I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
,
p
p
.
3
1
0
8
~
3
1
1
4
I
SS
N:
2088
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
9
i
4
.
p
p
3
1
0
8
-
3114
3108
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e
.
co
m/
jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
I
JE
C
E
E
m
piri
ca
l ana
ly
sis
of ense
m
bl
e
m
e
t
ho
ds for th
e clas
si
ficatio
n of
ro
bo
ca
lls in te
leco
mm
un
ica
tions
M
e
g
hn
a
G
ho
s
h,
P
ra
bu
P
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter S
c
ien
c
e
,
C
h
rist
(De
e
m
e
d
to
b
e
Un
iv
e
rsit
y
)
,
In
d
ia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Oct
25
,
2
0
1
8
R
ev
i
s
ed
A
p
r
1
,
2
0
1
9
A
cc
ep
ted
A
p
r
8
,
2
0
1
9
W
it
h
th
e
a
d
v
e
n
t
o
f
tec
h
n
o
lo
g
y
,
t
h
e
re
h
a
s
b
e
e
n
a
n
e
x
c
e
ss
iv
e
u
se
o
f
c
e
ll
u
lar
p
h
o
n
e
s.
Ce
ll
u
lar
p
h
o
n
e
s
h
a
v
e
m
a
d
e
li
f
e
c
o
n
v
e
n
ien
t
in
o
u
r
so
c
iety
.
Ho
w
e
v
e
r,
in
d
iv
id
u
a
ls
a
n
d
g
ro
u
p
s
h
a
v
e
su
b
v
e
rted
th
e
tele
c
o
m
m
u
n
ica
ti
o
n
d
e
v
ice
s
to
d
e
c
e
iv
e
u
n
w
a
r
y
v
icti
m
s.
Ro
b
o
c
a
ll
s
a
re
q
u
it
e
p
re
v
a
len
t
th
e
se
d
a
y
s
a
n
d
th
e
y
c
a
n
e
it
h
e
r
b
e
leg
a
l
o
r
u
se
d
b
y
sc
a
m
m
e
rs
t
o
tri
c
k
o
n
e
o
u
t
o
f
th
e
ir
m
o
n
e
y
.
T
h
e
p
ro
p
o
se
d
m
e
th
o
d
o
lo
g
y
in
t
h
e
p
a
p
e
r
is
to
e
x
p
e
rim
e
n
t
tw
o
e
n
se
m
b
le
m
o
d
e
ls
o
n
th
e
d
a
tas
e
t
a
c
q
u
ired
fro
m
th
e
F
e
d
e
ra
l
T
ra
d
e
Co
m
m
is
s
io
n
(DN
C
Da
tas
e
t).
It
is
im
p
e
ra
ti
v
e
to
a
n
a
ly
z
e
th
e
c
a
ll
re
c
o
rd
s
a
n
d
b
a
se
d
o
n
t
h
e
p
a
tt
e
rn
s
th
e
c
a
ll
s
c
a
n
c
las
si
fy
a
s
a
ro
b
o
c
a
ll
o
r
n
o
t
a
ro
b
o
c
a
ll
.
T
w
o
a
lg
o
rit
h
m
s
Ra
n
d
o
m
F
o
re
st
a
n
d
X
g
Bo
o
st
a
re
c
o
m
b
in
e
d
in
tw
o
w
a
y
s
a
n
d
c
o
m
p
a
re
d
in
th
e
p
a
p
e
r
in
term
s o
f
a
c
c
u
ra
c
y
,
se
n
siti
v
it
y
a
n
d
th
e
ti
m
e
tak
e
n
.
K
ey
w
o
r
d
s
:
E
n
s
e
m
b
le
m
et
h
o
d
Ma
ch
i
n
e
L
ea
r
n
i
n
g
R
an
d
o
m
Fo
r
est
R
o
b
o
ca
lls
XGB
o
o
s
t
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
:
Me
g
h
n
a
G
h
o
s
h
,
Dep
ar
t
m
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
,
C
h
r
i
s
t
(
Dee
m
ed
to
b
e
Un
i
v
er
s
i
t
y
)
,
Ho
s
u
r
Ma
in
R
o
ad
,
B
an
g
alo
r
e
-
5
6
0
0
2
9
,
I
n
d
ia
.
E
m
ail:
m
e
g
h
n
a.
g
h
o
s
h
@
cs.c
h
r
i
s
tu
n
i
v
er
s
it
y
.
i
n
,
p
r
ab
u
.
p
@
ch
r
is
tu
n
iv
er
s
it
y
.
i
n
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
Fed
er
al
T
r
ad
e
C
o
m
m
is
s
i
o
n
r
ec
eiv
ed
o
v
er
2
2
m
illi
o
n
c
o
m
p
lai
n
t
s
o
f
il
leg
al
a
n
d
u
n
w
a
n
ted
ca
lls
,
in
2
0
1
4
.
T
elep
h
o
n
e
s
p
a
m
m
er
s
to
d
ay
ar
e
le
v
er
ag
i
n
g
r
ec
en
t
t
ec
h
n
ica
l
ad
v
a
n
ce
s
i
n
t
h
e
telep
h
o
n
y
ec
o
s
y
s
te
m
to
d
is
tr
ib
u
te
m
a
s
s
i
v
el
y
a
u
to
m
ate
d
s
p
am
ca
ll
s
k
n
o
w
n
as
r
o
b
o
ca
lls
[
1
]
.
A
p
h
o
n
e
ca
ll
th
at
u
s
es
a
co
m
p
u
ter
ized
au
to
d
ialer
to
d
eliv
er
a
p
r
e
-
r
e
co
r
d
ed
m
ess
a
g
e
at
th
e
o
t
h
er
en
d
,
as
i
f
it
w
er
e
f
r
o
m
a
r
o
b
o
t
is
a
r
o
b
o
ca
ll
.
On
ce
v
ie
w
ed
a
s
a
n
i
n
co
n
v
e
n
ie
n
ce
t
h
e
y
h
av
e
r
ea
ch
ed
ep
id
e
m
ic
p
r
o
p
o
r
tio
n
s
.
Fe
w
r
o
b
o
ca
lls
ar
e
also
co
n
s
id
e
r
ed
leg
al.
T
h
e
ca
ll
s
p
er
m
itted
c
an
b
e
ca
m
p
aig
n
i
n
g
f
o
r
ca
n
d
id
ates,
aler
tin
g
s
tu
d
e
n
ts
to
ca
m
p
u
s
clo
s
u
r
es,
ap
p
o
in
t
m
en
t
r
e
m
i
n
d
er
s
,
f
li
g
h
t
ca
n
ce
llatio
n
etc.
An
illeg
a
l
r
o
b
o
ca
ll
is
a
n
o
n
-
e
m
er
g
e
n
c
y
ca
ll
co
n
tai
n
in
g
a
p
r
e
-
r
ec
o
r
d
ed
m
ess
a
g
e
w
i
th
o
u
t
th
e
co
n
s
e
n
t
o
f
th
e
co
n
s
u
m
er
.
I
t
ca
n
b
e
eit
h
er
f
r
o
m
a
r
eg
is
ter
ed
b
u
s
in
e
s
s
w
h
ic
h
co
n
tr
a
v
en
ed
t
h
e
la
w
o
r
f
r
o
m
a
s
ca
m
m
er
t
h
at
p
o
s
e
as
a
leg
al
o
r
g
a
n
izatio
n
i
n
o
r
d
er
to
s
teal
y
o
u
r
m
o
n
e
y
,
id
en
tit
y
o
r
b
o
th
.
T
ec
h
n
o
lo
g
y
h
as
m
ad
e
it
ea
s
y
to
f
i
n
d
w
a
y
s
s
cr
ap
e
p
er
s
o
n
al
in
f
o
r
m
atio
n
o
n
p
u
b
lic
d
atab
ases
o
r
in
ter
n
et
to
f
i
n
d
th
e
p
h
o
n
e
n
u
m
b
er
s
a
n
d
s
ell
t
h
e
m
to
b
o
th
leg
al
an
d
ille
g
al
s
p
a
m
ca
ller
s
.
I
n
C
a
n
ad
a,
d
u
r
in
g
th
e
C
an
a
d
ian
Fed
er
al
2
0
1
1
,
in
o
r
d
er
to
r
ea
ch
v
o
ter
s
,
th
e
p
o
liti
c
al
p
ar
ties
leg
iti
m
atel
y
u
s
ed
r
o
b
o
ca
lls
.
T
h
e
i
n
v
e
s
ti
g
atio
n
s
h
o
w
ed
th
a
t
t
h
e
r
o
b
o
ca
lls
w
er
e
u
s
ed
to
d
iv
er
t
th
e
p
eo
p
le
f
r
o
m
ca
s
tin
g
t
h
eir
b
allo
t
b
y
g
iv
i
n
g
t
h
e
m
i
n
ac
cu
r
ate
in
f
o
r
m
atio
n
o
f
th
e
ch
a
n
g
ed
lo
ca
tio
n
s
o
f
t
h
e
p
o
ll
s
ta
tio
n
s
.
T
h
er
e
h
as
b
ee
n
a
s
teep
r
is
e
in
th
e
au
to
m
ated
ca
lls
s
i
n
ce
2
0
0
9
.
A
cc
o
r
d
in
g
to
th
e
FT
C
r
ep
o
r
t,
an
ag
e
n
c
y
r
ec
eiv
ed
o
v
er
3
7
5
,
0
0
0
co
m
p
lain
ts
ab
o
u
t
a
u
to
m
ated
r
o
b
o
ca
lls
as
co
m
p
ar
ed
to
2
0
0
9
.
T
h
e
r
e
p
o
r
t
also
s
tated
th
at
th
e
in
cr
ea
s
e
i
n
th
e
n
u
m
b
er
o
f
r
o
b
o
ca
lls
is
d
u
e
to
th
e
f
r
ee
o
r
ch
ea
p
ac
ce
s
s
to
in
ter
n
et
ca
llin
g
s
er
v
ices
w
h
ic
h
also
h
elp
s
t
h
e
s
ca
m
m
er
s
h
id
e
t
h
eir
id
en
tit
y
.
Ma
ch
i
n
e
L
ea
r
n
i
n
g
i
s
an
ap
p
lic
atio
n
o
f
ar
ti
f
icial
i
n
tell
ig
e
n
ce
t
h
at
p
r
o
v
id
es
t
h
e
s
y
s
te
m
t
h
e
p
o
ten
tial
to
g
r
asp
p
atter
n
s
a
n
d
lear
n
f
r
o
m
d
ata
an
d
a
m
elio
r
ate
f
r
o
m
e
x
p
er
ien
ce
d
ep
en
d
in
g
o
n
s
o
m
e
task
,
w
it
h
o
u
t
b
ein
g
ex
p
licitl
y
co
d
ed
.
