al
s
e
b
u
t
ac
t
u
al
class
i
s
tr
u
e.
O
n
t
h
e
o
t
h
er
h
a
n
d
,
F1
s
co
r
e
i
s
u
s
ed
w
h
e
n
m
o
r
e
r
ea
lis
tic
m
ea
s
u
r
e
o
f
class
i
f
ier
’
s
p
er
f
o
r
m
a
n
ce
is
r
eq
u
ir
ed
as
ar
ith
m
etic
m
e
an
b
et
w
ee
n
a
p
o
o
r
p
r
ec
is
io
n
an
d
a
v
er
y
h
i
g
h
r
ec
a
ll c
an
b
e
av
o
id
ed
[
3
6
]
.
A
cc
u
r
ac
y
.
A
cc
u
r
ac
y
is
th
e
r
at
io
o
f
s
u
m
m
atio
n
o
f
tr
u
e
p
o
s
itiv
e
an
d
tr
u
e
n
e
g
ati
v
e
to
th
e
to
tal
ev
en
ts
.
T
h
e
f
o
r
m
u
la
f
o
r
ca
lcu
lati
n
g
ac
c
u
r
ac
y
is
s
h
o
w
n
in
(
3
)
.
A
c
c
ura
c
y
=
T
r
ue
Po
s
i
t
i
v
e
+
T
r
ue
N
eg
at
i
v
e
T
o
t
al
ev
en
t
s
100%
(
3
)
P
r
ec
is
io
n
.
P
r
ec
is
io
n
is
th
e
r
atio
o
f
tr
u
e
p
o
s
itiv
e
to
th
e
to
tal
p
r
ed
icted
p
o
s
itiv
e
o
b
s
er
v
atio
n
s
.
T
h
e
f
o
r
m
u
la
f
o
r
ca
lcu
lati
n
g
p
r
ec
is
io
n
is
s
h
o
w
n
i
n
(
4
)
.
Pr
e
c
ision
=
T
r
ue
Po
s
i
t
i
v
e
T
r
ue
Po
s
i
t
i
v
e
+
F
al
s
e
Po
s
i
t
i
v
e
100%
(
4
)
R
ec
all.
R
ec
a
ll
(
Se
n
s
it
iv
i
t
y
)
i
s
th
e
r
a
tio
o
f
co
r
r
ec
tl
y
p
r
ed
icted
p
o
s
itiv
e
o
b
s
er
v
atio
n
s
to
th
e
all
o
b
s
er
v
atio
n
s
in
ac
tu
a
l c
lass
.
T
h
e
f
o
r
m
u
la
f
o
r
ca
lcu
lati
n
g
r
ec
all
is
s
h
o
w
n
i
n
(
5
)
.
R
e
c
a
l
l
=
T
r
ue
Po
s
i
t
i
v
e
T
r
ue
Po
s
i
t
i
v
e
+
F
al
s
e
N
eg
at
i
v
e
100%
(
5
)
F1
s
co
r
e.
F1
s
co
r
e
is
th
e
av
er
ag
e
o
f
p
r
ec
is
io
n
a
n
d
r
ec
all,
it
r
ea
ch
es
it
s
b
est
v
al
u
e
at
1
an
d
w
o
r
s
t
a
t
0
.
T
h
e
f
o
r
m
u
la
f
o
r
ca
lcu
lati
n
g
F1
s
co
r
e
is
s
h
o
w
n
in
(
6
)
.
F1
s
c
or
e
=
2
x
p
r
ecis
i
o
n
x
r
ecal
l
p
r
e
cis
i
o
n
+
r
ecal
l
(
6
)
3.
RE
SU
L
T
S
A
ND
D
I
SCU
SS
I
O
N
T
h
e
p
u
r
p
o
s
e
o
f
t
h
e
e
x
p
er
i
m
e
n
ts
i
s
to
co
m
p
ar
e
t
h
e
p
er
f
o
r
m
an
ce
o
f
B
a
y
e
s
P
o
in
t
Ma
c
h
i
n
e
a
n
d
Su
p
p
o
r
t
Vec
to
r
Ma
ch
in
e
al
g
o
r
ith
m
s
i
n
s
h
ar
k
attac
k
d
ataset
f
o
r
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all
a
n
d
F
1
s
co
r
e.
T
h
e
r
esu
lts
s
h
o
w
ed
t
h
at
B
a
y
e
s
P
o
in
t
Ma
ch
in
e
p
er
f
o
r
m
s
b
etter
i
n
t
h
is
d
ataset
co
m
p
ar
ed
to
S
u
p
p
o
r
t
Vec
to
r
Ma
ch
i
n
e.
B
P
Ms
h
av
e
h
i
g
h
er
ac
cu
r
ac
y
th
an
SVMs.
T
h
is
is
b
ec
a
u
s
e
B
ay
e
s
-
o
p
ti
m
al
clas
s
i
f
ier
w
ill
m
i
n
i
m
ize
av
er
a
g
e
er
r
o
r
w
h
e
n
m
ar
g
i
n
alizi
n
g
o
v
e
r
all
p
o
s
s
ib
le
b
o
u
n
d
ar
ies
an
d
all
p
o
s
s
ib
le
s
a
m
p
li
n
g
s
o
f
t
h
e
d
ata
b
y
f
in
d
i
n
g
th
e
b
o
u
n
d
ar
y
in
a
f
ix
ed
s
p
ac
e
w
h
ich
is
clo
s
e
s
t
to
th
is
cla
s
s
i
f
ier
.
