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.
7
,
No
.
6
,
Dec
em
b
er
201
7
,
p
p
.
3
6
1
3
~
3
6
2
1
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v7
i
6
.
pp
3
6
1
3
-
3621
3613
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
jo
u
r
n
a
l.c
o
m/o
n
lin
e/in
d
ex
.
p
h
p
/I
JE
C
E
Finding
Ba
d Cod
e S
m
ells w
ith
Neu
ra
l Net
w
o
rk
Mo
dels
Do
ng
K
wa
n K
i
m
De
p
a
rtme
n
t
o
f
Co
m
p
u
ter E
n
g
in
e
e
rin
g
,
M
o
k
p
o
Na
ti
o
n
a
l
M
a
rit
im
e
U
n
iv
e
rsity
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
y
4
,
2
0
1
7
R
ev
i
s
ed
J
u
n
2
8
,
2
0
1
7
A
cc
ep
ted
J
u
l 1
4
,
2
0
1
7
Co
d
e
sm
e
ll
re
fe
rs
to
a
n
y
s
y
m
p
to
m
in
tro
d
u
c
e
d
in
d
e
sig
n
o
r
im
p
lem
e
n
tatio
n
p
h
a
se
s
in
th
e
so
u
rc
e
c
o
d
e
o
f
a
p
ro
g
ra
m
.
S
u
c
h
a
c
o
d
e
sm
e
ll
c
a
n
p
o
ten
ti
a
ll
y
c
a
u
se
d
e
e
p
e
r
a
n
d
se
ri
o
u
s
p
ro
b
lem
s
d
u
rin
g
s
o
f
twa
re
m
a
in
ten
a
n
c
e
.
T
h
e
e
x
isti
n
g
a
p
p
ro
a
c
h
e
s
to
d
e
tec
t
b
a
d
s
m
e
ll
s
u
se
d
e
te
c
ti
o
n
ru
les
o
r
sta
n
d
a
rd
s
u
sin
g
a
c
o
m
b
in
a
ti
o
n
o
f
d
if
fe
re
n
t
o
b
jec
t
-
o
rie
n
ted
m
e
tri
c
s.
A
lt
h
o
u
g
h
a
v
a
riety
o
f
so
f
t
w
a
re
d
e
tec
ti
o
n
to
o
ls
h
a
v
e
b
e
e
n
d
e
v
e
lo
p
e
d
,
t
h
e
y
stil
l
h
a
v
e
li
m
it
a
ti
o
n
s
an
d
c
o
n
stra
i
n
ts
in
th
e
ir
c
a
p
a
b
il
it
ies
.
In
th
is
p
a
p
e
r,
a
c
o
d
e
sm
e
ll
d
e
tec
ti
o
n
s
y
ste
m
is
p
re
se
n
ted
w
it
h
th
e
n
e
u
ra
l
n
e
tw
o
rk
m
o
d
e
l
th
a
t
d
e
li
v
e
rs
th
e
re
latio
n
sh
i
p
b
e
tw
e
e
n
b
a
d
sm
e
ll
s
a
n
d
o
b
jec
t
-
o
r
ien
ted
m
e
tri
c
s
b
y
tak
in
g
a
c
o
rp
u
s
o
f
Ja
v
a
p
ro
jec
ts
a
s e
x
p
e
rime
n
tal
d
a
tas
e
t.
T
h
e
m
o
st we
ll
-
k
n
o
w
n
o
b
jec
t
-
o
rien
ted
m
e
tri
c
s
a
re
c
o
n
sid
e
re
d
t
o
id
e
n
ti
fy
th
e
p
re
se
n
c
e
o
f
b
a
d
s
m
e
ll
s.
T
h
e
c
o
d
e
sm
e
ll
d
e
tec
ti
o
n
s
y
ste
m
u
se
s
th
e
twe
n
ty
Ja
v
a
p
ro
jec
ts
w
h
ich
a
re
sh
a
re
d
b
y
m
a
n
y
u
se
rs
in
th
e
G
it
Hu
b
re
p
o
sito
ries
.
T
h
e
d
a
tas
e
t
o
f
th
e
se
J
a
v
a
p
r
o
jec
ts
is
p
a
rti
ti
o
n
e
d
in
t
o
m
u
tu
a
ll
y
e
x
c
lu
siv
e
train
in
g
a
n
d
tes
t
se
ts.
T
h
e
train
in
g
d
a
tas
e
t
is
u
se
d
to
lea
rn
th
e
n
e
tw
o
rk
m
o
d
e
l
w
h
ich
w
il
l
p
re
d
ict
sm
e
l
ly
c
las
se
s
in
th
is
st
u
d
y
.
T
h
e
o
p
ti
m
ize
d
n
e
two
rk
m
o
d
e
l
w
il
l
b
e
c
h
o
se
n
t
o
b
e
e
v
a
lu
a
ted
o
n
th
e
tes
t
d
a
tas
e
t.
T
h
e
e
x
p
e
ri
m
e
n
tal
re
su
lt
s
sh
o
w
w
h
e
n
th
e
m
o
d
e
li
s
h
ig
h
ly
train
e
d
w
it
h
m
o
re
d
a
tas
e
t,
th
e
p
re
d
ictio
n
o
u
tco
m
e
s
a
r
e
i
m
p
ro
v
e
d
m
o
re
a
n
d
m
o
re
.
In
a
d
d
it
i
o
n
,
th
e
a
c
c
u
ra
c
y
o
f
th
e
m
o
d
e
l
in
c
re
a
se
s
w
h
e
n
it
p
e
r
f
o
r
m
s
w
it
h
h
ig
h
e
r
e
p
o
c
h
s an
d
m
a
n
y
h
id
d
e
n
l
a
y
e
rs.
K
ey
w
o
r
d
:
C
o
d
e
s
m
ell
s
Neu
r
al
n
et
w
o
r
k
s
Ob
j
ec
t
-
o
r
ien
ted
m
etr
ics
So
f
t
w
ar
e
m
a
in
te
n
a
n
ce
Co
p
y
rig
h
t
©
2
0
1
7
I
n
stit
u
te o
f
Ad
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
:
Do
n
g
K
w
a
n
Ki
m
,
Dep
ar
t
m
en
t o
f
C
o
m
p
u
ter
E
n
g
i
n
ee
r
in
g
,
Mo
k
p
o
Natio
n
al
Ma
r
iti
m
e
Un
iv
er
s
it
y
,
9
1
,
Hae
y
a
n
g
d
ae
h
a
k
-
r
o
,
Mo
k
p
o
-
s
i,
J
eo
llan
a
m
-
d
o
,
Ko
r
ea
.
E
m
ail: d
o
n
g
k
w
an
@
g
m
a
il.c
o
m
1.
I
NT
RO
D
UCT
I
O
N
So
f
t
w
ar
e
m
a
in
te
n
a
n
ce
is
t
h
e
m
o
d
i
f
icat
io
n
o
f
a
s
o
f
t
w
ar
e
p
r
o
d
u
ct
af
ter
d
eliv
er
y
to
co
r
r
ec
t
f
au
l
ts
,
to
i
m
p
r
o
v
ep
er
f
o
r
m
a
n
ce
o
r
o
th
er
attr
ib
u
tes,
o
r
to
ad
ap
t
th
e
p
r
o
d
u
ct
to
a
m
o
d
if
ied
e
n
v
ir
o
n
m
en
t
[
1
]
.
T
h
e
ac
tiv
it
ies
o
f
s
o
f
t
w
ar
e
m
ai
n
te
n
an
ce
ca
n
b
e
ca
teg
o
r
ized
in
to
f
o
u
r
t
y
p
es
:
ad
ap
tiv
e,
p
er
f
ec
ti
v
e,
co
r
r
ec
ti
v
e,
a
n
d
p
r
ev
en
t
iv
e
m
ai
n
ten
a
n
ce
s
.
Am
o
n
g
th
e
m
,
th
e
ad
ap
ti
v
e
m
ai
n
te
n
an
ce
in
v
o
lv
es
r
eo
r
g
a
n
izi
n
g
t
h
e
s
tr
u
ct
u
r
e
o
f
t
h
e
s
o
f
t
w
ar
e
s
y
s
te
m
s
to
co
p
e
w
ith
c
h
a
n
g
es
in
t
h
e
s
o
f
t
w
ar
e
e
n
v
ir
o
n
m
en
t
s
u
ch
a
s
u
s
er
r
eq
u
ir
e
m
e
n
ts
a
n
d
co
m
p
u
tin
g
tech
n
o
lo
g
ies
.
I
n
o
th
er
w
o
r
d
s
,
s
o
f
t
w
ar
e
c
h
an
g
es
ar
e
o
n
e
o
f
th
e
in
e
v
itab
le
c
h
ar
ac
ter
is
t
ics
in
th
e
s
o
f
t
w
ar
e
s
y
s
te
m
.
I
t
i
s
h
ar
d
to
esti
m
at
e
th
e
p
o
ten
tial
co
n
s
eq
u
e
n
ce
s
o
f
a
ch
an
g
e
a
f
ter
m
o
d
if
y
i
n
g
s
o
m
e
p
ar
ts
o
f
t
h
e
s
o
f
t
w
ar
e
s
y
s
te
m
b
ec
a
u
s
e
ev
e
n
a
m
in
o
r
c
h
a
n
g
e
co
u
ld
i
n
c
r
ea
s
e
th
e
co
m
p
le
x
it
y
o
f
t
h
e
s
y
s
te
m
a
n
d
ca
u
s
e
u
n
e
x
p
ec
ted
p
r
o
b
lem
s
.
So
m
eti
m
es
s
u
c
h
ch
a
n
g
es
ca
n
in
tr
o
d
u
ce
s
o
m
e
co
d
e
s
m
ell
s
ac
cid
en
t
all
y
i
n
th
e
s
y
s
te
m
.
C
o
d
e
s
m
ell
r
ef
er
s
to
an
y
s
y
m
p
to
m
i
n
tr
o
d
u
ce
d
in
d
esig
n
o
r
i
m
p
le
m
e
n
tatio
n
p
h
ases
i
n
th
e
s
o
u
r
ce
co
d
e
o
f
a
p
r
o
g
r
am
.