Ma
ch
in
e
L
e
ar
n
in
g
m
a
in
l
y
f
o
c
u
s
e
s
o
n
lear
n
in
g
f
r
o
m
in
p
u
t
d
ata
an
d
p
r
e
d
ictin
g
a
n
o
u
tco
m
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
E
mp
ir
ica
l a
n
a
lysi
s
o
f e
n
s
emb
le
meth
o
d
s
fo
r
th
e
cla
s
s
ifica
tio
n
o
f
r
o
b
o
c
a
lls
in
…
(
Meg
h
n
a
Gh
o
s
h
)
3109
alo
n
g
w
it
h
u
p
d
atin
g
th
e
r
es
u
lts
w
it
h
th
e
ar
r
iv
al
o
f
n
e
w
,
u
n
s
ee
n
d
ata.
E
f
f
icie
n
t
a
n
al
y
s
i
s
o
f
f
r
au
d
i
n
telec
o
m
m
u
n
icatio
n
w
it
h
t
h
e
h
elp
o
f
m
ac
h
i
n
e
lear
n
in
g
ca
n
h
elp
n
et
w
o
r
k
o
p
er
ato
r
s
s
av
e
a
m
o
n
e
y
r
esto
r
e
t
h
e
co
n
s
u
m
er
’
s
co
n
f
id
e
n
ce
in
t
h
e
ir
s
ec
u
r
it
y
.
T
h
e
a
n
al
y
s
is
i
n
m
ac
h
in
e
lear
n
i
n
g
h
as
a
lo
t
o
f
s
tep
s
,
th
e
f
ir
s
t
o
n
e
b
ein
g
id
e
n
ti
f
y
i
n
g
t
h
e
o
b
j
ec
tiv
e
o
f
th
e
ta
s
k
,
d
ep
en
d
in
g
o
n
wh
ich
th
e
m
o
d
el
is
c
h
o
s
en
.
C
o
ll
ec
tio
n
o
f
d
ata
is
t
h
e
n
ex
t
i
m
p
o
r
tan
t
s
tep
f
o
llo
w
ed
b
y
t
h
e
p
r
e
-
p
r
o
ce
s
s
in
g
o
f
d
ata
w
h
ic
h
is
th
e
tr
an
s
f
o
r
m
atio
n
s
ap
p
lied
to
th
e
d
ata.
B
ef
o
r
e
f
ee
d
in
g
th
e
p
r
e
-
p
r
o
ce
s
s
ed
d
ata
to
a
p
ar
ticu
lar
d
ata
it
i
s
r
eq
u
ir
ed
to
s
p
lit
t
h
e
d
at
a
in
to
tr
ain
i
n
g
a
n
d
test
i
n
g
d
ata.
Fe
w
s
tr
ate
g
ies
f
o
r
im
p
r
o
v
i
n
g
th
e
p
er
f
o
r
m
a
n
c
e
o
f
th
e
d
ata
co
u
ld
b
e
s
u
p
p
le
m
en
t
w
it
h
ad
d
itio
n
al
d
ata
o
r
s
w
itch
to
a
d
if
f
er
e
n
t
m
o
d
el.
I
n
m
ac
h
i
n
e
lear
n
i
n
g
,
e
n
s
e
m
b
l
e
m
o
d
els
u
s
e
m
u
ltip
le
lear
n
i
n
g
al
g
o
r
ith
m
s
an
d
co
m
b
i
n
e
t
h
e
m
i
n
o
r
d
er
to
b
u
ild
an
o
p
ti
m
al
p
r
ed
ictiv
e
m
o
d
el.
R
at
h
er
th
a
n
j
u
s
t
r
el
y
i
n
g
o
n
o
n
e
al
g
o
r
it
h
m
s
u
ch
a
s
a
Dec
is
io
n
T
r
ee
an
d
ex
p
ec
tin
g
t
h
at
t
h
e
r
i
g
h
t
d
ec
is
io
n
is
m
ad
e
at
ea
c
h
s
p
li
t,
en
s
e
m
b
le
m
eth
o
d
s
u
g
g
e
s
ts
a
s
s
e
m
b
lin
g
a
s
a
m
p
le
o
f
d
ec
is
io
n
tr
ee
s
an
d
m
a
k
e
a
f
i
n
al
p
r
ed
icto
r
b
ased
o
n
t
h
e
a
g
g
r
e
g
atio
n
o
f
th
e
r
esu
lts
.
T
h
e
p
ap
er
d
ef
in
es
t
h
e
en
s
e
m
b
le
m
e
th
o
d
s
,
b
ag
g
in
g
,
b
o
o
s
tin
g
a
n
d
s
tac
k
in
g
w
h
ic
h
is
d
escr
ib
ed
in
d
etail
b
elo
w
.
a.
B
ag
g
i
n
g
B
ag
g
i
n
g
is
a
n
e
n
s
e
m
b
le
m
eth
o
d
w
h
er
e
t
h
e
m
o
d
el
is
tr
ai
n
e
d
o
n
a
n
u
m
b
er
o
f
b
o
o
ts
tr
ap
p
e
d
s
a
m
p
le
s
an
d
t
h
e
f
i
n
al
m
o
d
el
i
s
t
h
e
ag
g
r
eg
ated
r
es
u
lt
o
f
t
h
e
s
a
m
p
le
m
o
d
els.
B
o
o
ts
tr
ap
p
i
n
g
i
s
t
h
e
p
r
o
ce
s
s
o
f
r
esa
m
p
li
n
g
w
it
h
r
ep
lace
m
e
n
t
w
h
er
e
s
m
a
ll
s
a
m
p
le
s
ar
e
r
ep
ea
ted
ly
c
h
o
s
en
f
r
o
m
th
e
o
r
i
g
i
n
al
s
a
m
p
l
e.
A
f
e
w
d
u
p
licate
s
a
m
p
les
ex
i
s
t
i
n
a
b
o
o
ts
tr
ap
p
ed
s
a
m
p
le
as
s
a
m
p
l
in
g
i
s
d
o
n
e
w
it
h
r
ep
lace
m
en
t.
I
n
t
h
e
ca
s
e
o
f
a
n
u
m
er
ic
tar
g
et
v
ar
iab
le
o
r
r
e
g
r
ess
io
n
p
r
o
b
lem
t
h
e
p
r
ed
icted
o
u
tco
m
e
i
s
th
e
a
v
er
ag
e
o
f
th
e
m
o
d
els
w
h
er
ea
s
i
n
ca
s
e
o
f
class
i
f
icatio
n
t
h
e
p
r
ed
icted
class
is
d
e
f
i
n
ed
b
ased
o
n
p
lu
r
al
it
y
.
B
ag
g
i
n
g
o
r
B
o
o
ts
tr
ap
Ag
g
r
eg
atio
n
g
e
n
er
all
y
r
ed
u
ce
s
v
ar
ian
ce
f
o
r
th
o
s
e
alg
o
r
ith
m
s
h
av
in
g
h
i
g
h
v
ar
ia
n
ce
.
T
h
e
B
ag
g
in
g
m
o
d
el
is
s
h
o
w
n
i
n
Fig
u
r
e
1
.
R
an
d
o
m
Fo
r
est
is
o
n
e
s
u
ch
al
g
o
r
ith
m
th
at
is
a
n
i
m
p
r
o
v
e
m
e
n
t
o
v
er
b
ag
g
ed
d
ec
is
io
n
tr
ee
s
.
R
a
n
d
o
m
Fo
r
est
i
s
th
e
co
m
b
i
n
atio
n
o
f
m
u
ltip
le
d
ec
is
io
n
tr
ee
s
i
n
w
h
ic
h
t
h
e
s
u
b
-
tr
ee
s
ar
e
lear
n
ed
in
s
u
c
h
a
w
a
y
s
o
t
h
at
t
h
e
r
esu
lti
n
g
p
r
ed
ictio
n
s
h
a
v
e
les
s
co
r
r
elatio
n
.
T
h
e
alg
o
r
ith
m
al
s
o
h
elp
s
ca
lc
u
late
t
h
e
er
r
o
r
f
u
n
ctio
n
f
o
r
a
v
ar
iab
le
at
ea
ch
s
p
lit.
B
ag
g
in
g
m
et
h
o
d
is
an
e
f
f
ic
ien
t
m
et
h
o
d
w
h
e
n
i
t
co
m
e
s
to
d
ea
lin
g
w
it
h
o
v
er
f
i
ttin
g
o
f
d
ata
alo
n
g
w
it
h
r
ed
u
ci
n
g
v
ar
ia
n
ce
.
b.
B
o
o
s
tin
g
B
o
o
s
tin
g
i
s
a
n
e
n
s
e
m
b
le
m
et
h
o
d
in
w
h
ic
h
a
s
et
o
f
w
ea
k
lear
n
er
s
i
s
co
n
v
er
ted
to
a
s
e
t
o
f
s
tr
o
n
g
lear
n
er
s
[
2
]
.
A
w
ea
k
lear
n
er
is
a
clas
s
i
f
ier
th
at
h
a
s
v
er
y
l
ess
o
r
s
li
g
h
t
co
r
r
elatio
n
to
th
e
tr
u
e
clas
s
i
f
icatio
n
w
h
er
ea
s
th
e
s
tr
o
n
g
lear
n
er
is
w
ell
co
r
r
elate
d
to
t
h
e
tr
u
e
c
lass
i
f
icatio
n
.
T
h
e
b
ag
g
i
n
g
m
o
d
el
is
d
ep
icted
i
n
Fig
u
r
e
2
.
I
n
t
h
e
tr
ai
n
i
n
g
p
h
a
s
e
w
ei
g
h
ts
ar
e
allo
ca
ted
to
e
ac
h
o
f
t
h
e
r
es
u
lt
in
g
m
o
d
els.
A
h
i
g
h
er
w
ei
g
h
t
i
s
ass
i
g
n
ed
to
t
h
e
lear
n
er
w
it
h
s
u
p
er
io
r
class
i
f
icatio
n
r
es
u
lt
s
a
s
co
m
p
ar
ed
to
a
lear
n
er
w
it
h
p
o
o
r
class
if
icat
io
n
r
esu
lt
s
.
B
o
o
s
tin
g
o
p
ti
m
ize
s
th
e
ad
v
an
ta
g
es
o
f
a
s
i
n
g
le
m
o
d
el
an
d
g
e
n
er
ates
a
co
m
b
in
ed
m
o
d
el
w
it
h
lo
w
er
er
r
o
r
s
.
B
o
o
s
tin
g
h
elp
s
r
ed
u
ce
b
ias.