B
esid
es,
B
P
Ms
also
h
as
a
h
ig
h
er
p
r
ec
is
io
n
t
h
a
n
SVMs.
T
h
is
is
b
ec
a
u
s
e
t
h
e
s
a
m
p
lin
g
s
c
h
e
m
e
u
s
ed
in
B
P
Ms
is
v
er
y
s
i
m
p
le
an
d
ef
f
icie
n
t,
t
h
u
s
m
a
k
i
n
g
it
to
b
e
ap
p
licab
le
to
lar
g
e
d
ata
s
ets s
u
ch
as s
h
ar
k
attac
k
d
ataset.
T
h
e
s
u
m
m
ar
y
ar
e
s
h
o
w
n
i
n
T
ab
le
2
.
T
ab
le
2
.
E
x
p
er
im
e
n
tal
r
esu
lts
A
l
g
o
r
i
t
h
m
A
c
c
u
r
a
c
y
P
r
e
c
i
si
o
n
R
e
c
a
l
l
F
-
M
e
a
su
r
e
Tw
o
-
C
l
a
ss B
a
y
e
s Po
i
n
t
M
a
c
h
i
n
e
0
.
9
5
2
0
.
8
9
9
0
.
9
8
3
0
.
9
3
9
Tw
o
-
C
l
a
ss Su
p
p
o
r
t
V
e
c
t
o
r
M
a
c
h
i
n
e
0
.
8
1
6
0
.
7
5
4
0
.
7
6
2
0
.
7
5
8
Tw
o
-
C
l
a
ss
L
o
g
i
st
i
c
R
e
g
r
e
ssi
o
n
0
.
8
0
3
0
.
7
4
0
0
.
7
4
0
0
.
7
4
0
Tw
o
-
C
l
a
ss B
o
o
st
e
d
D
e
c
i
si
o
n
T
r
e
e
0
.
8
5
4
0
.
7
9
4
0
.
8
2
9
0
.
8
1
1
Oth
er
th
a
n
t
h
at,
B
P
Ms h
av
e
a
h
ig
h
er
r
ec
all
v
a
lu
e
co
m
p
ar
ed
t
o
SVMs.
T
h
is
is
b
ec
au
s
e
B
P
Ms p
r
o
p
o
s
e
a
n
o
v
el
d
if
f
er
e
n
tiab
le
lo
s
s
f
u
n
ctio
n
ca
lled
tr
ig
o
n
o
m
etr
ic
lo
s
s
f
u
n
ctio
n
w
h
ic
h
w
ill
n
o
r
m
a
li
ze
th
e
li
k
eli
n
es
s
o
f
d
esira
b
le
ch
ar
ac
ter
is
tic
b
ef
o
r
e
s
ettin
g
u
p
a
B
a
y
e
s
ia
n
f
r
a
m
e
w
o
r
k
u
s
i
n
g
s
tan
d
ar
d
Gau
s
s
ia
n
p
r
o
ce
s
s
es
tech
n
iq
u
es.
B
P
Ms
h
a
v
e
h
i
g
h
er
ac
cu
r
ac
y
th
a
n
t
h
at
o
f
L
o
g
is
t
ic
R
e
g
r
es
s
io
n
.
T
h
i
s
i
s
b
ec
au
s
e
B
P
M
h
a
v
e
in
t
u
itio
n
s
t
h
at
ca
n
s
p
ec
if
y
t
h
e
p
r
io
r
in
th
e
s
h
ar
k
attac
k
d
atas
et.
W
ith
in
tu
i
tio
n
s
,
B
P
M
ca
n
m
ak
e
p
r
ed
ictio
n
o
n
th
e
m
o
d
el
th
r
o
u
g
h
th
e
p
o
s
ter
io
r
.
B
P
M
ca
n
s
im
p
l
y
w
o
r
k
b
y
id
e
n
ti
f
y
f
e
w
i
m
p
o
r
tan
t
in
d
ep
en
d
en
t
v
ar
iab
les
co
m
p
ar
ed
to
L
o
g
i
s
tic
R
e
g
r
es
s
io
n
w
h
ic
h
n
ee
d
to
i
n
cl
u
d
e
all
i
m
p
o
r
tan
t
i
n
d
ep
en
d
en
t
v
ar
i
ab
les.
T
h
is
en
ab
le
s
B
P
M
to
b
e
w
el
l
o
p
er
ate
w
h
e
n
ce
r
tain
cla
s
s
o
f
s
h
ar
k
a
ttack
d
ataset
is
c
h
a
n
g
ed
o
r
ed
ited
a
s
B
P
M
ca
n
ev
al
u
ate
Evaluation Warning : The document was created with Spire.PDF for Python.
IJ
-
AI
I
SS
N:
2252
-
8938
P
r
ed
ictin
g
fa
ta
liti
es a
mo
n
g
s
h
a
r
k
a
tta
ck
s
:
co
mp
a
r
is
o
n
o
f c
l
a
s
s
ifie
r
s
... (
Ya
n
a
Ma
z
w
in
Mo
h
ma
d
Ha
s
s
im
)
365
th
e
v
ar
iab
les
f
r
o
m
t
h
e
m
o
s
t
i
m
p
o
r
tan
t
v
ar
iab
le
b
y
it
s
el
f
.