Su
c
h
a
co
d
e
s
m
el
l
c
an
p
o
ten
tial
l
y
ca
u
s
e
d
ee
p
er
an
d
s
er
io
u
s
p
r
o
b
le
m
s
d
u
r
in
g
s
o
f
t
w
ar
e
m
ai
n
te
n
an
ce
.
A
cc
o
r
d
in
g
to
t
h
e
s
co
p
e
o
f
t
h
e
co
d
e
s
m
ell,
t
h
e
co
d
e
s
m
e
lls
ca
n
b
e
class
if
ied
in
to
t
h
r
ee
c
lass
es
:
ap
p
licatio
n
-
lev
el,
clas
s
-
le
v
el,
an
d
m
eth
o
d
-
lev
el
s
m
ell
s
.
Ma
n
y
s
o
f
t
w
ar
e
to
o
ls
h
av
e
b
ee
n
d
ev
elo
p
ed
to
d
etec
t
an
d
eli
m
i
n
ate
co
d
e
s
m
ell
s
u
s
in
g
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ics
an
d
co
d
e
r
ef
a
cto
r
in
g
m
e
th
o
d
s
[
2
]
.
T
h
e
co
d
e
s
m
e
ll
ca
n
b
e
a
n
ef
f
ec
tiv
e
in
d
icatio
n
o
f
w
h
et
h
er
o
r
n
o
ts
o
m
e
p
ar
ts
o
f
t
h
e
c
o
d
e
s
h
o
u
ld
b
e
r
ef
ac
to
r
ed
.
A
l
t
h
o
u
g
h
a
v
ar
iet
y
o
f
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2088
-
8708
I
J
E
C
E
Vo
l.
7
,
No
.
6
,
Dec
em
b
er
2
0
1
7
:
3
6
1
3
–
3
6
2
1
3614
s
o
f
t
w
ar
e
d
etec
tio
n
to
o
ls
h
av
e
b
ee
n
d
ev
elo
p
ed
,
th
e
y
s
ti
ll
h
a
v
e
li
m
itat
io
n
s
a
n
d
co
n
s
tr
ai
n
ts
i
n
th
eir
ca
p
ab
ilit
ie
s
.
I
n
t
h
is
p
ap
er
,
a
co
d
e
s
m
ell
d
etec
tio
n
s
y
s
te
m
is
p
r
esen
ted
w
it
h
th
e
n
e
u
r
al
n
et
w
o
r
k
m
o
d
el
th
at
d
eliv
er
s
t
h
e
r
elatio
n
s
h
ip
b
et
w
ee
n
b
ad
s
m
ell
s
an
d
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ics
b
y
ta
k
in
g
a
co
r
p
u
s
o
f
J
av
a
p
r
o
j
ec
ts
as
ex
p
er
i
m
e
n
tal
d
ata
s
et.
T
h
e
d
etec
tio
n
s
y
s
te
m
p
r
o
p
o
s
ed
in
t
h
e
p
ap
er
u
s
es
m
aj
o
r
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ic
s
t
o
ex
tr
ac
t
v
ar
io
u
s
f
ea
tu
r
e
s
f
o
r
b
ad
co
d
e
s
m
ell
s
b
y
ca
lc
u
lati
n
g
m
etr
ic
v
al
u
es.
Ma
n
y
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ics
h
a
v
e
b
ee
n
p
r
o
p
o
s
ed
in
th
e
p
ast
d
ec
ad
es.
T
h
e
w
e
ll
-
k
n
o
w
n
m
etr
ics
s
u
ch
a
s
L
in
e
o
f
C
o
d
e
(
L
OC
)
,
Dep
t
h
o
f
I
n
h
er
ita
n
ce
T
r
ee
(
DI
T
)
,
an
d
C
o
u
p
li
n
g
B
et
w
ee
n
Ob
j
ec
ts
(
C
B
O)
ar
e
ap
p
lied
to
ass
ess
t
h
e
q
u
ali
t
y
o
f
o
b
j
ec
t
-
o
r
ien
ted
p
r
o
g
r
am
s
at
d
if
f
er
e
n
t
lev
el
s
.
W
h
ile
s
o
m
e
m
etr
ics
ar
e
ap
p
lied
to
ass
ess
th
e
w
h
o
le
o
b
j
ec
t
-
o
r
ien
ted
s
y
s
te
m
s
,
o
th
er
s
ar
e
m
a
in
l
y
u
s
ed
to
ev
alu
ate
s
i
n
g
le
clas
s
es
o
r
m
e
th
o
d
s
.
I
n
th
e
p
r
e
v
io
u
s
s
tu
d
ies,
t
h
e
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ic
s
h
a
v
e
b
ee
n
u
s
ed
to
d
etec
t
s
m
ell
y
cla
s
s
es
b
y
s
u
p
p
o
r
tin
g
a
p
r
ed
ictio
n
m
o
d
el
f
o
r
t
h
e
co
d
e
s
m
el
ls
.
T
h
e
au
to
m
ated
d
etec
ti
o
n
to
o
ls
ca
lcu
la
te
m
e
tr
ic
v
al
u
es
o
v
er
an
i
n
s
p
ec
ted
s
o
f
t
w
ar
e
s
y
s
te
m
to
id
en
ti
f
y
s
p
ec
if
ic
c
h
ar
ac
ter
is
tic
s
o
f
b
a
d
s
m
ell
s
.
Fo
r
ex
a
m
p
le,
R
es
p
o
n
s
e
f
o
r
ac
las
s
(
R
F
C
)
an
d
C
o
u
p
li
n
g
B
et
w
ee
n
Ob
j
ec
ts
(
C
B
O)
m
etr
ic
s
ar
e
u
s
ed
to
ev
al
u
ate
t
h
e
d
eg
r
ee
o
f
co
u
p
lin
g
b
et
w
ee
n
clas
s
e
s
.
C
o
h
e
s
io
n
ca
n
b
e
m
ea
s
u
r
ed
b
y
L
ac
k
o
f
C
o
h
es
io
n
in
Me
th
o
d
s
(
L
C
O
M)
an
d
L
o
o
s
e
C
las
s
C
o
h
esio
n
(
L
C
C
)
.
T
h
e
r
est
o
f
th
is
p
ap
er
is
o
r
g
an
ized
as
f
o
llo
w
s
.
Sectio
n
2
d
escr
ib
es
th
e
p
r
o
p
o
s
ed
c
o
d
e
s
m
el
l
d
etec
to
r
i
n
d
etail.
Sectio
n
3
p
r
esen
ts
th
e
ex
p
er
i
m
en
ta
l
r
esu
lt
s
b
y
th
e
d
etec
tio
n
s
y
s
te
m
.
Sect
io
n
4
in
tr
o
d
u
ce
s
th
e
r
elate
d
w
o
r
k
a
n
d
Sectio
n
5
r
e
m
ar
k
s
t
h
e
co
n
clu
s
io
n
s
a
n
d
f
u
t
u
r
e
w
o
r
k
d
ir
ec
tio
n
s
.
2.
CO
DE
SM
E
L
L
D
E
T
E
C
T
O
R
Fig
u
r
e
1
.
T
h
e
Ov
er
all
W
o
r
k
f
l
o
w
o
f
t
h
e
P
r
o
p
o
s
ed
C
o
d
e
S
m
e
ll De
tectio
n
S
y
s
te
m
Fig
u
r
e
1
s
h
o
w
s
th
e
o
v
er
all
w
o
r
k
f
lo
w
o
f
t
h
e
p
r
o
p
o
s
ed
co
d
e
s
m
ell
d
etec
tio
n
s
y
s
te
m
to
f
i
n
d
co
d
e
s
m
el
ls
i
n
J
av
a
p
r
o
g
r
am
s
.
T
h
e
p
r
o
p
o
s
ed
d
etec
tio
n
s
y
s
te
m
u
s
es
th
e
t
w
e
n
t
y
J
av
a
p
r
o
j
ec
ts
w
h
ich
ar
e
d
o
w
n
lo
ad
ed
f
r
o
m
t
h
e
Git
Hu
b
r
ep
o
s
ito
r
ies.
T
h
e
d
etec
tio
n
s
y
s
te
m
co
n
s
is
t
s
o
f
th
r
ee
m
ai
n
co
m
p
o
n
e
n
t
s
-
O
O
Me
tr
ics
An
al
y
ze
r
,
C
o
d
e
S
m
ell
Dete
cto
r
,
an
d
Neu
r
al
Net
w
o
r
k
.
A
co
m
m
er
cial
co
d
e
an
al
y
s
i
s
to
o
l,
S
ciTo
o
ls
Un
d
ers
ta
n
d
is
u
s
ed
to
an
al
y
ze
t
h
e
J
av
a
p
r
o
j
ec
ts
d
o
w
n
lo
ad
ed
f
r
o
m
GitH
u
b
an
d
to
d
eliv
er
co
m
p
r
eh
e
n
s
i
v
e
m
etr
ics
v
al
u
es
f
o
r
J
av
a.
C
o
d
e
S
m
el
l
Dete
cto
r
s
u
p
p
o
r
ts
d
if
f
er
en
t
cr
iter
ia
a
n
d
th
r
e
s
h
o
ld
s
o
f
t
h
e
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ics
i
n
d
etec
ti
n
g
co
d
e
s
m
el
ls
.
C
o
d
e
S
m
el
l
Dete
cto
r
d
eter
m
in
e
s
w
h
e
th
er
o
r
n
o
t
th
e
class
e
s
in
t
h
e
J
av
a
p
r
o
j
ec
ts
ar
ea
s
m
ell
y
clas
s
b
y
c
h
ec
k
in
g
t
h
e
m
etr
ics
v
alu
e
s
o
f
ea
ch
cla
s
s
.
T
h
e
co
d
e
s
m
e
ll
r
esu
lt
s
ar
e
s
ep
ar
ated
in
to
th
e
tr
ain
in
g
a
n
d
tes
t
d
atasets
.
T
h
e
n
eu
r
al
n
et
w
o
r
k
m
o
d
el
i
n
t
h
e
d
etec
tio
n
s
y
s
te
m
is
tr
ain
ed
a
n
d
o
p
ti
m
ized
b
y
t
h
e
tr
ai
n
i
n
g
d
atase
t
an
d
th
e
n
is
e
v
alu
a
ted
b
y
t
h
e
te
s
t d
ataset
o
f
t
h
e
co
d
e
s
m
ell
r
es
u
lts
.