Gr
ad
ien
t
b
o
o
s
tin
g
i
s
o
n
e
b
o
o
s
tin
g
al
g
o
r
ith
m
th
at
g
r
asp
s
th
e
p
atter
n
s
i
n
o
r
d
er
t
o
s
tr
en
g
t
h
en
a
m
o
d
el
w
i
th
w
ea
k
p
r
ed
ictio
n
s
.
Fig
u
r
e
1
.
B
ag
g
i
n
g
m
o
d
el
Fig
u
r
e
2
.
B
o
o
s
tin
g
m
o
d
el
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
:
3
1
0
8
-
3114
3110
c.
Stack
i
n
g
Stack
i
n
g
i
s
a
w
a
y
to
en
s
e
m
b
l
e
m
u
ltip
le
r
e
g
r
ess
io
n
a
n
d
clas
s
if
ica
tio
n
p
r
o
b
le
m
s
w
h
ic
h
i
n
tr
o
d
u
ce
s
t
h
e
co
n
ce
p
t
o
f
a
m
e
ta
-
lear
n
er
.
St
ac
k
in
g
in
v
o
lv
e
s
tr
ai
n
i
n
g
a
s
e
t
o
f
b
ase
lear
n
er
s
o
n
th
e
d
at
a.
T
h
e
p
r
ed
ictio
n
s
o
b
tain
ed
f
r
o
m
th
e
b
ase
lear
n
e
r
s
ar
e
tak
e
n
a
s
a
n
i
n
p
u
t
f
o
r
t
h
e
m
eta
-
cla
s
s
i
f
ier
w
h
ic
h
g
i
v
es
th
e
f
i
n
al
p
r
ed
icted
o
u
tp
u
t.
Un
l
ik
e
b
ag
g
i
n
g
a
n
d
b
o
o
s
tin
g
,
s
tac
k
i
n
g
i
s
n
o
t
s
o
w
id
el
y
u
s
ed
.
I
t
h
elp
s
in
i
m
p
r
o
v
in
g
th
e
o
v
er
a
ll
p
er
f
o
r
m
a
n
ce
a
n
d
w
e
o
f
ten
e
n
d
u
p
w
i
th
a
m
o
d
el
t
h
at
is
b
etter
th
an
a
n
i
n
d
iv
id
u
al
in
ter
m
ed
iat
e
m
o
d
el.
d.
R
elate
d
w
o
r
k
Qa
y
y
u
m
et
al.
[
3
]
in
th
e
p
ap
er
,
h
as
f
o
cu
s
ed
o
n
d
etec
tin
g
f
r
au
d
u
len
t
b
eh
a
v
io
r
in
t
h
e
f
ield
o
f
s
u
b
s
cr
ip
tio
n
f
r
a
u
d
.
T
im
e
s
er
i
es
d
ata
is
ta
k
e
n
as
in
p
u
t
o
n
w
h
ic
h
n
e
u
r
al
n
et
w
o
r
k
is
i
m
p
l
e
m
en
ted
.
T
h
e
p
ap
er
p
r
o
p
o
s
ed
a
m
u
lti
la
y
er
p
er
ce
p
tr
o
n
w
h
ich
i
s
b
ased
o
n
a
n
i
n
p
u
t,
h
id
d
en
an
d
o
u
tp
u
t la
y
er
.
I
t i
s
u
s
ed
to
ass
o
ciate
a
f
r
au
d
r
ati
n
g
to
a
s
u
b
s
cr
ib
er
b
a
s
ed
o
n
h
is
/
h
er
ca
ll
in
g
p
r
o
f
ile.
T
w
o
tec
h
n
iq
u
es
ar
e
d
ev
el
o
p
e
d
,
o
n
e
in
w
h
ich
th
e
u
s
er
s
ar
e
allo
w
ed
to
ch
o
o
s
e
th
eir
o
w
n
f
r
a
u
d
f
ea
tu
r
e
s
w
h
i
ch
th
e
y
t
h
i
n
k
b
est
f
r
au
d
th
e
y
ar
e
d
ea
lin
g
w
it
h
,
an
o
th
er
i
n
w
h
ich
w
ei
g
h
t
s
p
r
io
r
ities
ar
e
a
s
s
o
ciate
d
b
ased
o
n
t
h
e
u
s
er
’
s
in
p
u
t
to
g
i
v
e
m
o
r
e
p
r
io
r
ity
to
o
n
e
f
r
a
u
d
co
m
p
ar
ed
to
o
th
er
s
.
A
s
i
n
g
le
th
r
es
h
o
ld
v
a
lu
e
is
u
s
ed
b
y
t
h
e
c
u
r
r
en
t
n
e
u
r
al
n
e
t
w
o
r
k
a
n
d
th
e
w
ei
g
h
ts
ar
e
ad
j
u
s
ted
b
ased
o
n
th
e
er
r
o
r
in
p
r
ed
ictio
n
.
Ho
w
e
v
er
,
f
o
r
b
etter
class
i
f
icatio
n
to
o
u
tp
u
t
n
e
u
tr
o
n
co
u
ld
b
e
u
s
ed
.
On
e
o
u
tp
u
t
n
eu
r
o
n
ca
n
b
e
a
s
s
i
g
n
ed
f
o
r
d
etec
tin
g
f
r
au
d
u
l
en
t
b
eh
a
v
io
r
an
d
t
h
e
o
th
er
f
o
r
n
o
n
-
f
r
au
d
u
le
n
t
b
eh
av
io
r
.
Hilas
et
al.
[
4
]
h
av
e
p
r
esen
ted
t
w
o
cl
u
s
ter
i
n
g
al
g
o
r
ith
m
s
to
id
en
ti
f
y
u
s
er
b
eh
a
v
io
r
p
r
o
f
iles
to
d
etec
t
f
r
au
d
u
len
t
ac
ti
v
it
ies
in
a
telec
o
m
m
u
n
icatio
n
o
r
g
a
n
izatio
n
.
Un
s
u
p
er
v
i
s
ed
lear
n
in
g
d
ea
ls
w
it
h
u
n
lab
elle
d
d
ata
an
d
h
elp
s
f
i
n
d
g
r
o
u
p
i
n
g
s
i
n
t
h
e
d
ata.
T
h
e
p
ar
titi
o
n
in
g
a
n
d
h
i
er
ar
ch
ical
clu
s
ter
in
g
ar
e
u
s
ed
i
n
t
h
e
p
r
esen
t
w
o
r
k
.
As
th
e
m
a
in
r
ep
r
esen
tat
iv
e
t
h
e
k
-
m
ea
n
s
alg
o
r
it
h
m
i
s
ap
p
lied
an
d
th
e
h
ier
ar
ch
ical
clu
s
te
r
in
g
is
u
s
ed
in
t
h
e
ag
g
lo
m
er
ati
v
e
cl
u
s
ter
in
g
.
I
n
t
h
e
p
ap
er
,
th
e
w
ell
estab
li
s
h
ed
u
n
s
u
p
er
v
i
s
ed
lear
n
i
n
g
tec
h
n
i
q
u
es
ar
e
ap
p
lied
o
n
telec
o
m
m
u
n
icatio
n
s
d
ata.
T
h
e
tech
n
iq
u
e
s
h
elp
co
m
p
r
e
h
en
d
o
n
a
f
r
a
u
d
u
le
n
t
b
e
h
av
io
u
r
f
r
o
m
a
leg
it
i
m
a
te
u
s
er
’
s
b
eh
a
v
io
r
.
R
a
w
u
s
a
g
e
d
ata
m
u
s
t
b
e
tr
an
s
f
o
r
m
ed
in
to
ap
p
r
o
p
r
iate
u
s
er
p
r
o
f
iles
.
So
m
e
o
f
t
h
e
c
h
alle
n
g
e
s
f
ac
ed
w
er
e
co
n
s
tr
u
ctio
n
a
n
d
s
elec
tio
n
.
I
t
is
co
n
clu
d
ed
f
r
o
m
th
e
a
n
al
y
s
is
th
at,
a
s
r
eg
ar
d
s
u
s
er
p
r
o
f
ile
b
u
ild
in
g
,
ac
cu
m
u
lated
c
h
ar
ac
ter
is
tic
s
o
f
a
u
s
er
y
ield
b
etter
d
is
cr
i
m
in
a
t
io
n
r
esu
l
ts
.
Ho
w
ev
er
,
i
n
o
r
d
er
to
p
r
eser
v
e
o
n
lin
e
d
etec
tio
n
ab
ilit
y
a
g
g
r
e
g
ati
n
g
u
s
er
’
s
b
eh
a
v
io
r
w
a
s
av
o
id
ed
f
o
r
lar
g
er
p
er
io
d
s
.
Misclass
i
f
i
ca
tio
n
in
cl
u
s
ter
i
n
g
o
cc
u
r
r
ed
d
u
e
to
m
i
x
ed
t
y
p
es o
f
b
eh
a
v
io
r
.
P
h
u
a
et
al.
[
5
]
t
h
e
p
ap
er
s
u
r
v
e
y
s
t
h
e
v
ar
io
u
s
tec
h
n
ical
ar
ti
cles
i
n
au
to
m
ated
f
r
au
d
d
etec
tio
n
f
o
r
a
p
er
io
d
o
f
1
0
y
ea
r
s
.
I
t
d
ef
i
n
es
a
n
d
f
o
r
m
al
izes
t
h
e
t
y
p
e
s
an
d
s
u
b
ty
p
e
s
o
f
f
r
au
d
.
T
h
is
r
esear
ch
p
ap
er
p
r
esen
ts
t
h
e
tech
n
iq
u
es
alo
n
g
w
ith
t
h
eir
p
r
o
b
lem
s
.
T
h
e
m
ai
n
o
b
j
ec
tiv
e
i
s
to
d
ef
in
e
cu
r
r
en
t
c
h
alle
n
g
e
s
in
th
is
d
o
m
ain
f
o
r
th
e
d
i
f
f
er
en
t
lar
g
e
d
ata
s
tr
ea
m
s
.
I
t
co
m
p
ar
es
a
n
d
s
u
m
m
ar
i
ze
s
t
h
e
v
ar
io
u
s
d
ata
m
i
n
in
g
b
ased
f
r
a
u
d
d
etec
tio
n
tech
n
iq
u
es.
T
h
e
d
if
f
er
en
t
d
ata
m
i
n
in
g
tec
h
n
iq
u
es
ar
e
s
elec
te
d
d
ep
en
d
in
g
o
n
th
e
p
r
ac
tical
i
s
s
u
es
o
f
o
p
er
atio
n
al
r
eq
u
ir
e
m
en
ts
,
r
eso
u
r
ce
co
n
s
tr
ain
ts
.