Hen
ce
,
L
o
g
is
t
ic
R
e
g
r
ess
io
n
i
s
o
u
tp
er
f
o
r
m
ed
b
y
B
P
M.
I
n
ad
d
itio
n
,
B
P
Ms
h
av
e
h
ig
h
er
ac
cu
r
ac
y
t
h
an
t
h
at
o
f
Dec
is
io
n
T
r
ee
.
T
h
is
is
b
ec
au
s
e
B
P
M
ca
n
h
av
e
b
ig
g
er
tr
ai
n
i
n
g
s
et
co
m
p
ar
ed
t
o
Dec
is
io
n
T
r
ee
.
T
h
is
e
n
s
u
r
e
t
h
e
lo
w
c
lass
if
ica
tio
n
er
r
o
r
r
at
e
as
b
i
g
g
er
tr
ain
i
n
g
s
et
co
n
s
is
t
s
o
f
m
o
r
e
n
u
m
b
er
o
f
class
e
s
.
B
P
M
ca
n
ad
m
it
t
h
e
tr
ain
i
n
g
er
r
o
r
in
s
h
ar
k
attac
k
d
ataset
to
a
v
o
id
ex
is
t
in
g
o
f
n
o
is
y
d
ata.
4.
CO
NC
L
U
SI
O
N
S
I
n
th
i
s
p
ap
er
,
th
e
co
m
p
ar
is
o
n
o
f
t
w
o
cla
s
s
i
f
ier
s
’
p
er
f
o
r
m
a
n
c
e
o
n
f
atalit
y
o
f
s
h
ar
k
attac
k
v
icti
m
w
a
s
ca
r
r
ied
o
u
t.
T
h
e
d
ataset
w
as
r
u
n
o
n
t
w
o
class
if
ier
s
;
S
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
s
(
SVMs
)
an
d
B
ay
es
P
o
in
t
Ma
ch
i
n
es
(
B
P
Ms)
an
d
th
eir
p
er
f
o
r
m
an
ce
w
er
e
a
n
al
y
s
ed
.
B
ased
o
n
th
e
r
es
u
lt,
it
i
s
s
h
o
w
n
t
h
at
B
P
Ms
w
a
s
ab
le
to
p
r
ed
ict
th
e
r
es
u
lt
w
it
h
h
ig
h
er
ac
cu
r
ac
y
a
n
d
p
r
ec
is
io
n
as
co
m
p
ar
e
to
SVMs
d
u
e
to
th
e
ab
ilit
y
o
f
B
P
Ms
to
m
i
n
i
m
ize
t
h
e
a
v
er
ag
e
er
r
o
r
w
h
en
m
ar
g
i
n
alizi
n
g
o
v
er
all
p
o
s
s
ib
le
b
o
u
n
d
ar
ies
an
d
p
o
s
s
ib
le
s
a
m
p
li
n
g
s
o
f
t
h
e
d
ata.
Fro
m
t
h
i
s
w
o
r
k
,
w
e
ca
n
co
n
cl
u
d
e
t
h
at
b
et
w
ee
n
th
e
s
e
t
w
o
cla
s
s
i
f
ier
s
,
th
e
B
P
Ms
ar
e
m
o
r
e
s
u
i
tab
le
i
n
p
r
ed
ictin
g
th
e
f
atali
t
y
o
f
s
h
ar
k
attac
k
v
icti
m
.
A
f
u
t
u
r
e
w
o
r
k
m
a
y
b
e
ca
r
r
ied
o
u
t
to
s
ee
k
a
b
etter
class
if
er
t
h
at
ca
n
b
e
ef
f
icie
n
tl
y
u
s
ed
to
p
r
ed
ict
th
e
f
atalit
y
o
f
s
h
ar
k
att
ac
k
v
icti
m
in
o
r
d
er
to
av
o
id
s
u
c
h
an
u
n
w
a
n
ted
in
cid
en
t i
n
t
h
e
f
u
t
u
r
e
.
ACK
NO
WL
E
D
G
E
M
E
NT
S
T
h
is
r
esear
ch
is
s
u
p
p
o
r
ted
b
y
Un
i
v
er
s
iti T
u
n
H
u
s
s
ei
n
On
n
Ma
la
y
s
ia.
RE
F
E
R
E
NC
E
S
[1
]
“
Ye
a
rl
y
W
o
rld
w
id
e
S
h
a
rk
Attac
k
S
u
m
m
a
r
y
”
,
re
tri
e
v
e
d
f
r
o
m
h
tt
p
s://
ww
w
.
f
lo
rid
a
m
u
se
u
m
.
u
f
l.
e
d
u
/sh
a
rk
-
a
tt
a
c
k
s/
y
e
a
rl
y
-
w
o
rld
w
id
e
-
su
m
m
a
ry
/
,
2
0
1
8
.
[2
]
D.
G
.
C
a
ld
ico
tt
,
R.
M
a
h
a
jan
i
,
M
.
Ku
h
n
,
“
T
h
e
a
n
a
to
m
y
o
f
a
sh
a
rk
a
tt
a
c
k
:
A
c
a
se
re
p
o
rt
a
n
d
re
v
ie
w
o
f
th
e
li
tera
tu
re
,
”
In
j
u
ry
,
v
o
l.
3
2
,
n
o
.
6
,
p
p
.
4
4
5
-
4
5
3
,
2
0
1
1
.
[3
]
C.
M
c
Ca
g
h
,
J.
S
n
e
d
d
o
n
,
D.