2
.
1
.
So
f
t
wa
re
M
e
t
rics
a
nd
Co
de
S
m
ells
S
ciTo
o
ls
Un
d
ers
ta
n
d
,
a
s
tatic
an
al
y
s
is
to
o
l,
ca
n
ex
tr
ac
t
a
w
id
e
r
an
g
e
o
f
m
e
tr
ics
an
d
g
en
er
ate
a
cu
s
to
m
izab
le
r
ep
o
r
t
f
o
r
th
e
J
a
v
a
p
r
o
j
ec
ts
.
T
h
e
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ic
s
f
o
r
th
i
s
r
esear
ch
ar
e
s
u
m
m
ar
ized
i
n
th
e
f
o
llo
w
in
g
:
L
i
n
e
o
f
C
o
d
e
(
L
O
C
)
:
L
i
n
e
o
f
co
d
e
m
etr
ic
is
u
s
ed
to
co
u
n
t t
h
e
lin
es
o
f
t
h
e
s
o
u
r
ce
co
d
e
w
i
th
o
u
t c
o
n
s
id
er
in
g
th
e
co
m
m
e
n
t
li
n
es
a
n
d
b
la
n
k
lin
es.
L
OC
is
t
y
p
ical
s
o
f
t
w
ar
e
m
etr
ic
th
a
t
is
u
s
ed
to
m
ea
s
u
r
e
t
h
e
p
r
o
g
r
a
m
s
ize.
Ma
n
y
r
esear
ch
f
i
n
d
in
g
s
d
em
o
n
s
tr
ate
t
h
at
la
r
g
er
L
O
C
v
alu
e
s
ta
k
e
m
o
r
e
ti
m
e
to
d
ev
e
lo
p
an
d
m
a
in
ta
in
a
p
r
o
g
r
am
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
F
in
d
in
g
B
a
d
C
o
d
e
S
mells w
ith
N
eu
r
a
l Netw
o
r
k
Mo
d
els
(
Do
n
g
K
w
a
n
K
im
)
3615
Dep
th
o
f
I
n
h
er
ita
n
ce
T
r
ee
(
DI
T
)
:
T
h
e
d
ep
th
o
f
in
h
er
itan
ce
tr
ee
ca
n
b
e
d
ef
i
n
ed
as t
h
e
m
a
x
i
m
u
m
le
n
g
t
h
o
f
a
class
i
n
th
e
in
h
er
ita
n
ce
tr
ee
.
DI
T
c
o
u
n
ts
t
h
e
d
ep
th
f
r
o
m
a
s
p
ec
if
ic
cla
s
s
to
ato
p
-
l
e
v
el
r
o
o
t
class
in
t
h
e
h
ier
ar
ch
ical
s
tr
u
ct
u
r
e
tr
ee
.
T
h
e
J
av
a
p
r
o
g
r
a
m
m
in
g
allo
w
s
f
o
r
o
n
l
y
s
in
g
le
cla
s
s
in
h
er
ita
n
ce
.
I
n
g
en
er
al,
as
a
p
ar
ticu
lar
class
h
as
m
o
r
e
s
u
p
er
-
class
e
s
,
t
h
e
class
in
h
er
its
m
o
r
e
f
ield
s
a
n
d
m
e
th
o
d
s
.
Si
n
ce
th
e
p
o
ten
tia
l
r
eu
s
e
o
f
i
n
h
er
ited
m
e
m
b
er
s
c
an
m
a
k
e
it
m
o
r
e
co
m
p
le
x
a
s
o
f
t
w
ar
e
s
y
s
te
m
,
th
e
DI
T
m
e
tr
ic
ca
n
b
e
o
n
e
o
f
th
e
cr
iter
iato
p
r
ed
ict
th
e
d
eg
r
ee
o
f
th
e
co
m
p
le
x
it
y
o
f
an
o
b
j
ec
t
-
o
r
ien
ted
s
y
s
te
m
.
C
o
u
p
li
n
g
B
et
w
ee
n
Ob
j
ec
t
class
es
(
C
B
O)
:
C
B
O
m
ea
s
u
r
es
t
h
e
n
u
m
b
er
o
f
o
th
er
clas
s
es
w
h
ich
a
p
ar
tic
u
lar
class
i
s
co
u
p
led
.
W
e
ca
n
s
a
y
t
h
at
cla
s
s
es
C
1
a
n
d
C
2
ar
e
co
u
p
led
if
clas
s
C
1
u
s
e
s
a
t
y
p
e,
d
ata,
o
r
m
eth
o
d
f
r
o
m
cla
s
s
C
2
.
B
y
th
e
UM
L
ter
m
i
n
o
lo
g
ies,
clas
s
es
C
1
an
d
C
2
h
a
v
e
cla
s
s
r
elat
io
n
s
h
ip
s
s
u
c
h
a
s
ass
o
ciatio
n
,
a
g
g
r
eg
atio
n
,
an
d
co
m
p
o
s
i
tio
n
.
L
ar
g
er
C
B
O
v
al
u
es
ca
n
d
ec
r
ea
s
e
t
h
e
m
o
d
u
lar
it
y
a
n
d
p
r
ev
en
t
r
ea
s
o
n
ab
le
r
eu
s
e
o
f
a
s
o
f
t
w
ar
e
s
y
s
te
m
.
T
h
er
ef
o
r
e,
w
e
n
ee
d
to
k
ee
p
C
B
O
v
alu
e
s
to
a
m
i
n
i
m
u
m
f
o
r
th
e
b
etter
m
a
in
tai
n
ab
ilit
y
.
R
esp
o
n
s
e
f
o
r
a
C
lass
(
R
F
C
)
:
R
F
C
co
n
s
id
er
s
t
h
e
n
u
m
b
er
o
f
ac
ce
s
s
ib
le
m
et
h
o
d
s
an
d
co
n
s
t
r
u
cto
r
s
in
v
o
k
ed
b
y
a
class
.
W
h
e
n
an
o
b
j
ec
t
o
f
th
at
clas
s
r
ec
eiv
es
a
m
es
s
a
g
e,
th
e
R
F
C
m
etr
ic
co
u
n
ts
t
h
e
n
u
m
b
er
o
f
d
is
ti
n
ct
m
et
h
o
d
s
in
cl
u
d
in
g
i
n
h
er
ited
m
et
h
o
d
s
.
T
h
e
J
av
a
p
r
o
g
r
a
m
m
in
g
la
n
g
u
a
g
e
d
eter
m
i
n
e
s
w
h
et
h
er
o
r
n
o
t
o
th
er
class
es
ca
n
ac
ce
s
s
a
p
ar
ticu
lar
d
ata
o
r
in
v
o
k
e
a
p
ar
ticu
lar
m
eth
o
d
in
a
clas
s
.
T
h
e
R
FC
m
e
tr
ic
is
r
elate
d
to
th
e
J
av
a
ac
ce
s
s
m
o
d
if
ier
s
s
u
c
h
as
p
u
b
lic,
p
ac
k
ag
e
-
p
r
iv
ate,
p
r
o
tecte
d
,
an
d
p
r
iv
ate.
I
f
th
e
R
F
C
v
al
u
e
f
o
r
a
class
is
u
n
ac
ce
p
tab
l
y
lar
g
e,
it
m
ea
n
s
t
h
at
clas
s
h
as p
o
ten
tial
in
ter
ac
tio
n
s
w
it
h
th
e
r
es
t o
f
a
p
r
o
g
r
am
.
W
eig
h
ted
Me
t
h
o
d
s
p
er
C
la
s
s
(
W
MC):
T
h
e
W
MC
m
e
tr
ic
is
t
h
e
s
u
m
o
f
t
h
e
co
m
p
le
x
i
ties
o
f
all
cla
s
s
m
et
h
o
d
s
.
S
ciTo
o
ls
Un
d
er
s
ta
n
d
ca
lcu
late
s
t
h
e
s
u
m
o
f
c
y
c
l
o
m
a
tic
co
m
p
lex
i
t
y
o
f
all
n
es
t
ed
f
u
n
ct
i
o
n
s
o
r
m
et
h
o
d
s
in
a
cla
s
s
,
n
o
t c
o
n
s
id
er
in
g
in
h
er
ited
m
et
h
o
d
s
.
I
n
g
e
n
er
al,
a
clas
s
w
i
th
a
lar
g
e
W
MC
v
alu
e
i
m
p
l
ie
s
th
at
t
h
at
clas
s
ca
n
b
e
co
m
p
lex
an
d
h
ar
d
er
to
m
ai
n
tai
n
.
Nu
m
b
er
o
f
C
h
ild
r
en
(
N
OC
)
:
T
h
e
NOC
m
etr
ic
m
ea
s
u
r
es
t
h
e
n
u
m
b
er
o
f
i
m
m
ed
iate
s
u
b
class
es
i
.
e.
,
th
e
n
u
m
b
er
o
f
clas
s
es
o
n
e
le
v
el
d
o
w
n
th
e
i
n
h
er
i
tan
ce
tr
ee
f
r
o
m
a
tar
g
et
class
.
T
h
e
NOC
v
al
u
e
in
d
icate
s
t
h
at
th
e
d
ata
an
d
m
et
h
o
d
s
o
f
a
cl
ass
ca
n
b
e
r
eu
s
ed
in
its
s
u
b
c
lass
es.
T
h
er
ef
o
r
e,
if
a
clas
s
h
as
a
h
i
g
h
NO
C
v
alu
e,
t
h
e
clas
s
is
m
o
r
e
r
esp
o
n
s
ib
le
in
a
s
o
f
t
w
ar
e
s
y
s
te
m
.
C
y
clo
m
atic
C
o
m
p
le
x
it
y
(
C
C
)
:
C
y
clo
m
at
ic
co
m
p
le
x
it
y
,
as
k
n
o
w
n
as
Mc
C
ab
e
'
s
C
y
clo
m
a
tic
C
o
m
p
le
x
it
y
,
m
ea
s
u
r
es
t
h
e
n
u
m
b
er
o
f
li
n
e
ar
l
y
in
d
ep
en
d
en
t
p
ath
s
t
h
r
o
u
g
h
a
p
r
o
g
r
a
m
m
o
d
u
le.