Gr
ap
h
-
th
eo
r
etic
a
n
o
m
a
l
y
d
etec
tio
n
an
d
I
n
d
u
c
ti
v
e
L
o
g
ic
P
r
o
g
r
am
m
i
n
g
ar
e
s
o
m
e
o
f
t
h
e
co
m
m
er
cial
f
r
au
d
d
etec
tio
n
tec
h
n
iq
u
es.
No
n
-
l
in
ea
r
s
u
p
er
v
i
s
ed
alg
o
r
it
h
m
s
w
h
ic
h
ar
e
co
m
p
le
x
,
s
u
c
h
a
s
,
s
u
p
p
o
r
t
v
ec
to
r
m
a
ch
in
e
s
a
n
d
n
eu
r
al
n
e
t
w
o
r
k
s
is
g
i
v
e
n
m
o
r
e
i
m
p
o
r
ta
n
ce
.
I
n
t
h
e
lo
n
g
ter
m
,
less
co
m
p
le
x
alg
o
r
ith
m
s
s
u
c
h
as
n
a
ï
v
e
B
a
y
es
a
n
d
lo
g
is
t
ic
r
eg
r
ess
io
n
w
ill
p
r
o
d
u
c
e
f
aster
r
esu
lt
s
if
n
o
t
b
etter
.
Oth
er
r
elate
d
d
ata
m
i
n
i
n
g
tec
h
n
iq
u
es
co
v
er
ed
b
y
s
u
r
v
e
y
p
ap
er
s
in
c
lu
d
e
o
u
tlier
d
etec
tio
n
,
s
k
e
w
ed
/i
m
b
ala
n
ce
d
/r
ar
e
class
es,
s
a
m
p
li
n
g
,
co
s
t
-
s
en
s
iti
v
e
lear
n
in
g
,
s
tr
ea
m
m
i
n
i
n
g
,
g
r
ap
h
m
in
i
n
g
,
an
d
s
ca
lab
ilit
y
.
Fro
m
t
h
i
s
p
ap
er
,
it
is
co
n
clu
d
ed
th
at
u
n
s
u
p
er
v
is
ed
ap
p
r
o
ac
h
es
th
at
ca
n
co
n
tr
ib
u
te
to
f
u
tu
r
e
f
r
au
d
d
etec
tio
n
r
esear
ch
in
cl
u
d
e
ac
tu
al
m
o
n
ito
r
i
n
g
s
y
s
te
m
s
an
d
tex
t
m
in
in
g
f
r
o
m
la
w
en
f
o
r
ce
m
e
n
t
a
n
d
s
e
m
i
-
s
u
p
er
v
i
s
ed
an
d
g
a
m
e
-
th
eo
r
eti
c
ap
p
r
o
ac
h
es f
r
o
m
i
n
tr
u
s
io
n
an
d
s
p
am
d
etec
tio
n
.
Xu
et
a
l.
[
6
]
w
o
r
k
ed
o
n
t
h
e
d
e
tectio
n
o
f
f
r
a
u
d
i
n
t
h
e
3
G
tele
co
m
m
u
n
icatio
n
n
et
w
o
r
k
.
A
r
o
u
g
h
f
u
zz
y
s
et
b
ased
ap
p
r
o
ac
h
w
as
u
s
ed
.
T
h
e
3
G
n
et
w
o
r
k
is
a
n
al
y
ze
d
in
cl
u
d
in
g
t
h
e
s
u
b
s
cr
ip
tio
n
an
d
s
u
p
er
i
m
p
o
s
ed
f
r
au
d
.
T
h
e
p
r
o
f
ile
s
a
n
d
v
ar
i
o
u
s
p
ar
a
m
e
ter
s
ar
e
d
e
f
i
n
ed
in
o
r
d
er
to
p
r
esen
t
t
h
e
f
r
a
m
e
w
o
r
k
.
C
it
i
FM
S,
a
r
u
le
-
b
ased
s
y
s
te
m
,
w
a
s
d
ev
e
lo
p
ed
f
o
r
th
e
d
etec
tio
n
o
f
ab
n
o
r
m
al
ities
a
n
d
alar
m
.
Far
v
ar
es
h
,
H.
et
al.
[
7
]
aim
s
at
id
en
tify
i
n
g
th
e
s
u
b
s
cr
ip
tio
n
f
r
au
d
b
y
an
al
y
zi
n
g
th
e
u
s
er
p
r
o
f
iles
in
th
e
p
ap
er
.
A
v
ar
iet
y
o
f
h
y
b
r
id
alg
o
r
ith
m
s
ar
e
ap
p
lied
to
th
e
d
ataset
ac
q
u
ir
ed
f
r
o
m
t
h
e
T
elec
o
m
m
u
n
icatio
n
C
o
m
p
an
y
o
f
T
eh
r
an
.
I
n
clu
s
te
r
in
g
SOM
a
n
d
K
m
ea
n
s
is
co
m
b
in
ed
w
h
er
ea
s
in
clas
s
if
icat
io
n
SVM,
d
ec
is
io
n
tr
ee
s
,
n
eu
r
al
n
et
w
o
r
k
s
,
b
ag
g
in
g
,
b
o
o
s
tin
g
a
n
d
v
ar
io
u
s
o
th
er
en
s
e
m
b
les
w
er
e
ap
p
lied
.
T
h
e
r
esu
lt
s
r
ev
ea
led
t
h
at
SVM
a
m
o
n
g
s
in
g
le
clas
s
if
ie
r
s
an
d
b
o
o
s
ti
n
g
en
s
e
m
b
le
h
ad
h
i
g
h
er
ac
c
u
r
ac
y
as
co
m
p
ar
ed
to
th
e
o
th
er
alg
o
r
ith
m
s
.
I
n
th
e
p
ap
er
,
th
e
au
th
o
r
d
is
cu
s
s
es
th
e
r
o
le
o
f
n
e
u
r
al
n
et
w
o
r
k
s
f
o
r
p
att
er
n
r
ec
o
g
n
i
ti
o
n
in
th
e
p
r
ev
en
tio
n
o
f
te
leco
m
m
u
n
icat
io
n
f
r
a
u
d
.
A
k
h
ter
et
al.
[
8
]
h
a
s
co
llected
d
ata
o
n
f
r
au
d
u
le
n
t
an
d
n
o
n
-
f
r
au
d
u
le
n
t
ca
lls
w
h
ic
h
ar
e
p
r
ep
r
o
ce
s
s
ed
f
o
r
s
u
itab
le
n
eu
r
al
n
et
w
o
r
k
le
ar
n
in
g
.
A
m
o
d
el
is
b
u
i
lt
f
r
o
m
th
e
p
r
ep
r
o
ce
s
s
ed
d
ata
w
h
ic
h
in
co
r
p
o
r
ates
v
ar
i
o
u
s
p
atter
n
s
o
f
f
r
a
u
d
u
le
n
t
b
eh
av
io
r
.
T
h
e
co
m
b
in
at
io
n
o
f
n
e
u
r
al,
r
u
le
-
b
ased
,
ca
s
e
-
b
ased
tec
h
n
o
lo
g
ies
p
r
o
v
i
d
e
a
f
r
au
d
d
etec
tio
n
r
ate
s
u
p
er
io
r
to
th
at
o
f
co
n
v
en
t
io
n
al
s
y
s
t
e
m
s
a
n
d
th
e
m
u
lti
-
s
tr
ea
m
a
n
al
y
s
is
ca
p
ab
ilit
y
m
ak
es
it
e
x
tr
e
m
el
y
ac
c
u
r
ate.
Du
e
to
th
e
i
n
h
er
e
n
t
ab
ilit
y
t
o
ad
ap
t
alo
n
g
w
it
h
th
e
s
p
ee
d
an
d
ef
f
icie
n
c
y
,
A
r
ti
f
icial
Neu
r
al
Net
w
o
r
k
i
s
a
b
ette
r
m
e
th
o
d
f
o
r
d
etec
tin
g
telep
h
o
n
e
f
r
a
u
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
E
mp
ir
ica
l a
n
a
lysi
s
o
f e
n
s
emb
le
meth
o
d
s
fo
r
th
e
cla
s
s
ifica
tio
n
o
f
r
o
b
o
c
a
lls
in
…
(
Meg
h
n
a
Gh
o
s
h
)
311
1
Su
b
s
cr
ip
tio
n
Fra
u
d
,
C
al
l
Fo
r
w
ar
d
in
g
,
C
alli
n
g
B
y
p
a
s
s
,
R
o
a
m
i
n
g
Fra
u
d
,
an
d
C
lo
n
i
n
g
Fr
au
d
is
t
h
e
d
if
f
er
e
n
t
t
y
p
es
o
f
f
r
au
d
i
n
tele
co
m
m
u
n
ica
tio
n
.
A
d
eb
is
i
et
al.
[
9
]
d
ev
elo
p
ed
a
m
o
d
el
th
a
t
d
etec
ts
telec
o
m
m
u
n
icatio
n
f
r
au
d
b
as
ed
o
n
a
n
e
u
r
al
n
et
w
o
r
k
e
n
s
e
m
b
le
m
eth
o
d
.
A
r
a
n
d
o
m
r
o
u
g
h
s
u
b
s
p
ac
e
b
ased
n
eu
r
al
n
et
w
o
r
k
e
n
s
e
m
b
le
m
et
h
o
d
w
as
e
m
p
lo
y
ed
in
t
h
e
d
ev
elo
p
m
e
n
t
o
f
th
e
m
o
d
el.
T
h
e
m
o
d
el
w
as
d
esi
g
n
e
d
to
d
etec
t
s
u
b
s
cr
ip
tio
n
f
r
au
d
.
I
t
p
r
esen
ts
t
h
e
d
ev
elo
p
m
en
t
o
f
p
atter
n
s
th
at
p
o
r
tr
a
y
s
th
e
c
u
s
to
m
er
’
s
b
eh
a
v
io
r
f
o
cu
s
in
g
o
n
t
h
e
id
e
n
ti
f
icatio
n
o
f
n
o
n
-
p
a
y
m
en
t
e
v
en
t
s
.
R
u
le
s
w
er
e
f
o
r
m
ed
b
ased
o
n
t
h
e
i
n
f
o
r
m
atio
n
in
ter
r
elate
d
w
ith
o
t
h
er
f
ea
t
u
r
es.
T
h
is
lead
to
f
aster
p
r
ed
ictio
n
s
to
p
r
ev
en
t
r
ev
en
u
e
lo
s
s
f
o
r
th
e
co
m
p
an
y
.