Bla
c
h
e
,
“
Kill
in
g
sh
a
rk
s:
T
h
e
m
e
d
ia
’s
ro
le
in
p
u
b
li
c
a
n
d
p
o
l
it
ica
l
re
sp
o
n
se
t
o
f
a
tal
h
u
m
a
n
–
sh
a
rk
in
tera
c
ti
o
n
s,”
M
a
ri
n
e
Po
li
c
y
,
v
o
l.
62
,
p
p
.
271
-
2
7
8
,
2
0
1
5
.
[4
]
E.
Clu
a
,
B
.
S
é
re
t,
“
Un
p
ro
v
o
k
e
d
f
a
tal
sh
a
rk
a
tt
a
c
k
in
L
i
f
o
u
Isla
n
d
(
L
o
y
a
lt
y
Isla
n
d
s,
Ne
w
Ca
led
o
n
ia,
S
o
u
t
h
P
a
c
if
ic)
b
y
a
g
r
e
a
t
w
h
it
e
sh
a
rk
,
C
a
rc
h
a
ro
d
o
n
c
a
rc
h
a
rias
,
”
T
h
e
Ame
ric
a
n
J
o
u
rn
a
l
o
f
F
o
re
n
sic
M
e
d
icin
e
a
n
d
Pa
th
o
lo
g
y
,
v
o
l.
31
,
n
o
.
3
,
p
p
.
2
8
1
-
2
8
6
,
2
0
1
0
.
[5
]
A
.
Ko
c
k
,
R.
Jo
h
n
so
n
,
“
W
h
it
e
sh
a
rk
a
b
u
n
d
a
n
c
e
:
No
t
a
c
a
u
sa
ti
v
e
f
a
c
to
r
in
n
u
m
b
e
rs
o
f
sh
a
r
k
b
it
e
in
c
id
e
n
ts.
F
i
n
d
i
n
g
a
b
a
lan
c
e
:
W
h
it
e
sh
a
rk
c
o
n
se
rv
a
ti
o
n
a
n
d
re
c
re
a
ti
o
n
a
l
sa
f
e
t
y
in
th
e
in
sh
o
re
w
a
ters
o
f
Ca
p
e
T
o
w
n
,
”
p
p
.
1
-
1
9
,
2
0
0
6
.
[6
]
Z.
L
u
c
a
s,
W
.
T
.
S
to
b
o
,
“
S
h
a
rk
‐in
f
li
c
ted
m
o
rtalit
y
o
n
a
p
o
p
u
latio
n
o
f
h
a
rb
o
u
r
se
a
ls
(P
h
o
c
a
v
it
u
li
n
a
)
a
t
S
a
b
le
Isla
n
d
,
No
v
a
S
c
o
ti
a
,
”
J
o
u
r
n
a
l
o
f
Z
o
o
lo
g
y
,
v
o
l.
2
5
2
,
n
o
.
3
,
p
p
.
4
0
5
-
4
1
4
,
2
0
1
0
.
[7
]
J.
S
e
a
rin
g
,
“
T
h
e
Big
Nu
m
b
e
r:
5
3
sh
a
rk
a
tt
a
c
k
s
in
U.S
.
w
a
ters
,
”
Re
t
riev
e
d
f
ro
m
h
tt
p
s:/
/w
ww
.
w
a
sh
in
g
to
n
p
o
st.co
m
/n
a
ti
o
n
a
l/
h
e
a
lt
h
-
sc
ien
c
e
/t
h
e
-
b
ig
-
n
u
m
b
e
r
-
53
-
s
h
a
r
k
-
a
t
t
a
c
k
s
-
in
-
us
-
w
a
t
e
r
s
/
2
0
1
8
/
0
6
/
2
9
/
a
d
1
f
7
5
d
0
-
7
a
e
c
-
1
1
e
8
-
aeee
-
4
d
0
4
c
8
a
c
6
1
5
8
_
st
o
ry
.
h
tm
l?n
o
re
d
ire
c
t=o
n
&
u
tm
_
term
=
.
3
8
a
0
3
a
3
e
f
9
5
8
,
2
0
1
8
.
[8
]
R.
Cro
ss
ley
,
C.
M
.
Co
ll
in
s,
S
.
G
.
S
u
tt
o
n
,
C
.
Hu
v
e
n
e
e
rs,
“
P
u
b
l
ic
p
e
rc
e
p
ti
o
n
a
n
d
u
n
d
e
rsta
n
d
i
n
g
o
f
sh
a
rk
a
tt
a
c
k
m
it
ig
a
ti
o
n
m
e
a
su
re
s in
A
u
stra
li
a
,
”
Hu
ma
n
d
ime
n
si
o
n
s
o
f
wil
d
li
fe
,
v
o
l.
19
,
n
o
.
2
,
p
p
.
1
5
4
-
1
6
5
,
2
0
1
4
.
[9
]
C.
Ne
ff
,
R.
Hu
e
ter,
“
S
c
ien
c
e
,
p
o
li
c
y
,
a
n
d
th
e
p
u
b
l
ic
d
isc
o
u
rse
o
f
sh
a
rk
“
a
tt
a
c
k
”
:
A
p
ro
p
o
sa
l
fo
r
re
c
las
si
fy
in
g
h
u
m
a
n
–
sh
a
rk
in
tera
c
ti
o
n
s,”
J
o
u
rn
a
l
o
f
e
n
v
iro
n
me
n
ta
l
stu
d
ies
a
n
d
s
c
ien
c
e
s
,
v
o
l.