S
ciTo
o
ls
Un
d
ers
ta
n
d
co
u
n
t
s
t
h
e
k
e
y
w
o
r
d
s
f
o
r
d
ec
i
s
io
n
p
o
in
ts
(
FOR
,
W
HI
L
E
,
e
tc.
)
an
d
t
h
en
ad
d
s
1
.
Fo
r
a
s
w
itc
h
s
tate
m
e
n
t,
ea
ch
‘
ca
s
e’
i
s
co
u
n
ted
as 1
an
d
th
e
‘
s
w
itch
’
it
s
el
f
ad
d
s
o
n
et
o
th
e
f
i
n
al
C
y
clo
m
atic
C
o
m
p
l
ex
it
y
co
u
n
t.
L
ac
k
o
f
co
h
esio
n
in
m
et
h
o
d
s
(
L
C
OM
)
:
C
o
h
e
s
io
n
r
e
f
er
s
to
th
e
d
eg
r
ee
o
f
th
e
in
tr
a
-
r
elatio
n
s
h
ip
b
et
w
ee
n
t
h
e
ele
m
e
n
ts
i
n
a
s
o
f
t
w
ar
e
m
o
d
u
l
e
s
u
c
h
as
p
ac
k
a
g
es
a
n
d
clas
s
es.
I
t
is
id
ea
l
t
h
at
ea
c
h
ele
m
e
n
t
h
as
a
s
tr
o
n
g
r
elatio
n
s
h
ip
i
n
t
h
e
m
o
d
u
le
b
y
ac
h
ie
v
i
n
g
a
p
ar
ticu
lar
f
u
n
cti
o
n
alit
y
.
T
h
e
L
C
OM
m
etr
ic
i
n
d
icate
s
a
s
et
o
f
m
et
h
o
d
s
in
a
clas
s
is
n
o
t stro
n
g
l
y
co
n
n
e
cted
to
o
th
er
m
eth
o
d
s
.
C
o
d
e
s
m
ell
s
r
ef
er
to
a
n
y
s
y
m
p
to
m
in
th
e
s
o
u
r
ce
co
d
e
o
f
a
p
r
o
g
r
a
m
th
a
t
p
o
s
s
ib
l
y
in
d
icate
s
a
d
ee
p
er
p
r
o
b
lem
.
B
ad
co
d
e
s
m
el
ls
ar
e
u
s
u
all
y
p
lace
d
i
n
s
o
f
t
w
ar
e
s
y
s
te
m
s
d
u
e
to
p
o
o
r
d
esig
n
a
n
d
i
m
p
le
m
e
n
tatio
n
ch
o
ices.
Du
r
i
n
g
d
esi
g
n
an
d
im
p
le
m
e
n
tat
io
n
p
h
ase
s
,
s
o
f
t
w
a
r
e
d
ev
elo
p
er
s
m
a
y
i
g
n
o
r
e
th
e
p
o
ten
tial
p
r
o
b
lem
s
s
u
c
h
as
co
d
e
d
u
p
licatio
n
,
u
n
c
lear
co
d
e,
co
m
p
licated
co
d
e,
an
d
d
ea
d
co
d
e.
Su
ch
d
esi
g
n
a
n
d
i
m
p
le
m
e
n
tatio
n
is
s
u
es a
r
e
ab
le
to
m
ak
e
s
o
f
t
w
a
r
e
s
y
s
te
m
s
m
o
r
e
an
d
m
o
r
e
m
e
s
s
y
o
v
er
ti
m
e.
T
ab
le
1
s
h
o
w
t
h
e
s
i
x
co
d
e
s
m
ells
w
h
ic
h
w
ill
b
e
co
n
s
id
er
ed
i
n
t
h
is
s
t
u
d
y
.
T
h
ese
co
d
e
s
m
ell
s
ar
e
m
aj
o
r
an
d
f
r
eq
u
en
t
l
y
o
cc
u
r
r
ed
b
ad
c
o
d
e
s
m
ell
s
i
n
th
e
o
b
j
ec
t
-
o
r
ien
ted
s
y
s
te
m
s
.
So
m
e
s
m
ell
s
s
u
c
h
a
s
Go
d
C
las
s
a
n
d
L
ar
g
e
C
lass
ar
e
clo
s
el
y
r
elate
d
to
th
e
class
s
tr
u
ctu
r
e
s
an
d
o
th
er
s
ar
e
f
in
e
-
g
r
ain
ed
co
d
e
s
m
ell
s
in
th
e
s
y
s
te
m
lik
e
Featu
r
e
E
n
v
y
a
n
d
L
o
n
g
Me
th
o
d
.
A
s
co
n
s
id
er
in
g
t
h
e
class
le
v
elas
w
ell
a
s
t
h
e
m
et
h
o
d
lev
el,
w
e
ca
n
i
m
p
r
o
v
e
th
e
ac
c
u
r
ac
y
o
f
th
e
p
r
o
p
o
s
ed
n
eu
r
al
n
et
w
o
r
k
m
o
d
el.
T
ab
le
1
.
C
o
d
e
s
m
ell
s
co
n
s
id
er
ed
in
th
e
s
t
u
d
y
C
o
d
e
S
me
l
l
L
e
v
e
l
D
e
scri
p
t
i
o
n
G
o
d
C
l
a
ss
C
l
a
ss
C
l
a
ss
es
ha
ve
ma
n
y
me
mb
e
r
s a
n
d
i
mp
l
e
me
n
t
d
i
f
f
e
r
e
n
t
b
e
h
a
v
i
o
r
s
L
a
r
g
e
C
l
a
ss
C
l
a
ss
C
l
a
sse
s d
o
t
o
o
m
a
n
y
t
a
s
k
s w
i
t
h
m
a
n
y
me
t
h
o
d
s
a
n
d
d
a
t
a
me
mb
e
r
s
F
e
a
t
u
r
e
En
v
y
C
l
a
ss,
M
e
t
h
o
d
M
e
t
h
o
d
s
a
c
c
e
ss
t
h
e
d
a
t
a
o
f
a
n
o
t
h
e
r
o
b
j
e
c
t
mo
r
e
t
h
a
n
i
t
s o
w
n
d
a
t
a
P
a
r
a
l
l
e
l
I
n
h
e
r
i
t
a
n
c
e
H
i
e
r
a
r
c
h
i
e
s
C
l
a
ss
W
e
n
e
e
d
t
o
c
r
e
a
t
e
a
su
b
c
l
a
ss fo
r
a
n
o
t
h
e
r
c
l
a
ss
w
h
e
n
e
v
e
r
we
c
r
e
a
t
e
a
su
b
c
l
a
ss fo
r
a
c
l
a
ss
D
a
t
a
C
l
a
ss
C
l
a
ss
C
l
a
sse
s c
o
n
t
a
i
n
o
n
l
y
f
i
e
l
d
s(
d
a
t
a
)
a
n
d
me
t
h
o
d
s fo
r
a
c
c
e
ssi
n
g
t
h
e
m
L
a
z
y
C
l
a
ss
C
l
a
ss
C
l
a
ss
es
do
n
o
t
d
o
e
n
o
u
g
h
w
o
r
k
2
.
2
.
Neura
l
Net
w
o
rk
M
o
del
Fig
u
r
e
2
d
ep
icts
a
s
i
m
p
li
f
ied
n
eu
r
al
n
et
w
o
r
k
m
o
d
el
t
h
at
i
s
u
s
ed
in
th
e
p
r
o
p
o
s
ed
co
d
e
s
m
ell
d
etec
tio
n
s
y
s
te
m
.
T
h
e
n
et
w
o
r
k
m
o
d
el
i
s
i
m
p
le
m
e
n
ted
i
n
T
en
s
o
r
Flo
w
a
n
d
co
n
s
is
t
s
o
f
th
r
ee
la
y
er
s
-
i
n
p
u
t
la
y
er
,
h
id
d
en
la
y
er
,
an
d
o
u
tp
u
t
la
y
er
.
T
h
er
e
ar
e
th
e
ei
g
h
t
o
b
j
ec
t
-
o
r
ien
te
d
m
etr
ics
a
s
i
n
p
u
t
v
al
u
e
i
n
t
h
e
i
n
p
u
t
la
y
er
-
L
OC
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2088
-
8708
I
J
E
C
E
Vo
l.
7
,
No
.
6
,
Dec
em
b
er
2
0
1
7
:
3
6
1
3
–
3
6
2
1
3616
DI
T
,
C
B
O,
R
FC
,
W
MC,
NOC,
C
C
,
a
n
d
L
C
OM
.
T
h
e
n
e
u
r
o
n
s
in
t
h
e
d
if
f
er
e
n
t
la
y
er
s
ar
e
co
n
n
ec
ted
w
it
h
co
n
n
ec
tio
n
w
ei
g
h
ts
a
n
d
b
iase
s
.
Fu
r
t
h
er
m
o
r
e,
t
h
e
p
r
o
p
o
s
ed
n
et
w
o
r
k
m
o
d
el
s
u
p
p
o
r
ts
m
u
ltip
le
h
id
d
en
la
y
er
s
to
i
m
p
r
o
v
e
t
h
e
p
er
f
o
r
m
a
n
ce
o
f
t
h
e
m
o
d
el
w
h
en
it
p
r
ed
icts
t
h
e
s
m
ell
y
cla
s
s
.
I
n
t
h
e
n
e
u
r
al
n
et
w
o
r
k
m
o
d
el,
t
h
e
tan
h
f
u
n
ctio
n
is
u
s
ed
as
ac
ti
v
atio
n
f
u
n
ctio
n
f
o
r
t
h
e
h
id
d
en
la
y
er
s
a
n
d
t
h
e
s
o
f
t
m
a
x
f
u
n
ct
io
n
i
s
u
s
ed
a
s
ac
tiv
atio
n
f
u
n
ctio
n
f
o
r
th
e
o
u
t
p
u
t
la
y
er
.
I
n
t
h
i
s
s
t
u
d
y
,
A
d
a
m
Op
ti
m
izer
i
n
T
en
s
o
r
Flo
w
i
s
u
s
ed
to
o
p
tim
ize
t
h
e
lear
n
in
g
r
ate.
Fig
u
r
e
2
.
Neu
r
al
Net
w
o
r
k
Mo
d
el
3.