T
h
e
r
esu
lts
s
h
o
w
ed
t
h
at
n
e
u
r
a
l
n
et
w
o
r
k
clas
s
i
f
ier
1
g
av
e
7
w
r
o
n
g
clas
s
i
f
icatio
n
s
,
th
e
s
ec
o
n
d
class
i
f
ier
g
av
e
1
2
w
r
o
n
g
clas
s
i
f
icatio
n
s
,
th
e
th
ir
d
class
i
f
ier
g
a
v
e
1
w
r
o
n
g
class
if
ica
tio
n
an
d
t
h
e
f
o
u
r
t
h
class
if
ier
g
av
e
7
w
r
o
n
g
c
lass
if
ica
tio
n
s
.
T
h
e
n
e
u
r
al
n
et
w
o
r
k
en
s
e
m
b
le
o
u
tp
e
r
f
o
r
m
ed
f
u
r
t
h
er
e
n
h
a
n
ci
n
g
t
h
e
ef
f
icie
n
c
y
o
f
t
h
e
p
r
o
p
o
s
ed
m
o
d
el.
C
o
x
e
t
al.
[
1
0
]
i
n
t
h
e
p
ap
er
d
o
m
ain
-
s
p
ec
if
ic
i
n
ter
f
ac
e
s
a
r
e
b
u
ilt
f
o
r
telep
h
o
n
e
f
r
a
u
d
d
etec
tio
n
.
Hu
m
an
r
ec
o
g
n
itio
n
s
k
ill
s
ex
ce
ed
au
to
m
ated
m
i
n
i
n
g
alg
o
r
it
h
m
s
.
E
x
p
lo
itin
g
p
eo
p
le’
s
ab
ilit
y
to
d
ea
l
w
it
h
v
is
u
al
r
ep
r
esen
tatio
n
s
w
e
m
a
y
r
ev
o
l
u
tio
n
ize
t
h
e
w
a
y
w
e
u
n
d
er
s
ta
n
d
a
lar
g
e
a
m
o
u
n
t
o
f
d
ata.
Di
f
f
er
en
t
v
ie
w
s
o
f
th
e
s
a
m
e
d
ata
ca
n
b
e
i
n
ter
li
n
k
ed
.
T
h
e
v
is
u
ali
za
tio
n
ap
p
r
o
ac
h
t
o
d
etec
tin
g
ca
lli
n
g
f
r
a
u
d
in
v
o
lv
es
a
d
is
p
la
y
o
f
ca
llin
g
ac
tiv
it
y
t
h
at
d
is
p
la
y
s
th
e
u
n
u
s
u
al
p
atter
n
s
an
d
with
th
e
h
elp
o
f
o
n
e
o
r
m
o
r
e
d
r
ill
-
d
o
w
n
v
ie
w
s
,
s
u
s
p
icio
u
s
p
atter
n
s
m
a
y
b
e
f
u
r
th
er
in
v
es
tig
a
ted
.
P
atter
n
r
ec
o
g
n
itio
n
i
s
p
er
f
o
r
m
ed
o
n
th
e
A
M
A
ca
ll
r
ec
o
r
d
s
.
T
h
e
s
ca
tter
p
lo
t
an
d
b
ar
p
lo
t
s
h
o
w
t
h
e
v
ar
io
u
s
an
o
m
alie
s
.
T
h
e
ad
v
an
ta
g
es
o
f
v
i
s
u
al
d
ata
m
in
i
n
g
lie
i
n
t
h
e
f
ac
t
th
at
p
eo
p
le
ex
ce
l
at
d
etec
tin
g
p
atter
n
s
,
th
e
d
y
n
a
m
ic
n
at
u
r
e
o
f
f
r
a
u
d
m
ak
e
s
it
a
ch
alle
n
g
i
n
g
d
etec
tio
n
p
r
o
b
lem
f
o
r
s
tatic
a
lg
o
r
it
h
m
s
.
I
n
th
i
s
p
ap
er
t
w
o
r
ep
r
esen
tatio
n
s
,
o
n
e
f
o
r
ca
llin
g
co
m
m
u
n
itie
s
a
n
d
t
h
e
o
th
er
f
o
r
s
h
o
w
in
g
in
d
iv
id
u
al
ca
lls
h
a
v
e
p
r
o
v
e
n
to
b
e
ef
f
ec
t
iv
e
to
d
etec
t
f
r
au
d
i
n
telec
o
m
m
u
n
icatio
n
s
.
P
eo
p
le
co
m
p
le
m
e
n
t
m
ac
h
in
e
s
an
d
b
etter
ex
p
lo
it th
e
ca
p
ab
ilit
ies f
o
r
k
n
o
w
led
g
e
d
is
co
v
er
y
.
W
u
et
al.
[
1
1
]
id
en
ti
f
ie
s
t
h
e
s
ta
n
d
ar
d
f
ea
tu
r
e
s
o
f
f
r
a
u
d
u
l
en
t
b
eh
a
v
io
r
o
f
cu
s
to
m
er
s
i
n
telec
o
m
in
d
u
s
tr
y
s
y
s
te
m
atica
l
l
y
.
T
h
e
o
u
tlier
s
i
n
d
ata
ar
e
id
en
ti
f
ied
b
y
t
h
e
cl
u
s
ter
i
n
g
tec
h
n
iq
u
es.
T
h
e
w
o
r
k
i
n
g
iv
e
s
d
ef
in
i
tio
n
o
f
tar
g
et
cu
s
to
m
er
s
w
h
o
ar
e
m
a
l
icio
u
s
l
y
b
ased
o
n
t
h
ese
s
p
ec
i
f
ic
m
et
h
o
d
s
ar
e
p
r
o
p
o
s
ed
to
b
u
ild
,
ev
alu
a
te,
an
d
ap
p
l
y
t
h
e
m
o
d
el
f
o
r
id
en
ti
f
y
i
n
g
f
r
au
d
u
len
t
b
eh
av
io
r
.
Ko
h
o
n
e
n
n
eu
r
al
n
et
w
o
r
k
an
d
cl
u
s
ter
in
g
alg
o
r
ith
m
ar
e
ef
f
icie
n
tl
y
u
s
ed
f
o
r
th
e
d
etec
tio
n
o
f
o
u
tlier
s
.
2.
RE
S
E
ARCH
M
E
T
H
O
D
T
h
e
s
tu
d
y
f
o
cu
s
ed
o
n
f
o
u
r
d
if
f
er
e
n
t
p
ar
ts
w
h
ich
ar
e,
ac
q
u
is
itio
n
o
f
d
ata,
d
ata
p
r
e
-
p
r
o
ce
s
s
in
g
,
d
ev
elo
p
m
en
t
o
f
t
h
e
m
o
d
el
b
y
co
m
b
i
n
i
n
g
b
ag
g
i
n
g
an
d
b
o
o
s
tin
g
alg
o
r
it
h
m
s
u
s
in
g
v
o
tin
g
clas
s
i
f
ier
a
n
d
d
ev
elo
p
m
en
t o
f
a
s
tac
k
in
g
m
o
d
el.
a.
Data
s
o
u
r
ce
T
h
e
d
ataset
u
s
ed
f
o
r
th
e
ex
p
e
r
i
m
en
t
is
ac
q
u
ir
ed
f
r
o
m
t
h
e
F
ed
er
al
T
r
ad
e
C
o
m
m
i
s
s
io
n
,
Do
No
t
C
all
(
DNC)
R
ep
o
r
ted
C
alls
Da
ta.
T
h
e
d
ataset
co
n
tain
s
in
f
o
r
m
a
t
io
n
ab
o
u
t
th
e
r
o
b
o
ca
lls
t
h
at
w
er
e
r
ep
o
r
ted
to
th
e
Fed
er
al
T
r
a
d
e
C
o
m
m
is
s
io
n
.
T
h
e
d
ataset
in
cl
u
d
es
i
n
f
o
r
m
atio
n
ab
o
u
t
t
h
e
p
h
o
n
e
n
u
m
b
er
o
r
ig
in
ati
n
g
th
e
u
n
w
an
ted
ca
ll,
th
e
d
ate
an
d
tim
e
t
h
e
ca
ll
w
a
s
m
ad
e,
th
e
d
ate
a
n
d
tim
e
t
h
e
co
m
p
lain
t
w
a
s
m
ad
e,
th
e
co
n
s
u
m
er
's
ci
t
y
,
s
tate
an
d
ar
ea
co
d
e
an
d
th
e
s
u
b
j
ec
t
o
f
th
e
ca
l
ls
.
T
h
e
d
ata
s
et
al
s
o
co
n
tai
n
s
a
co
l
u
m
n
o
f
d
ata
s
tatin
g
w
h
eth
er
t
h
e
ca
ll is
a
r
o
b
o
ca
ll/re
co
r
d
e
d
m
es
s
a
g
e
o
r
n
o
t.
b.
Data
p
r
e
p
r
o
ce
s
s
in
g
T
h
e
g
iv
en
d
ataset
i
s
alr
ea
d
y
co
n
g
r
eg
a
ted
in
th
e
f
o
r
m
o
f
co
lu
m
n
s
w
i
th
ea
c
h
r
o
w
h
a
v
i
n
g
a
u
n
iq
u
e
id
en
tit
y
.
T
h
e
f
ir
s
t
s
tep
in
t
h
e
p
r
ep
r
o
ce
s
s
in
g
o
f
d
ata
in
cl
u
d
es
h
an
d
li
n
g
th
e
m
is
s
i
n
g
v
al
u
es
i
n
th
e
d
ataset.
T
h
e
m
i
s
s
i
n
g
v
al
u
es
ca
n
b
e
i
m
p
u
ted
u
s
i
n
g
s
tatis
tics
(
m
ea
n
,
m
ed
ian
,
m
o
d
e)
o
f
ea
ch
o
f
t
h
e
co
lu
m
n
s
o
r
b
y
u
s
i
n
g
a
co
n
s
tan
t
v
al
u
e.
Feat
u
r
e
Sele
ctio
n
is
a
n
o
th
er
v
ital
s
tep
i
n
th
e
p
r
e
-
p
r
o
ce
s
s
in
g
o
f
d
ata
w
h
ic
h
in
c
lu
d
es
c
h
o
o
s
in
g
th
e
s
u
b
s
e
t
o
f
f
ea
tu
r
e
s
t
h
at
ar
e
r
ele
v
an
t
to
t
h
e
p
r
ed
ictiv
e
m
o
d
elin
g
p
r
o
b
le
m
.
A
s
et
o
f
s
i
x
f
e
atu
r
es
ar
e
s
elec
ted
f
o
r
th
e
ex
p
er
i
m
en
t.