3
,
n
o
.
1
,
6
5
-
7
3
,
2
0
1
3
.
[1
0
]
R.
He
rb
rich
,
T
.
G
ra
e
p
e
l,
C.
Ca
m
p
b
e
ll
,
“
Ba
y
e
s
p
o
in
t
m
a
c
h
in
e
s,”
J
o
u
rn
a
l
o
f
M
a
c
h
in
e
L
e
a
r
n
in
g
Res
e
a
rc
h
,
v
o
l.
1
,
p
p
.
245
-
2
7
9
,
2
0
0
1
.
[1
1
]
J.
F
ro
st,
“
Ov
e
rf
it
ti
n
g
Re
g
r
e
ss
io
n
M
o
d
e
ls:
P
ro
b
lem
s,
De
tec
ti
o
n
,
a
n
d
A
v
o
id
a
n
c
e
,
”
Re
tri
e
v
e
d
De
c
e
m
b
e
r
9
,
2
0
1
8
f
ro
m
h
tt
p
:
//
ww
w
.
sta
t.
y
a
le.ed
u
/Co
u
rse
s/1
9
9
7
-
9
8
/
1
0
1
/l
i
n
re
g
.
h
tm
,
2
0
1
8
.
[1
2
]
N.
Oliv
o
,
“
Us
in
g
S
h
a
rk
A
tt
a
c
k
s
to
Un
d
e
rsta
n
d
Ba
y
e
sia
n
Ne
tw
o
rk
s,”
R
e
tri
e
v
e
d
De
c
e
m
b
e
r
9
,
2
0
1
9
f
ro
m
h
tt
p
s:/
/m
e
d
iu
m
.
c
o
m
/@Na
t
a
li
e
Oliv
o
/sh
a
rk
-
b
it
e
s
-
9
2
0
2
9
9
b
9
0
8
b
2
,
2
0
1
8
.
[1
3
]
D.
G
.
Na
b
i,
S
.
R.
T
a
k
,
K.
A
.
Ka
n
g
o
o
,
M
.
A
.
Ha
lw
a
i,
“
In
ju
ries
f
ro
m
leo
p
a
rd
a
tt
a
c
k
s in
Ka
sh
m
ir,
”
In
ju
ry
,
v
o
l.
4
0
,
n
o
.
1
,
p
p
.
9
0
-
9
2
,
2
0
0
9
.
[1
4
]
S
.
K.
Da
s,
S
.
Ch
a
tt
o
p
a
d
h
y
a
y
,
“
H
u
m
a
n
f
a
talit
ies
f
ro
m
w
il
d
e
lep
h
a
n
t
a
tt
a
c
k
s:
A
stu
d
y
o
f
f
o
u
rtee
n
c
a
se
s,”
J
o
u
rn
a
l
o
f
Fo
re
n
sic
a
n
d
L
e
g
a
l
M
e
d
icin
e
,
v
o
l
.
1
8
,
p
p
.
1
5
4
-
1
5
7
,
2
0
1
1
.
[1
5
]
M
.
K.
S
a
d
o
o
n
,
“
S
n
a
k
e
b
it
e
e
n
v
e
n
o
m
a
ti
o
n
in
Riy
a
d
h
p
ro
v
in
c
e
o
f
S
a
u
d
i
A
ra
b
ia
o
v
e
r
th
e
p
e
rio
d
(2
0
0
5
–
2
0
1
0
),
”
S
a
u
d
i
J
o
u
rn
a
l
o
f
Bi
o
lo
g
ica
l
S
c
ien
c
e
s
,
v
o
l.
2
2
,
n
o
.
2
,
p
p
.
1
9
8
-
2
0
3
,
2
0
1
5
.
[1
6
]
C.
P
e
p
i
n
-
Ne
ff
,
T
.
Wy
n
ter,
“
S
h
a
rk
Bit
e
s
a
n
d
S
h
a
rk
Co
n
se
rv
a
ti
o
n
:
A
n
A
n
a
ly
sis
o
f
Hu
m
a
n
A
tt
it
u
d
e
s
F
o
ll
o
w
in
g
S
h
a
rk
Bit
e
In
c
id
e
n
ts i
n
T
w
o
L
o
c
a
ti
o
n
s i
n
A
u
stra
li
a
,
”
Co
n
se
rv
a
ti
o
n
L
e
tt
e
r
s
,
v
o
l.
1
1
,
n
o
.
2
,
2
0
1
7
.
[1
7
]
C.
Ne
ff
,
“
T
h
e
Ja
ws
e
ff
e
c
t:
h
o
w
m
o
v
ie
n
a
rra
ti
v
e
s
a
re
u
se
d
to
i
n
f
lu
e
n
c
e
p
o
li
c
y
re
sp
o
n
se
s
to
sh
a
rk
b
it
e
s
in
W
e
ste
rn
A
u
stra
li
a
,
”
Au
stra
li
a
n
J
o
u
r
n
a
l
o
f
Po
li
ti
c
a
l
S
c
ie
n
c
e
,
v
o
l.
5
0
,
n
o
.
1
,
p
p
.
1
1
4
-
1
2
7
,
2
0
1
5
.
[1
8
]
N.
K.
Du
lv
y
,
S
.
L
.
F
o
w
l
e
r,
J.
A
.
M
u
sic
k
,
R.
D.
Ca
v
a
n
a
g
h
,
P
.