E
VA
L
UA
T
I
O
N
C
ase
s
t
u
d
ies
h
a
v
e
b
ee
n
co
n
d
u
cted
to
d
e
m
o
n
s
tr
ate
t
h
e
e
f
f
e
ctiv
e
n
ess
a
n
d
ac
cu
r
ac
y
o
f
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
f
o
r
co
d
e
s
m
ell
d
ete
ctio
n
w
it
h
a
co
r
p
u
s
o
f
m
as
s
i
v
e
J
av
a
p
r
o
j
ec
ts
.
Du
r
i
n
g
t
h
es
e
ca
s
e
s
tu
d
ie
s
,
t
h
e
p
r
o
p
o
s
ed
n
eu
r
al
n
et
w
o
r
k
m
o
d
el
h
as
b
ee
n
tr
ain
ed
an
d
te
s
t
ed
to
p
r
ed
ict
th
e
co
d
e
s
m
el
l
s
in
t
h
e
p
r
ec
ed
in
g
s
ec
tio
n
.
3
.
1
.
J
a
v
a
G
it
H
ub
pro
j
ec
t
s
T
h
e
J
av
a
p
r
o
g
r
am
m
i
n
g
la
n
g
u
ag
e
h
as
co
n
tin
u
ed
to
e
v
o
lv
e
with
t
h
e
r
ap
id
s
p
r
ea
d
o
f
t
h
e
I
n
t
er
n
et
s
i
n
c
e
its
r
elea
s
e
i
n
th
e
w
o
r
ld
.
J
av
a
also
h
as
b
ee
n
u
s
ed
as
t
h
e
p
r
i
m
ar
y
d
ev
e
lo
p
m
en
t
lan
g
u
ag
e
f
o
r
A
n
d
r
o
id
w
h
ic
h
is
o
n
e
o
f
th
e
m
o
s
t
p
o
p
u
lar
m
o
b
i
le
p
latf
o
r
m
s
.
GitH
u
b
,
a
p
u
b
lic
co
d
e
r
ep
o
s
ito
r
y
,
s
to
r
es
an
d
m
an
a
g
e
s
m
a
n
y
J
av
a
p
r
o
j
ec
ts
in
d
if
f
er
e
n
t
ap
p
licatio
n
d
o
m
ai
n
s
.
T
h
e
p
r
o
p
o
s
ed
c
o
d
e
s
m
el
l
d
etec
tio
n
s
y
s
te
m
a
i
m
s
at
lo
ca
ti
n
g
b
ad
s
m
el
ls
in
t
h
e
J
av
a
p
r
o
j
ec
ts
.
W
h
ile
ex
i
s
ti
n
g
r
esear
ch
h
as
d
etec
ted
co
d
e
s
m
e
lls
i
n
r
elati
v
el
y
s
m
all
-
s
ca
le
J
av
a
p
r
o
j
ec
ts
,
th
e
p
r
o
p
o
s
ed
ap
p
r
o
a
ch
u
s
e
s
o
p
en
lar
g
e
-
s
ca
le
J
av
a
p
r
o
g
r
am
s
to
in
cr
ea
s
e
t
h
e
r
eliab
ilit
y
an
d
p
r
ec
is
io
n
o
f
th
e
co
d
e
s
m
el
l
d
etec
tio
n
s
t
r
ateg
ies.
T
h
e
Git
Hu
b
co
d
e
r
ep
o
s
ito
r
ies
m
a
y
i
n
cl
u
d
e
lo
w
-
q
u
alit
y
J
av
a
p
r
o
j
ec
ts
,
th
u
s
r
eq
u
ir
i
n
g
t
h
e
ab
ilit
y
to
r
e
m
o
v
e
t
h
e
s
o
u
r
ce
co
d
e
f
o
r
th
ese
p
r
o
j
ec
ts
.
T
h
e
f
o
r
k
s
y
s
te
m
p
r
o
v
id
ed
b
y
t
h
e
GitHu
b
co
d
e
r
ep
o
s
ito
r
y
ca
n
b
e
u
s
ed
to
ef
f
ec
ti
v
el
y
ex
tr
ac
t
l
o
w
-
q
u
alit
y
J
av
a
p
r
o
j
ec
ts
w
it
h
lo
w
f
o
r
k
r
atio
s
.
A
J
av
a
p
r
o
j
ec
t w
i
th
a
lo
w
f
o
r
k
f
r
eq
u
en
c
y
is
as
s
u
m
ed
to
b
e
a
lo
w
-
q
u
al
it
y
p
r
o
j
ec
t.
T
ab
le
2
.
J
av
a
GitHu
b
p
r
o
j
ec
ts
N
o
.
P
r
o
j
e
c
t
N
a
me
s
#
o
f
C
l
a
sse
s
L
O
C
D
e
scri
p
t
i
o
n
s
1
A
n
d
r
o
i
d
-
U
n
i
v
e
r
sal
-
I
mag
e
-
L
o
a
d
e
r
1
3
8
6
,
8
2
0
A
n
d
r
o
i
d
L
i
b
r
a
r
y
2
b
i
g
b
l
u
e
b
u
t
t
o
n
9
5
9
59
,
099
W
e
b
C
o
n
f
e
r
e
n
c
i
n
g
3
b
u
k
k
i
t
7
8
5
32
,
277
M
i
n
c
r
a
f
t
M
o
d
A
P
I
4
c
l
o
j
u
r
e
7
5
4
38
,
389
P
r
o
g
r
a
mm
i
n
g
L
a
n
g
u
a
g
e
5
d
r
o
p
w
i
z
a
r
d
6
1
7
22
,
687
R
EST
f
u
l
w
e
b
se
r
v
e
r
6
e
l
a
st
i
c
se
a
r
c
h
7
,
3
9
1
4
2
0
,
8
3
4
R
EST
S
e
a
r
c
h
En
g
i
n
e
7
j
u
n
i
t
1
,
2
2
4
2
5
3
,
79
T
e
st
i
n
g
F
r
a
mew
o
r
k
8
l
i
b
g
d
x
3
,
2
2
4
2
1
7
,
7
7
5
G
a
me
D
e
v
F
r
a
mew
o
r
k
9
me
t
r
i
c
s
4
0
1
14
,
904
M
e
t
r
i
c
s
F
r
a
mew
o
r
k
10
n
e
t
t
y
2
,
6
7
4
1
6
1
,
4
0
6
N
e
t
w
o
r
k
A
p
p
F
r
a
mew
o
r
k
11
n
o
k
o
g
i
r
i
1
5
4
12
,
300
H
T
M
L
/
X
M
L
/
C
S
S
p
a
r
se
r
12
o
k
h
t
t
p
5
1
5
37
,
264
H
TT
P
&
H
T
TP/2
C
l
i
e
n
t
13
p
l
a
t
f
o
r
m_
f
r
a
me
w
o
r
k
s_
b
a
se
12
,
959
1
,
3
7
4
,
8
5
2
A
n
d
r
o
i
d
B
a
se
F
r
a
mew
o
r
k
14
p
r
e
st
o
2
,
8
4
1
2
0
9
,
2
9
5
D
i
st
r
i
b
u
t
e
d
S
Q
L
e
n
g
i
n
e
15
r
e
t
r
o
f
i
t
3
6
2
9
,
0
9
8
R
EST
c
l
i
e
n
t
16
r
x
j
a
v
a
2
,
1
0
9
49
,
527
R
e
a
c
t
i
v
e
JV
M
e
x
t
e
n
si
o
n
s
17
sp
r
i
n
g
-
b
o
o
t
2
,
6
6
6
93
,
347
A
p
p
F
r
a
mew
o
r
k
W
r
a
p
p
e
r
18
sp
r
i
n
g
-
f
r
a
me
w
o
r
k
11
,
651
4
8
4
,
2
2
4
A
p
p
l
i
c
a
t
i
o
n
F
r
a
mew
o
r
k
19
st
o
r
m
1
,
2
0
5
64
,
641
D
i
st
r
i
b
u
t
e
d
C
o
mp
u
t
a
t
i
o
n
20
z
x
i
n
g
5
4
4
39
,
873
B
a
r
c
o
d
e
i
mag
e
p
r
o
c
e
ssi
n
g
T
o
t
a
l
53
,
173
3
,
3
4
8
,
6
1
2
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
F
in
d
in
g
B
a
d
C
o
d
e
S
mells w
ith
N
eu
r
a
l Netw
o
r
k
Mo
d
els
(
Do
n
g
K
w
a
n
K
im
)
3617
T
ab
le
2
s
h
o
w
s
th
e
Git
Hu
b
J
av
a
co
r
p
u
s
[
3
]
w
it
h
s
h
o
r
t
d
escr
ip
tio
n
s
th
at
is
u
s
ed
to
tr
ain
an
d
test
th
e
p
r
o
p
o
s
ed
n
eu
r
al
n
et
w
o
r
k
m
o
d
el.
T
h
e
d
ata
s
et
is
p
ick
ed
f
r
o
m
th
e
to
p
ac
ti
v
e
J
av
a
GitH
u
b
p
r
o
j
ec
ts
an
d
in
cl
u
d
es
d
if
f
er
e
n
t
ap
p
licatio
n
d
o
m
ain
s
f
o
r
d
iv
er
s
it
y
.
T
h
e
d
ata
s
et
i
n
cl
u
d
es
2
0
p
r
o
j
ec
ts
,
m
o
r
e
th
a
n
5
3
,
0
0
0
J
av
a
class
es,
an
d
m
o
r
e
th
a
n
3
,
0
0
0
,
0
0
0
s
o
u
r
ce
co
d
e
lin
e
s
.
S
u
c
h
a
co
r
p
u
s
s
t
atis
tics
tell
s
t
h
e
d
ata
s
et
is
b
ig
en
o
u
g
h
to
tr
ai
n
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
to
ca
tch
b
ad
s
m
el
ls
i
n
co
d
e.
T
h
e
w
h
o
le
J
av
a
co
r
p
u
s
is
s
p
lit
u
n
i
f
o
r
m
l
y
i
n
to
a
tr
ain
i
n
g
an
d
a
test
s
et.
T
ab
le
3
.