T
ab
le
1
s
h
o
w
s
t
h
e
lis
t
o
f
d
escr
ip
to
r
s
u
s
ed
f
o
r
th
e
p
u
r
p
o
s
e
o
f
d
esig
n
in
g
th
e
cla
s
s
i
f
icatio
n
m
o
d
el.
T
h
e
d
ata
ac
q
u
ir
ed
f
r
o
m
t
h
e
m
o
n
t
h
o
f
J
u
l
y
i
s
u
s
ed
as
th
e
tr
ain
i
n
g
d
ata.
T
h
e
m
o
d
el
is
tr
ai
n
ed
o
n
t
h
e
tr
ain
i
n
g
d
ata.
T
h
e
d
ata
ac
q
u
ir
e
d
f
r
o
m
t
h
e
m
o
n
t
h
o
f
A
u
g
u
s
t
is
u
s
ed
as
th
e
tes
t
d
ata.
T
h
e
m
o
d
el
is
f
it
o
n
th
e
te
s
t
d
ata
to
g
et
th
e
d
esire
d
o
u
tco
m
es.
T
ab
le
1
.
Featu
r
e
d
escr
ip
to
r
s
F
i
e
l
d
N
a
me
D
e
scri
p
t
i
o
n
D
a
t
a
T
y
p
e
C
o
mp
a
n
y
_
P
h
o
n
e
_
N
u
m
b
e
r
T
e
l
e
p
h
o
n
e
n
u
m
b
e
r
o
r
i
g
i
n
a
t
i
n
g
t
h
e
c
a
l
l
.
I
n
t
6
4
V
i
o
l
a
t
i
o
n
_
D
a
t
e
T
h
e
d
a
t
e
a
n
d
t
i
me
t
h
e
c
a
l
l
w
a
s ma
d
e
.
O
b
j
e
c
t
C
o
n
s
u
me
r
_
S
t
a
t
e
T
h
e
c
o
n
su
me
r
’
s st
a
t
e
l
o
c
a
t
i
o
n
s.
O
b
j
e
c
t
C
o
n
s
u
me
r
_
C
i
t
y
T
h
e
c
o
n
su
me
r
’
s c
i
t
y
l
o
c
a
t
i
o
n
s
.
O
b
j
e
c
t
C
o
n
s
u
me
r
_
A
r
e
a
_
C
o
d
e
T
h
e
a
r
e
a
c
o
d
e
o
f
t
h
e
c
o
n
s
u
me
r
’
s c
i
t
y
.
I
n
t
6
4
S
u
b
j
e
c
t
T
h
e
su
b
j
e
c
t
o
f
t
h
e
c
a
l
l
.
O
b
j
e
c
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
:
3
1
0
8
-
3114
3112
c.
R
an
d
o
m
Fo
r
est
R
an
d
o
m
Fo
r
est
i
s
a
s
u
p
er
v
is
ed
lear
n
in
g
tec
h
n
iq
u
e
th
at
es
s
en
tia
ll
y
u
s
es
a
n
en
s
e
m
b
le
o
f
d
ec
o
r
r
elate
d
d
ec
is
io
n
tr
ee
s
.
T
h
e
b
asic
id
ea
o
f
R
an
d
o
m
Fo
r
est
i
s
t
h
at
t
h
e
y
u
s
e
b
a
g
g
i
n
g
o
r
b
o
o
ts
tr
ap
ag
g
r
e
g
ati
n
g
.
I
t
u
s
es
s
a
m
p
li
n
g
w
it
h
r
ep
lace
m
en
t
to
cr
ea
te
m
u
ltip
le
d
ataset
s
a
n
d
o
n
ea
c
h
o
f
th
e
d
ata
s
et,
t
h
e
d
ec
i
s
io
n
tr
ee
al
g
o
r
ith
m
is
ap
p
lied
.
E
ac
h
tr
ee
co
m
e
s
u
p
w
it
h
a
p
r
ed
ictio
n
w
h
ich
c
o
u
ld
b
e
a
clas
s
i
f
icatio
n
o
r
a
r
eg
r
ess
io
n
.
R
a
n
d
o
m
Fo
r
est
co
n
s
is
t
s
o
f
a
co
llect
io
n
o
f
tr
ee
-
s
tr
u
ct
u
r
ed
clas
s
i
f
ier
s
{
h
(
x
,
Θk
)
k
=1
,
2
,
…
.
}
w
h
er
e
Θ
k
ar
e
iid
(
in
d
ep
en
d
en
t
id
en
tical
l
y
d
is
tr
ib
u
ted
r
an
d
o
m
v
ec
to
r
s
a
n
d
ea
ch
tr
ee
ca
s
ts
a
v
o
te
f
o
r
th
e
m
o
s
t
f
av
o
r
ed
class
[
1
2
]
.
T
h
e
r
esu
lts
ar
e
f
in
a
ll
y
ag
g
r
e
g
ated
.
On
ap
p
r
o
ac
h
is
r
an
d
o
m
s
u
b
s
p
ac
e
m
et
h
o
d
w
h
er
e
at
ea
ch
s
p
lit
th
e
alg
o
r
ith
m
is
p
r
o
v
id
ed
w
ith
a
s
u
b
s
et
o
f
f
ea
t
u
r
es
o
n
w
h
ic
h
it
ca
n
b
e
s
p
lit.
I
t
r
an
d
o
m
l
y
s
elec
ts
t
h
e
s
u
b
s
ets
t
h
at
ar
e
g
iv
e
n
at
ea
c
h
s
te
p
.
T
h
er
e
is
also
t
h
e
r
an
d
o
m
s
p
lit
s
elec
tio
n
in
w
h
i
ch
a
d
ata
p
o
in
t
is
s
elec
ted
r
an
d
o
m
l
y
o
n
w
h
ic
h
t
h
e
s
p
lit
ca
n
o
cc
u
r
.
Du
r
in
g
t
h
e
tr
ee
b
u
ild
in
g
p
r
o
ce
s
s
u
s
i
n
g
b
o
o
ts
tr
ap
s
am
p
le
s
a
s
m
a
ll
p
ar
t
o
f
th
e
o
r
ig
i
n
al
in
s
tan
ce
s
ar
e
lef
t
o
u
t.
T
h
e
s
et
o
f
in
s
ta
n
ce
s
is
ca
lled
t
h
e
OOB
(
Ou
t
O
f
B
a
g
)
d
ata
w
h
ich
is
u
s
ed
f
o
r
th
e
er
r
o
r
esti
m
a
tio
n
in
i
n
d
iv
id
u
al
tr
ee
s
.
R
an
d
o
m
Fo
r
est
h
elp
s
in
cr
ea
s
e
class
if
icatio
n
ac
cu
r
ac
y
b
y
av
er
ag
i
n
g
th
e
n
o
is
y
an
d
u
n
b
iased
m
o
d
els
an
d
b
u
ild
in
g
a
m
o
d
el
w
it
h
lo
w
v
ar
ian
ce
.
d.
XGB
o
o
s
t
Gr
ad
ien
t
B
o
o
s
tin
g
is
a
s
u
p
er
v
is
ed
lear
n
i
n
g
alg
o
r
it
h
m
t
h
at
w
o
r
k
s
w
ell
f
o
r
b
o
th
cl
ass
i
f
icatio
n
a
n
d
r
eg
r
ess
io
n
p
r
o
b
le
m
s
.
Gr
ad
ien
t
B
o
o
s
tin
g
ai
m
s
at
b
u
ild
i
n
g
a
p
r
ed
ictiv
e
m
o
d
el.
T
h
e
p
r
ed
ictiv
e
m
o
d
el
i
s
a
n
en
s
e
m
b
le
o
f
v
ar
io
u
s
w
ea
k
le
ar
n
er
s
s
u
c
h
a
s
d
ec
is
io
n
tr
ee
s
.
T
h
e
g
o
al
i
s
to
m
i
n
i
m
ize
t
h
e
er
r
o
r
f
u
n
ctio
n
f
o
r
o
p
tim
izatio
n
p
u
r
p
o
s
es.
Gr
ad
i
en
t
d
escen
t
is
o
n
e
w
a
y
o
f
m
i
n
i
m
izi
n
g
t
h
e
lo
s
s
f
u
n
ct
io
n
b
y
u
p
d
ati
n
g
t
h
e
p
r
ed
ictio
n
s
b
ased
o
n
a
lear
n
i
n
g
r
ate.
XGB
o
o
s
t
s
y
s
te
m
i
s
av
ailab
le
as
an
o
p
en
s
o
u
r
ce
p
ac
k
ag
e
[
1
3
]
.
XGB
o
o
s
t
is
a
p
o
w
er
f
u
l,
o
p
ti
m
ized
g
r
ad
ien
t
b
o
o
s
ti
n
g
l
ib
r
ar
y
th
a
t
p
r
o
v
id
es
p
ar
allel
tr
ee
b
o
o
s
tin
g
.
XGB
o
o
s
t
is
a
r
o
b
u
s
t
lib
r
ar
y
a
n
d
is
ef
f
icac
io
u
s
i
n
m
o
d
el
p
er
f
o
r
m
an
ce
a
n
d
ex
ec
u
ti
o
n
s
p
ee
d
.
e.
Vo
tin
g
cla
s
s
i
f
ier
an
d
s
tack
i
n
g
class
i
f
ier
T
w
o
m
e
th
o
d
s
ar
e
i
m
p
le
m
e
n
te
d
o
n
th
e
d
ata
an
d
th
e
r
esu
lts
p
r
o
cu
r
ed
f
r
o
m
t
h
e
ex
p
er
i
m
e
n
t
s
ar
e
co
m
p
ar
ed
in
ter
m
s
o
f
ac
c
u
r
ac
y
.
T
h
e
f
ir
s
t
m
eth
o
d
o
lo
g
y
in
v
o
l
v
es a
p
p
l
y
i
n
g
t
h
e
R
a
n
d
o
m
Fo
r
est a
n
d
XG
B
o
o
s
t a
lg
o
r
ith
m
o
n
th
e
d
ata.
T
h
e
r
esu
lt
s
ar
e
th
e
n
i
n
teg
r
ated
w
i
th
t
h
e
h
elp
o
f
t
h
e
Vo
tin
g
C
la
s
s
i
f
ier
.
X
u
e
y
i
W
a
n
g
en
t
itled
“A
Ne
w
Mo
d
el
f
o
r
Me
asu
r
in
g
t
h
e
A
cc
u
r
ac
ies
o
f
Ma
j
o
r
it
y
Vo
ti
n
g
E
n
s
e
m
b
les
“
a
n
e
w
m
o
d
el
ca
lled
C
OB
(
co
r
e,
o
u
tlier
,
an
d
b
o
u
n
d
ar
y
)
w
h
ich
q
u
a
n
ti
tativ
el
y
m
ea
s
u
r
es
t
h
e
ac
c
u
r
ac
ies
o
f
m
aj
o
r
ity
v
o
ti
n
g
en
s
e
m
b
les
f
o
r
b
in
ar
y
class
i
f
icatio
n
[
1
4
]
.