M
.
K
y
n
e
,
L
.
R.
Ha
rriso
n
,
C.
M
.
P
o
ll
o
c
k
,
“
Ex
ti
n
c
ti
o
n
risk
a
n
d
c
o
n
se
rv
a
ti
o
n
o
f
th
e
w
o
rl
d
’s sh
a
rk
s a
n
d
ra
y
s,”
e
L
if
e
,
v
o
l.
3
,
2
0
1
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8938
IJ
-
AI
Vo
l.
8
,
No
.
4
,
Dec
em
b
er
20
1
9
:
3
6
0
–
3
66
366
[1
9
]
R.
A
.
M
a
rti
n
,
D.
K.
Ro
ss
m
o
,
N.
Ha
m
m
e
r
sc
h
lag
,
“
Hu
n
ti
n
g
p
a
tt
e
rn
s
a
n
d
g
e
o
g
ra
p
h
ic
p
r
o
f
il
in
g
o
f
w
h
it
e
sh
a
rk
p
re
d
a
ti
o
n
,
”
J
o
u
rn
a
l
o
f
Z
o
o
l
o
g
y
,
v
o
l.
2
7
9
,
n
o
.
2
,
p
p
.
1
1
1
-
1
1
8
,
2
0
0
9
.
[2
0
]
G
.
Bian
u
c
c
i,
M
.
Bisc
o
n
ti
,
W
.
L
a
n
d
i
n
i,
T
.
S
to
ra
i,
M
.
Zu
f
f
a
,
S
.
G
i
u
li
a
n
i,
A
.
M
o
jetta,
“
T
ro
p
h
ic
in
te
ra
c
ti
o
n
b
e
tw
e
e
n
w
h
it
e
sh
a
rk
,
c
a
rc
h
a
ro
d
o
n
c
a
rc
h
a
rias
,
a
n
d
c
e
tac
e
a
n
s:
A
c
o
m
p
a
riso
n
b
e
tw
e
e
n
P
li
o
c
e
n
e
a
n
d
re
c
e
n
t
d
a
t
a
f
ro
m
c
e
n
tral
M
e
d
it
e
rra
n
e
a
n
S
e
a
,
”
In
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
4
t
h
E
u
ro
p
e
a
n
El
a
sm
o
b
ra
n
c
h
Ass
o
c
ia
ti
o
n
M
e
e
ti
n
g
,
p
p
.
3
3
-
4
8
,
2
0
0
0
.
[2
1
]
A
.
I.
R.
L
.
A
z
e
v
e
d
o
,
M
.
F
.
S
a
n
t
o
s,
“
KD
D,
S
EM
M
A
a
n
d
CRI
S
P
-
D
M
:
A
p
a
ra
ll
e
l
o
v
e
rv
ie
w
,
”
IADS
-
DM
,
2
0
0
8
.
[2
2
]
D.
M
o
h
a
m
m
e
d
,
K.
A
.
K
a
ra
w
i,
P
.
Du
n
c
a
n
,
F
.
L
.
F
ra
n
c
is,
“
Ov
e
rlap
p
e
d
M
u
sic
S
e
g
m
e
n
tatio
n
u
si
n
g
a
Ne
w
E
ff
e
c
ti
v
e
F
e
a
tu
re
a
n
d
Ra
n
d
o
m
F
o
re
sts,”
AI
ES
I
n
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Arti
fi
c
ia
l
In
tell
ig
e
n
c
e
(
IJ
-
AI)
,
v
o
l.
8
,
n
o
2
,
2
0
1
9
.
[2
3
]
H.
Ka
ri
m
,
S
.
R.
Nia
k
a
n
,
R.
S
a
fd
a
ri,
“
Co
m
p
a
riso
n
o
f
Ne
u
ra
l
Ne
tw
o
rk
T
ra
in
in
g
A
l
g
o
rit
h
m
s
f
o
r
Clas
sif
ic
a
ti
o
n
o
f
He
a
rt
Dise
a
s
e
s,”
AIE
S
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Art
if
icia
l
In
tell
ig
e
n
c
e
(
I
J
-
AI)
,
v
o
l.
7
,
n
o
4
,
2
0
1
8
.
[2
4
]
P
.
R.
Iy
e
r,
S
.
R.
I
y
e
r,
R.
Ra
m
e
sh
,
M
.
R.
A
n
a
la,
K.N.
S
u
b
ra
m
a
n
y
a
,
“
A
d
a
p
ti
v
e
r
e
a
l
ti
m
e
tra
ff
ic
p
re
d
ic
ti
o
n
u
sin
g
d
e
e
p
n
e
u
ra
l
n
e
tw
o
rk
s,”
AIE
S
I
n
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Arti
fi
c
i
a
l
I
n
telli
g
e
n
c
e
(
IJ
-
AI)
,
v
o
l.
8
,
n
o
.
2
,
2
0
1
9
.
[2
5
]
E.
Rit
ter,
M
.
L
e
v
in
e
,
“
Us
e
o
f
f
o
re
n
sic
a
n
a
l
y
sis
to
b
e
tt
e
r
u
n
d
e
rsta
n
d
sh
a
rk
a
tt
a
c
k
b
e
h
a
v
io
r,
”
J
o
u
rn
a
l
o
f
F
o
re
n
sic
Od
o
n
t
o
sto
m
a
to
lo
g
y
,
v
o
l.
22
,
n
o
.