I
d
en
tif
icatio
n
R
u
les
f
o
r
Dete
ctin
g
B
ad
S
m
ells
B
a
d
S
me
l
l
R
u
l
e
s
L
a
r
g
e
C
l
a
ss
L
O
C
>
3
0
0
o
r
D
I
T
>
5
o
r
R
F
C
>
2
0
L
a
z
y
C
l
a
ss
R
F
C
=
=
0
o
r
L
O
C
<
1
0
0
o
r
W
M
C
<
=
2
D
a
t
a
C
l
a
ss
L
C
O
M
>
8
0
o
r
W
M
C
>
5
0
P
a
r
a
l
l
e
l
I
n
h
e
r
i
t
a
n
c
e
H
i
e
r
a
r
c
h
i
e
s
D
I
T
>
3
o
r
N
O
C
>
4
G
o
d
C
l
a
ss
W
M
C
>
=
4
7
F
e
a
t
u
r
e
En
v
y
C
B
O
>
5
o
r
L
C
O
M
>
5
0
Fo
r
th
e
ca
s
e
s
t
u
d
ies,
s
ix
d
esi
g
n
co
d
e
s
m
ell
s
ar
e
co
n
s
id
er
ed
t
o
id
en
ti
f
y
t
h
e
p
r
ese
n
ce
o
f
b
ad
s
m
ells
i
n
th
e
g
i
v
e
n
J
av
a
co
d
e.
T
h
e
g
iv
e
n
J
av
a
p
r
o
j
ec
ts
h
av
e
b
ee
n
an
a
l
y
ze
d
an
d
test
ed
to
d
etec
t
ea
ch
o
f
th
e
s
i
x
b
ad
co
d
e
s
m
el
ls
.
I
f
a
cla
s
s
o
r
a
m
et
h
o
d
s
atis
f
ies
t
h
e
id
en
t
if
ied
th
r
es
h
o
ld
s
s
h
o
w
n
in
T
ab
le
3
,
it
is
m
ar
k
ed
as
s
m
e
ll
y
clas
s
o
r
m
eth
o
d
.
T
ab
le
3
p
r
esen
ts
ex
p
er
i
m
e
n
tal
t
h
r
es
h
o
ld
s
to
ap
p
ly
th
e
o
b
j
ec
ted
-
o
r
ien
ted
m
etr
ics.
Di
f
f
er
e
n
t
id
en
ti
f
icatio
n
r
u
les ca
n
b
e
d
ef
i
n
ed
as f
o
llo
w
i
n
g
:
L
ar
g
e
C
las
s
:
Si
n
ce
m
ea
s
u
r
i
n
g
th
e
s
ize
o
f
a
clas
s
is
a
s
i
m
p
le
b
u
t
ess
e
n
tia
l
ele
m
e
n
t,
it
h
a
s
b
ee
n
u
s
ed
to
ev
alu
a
te
th
e
q
u
alit
y
o
f
s
o
f
t
war
e
s
y
s
te
m
s
.
T
h
e
class
ic
a
n
d
co
m
m
o
n
w
a
y
o
f
m
ea
s
u
r
i
n
g
class
s
ize
is
to
m
ea
s
u
r
e
t
h
e
n
u
m
b
er
o
f
li
n
es
o
f
co
d
e,
i.e
.
L
OC
,
o
r
t
h
e
n
u
m
b
er
o
f
attr
ib
u
tes
a
n
d
o
p
er
atio
n
s
.
Ho
w
e
v
er
,
to
co
n
s
id
er
o
n
l
y
L
O
C
i
s
n
o
t
en
o
u
g
h
to
d
etec
t
t
h
e
lar
g
e
cla
s
s
es.
T
h
e
p
r
o
p
o
s
ed
n
eu
r
al
n
et
wo
r
k
m
o
d
el
w
il
l
m
ea
s
u
r
e
m
o
r
e
p
r
o
p
er
ties
:
Dep
th
o
f
I
n
h
er
ita
n
ce
T
r
ee
(
DI
T
)
an
d
R
esp
o
n
s
e
f
o
r
a
C
las
s
(
R
FC
)
.
T
h
e
m
o
d
el
u
s
e
s
a
th
r
es
h
o
ld
o
f
3
00
f
o
r
L
OC
,
a
th
r
e
s
h
o
ld
o
f
5
f
o
r
DI
T
an
d
a
th
r
esh
o
ld
o
f
2
0
f
o
r
R
F
C
in
o
r
d
er
to
d
eter
m
in
e
i
f
clas
s
es
ar
e
s
tr
u
ct
u
r
al
ch
ar
ac
ter
is
t
ics d
u
e
to
th
e
L
ar
g
e
C
las
s
co
d
e
s
m
ell.
L
az
y
C
las
s
:
So
f
t
w
ar
e
d
ev
elo
p
er
s
m
ak
e
ef
f
o
r
ts
to
m
ai
n
tai
n
class
es
i
n
an
o
b
j
ec
t
-
o
r
ien
ted
s
y
s
te
m
o
v
er
ti
m
e.
Sin
ce
u
n
d
er
s
ta
n
d
in
g
a
n
d
m
ai
n
tain
i
n
g
cla
s
s
e
s
is
a
ti
m
e
-
co
n
s
u
m
i
n
g
ta
s
k
,
t
h
e
y
w
an
t
to
k
ee
p
m
ea
n
in
g
f
u
l
an
d
f
u
n
ctio
n
al
cla
s
s
es
in
t
h
e
s
o
f
t
w
ar
e
s
y
s
te
m
.
E
x
tr
a
cla
s
s
es
c
an
i
n
cr
ea
s
e
t
h
e
co
m
p
le
x
it
y
o
f
t
h
e
s
o
f
t
w
ar
e
s
y
s
te
m
.
I
n
ter
m
s
o
f
s
o
f
t
w
ar
e
m
ai
n
ten
a
n
ce
,
n
ea
r
l
y
u
s
ele
s
s
c
lass
es
ca
n
b
e
s
u
b
j
ec
t
to
co
d
e
r
ef
ac
to
r
in
g
[
2
]
.
L
az
y
C
la
s
s
s
i
m
p
l
y
r
e
f
er
s
to
a
class
th
a
t
d
o
es
n
o
t
d
o
en
o
u
g
h
an
d
d
o
es
n
o
t
ea
r
n
y
o
u
r
atte
n
tio
n
.
L
O
C
is
a
b
asic
m
etr
ic
to
d
etec
t
th
e
L
az
y
C
la
s
s
s
m
ells
.
R
F
C
a
n
d
W
MC
w
ill
b
e
co
n
s
id
er
ed
f
o
r
th
e
p
r
ec
is
e
m
ea
s
u
r
e
m
e
n
t.
T
h
e
p
r
o
p
o
s
ed
co
d
e
s
m
ell
d
etec
tio
n
s
y
s
te
m
tr
i
es
to
f
in
d
th
e
laz
y
clas
s
es
w
it
h
t
h
e
f
o
llo
w
i
n
g
co
n
d
itio
n
s
:
R
F
C
==
0
o
r
L
OC
<
1
0
0
o
r
W
MC <
=
2.
Data
C
las
s
:
A
cla
s
s
ca
n
b
e
d
ef
in
ed
as
a
s
p
ec
if
icatio
n
o
f
attr
i
b
u
tes
an
d
p
er
m
is
s
ib
le
o
p
er
atio
n
s
.
T
h
er
ef
o
r
e,
a
class
s
h
o
u
ld
p
r
o
v
id
e
p
ar
ticu
la
r
s
er
v
ices
u
s
i
n
g
o
p
er
atio
n
s
f
o
r
o
th
er
clas
s
es.
Ho
w
ev
er
,
Data
C
las
s
i
s
a
c
la
s
s
th
at
d
ef
i
n
es
v
ar
iab
les
f
o
r
m
an
ag
i
n
g
d
ata
w
it
h
o
u
t
m
ea
n
in
g
f
u
l
o
p
er
atio
n
s
.
I
t
d
o
es
n
o
t
r
ep
r
es
en
t
u
s
e
f
u
l
an
d
in
d
ep
en
d
en
t
b
eh
a
v
io
r
s
i
n
a
s
o
f
t
w
ar
e
s
y
s
te
m
.
T
h
e
d
ata
cla
s
s
i
s
clo
s
er
to
a
s
i
m
p
le
d
ata
s
tr
u
ct
u
r
e
t
h
an
a
r
esp
o
n
s
ib
le
class
.
Fo
r
th
e
b
etter
m
ai
n
tai
n
ab
ilit
y
,
co
d
e
s
m
ell
d
etec
to
r
s
lo
ca
te
th
e
d
ata
class
es
if
p
o
s
s
ib
l
e
an
d
g
u
id
e
p
r
o
g
r
am
m
er
s
to
ap
p
ly
r
e
f
ac
to
r
in
g
tech
n
iq
u
es
s
u
c
h
as
Mo
v
e
Me
t
h
o
d
an
d
E
x
tr
a
ct
Me
th
o
d
.
T
h
e
t
w
o
m
etr
ics
L
C
O
M
an
d
W
MC
is
u
s
ed
to
f
in
d
th
e
d
ata
clas
s
es
w
it
h
th
e
f
o
llo
w
in
g
co
n
d
iti
o
n
s
:
L
C
OM
>
8
0
o
r
W
MC >
5
0
.
P
ar
allel
I
n
h
er
ita
n
ce
Hier
ar
ch
i
es:
I
t
i
s
n
o
d
o
u
b
t
t
h
at
t
h
e
co
n
ce
p
t
o
f
i
n
h
er
ita
n
ce
in
o
b
j
ec
t
-
o
r
ien
ted
s
y
s
te
m
s
p
r
o
v
id
es
m
a
n
y
ad
v
an
tag
e
s
s
u
ch
as
co
d
e
r
eu
s
e.
Ho
w
ev
er
,
t
h
e
u
s
e
o
f
m
is
u
s
ed
in
h
er
ita
n
c
e
s
tr
u
ctu
r
e
s
n
o
t
o
n
l
y
co
m
p
l
icate
s
t
h
e
s
o
f
t
war
e
s
y
s
te
m
,
b
u
t
al
s
o
m
a
k
es
s
o
f
t
w
ar
e
m
ain
t
en
a
n
ce
h
ar
d
e
r
.