T
h
e
Vo
tin
g
C
las
s
if
ier
cr
ea
te
s
a
n
e
n
s
e
m
b
le
b
y
c
o
m
b
in
in
g
t
w
o
s
tan
d
alo
n
e
m
o
d
els.
T
h
e
Vo
tin
g
C
las
s
i
f
ier
co
m
b
in
es
th
e
p
r
ed
ictio
n
s
f
r
o
m
th
e
b
a
g
g
i
n
g
a
n
d
b
o
o
s
tin
g
al
g
o
r
it
h
m
.
I
t
p
r
o
v
id
es
u
s
w
i
t
h
th
e
av
er
a
g
e
o
f
th
e
p
r
ed
ictio
n
s
.
A
r
eq
u
is
ite
f
o
r
an
en
s
e
m
b
le
o
f
class
i
f
ier
s
i
s
to
b
e
p
r
ec
is
e
th
an
t
h
e
in
d
i
v
id
u
al
class
i
f
ier
s
i
s
i
f
t
h
e
in
d
ep
en
d
e
n
t
class
i
f
ier
s
ar
e
d
i
v
er
s
e
an
d
ac
cu
r
ate
[
1
5
]
.
T
h
e
f
o
llo
w
i
n
g
r
es
u
lts
w
er
e
o
b
tain
ed
o
n
th
e
s
u
cc
es
s
f
u
l b
u
ild
o
f
th
e
en
s
e
m
b
le
w
h
ich
i
s
s
h
o
w
n
i
n
T
ab
le
2
.
T
ab
le
2
.
C
lass
if
icatio
n
r
ep
o
r
t
P
r
e
c
i
si
o
n
F1
-
sco
r
e
r
e
c
a
l
l
su
p
p
o
r
t
N
0
.
3
4
0
.
0
3
0
.
0
5
7
0
9
5
Y
0
.
7
0
0
.
9
8
0
.
8
2
1
6
7
2
3
A
v
g
/
t
o
t
a
l
0
.
5
9
2
3
8
1
8
0
.
5
9
2
3
8
1
8
3.
RE
SU
L
T
S
A
ND
AN
AL
Y
SI
S
T
h
e
ev
alu
atio
n
s
o
f
th
e
Vo
ti
n
g
C
las
s
if
ier
m
o
d
el
an
d
Stack
in
g
C
la
s
s
i
f
ier
m
o
d
els
ar
e
co
m
p
ar
ed
.
T
h
e
in
ter
p
r
etatio
n
o
f
r
esu
lts
t
h
at
th
e
Vo
ti
n
g
C
las
s
i
f
ier
m
o
d
el
is
b
etter
th
a
n
th
e
S
tack
i
n
g
C
l
ass
i
f
ier
in
ter
m
s
o
f
ac
cu
r
ac
y
,
ti
m
e
a
n
d
g
e
n
er
aliza
tio
n
er
r
o
r
.
T
h
er
e
is
n
o
o
v
er
f
itt
in
g
in
th
e
d
ata.
C
o
m
p
ar
in
g
t
h
e
tr
u
e
n
e
g
ati
v
e
a
n
d
tr
u
e
p
o
s
iti
v
e
o
f
th
e
co
n
f
u
s
io
n
m
atr
i
x
d
er
iv
ed
f
r
o
m
b
o
t
h
t
h
e
m
o
d
els,
t
h
e
ac
c
u
r
a
c
y
f
o
r
Vo
t
in
g
C
la
s
s
i
f
ier
w
as
ca
lcu
lated
to
b
e
7
0
.
3
5
%
as
o
p
p
o
s
ed
t
o
th
e
Stack
in
g
C
la
s
s
i
f
ier
w
h
ic
h
s
h
o
w
ed
th
e
ac
c
u
r
ac
y
o
f
o
n
l
y
6
1
.
7
%.
T
h
e
ti
m
e
tak
e
n
b
y
t
h
e
Vo
tin
g
C
la
s
s
i
f
ier
w
as
al
s
o
co
n
s
id
e
r
ab
ly
lo
w
,
1
.
9
8
8
8
s
ec
s
as
co
m
p
ar
ed
to
Stac
k
in
g
C
las
s
i
f
ier
w
h
ic
h
tak
e
s
1
1
.
8
8
5
s
ec
s
.
T
h
er
ef
o
r
e,
th
e
h
y
b
r
id
ap
p
r
o
ac
h
o
f
co
m
b
in
i
n
g
th
e
b
o
o
s
tin
g
an
d
b
ag
g
in
g
alg
o
r
ith
m
u
s
i
n
g
th
e
a
g
g
r
a
v
at
io
n
tech
n
iq
u
e
o
u
tp
er
f
o
r
m
ed
t
h
e
s
tac
k
i
n
g
m
et
h
o
d
.
T
a
b
le
s
3
an
d
4
s
h
o
w
t
h
e
co
n
f
u
s
io
n
m
atr
i
x
f
o
r
th
e
p
r
o
p
o
s
ed
m
o
d
els.
C
o
m
ap
r
is
o
n
o
f
m
o
d
el
ac
cu
r
ac
y
a
s
s
h
o
w
n
i
n
Fi
g
u
r
e
3
.
T
ab
le
3
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
v
o
tin
g
clas
s
i
f
ier
P
r
e
d
i
c
t
e
d
O
b
se
r
v
e
d
P
o
si
t
i
v
e
P
o
si
t
i
v
e
N
e
g
a
t
i
v
e
2
0
9
6
8
8
6
N
e
g
a
t
i
v
e
4
1
3
1
6
3
1
0
T
ab
le
4
.
C
o
n
f
u
s
io
n
m
atr
i
x
o
f
s
tack
i
n
g
cla
s
s
i
f
ier
P
r
e
d
i
c
t
e
d
O
b
se
r
v
e
d
P
o
si
t
i
v
e
P
o
si
t
i
v
e
N
e
g
a
t
i
v
e
3
0
1
8
5
7
4
N
e
g
a
t
i
v
e
5
3
8
1
4
3
8
5
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8708
E
mp
ir
ica
l a
n
a
lysi
s
o
f e
n
s
emb
le
meth
o
d
s
fo
r
th
e
cla
s
s
ifica
tio
n
o
f
r
o
b
o
c
a
lls
in
…
(
Meg
h
n
a
Gh
o
s
h
)
3113
Fig
u
r
e
3
.
C
o
m
ap
r
is
o
n
o
f
m
o
d
el
ac
cu
r
ac
y
4.
CO
NCLU
SI
O
N
T
h
e
m
ai
n
f
o
cu
s
o
f
th
e
p
ap
er
w
a
s
to
co
m
e
u
p
w
it
h
a
h
y
b
r
id
ap
p
r
o
ac
h
o
f
co
m
b
in
in
g
th
e
t
w
o
d
ef
i
n
ed
en
s
e
m
b
les,
B
ag
g
i
n
g
an
d
B
o
o
s
tin
g
.
T
h
e
p
o
ten
tialiti
e
s
o
f
th
e
lear
n
in
g
m
o
d
els
w
er
e
e
x
p
lo
r
ed
.
T
h
e
co
n
j
u
n
ctio
n
o
f
B
ag
g
i
n
g
an
d
B
o
o
s
tin
g
a
n
d
B
o
o
s
tin
g
al
g
o
r
ith
m
s
d
ec
r
ea
s
e
th
e
b
ias
a
n
d
v
ar
ian
ce
f
r
o
m
th
e
d
ataset
w
h
ich
lead
s
to
a
m
o
r
e
ef
f
icien
t
m
o
d
e
l
w
i
th
le
s
s
g
en
er
aliza
tio
n
er
r
o
r
.
T
h
e
Vo
tin
g
C
la
s
s
i
f
ier
clas
s
i
f
ied
th
e
r
esu
l
t
m
o
r
e
ef
f
ec
tiv
e
l
y
w
it
h
t
h
e
m
o
d
el
h
a
v
in
g
h
i
g
h
er
tr
u
e
n
e
g
ati
v
e
r
ate
s
as
co
m
p
ar
ed
to
tr
u
e
p
o
s
iti
v
e
.
T
h
er
e
ar
e
v
ar
io
u
s
o
th
er
en
s
e
m
b
les t
h
at
co
u
ld
b
e
co
m
b
i
n
ed
in
to
a
h
y
b
r
id
m
o
d
el
to
ch
ec
k
th
e
p
r
o
f
icie
n
c
y
o
f
t
h
e
m
o
d
el.
ACK
NO
WL
E
D
G
E
M
E
NT
S
T
h
e
au
t
h
o
r
w
o
u
ld
l
ik
e
to
t
h
a
n
k
g
u
id
e
an
d
m
e
n
to
r
,
P
r
o
f
.
P
r
ab
u
P
,
f
o
r
h
is
co
n
s
tan
t
s
u
p
p
o
r
t
th
r
o
u
g
h
o
u
t
th
e
cr
ea
tio
n
o
f
t
h
is
r
esear
ch
d
o
cu
m
e
n
t.
O
th
er
f
ac
il
itato
r
s
o
f
th
e
u
n
iv
er
s
it
y
h
av
e
also
b
ee
n
a
m
o
s
t
h
elp
f
u
l
a
ll
y
in
s
tr
u
c
tu
r
i
n
g
t
h
e
d
ata
an
d
u
n
d
er
s
tan
d
i
n
g
t
h
e
p
r
o
b
lem
d
o
m
ai
n
.
T
h
e
p
a
p
er
w
o
u
ld
n
o
t
h
av
e
b
ee
n
co
m
p
lete
w
it
h
o
u
t
t
h
e
m
ass
iv
e
co
n
tr
ib
u
tio
n
s
b
y
a
ll
t
h
e
f
ac
u
l
ties
a
n
d
r
eso
u
r
ce
s
in
leu
o
f
C
h
r
i
s
t
(
Dee
m
ed
to
b
e
Un
i
v
er
s
it
y
)
,
f
o
r
b
o
th
r
esear
ch
an
d
r
ev
ie
w
.
T
h
e
d
o
cu
m
en
ts
p
r
esen
ted
an
d
p
er
s
is
te
n
t
i
n
t
h
e
p
ap
er
w
er
e
w
o
r
k
ed
o
n
a
Ma
cb
o
o
k
A
ir
.