2
,
p
p
.
4
0
-
4
6
,
2
0
0
4
.
[2
6
]
E.
Ra
h
m
,
H.
H.
Do
,
“
Da
ta
c
lea
n
in
g
:
P
ro
b
lem
s
a
n
d
c
u
rre
n
t
a
p
p
ro
a
c
h
e
s,”
IEE
E
Da
ta
En
g
.
B
u
ll
.
,
v
o
l.
23
.
n
o
.
4
,
p
p
.
3
-
1
3
,
2
0
2
0
.
[2
7
]
N.
Qa
z
i,
K.
Ra
z
a
,
“
Ef
f
e
c
t
o
f
f
e
a
tu
re
se
lec
ti
o
n
,
S
M
OT
E
a
n
d
u
n
d
e
r
-
sa
m
p
li
n
g
o
n
c
las
s
im
b
a
lan
c
e
c
las
sif
ic
a
ti
o
n
,
”
In
Pro
c
e
e
d
i
n
g
s
o
f
2
0
1
2
UKS
im
1
4
t
h
I
n
ter
n
a
t
io
n
a
l
Co
n
fer
e
n
c
e
on
Co
m
p
u
ter
M
o
d
e
ll
in
g
a
n
d
S
imu
la
ti
o
n
(
UKS
im),
p
p
.
1
4
5
-
1
5
0
,
2
0
1
2
.
[2
8
]
J
.
Ha
n
,
J.
P
e
i
,
M
.
Ka
m
b
e
r,
“
Da
ta
min
in
g
:
c
o
n
c
e
p
ts a
n
d
tec
h
n
i
q
u
e
s
”
,
2
0
1
1
.
[2
9
]
E.
Na
m
e
y
,
G
.
G
u
e
st,
L
.
T
h
a
iru
,
L
.
Jo
h
n
so
n
,
“
Da
ta
re
d
u
c
ti
o
n
tec
h
n
iq
u
e
s
f
o
r
larg
e
q
u
a
li
tativ
e
d
a
ta
s
e
ts,”
Ha
n
d
b
o
o
k
fo
r tea
m
-
b
a
se
d
q
u
a
li
t
a
ti
v
e
re
se
a
rc
h
,
v
o
l.
2
,
n
o
.
1
,
1
3
7
-
1
6
1
,
2
0
0
8
.
[3
0
]
R.
Jin
,
Y.
Bre
it
b
a
rt,
C.
M
u
o
h
,
“
D
a
ta d
isc
re
ti
z
a
ti
o
n
u
n
if
ica
ti
o
n
,
”
Kn
o
wled
g
e
a
n
d
I
n
fo
rm
a
ti
o
n
S
y
ste
ms
,
v
o
l.
1
9
,
n
o
.
1
,
2
0
0
9
.
[3
1
]
M
.
Bih
is,
S
.
R
o
y
c
h
o
w
d
h
u
ry
,
“
A
g
e
n
e
ra
li
z
e
d
f
lo
w
f
o
r
m
u
lt
i
-
c
las
s
a
n
d
b
in
a
ry
c
las
sif
ic
a
ti
o
n
tas
k
s:
A
n
A
z
u
re
M
L
a
p
p
ro
a
c
h
,
”
I
n
2
0
1
5
I
EE
E
I
n
ter
n
a
t
io
n
a
l
C
o
n
fer
e
n
c
e
o
n
Bi
g
Da
t
a
(
Bi
g
Da
t
a
),
p
p
.
1
7
2
8
-
1
7
3
7
,
2
0
1
5
.
[3
2
]
T
.
P
.
M
i
n
k
a
,
“
Ex
p
e
c
tatio
n
p
ro
p
a
g
a
ti
o
n
f
o
r
a
p
p
ro
x
im
a
te
B
a
y
e
sia
n
in
f
e
re
n
c
e
,
”
In
Pro
c
e
e
d
in
g
s
o
f
t
h
e
S
e
v
e
n
tee
n
th
c
o
n
fer
e
n
c
e
o
n
U
n
c
e
rta
in
ty i
n
Arti
fi
c
ia
l
I
n
telli
g
e
n
c
e
,
p
p
.
3
6
2
-
3
6
9
,
M
o
rg
a
n
Ka
u
f
m
a
n
n
P
u
b
li
sh
e
rs I
n
c
.
,
2
0
0
1
.
[3
3
]
C.
W
.
Hs
u
,
C.
J.
L
in
,
“
A
c
o
m
p
a
riso
n
o
f
m
e
th
o
d
s
f
o
r
m
u
lt
icla
ss
s
u
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
s,”
IEE
E
t
ra
n
sa
c
ti
o
n
s
o
n
Ne
u
ra
l
Ne
two
rk
s
,
v
o
l.
13
,
n
o
.
2
,
p
p
.
4
1
5
-
4
2
5
,
2
0
0
2
.
[3
4
]
O.
Be
n
a
rc
h
id
a
n
d
N.
Ra
isso
u
n
i,
“
S
u
p
p
o
rt
V
e
c
to
r
M
a
c
h
i
n
e
s
f
o
r
Ob
jec
t
Ba
se
d
Bu
il
d
in
g
Ex
trac
ti
o
n
in
S
u
b
u
rb
a
n
A
re
a
u
sin
g
V
e
ry
Hig
h
Re
so
lu
ti
o
n
S
a
t
e
ll
it
e
Im
a
g
e
s,
a
Ca
s
e
S
tu
d
y
:
T
e
t
u
a
n
,
M
o
ro
c
c
o
.