T
h
e
p
ar
allel
in
h
er
i
tan
ce
h
ier
ar
ch
ie
s
s
m
ell
h
ap
p
en
s
w
h
en
w
e
h
a
v
e
t
w
o
p
ar
allel
in
h
er
ita
n
ce
h
ier
ar
c
h
ie
s
ass
o
ciate
d
b
y
co
m
p
o
s
i
tio
n
.
E
v
er
y
ti
m
e
we
c
r
ea
te
a
s
u
b
class
o
f
a
cla
s
s
,
w
e
n
ee
d
to
cr
ea
te
a
s
u
b
clas
s
f
o
r
a
n
o
th
er
cla
s
s
d
u
e
to
th
e
s
tr
u
ct
u
r
e
p
r
o
b
le
m
.
I
t
is
n
ec
es
s
ar
y
to
c
h
a
n
g
e
t
h
e
p
ar
all
el
in
h
er
ita
n
ce
s
tr
u
ct
u
r
e
to
s
o
lv
e
th
e
p
r
o
b
le
m
s
ca
u
s
ed
b
y
t
h
e
i
n
h
er
itan
ce
h
ier
ar
ch
y
.
I
n
t
h
i
s
p
ap
er
,
DI
T
an
d
NOC
ar
e
co
n
s
id
er
ed
to
d
eter
m
in
e
i
f
a
s
o
f
t
w
ar
e
s
y
s
te
m
co
n
tain
s
t
h
e
P
ar
allel
I
n
h
er
ita
n
ce
H
ier
ar
ch
ies
p
r
o
b
lem
s
w
i
th
th
e
f
o
llo
w
i
n
g
co
n
d
itio
n
s
:
DI
T
>
3
o
r
NOC >
4
.
Go
d
C
las
s
:
Go
d
C
las
s
r
e
f
er
s
t
o
a
clas
s
t
h
at
d
o
es
to
o
m
u
c
h
.
Ma
n
y
b
eh
a
v
io
r
s
ar
e
ce
n
tr
alize
d
in
a
p
ar
ticu
la
r
class
w
it
h
u
n
m
a
n
ag
ea
b
le
attr
ib
u
tes
a
n
d
o
p
er
atio
n
s
.
T
o
a
d
d
r
ess
s
u
ch
p
r
o
b
le
m
s
d
u
e
to
th
e
g
o
d
class
,
a
co
m
m
o
n
an
d
i
n
tu
iti
v
e
tech
n
i
q
u
e
is
to
s
ep
ar
ate
th
e
g
o
d
class
i
n
to
s
ev
er
al
s
m
a
ller
class
es.
T
h
e
o
b
j
ec
t
-
o
r
ien
ted
m
e
tr
ic
W
MC
is
u
s
ed
to
f
in
d
th
e
Go
d
C
lass
s
m
ell
th
r
o
u
g
h
t
h
e
s
u
m
o
f
th
e
s
tatis
t
ica
l
co
m
p
le
x
it
y
o
f
all
m
e
th
o
d
s
i
n
a
clas
s
.
A
cla
s
s
is
a
Go
d
C
las
s
w
h
e
n
W
MC
o
f
t
h
e
cla
s
s
is
eq
u
al
to
o
r
g
r
ea
t
er
th
an
4
7
.
E
v
e
n
if
w
e
ca
n
ap
p
l
y
o
t
h
er
m
etr
ics
f
o
r
Go
d
C
la
s
s
,
W
MC
is
o
n
e
o
f
t
h
e
m
aj
o
r
m
ea
s
u
r
es
to
d
etec
t
th
e
Go
d
C
las
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2088
-
8708
I
J
E
C
E
Vo
l.
7
,
No
.
6
,
Dec
em
b
er
2
0
1
7
:
3
6
1
3
–
3
6
2
1
3618
s
m
el
ls
.
I
n
g
e
n
er
al,
Go
d
C
la
s
s
ca
n
b
e
d
ec
o
m
p
o
s
ed
i
n
o
th
e
r
m
u
ltip
le
cla
s
s
es
b
y
u
s
i
n
g
c
o
d
e
r
ef
ac
to
r
in
g
tech
n
iq
u
es.
Featu
r
e
E
n
v
y
:
Af
ter
s
o
f
t
w
ar
e
s
y
s
te
m
s
ar
e
r
elea
s
ed
,
t
h
eir
co
d
e
s
tr
u
ct
u
r
e
ca
n
ch
a
n
g
e
d
u
r
i
n
g
o
p
er
atio
n
.
Du
e
to
th
e
v
ar
io
u
s
co
d
e
r
ef
ac
to
r
in
g
,
clas
s
es
m
a
y
b
e
i
n
te
g
r
ated
o
r
s
ep
ar
ated
w
it
h
o
t
h
er
clas
s
e
s
.
T
h
e
s
tr
u
c
tu
r
e
ch
an
g
e
o
f
th
e
clas
s
ca
n
ca
u
s
e
its
attr
ib
u
teso
r
o
p
er
atio
n
s
to
b
e
p
lace
d
in
d
if
f
er
en
t
c
lass
e
s
.
I
n
t
h
is
ca
s
e,
a
s
p
ec
if
ic
o
p
er
atio
n
n
ee
d
s
to
f
r
eq
u
en
tl
y
ac
ce
s
s
s
o
m
e
d
ata
o
f
an
o
th
er
clas
s
.
T
h
is
is
ca
lled
Featu
r
e
E
n
v
y
,
w
h
ic
h
m
a
y
in
cr
ea
s
e
t
h
e
co
m
p
lex
it
y
o
f
th
e
s
o
f
t
w
ar
e
s
y
s
te
m
.
T
h
e
id
en
tif
icatio
n
o
f
Feat
u
r
e
E
n
v
y
in
s
o
u
r
ce
co
d
e
ca
n
b
e
p
er
f
o
r
m
ed
b
y
m
e
asu
r
i
n
g
t
h
e
s
tr
en
g
th
o
f
co
u
p
li
n
g
t
h
at
a
m
et
h
o
d
h
as
to
m
e
th
o
d
s
b
elo
n
g
i
n
g
to
o
th
er
clas
s
es.
Als
o
,
t
h
is
co
d
e
s
m
ell
ca
n
b
e
d
etec
ted
b
y
m
ea
s
u
r
in
g
th
e
d
eg
r
ee
o
f
l
ac
k
o
f
co
h
e
s
io
n
i
n
me
t
h
o
d
s
.
T
h
e
t
w
o
m
etr
ics
C
B
O
an
d
L
C
OM
ar
e
co
n
s
i
d
er
ed
f
o
r
Featu
r
e
E
n
v
y
w
i
t
h
t
h
e
f
o
llo
w
i
n
g
co
n
d
itio
n
s
:
C
B
O
>
5
o
r
L
C
O
M
>
5
0
.
3
.
2
.
E
x
peri
m
ent
a
l
Res
ults
a
n
d
Dis
cu
s
s
io
n
T
h
e
p
r
o
p
o
s
ed
n
eu
r
al
n
et
w
o
r
k
m
o
d
el
h
as
b
ee
n
tr
ai
n
ed
an
d
test
ed
w
it
h
a
co
r
p
u
s
o
f
t
h
e
t
w
e
n
t
y
J
av
a
p
r
o
j
ec
ts
w
h
ich
co
n
tai
n
m
o
r
e
t
h
an
5
3
,
0
0
0
J
av
a
class
es
an
d
m
o
r
e
t
h
a
n
3
,
0
0
0
,
0
0
0
s
o
u
r
ce
c
o
d
e
lin
es.
T
h
ese
d
ata
h
as
b
ee
n
u
s
ed
i
n
p
r
ev
io
u
s
s
tu
d
ies
a
n
d
d
o
n
o
t
i
n
clu
d
e
a
s
p
ec
if
ic
ap
p
licatio
n
d
o
m
ai
n
,
b
u
t
a
v
ar
iet
y
o
f
d
o
m
ai
n
s
.
I
n
th
i
s
s
t
u
d
y
,
t
h
e
d
ataset
s
ize
is
lar
g
e
en
o
u
g
h
to
tr
ain
th
e
p
r
o
p
o
s
ed
n
et
w
o
r
k
m
o
d
el
an
d
p
r
o
d
u
ce
m
o
r
e
ac
cu
r
ate
p
r
ed
ictio
n
r
esu
lt
s
w
h
en
th
e
m
o
d
el
is
u
s
ed
to
d
etec
t
th
e
co
d
e
s
m
el
ls
.
Fi
g
u
r
es
3
t
h
r
o
u
g
h
8
s
h
o
w
th
e
ex
p
er
i
m
e
n
tal
r
es
u
lt
s
ac
co
r
d
in
g
to
t
h
e
s
ix
co
d
e
s
m
el
ls
w
h
ich
ar
e
co
n
s
id
er
ed
in
t
h
i
s
s
t
u
d
y
.
T
h
e
p
r
o
p
o
s
ed
n
et
w
o
r
k
m
o
d
el
h
as
b
ee
n
e
v
alu
ated
w
it
h
d
if
f
er
en
t
iter
at
io
n
s
w
h
e
n
f
i
x
i
n
g
t
h
e
n
u
m
b
er
o
f
th
e
h
id
d
en
la
y
er
s
.
T
h
e
n
u
m
b
er
o
f
th
e
ep
o
ch
s
i
n
t
h
e
n
et
w
o
r
k
m
o
d
el
v
ar
ies
2
0
0
th
r
o
u
g
h
5
0
,
0
0
0
.
I
t
is
o
b
s
er
v
ed
th
at
th
e
ac
c
u
r
ac
y
o
f
t
h
e
n
et
w
o
r
k
m
o
d
el
r
an
g
e
s
f
r
o
m
9
1
%
to
9
9
%
ac
co
r
d
in
g
to
t
h
e
n
u
m
b
er
o
f
th
e
ep
o
ch
s
.
As
w
e
ca
n
s
ee
in
t
h
e
f
i
g
u
r
es,
t
h
e
ac
c
u
r
ac
y
o
f
t
h
e
m
o
d
el
is
m
o
v
i
n
g
u
p
w
ar
d
s
w
h
en
th
e
n
u
m
b
er
o
f
th
e
ep
o
ch
s
is
i
n
cr
ea
s
i
n
g
.