RE
F
E
R
E
NC
E
S
[1
]
Hu
a
h
o
n
g
T
u
,
A
d
a
m
Do
u
p
é
,
Zi
m
in
g
Zh
a
o
,
G
a
il
-
Jo
o
n
A
h
n
,
“
S
o
K:
Ev
e
ry
o
n
e
Ha
tes
Ro
b
o
c
a
ll
s:
A
S
u
rv
e
y
of
T
e
c
h
n
iq
u
e
s
a
g
a
in
st
Tele
p
h
o
n
e
S
p
a
m
,
”
IEE
E
S
y
mp
o
siu
m
on
S
e
c
u
rity
and
Priva
c
y
,
A
rizo
n
a
S
tate
Un
iv
e
rsit
y
,
p
p
.
3
2
0
-
3
3
8
,
A
u
g
u
st
2
0
1
6
.
[2
]
Ro
b
e
rt
E.
S
c
h
a
p
ire,
“
T
h
e
Bo
o
stin
g
A
p
p
ro
a
c
h
to
M
a
c
h
in
e
L
e
a
r
n
in
g
An
Ov
e
r
v
ie
w
,
”
No
n
li
n
e
a
r
Esti
ma
ti
o
n
and
Cla
ss
if
ica
ti
o
n
,
S
p
rin
g
e
r
,
p
p
.
1
-
23
2
0
0
3
.
[3
]
S
a
m
e
e
r
Qa
y
y
u
m
,
S
h
a
h
e
e
r
M
a
n
so
o
r,
A
d
e
e
l
Kh
a
li
d
,
Kh
u
sh
b
a
k
h
t,
Zah
id
Ha
li
m
a
n
d
A
.
Ra
u
f
Ba
ig
,
“
F
ra
u
d
u
len
t
Ca
ll
De
tec
ti
o
n
F
o
r
M
o
b
i
le
Ne
tw
o
rk
s,”
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
on
I
n
f
o
rm
a
ti
o
n
a
n
d
Eme
rg
i
n
g
T
e
c
h
n
o
lo
g
ies
,
Isla
m
a
b
a
d
,
P
a
k
istan
,
p
p
.
1
-
5
2
0
1
0
.
[4
]
Co
n
sta
n
ti
n
o
s
S.
Hilas
1
,
P
a
ris
A.
M
a
sto
ro
c
o
sta
s,
Io
a
n
n
is
T.
Re
k
a
n
o
s,
“
Clu
ste
rin
g
of
T
e
lec
o
m
m
u
n
ica
ti
o
n
s
Us
er
P
r
o
f
il
e
s
f
o
r
F
ra
u
d
De
tec
ti
o
n
a
n
d
S
e
c
u
rit
y
En
h
a
n
c
e
m
e
n
t
in
L
a
rg
e
Co
rp
o
ra
te
Ne
tw
o
rk
s:
A
c
a
s
e
S
tu
d
y
,
”
Ap
p
li
e
d
M
a
th
e
ma
ti
c
s
&
In
fo
rm
a
ti
o
n
S
c
ien
c
e
s
an
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
,
v
o
l
.
9,
no.
4,
p
p
.
1
7
0
9
-
1
7
1
8
,
2
0
1
5
.
[5
]
Cli
f
to
n
P
h
u
a
,
Vin
c
e
n
t
L
e
e
,
Ka
te
S
m
it
h
Ro
ss
G
a
y
ler,
“A
Co
m
p
re
h
e
n
siv
e
S
u
rv
e
y
of
Da
ta
M
in
in
g
b
a
se
d
F
ra
u
d
De
tec
ti
o
n
Re
se
a
rc
h
,
”
S
c
h
o
o
l
of
Bu
sin
e
ss
S
y
ste
ms
,
Fa
c
u
l
ty
of
In
fo
rm
a
t
io
n
T
e
c
h
n
o
l
o
g
y
,
M
e
l
b
o
u
rn
e
,
A
u
stra
li
a
,
pp.
1
-
14,
M
a
rc
h
2
0
0
7
.
[6
]
W
.
X
u
,
Y.
P
a
n
g
,
J.
M
a
,
S.
W
a
n
g
,
G.
Ha
o
,
S.
Zen
g
,
Y.
Qa
in
,
“
F
ra
u
d
d
e
tec
ti
o
n
in
tele
c
o
m
m
u
n
ica
ti
o
n
:
a
ro
u
g
h
f
u
z
z
y
se
t
b
a
se
d
a
p
p
r
o
a
c
h
,
”
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
of
M
a
c
h
in
e
L
e
a
rn
i
n
g
a
n
d
Cy
b
e
rn
e
ti
c
s
,
v
o
l.
3,
p
p
.
1
7
7
7
-
1
7
8
7
,
Ju
ly
2008.
[7
]
F
a
rv
a
re
sh
,
H.
a
n
d
S
e
p
e
h
ri,
“A
D
a
ta
M
in
in
g
F
ra
m
e
w
o
rk
f
o
r
De
tec
ti
n
g
S
u
b
sc
ri
p
ti
o
n
F
ra
u
d
in
T
e
lec
o
m
m
u
n
ica
ti
o
n
,
”
En
g
i
n
e
e
rin
g
Ap
p
li
c
a
ti
o
n
s
of
Arti
fi
c
ia
l
In
tell
ig
e
n
c
e
,
v
o
l.
24,
n
o
.
1,
p
p
.
1
8
2
–
1
9
4
,
2
0
1
1
.
[8
]
M
o
h
a
m
m
a
d
Iq
u
e
b
a
l
Ak
h
ter,
M
o
h
a
m
m
a
d
G
u
la
m
A
h
a
m
a
d
,
“
De
tec
ti
n
g
Tele
c
o
m
m
u
n
ica
ti
o
n
F
ra
u
d
u
sin
g
Ne
u
ra
l
Ne
tw
o
rk
s
th
ro
u
g
h
Da
ta
M
i
n
in
g
,
”
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
of
S
c
ien
ti
fi
c
&
En
g
in
e
e
rin
g
Res
e
a
rc
h
,
v
o
l
.
3,
n
o
.
3,
p
p
.
1
-
5,
M
a
rc
h
2
0
1
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
9
,
No
.
4
,
A
u
g
u
s
t
201
9
:
3
1
0
8
-
3114
3114
[9
]
F
a
y
e
m
i
w
o
M
ich
a
e
l
A
d
e
b
isi
a
n
d
Ola
so
ji
Ba
b
a
tu
n
d
e
,
“
F
ra
u
d
De
tec
ti
o
n
in
M
o
b
il
e
T
e
lec
o
m
m
u
n
ica
ti
o
n
s
,”
In
ter
n
a
t
io
n
a
l
J
o
u
r
n
a
l
of
In
n
o
v
a
t
ive
Res
e
a
rc
h
in
S
c
ien
c
e
En
g
in
e
e
rin
g
a
n
d
tec
h
n
o
lo
g
y
,
v
o
l.
3,
n
o
.
4,
pp
.
1
1
6
1
2
-
1
1
6
2
0
,
A
p
ril
2
0
1
4
.
[1
0
]
Ke
n
n
e
th
C
Co
x
,
S
te
p
h
e
n
G,
G
r
a
h
a
m
J
W
il
ls,
“
Brie
f
A
p
p
li
c
a
ti
o
n
De
sc
rip
ti
o
n
Visu
a
l
Da
ta
M
in
i
n
g
:
Re
c
o
g
n
izin
g
T
e
lep
h
o
n
e
Ca
ll
in
g
F
ra
u
d
,
”
Da
ta
M
in
i
n
g
a
n
d
K
n
o
w
led
g
e
Disc
o
v
e
r
y
,
v
o
l
1,
n
o
.
2,
p
p
.
2
2
5
–
2
3
1
,
J
u
n
e
.
[1
1
]
S.
W
u
,
N.
Ka
n
g
,
L.
Ya
n
g
,
“
F
r
a
u
d
u
len
t
Be
h
a
v
io
r
F
o
re
c
a
st
in
T
e
lec
o
m
In
d
u
stry
Ba
s
e
d
on
Da
ta
M
in
in
g
T
e
c
h
n
o
lo
g
y
,
”
Co
mm
u
n
ica
ti
o
n
s
of
th
e
IIM
A
,
v
o
l
.
7,
n
o
.
4,
p
p
.
1
-
6,
2
0
0
7
.
[1
2
]
V
ru
s
h
a
li
Y
K
u
lk
a
rn
i,
P
ra
d
e
e
p
K
S
in
h
a
,
“
Ef
fe
c
ti
v
e
L
e
a
rn
in
g
a
n
d
Clas
sif
ic
a
ti
o
n
u
si
n
g
Ra
n
d
o
m
F
o
r
e
st
A
lg
o
rit
h
m
,
”
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
of
E
n
g
in
e
e
rin
g
a
n
d
I
n
n
o
v
a
ti
v
e
T
e
c
h
n
o
l
o
g
y
,
v
o
l.
3,
no.
1
1
,
pp.
2
6
7
-
2
7
3
,
M
a
y
2
0
1
4
.
[1
3
]
T
ian
q
i
Ch
e
n
,
Ca
rlo
s
G
u
e
strin
,
“
X
G
Bo
o
st:
A
S
c
a
lab
le
T
re
e
Bo
o
stin
g
S
y
ste
m
,
”
A
CM
S
IG
KD
D
,
p
p
.
7
8
5
-
7
9
4
,
A
u
g
u
st
13
-
17,
2
0
1
6
.
[1
4
]
X
u
e
y
i
W
a
n
g
“A
Ne
w
M
o
d
e
l
f
o
r
M
e
a
su
rin
g
th
e
A
c
c
u
ra
c
ies
of
m
a
j
o
rit
y
v
o
ti
n
g
e
n
se
m
b
les
,
”
IEE
E
W
o
rld
C
o
n
g
re
ss
on
C
o
mp
u
ta
t
io
n
a
l
I
n
telli
g
e
n
c
e
,
2
0
1
2
.
[1
5
]
S
a
rw
e
sh
S
it
e
,
Dr.
S
a
d
h
n
a
K.
M
ish
ra
,
“A
Re
v
ie
w
of
En
se
m
b
le
T
e
c
h
n
iq
u
e
f
o
r
Im
p
ro
v
in
g
M
a
jo
r
it
y
V
o
ti
n
g
f
o
r
Clas
sif
ier,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
of
Ad
v
a
n
c
e
d
Res
e
a
rc
h
in
Co
m
p
u
ter
S
c
ien
c
e
and
S
o
ft
wa
re
En
g
in
e
e
rin
g
,
v
o
l.
3,
n
o
.
1
,
pp.
1
7
7
-
1
8
0
,
Ja
n
u
a
ry
2013.
Evaluation Warning : The document was created with Spire.PDF for Python.