”
AIE
S
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Arti
fi
c
ia
l
I
n
telli
g
e
n
c
e
(
IJ
-
AI)
,
v
o
l.
2
,
n
o
.
1
,
2
0
1
3.
[3
5
]
T
.
F
a
wc
e
tt
,
“
A
n
in
tro
d
u
c
ti
o
n
to
R
OC an
a
ly
sis,”
Pa
tt
e
rn
re
c
o
g
n
it
i
o
n
letter
s
,
v
o
l.
27
,
n
o
.
8
,
p
p
.
8
6
1
-
8
7
4
,
2
0
0
6
.
[3
6
]
G
.
Tso
u
m
a
k
a
s,
I.
Ka
tak
is,
“
M
u
lt
i
-
lab
e
l
c
las
sif
ica
ti
o
n
:
A
n
o
v
e
rv
ie
w
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
Da
t
a
W
a
re
h
o
u
si
n
g
a
n
d
M
in
i
n
g
,
v
o
l.
3
,
n
o
.
3
,
p
p
.
1
-
1
3
,
2
0
0
7
.
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
L
e
e
Ju
n
M
e
i
re
c
e
iv
e
d
th
e
se
c
o
n
d
a
r
y
e
d
u
c
a
ti
o
n
i
n
S
c
ien
c
e
S
trea
m
f
ro
m
S
e
k
o
lah
M
e
n
e
n
g
a
h
Je
n
is
Ke
b
a
n
g
sa
a
n
(C)
Ch
a
n
W
a
II,
S
e
re
m
b
a
n
,
Ne
g
e
ri
S
e
m
b
il
a
n
.
Th
e
n
,
sh
e
c
o
n
ti
n
u
e
s
h
e
r
p
re
-
u
n
iv
e
rsity
stu
d
y
a
t
S
t.
P
a
u
l
In
sti
t
u
ti
o
n
,
S
e
re
m
b
a
n
,
M
a
la
y
sia
.
In
S
e
p
tem
b
e
r
5
,
2
0
1
6
,
S
h
e
re
c
e
iv
e
d
B.
d
e
g
re
e
in
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
lo
g
y
f
ro
m
U
T
HM.
S
h
e
h
o
p
e
s
th
a
t
sh
e
c
a
n
b
e
c
o
m
e
a
n
e
x
c
e
ll
e
n
t
s
y
ste
m
a
n
a
l
y
st i
n
th
e
f
u
tu
re
.
A
id
a
M
u
sta
p
h
a
re
c
e
iv
e
d
th
e
B.
S
c
.
d
e
g
re
e
in
Co
m
p
u
ter
S
c
ien
c
e
fro
m
M
ich
ig
a
n
T
e
c
h
n
o
lo
g
ica
l
Un
iv
e
rsit
y
a
n
d
th
e
M
.
IT
d
e
g
re
e
i
n
Co
m
p
u
ter
S
c
ien
c
e
f
ro
m
UK
M
,
M
a
la
y
sia
in
1
9
9
8
a
n
d
2
0
0
4
,
re
sp
e
c
ti
v
e
l
y
.
S
h
e
re
c
e
iv
e
d
h
e
r
P
h
.
D.
in
A
rti
f
icia
l
In
telli
g
e
n
c
e
f
o
c
u
sin
g
o
n
d
ial
o
g
u
e
sy
ste
m
s.
S
h
e
is
c
u
rre
n
tl
y
a
n
a
c
ti
v
e
re
se
a
rc
h
e
r
i
n
th
e
a
re
a
o
f
Co
m
p
u
tatio
n
a
l
L
in
g
u
isti
c
s,
S
o
f
t
Co
m
p
u
ti
n
g
,
Da
ta
M
in
i
n
g
,
a
n
d
A
g
e
n
t
-
b
a
se
d
S
y
ste
m
s.
Ya
n
a
M
a
z
w
in
M
o
h
a
m
a
d
Ha
ss
i
m
g
ra
d
u
a
ted
w
it
h
a
P
h
D
d
e
g
re
e
f
ro
m
Un
iv
e
rsiti
T
u
n
Hu
ss
e
in
On
n
M
a
la
y
sia
(U
T
HM)
in
2
0
1
6
.
Earl
ier,
in
2
0
0
6
sh
e
c
o
m
p
lete
d
h
e
r
M
a
ste
r'
s
d
e
g
re
e
in
Co
m
p
u
ter
S
c
ien
c
e
f
ro
m
Un
iv
e
rsiti
o
f
M
a
la
y
a
(UM)
.
S
h
e
re
c
e
iv
e
d
h
e
r
Ba
c
h
e
l
o
r
o
f
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
lo
g
y
(Ho
n
s)
d
e
g
re
e
m
a
jo
rin
g
in
I
n
d
u
st
rial
Co
m
p
u
ti
n
g
f
ro
m
Un
iv
e
rsiti
Ke
b
a
n
g
sa
a
n
M
a
la
y
sia
(UK
M
)
in
2
0
0
1
.
He
r
re
se
a
rc
h
a
r
e
a
in
c
lu
d
e
s
n
e
u
ra
l
n
e
tw
o
rk
s,
sw
a
r
m
i
n
telli
g
e
n
c
e
,
o
p
ti
m
iza
ti
o
n
a
n
d
c
las
si
f
ica
ti
o
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.