I
n
o
th
er
w
o
r
d
s
,
t
h
e
ac
c
u
r
ac
y
o
f
th
e
m
o
d
el
o
n
t
h
e
te
s
t d
ata
is
p
r
o
p
o
r
ti
o
n
al
to
t
h
e
n
u
m
b
er
o
f
t
h
e
ep
o
c
h
s
.
As
th
e
m
o
d
el
i
s
h
ig
h
l
y
tr
ai
n
ed
o
n
th
e
s
a
m
e
d
ataset,
th
e
r
es
u
lt
s
w
ill
b
e
b
ette
r
.
Ho
w
ev
er
,
t
h
e
t
y
p
e
o
f
t
h
e
c
o
d
e
s
m
ell
d
o
es
n
o
t
af
f
ec
t
th
e
ac
c
u
r
ac
y
o
f
t
h
e
p
r
o
p
o
s
ed
m
o
d
el.
T
h
e
p
r
o
p
o
s
ed
n
et
w
o
r
k
m
o
d
el
is
tr
ain
ed
an
d
ev
alu
a
ted
w
it
h
t
h
e
d
if
f
er
e
n
t
n
u
m
b
er
o
f
th
e
h
id
d
en
la
y
er
s
to
d
e
m
o
n
s
tr
ate
t
h
e
ef
f
ec
ti
v
en
e
s
s
o
f
t
h
e
h
id
d
en
la
y
e
r
.
A
s
c
h
an
g
i
n
g
t
h
e
n
u
m
b
er
o
f
t
h
e
h
id
d
en
la
y
er
s
,
w
e
ca
n
o
b
s
er
v
e
h
o
w
m
u
ch
t
h
e
h
id
d
en
la
y
er
af
f
ec
t
s
t
h
e
p
r
e
d
ictio
n
r
es
u
lt
o
f
t
h
e
m
o
d
el.
T
h
e
n
u
m
b
er
o
f
t
h
e
h
id
d
en
la
y
er
s
r
a
n
g
e
s
f
r
o
m
1
to
6
0
.
T
h
e
lar
g
er
th
e
n
u
m
b
er
o
f
t
h
e
h
id
d
en
la
y
er
s
is
,
th
e
b
etter
t
h
e
p
r
ed
ictio
n
r
es
u
lt
s
ar
e.
T
h
e
ac
cu
r
ac
y
o
f
t
h
e
m
o
d
el
r
ea
ch
es
u
p
to
t
h
e
9
9
%
w
h
e
n
t
h
e
n
u
m
b
er
o
f
th
e
h
id
d
en
la
y
er
s
v
ar
ies
r
eg
ar
d
le
s
s
o
f
t
h
e
t
y
p
e
o
f
th
e
co
d
e
s
m
ell.
I
t
is
o
b
s
er
v
ed
th
at
w
h
e
n
t
h
e
m
o
d
elis
h
i
g
h
l
y
tr
ain
ed
w
it
h
m
o
r
e
d
ataset,
t
h
e
p
r
ed
ictio
n
o
u
tco
m
esar
ei
m
p
r
o
v
ed
m
o
r
e
an
d
m
o
r
e.
I
n
ad
d
itio
n
,
th
e
ac
c
u
r
ac
y
o
f
th
e
m
o
d
el
in
cr
ea
s
es
w
h
e
n
it
p
er
f
o
r
m
s
w
i
th
h
i
g
h
er
ep
o
ch
s
an
d
m
a
n
y
h
id
d
en
la
y
er
s
.
T
h
e
h
ig
h
er
v
al
u
e
o
f
th
e
ac
cu
r
ac
y
i
n
d
icate
s
th
a
t
th
e
p
r
o
p
o
s
ed
n
et
w
o
r
k
m
o
d
el
ef
f
ec
t
iv
el
y
d
etec
t
s
b
ad
co
d
e
s
m
ells
in
a
p
r
o
g
r
am
.
T
h
e
ar
tif
icial
n
e
u
r
al
n
et
w
o
r
k
i
s
f
r
e
q
u
en
tl
y
u
s
ed
i
n
co
m
p
u
ter
v
is
i
o
n
an
d
h
a
s
n
o
t
b
ee
n
u
s
ed
f
o
r
ad
d
r
ess
in
g
s
o
f
t
w
ar
e
en
g
i
n
ee
r
i
n
g
p
r
o
b
le
m
s
.
I
n
p
ar
ticu
lar
,
th
er
e
is
litt
le
e
f
f
o
r
tto
ap
p
ly
t
h
e
n
e
u
r
al
n
et
w
o
r
k
m
o
d
el
in
to
co
d
e
s
m
el
l
d
etec
tio
n
w
ith
b
i
g
en
o
u
g
h
p
r
o
g
r
a
m
co
d
e.
Fig
u
r
e
3
.
A
cc
u
r
ac
y
o
f
t
h
e
Net
w
o
r
k
Mo
d
el
w
h
e
n
Dete
ct
in
g
t
h
e
L
ar
g
e
C
la
s
s
C
o
d
e
S
m
ell
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
F
in
d
in
g
B
a
d
C
o
d
e
S
mells w
ith
N
eu
r
a
l Netw
o
r
k
Mo
d
els
(
Do
n
g
K
w
a
n
K
im
)
36
19
Fig
u
r
e
4
.
A
cc
u
r
ac
y
o
f
t
h
e
Net
w
o
r
k
Mo
d
el
w
h
e
n
Dete
ct
in
g
t
h
e
L
az
y
C
lass
C
o
d
e
S
m
el
l
Fig
u
r
e
5
.
A
cc
u
r
ac
y
o
f
t
h
e
Net
w
o
r
k
Mo
d
el
w
h
e
n
Dete
ct
in
g
t
h
e
Data
C
la
s
s
C
o
d
e
S
m
ell
Fig
u
r
e
6
.
A
cc
u
r
ac
y
o
f
t
h
e
Net
w
o
r
k
Mo
d
el
w
h
e
n
Dete
ct
in
g
t
h
e
P
ar
allel
I
n
h
er
ita
n
ce
Hier
ar
ch
ies
C
o
d
e
S
m
ell
Fig
u
r
e
7
.
A
cc
u
r
ac
y
o
f
t
h
e
Net
w
o
r
k
Mo
d
el
w
h
e
n
Dete
ct
in
g
t
h
e
Go
d
C
lass
C
o
d
e
S
m
el
l
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2088
-
8708
I
J
E
C
E
Vo
l.
7
,
No
.
6
,
Dec
em
b
er
2
0
1
7
:
3
6
1
3
–
3
6
2
1
3620
Fig
u
r
e
8
.
A
cc
u
r
ac
y
o
f
t
h
e
Net
w
o
r
k
Mo
d
el
w
h
e
n
Dete
ct
in
g
t
h
e
Feat
u
r
e
E
n
v
y
C
o
d
e
S
m
ell
4.
RE
L
AT
E
D
WO
RK
P
r
ev
io
u
s
s
t
u
d
ies
h
a
v
e
p
r
esen
t
ed
m
et
h
o
d
o
lo
g
ies
an
d
s
tr
ate
g
ies
to
d
etec
t
b
ad
co
d
e
s
m
ells
in
o
b
j
ec
t
-
o
r
ien
ted
s
o
f
t
w
ar
e
s
y
s
te
m
s
[
4
,
5
]
.
Mo
s
t
ap
p
r
o
ac
h
es
f
o
r
co
d
e
s
m
el
l
d
etec
tio
n
u
s
e
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ics
[
6
,
7
]
to
d
eter
m
i
n
e
i
f
a
s
o
f
t
w
ar
e
s
y
s
te
m
co
n
ta
in
s
b
ad
s
m
ell
s
o
r
n
o
t.
T
h
ey
h
a
v
e
id
e
n
ti
f
ied
e
m
p
i
r
ical
th
r
es
h
o
ld
s
f
o
r
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ic
s
b
y
co
n
d
u
cti
n
g
ca
s
e
s
t
u
d
ies
o
n
o
b
j
ec
t
-
o
r
ien
te
d
p
r
o
g
r
a
m
s
[
8
]
.
As
c
o
d
e
s
ize
in
cr
ea
s
es,
au
to
m
at
ic
d
etec
tio
n
to
o
ls
[
9
]
ar
e
n
ee
d
ed
to
h
elp
th
e
d
ev
elo
p
er
b
y
f
i
n
d
in
g
co
d
e
s
m
ells
s
y
s
te
m
atica
ll
y
.
Au
to
m
a
tic
s
o
f
t
w
ar
e
to
o
ls
[
1
0
]
h
av
e
b
ee
n
i
n
tr
o
d
u
ce
d
f
o
r
v
is
u
al
l
y
lo
ca
tin
g
co
d
e
s
m
ell
s
in
s
o
u
r
ce
co
d
e
b
y
h
ig
h
li
g
h
ti
n
g
s
u
s
p
icio
u
s
co
d
e
s
n
ip
p
ets.
T
h
e
y
al
lo
w
th
e
d
e
v
elo
p
er
to
ap
p
ly
co
d
e
r
ef
ac
to
r
in
g
tech
n
iq
u
es
f
o
r
id
en
ti
f
ied
co
d
e
s
m
ells
.
Mo
s
t d
etec
tio
n
s
y
s
te
m
s
o
r
to
o
ls
tr
y
t
o
f
i
n
d
p
r
i
m
ar
y
co
d
e
s
m
ell
in
s
t
an
ce
s
s
u
ch
as
L
ar
g
e
C
las
s
,
L
az
y
C
la
s
s
,
Data
C
la
s
s
,
P
ar
allel
I
n
h
er
ita
n
ce
Hier
ar
ch
ies,
Go
d
C
las
s
,
a
n
d
Feat
u
r
e
E
n
v
y
.
T
h
e
p
r
o
p
o
s
ed
d
etec
tio
n
s
y
s
te
m
al
s
o
atte
m
p
t
s
to
f
in
d
all
o
f
th
e
m
i
n
J
av
a
p
r
o
j
ec
ts
.
So
m
e
s
tu
d
ies
h
av
e
d
i
v
ed
in
to
ev
al
u
ati
n
g
co
d
e
s
m
ell
d
etec
tio
n
to
o
ls
w
it
h
th
e
ir
o
w
n
as
s
es
s
m
e
n
t
s
ta
n
d
ar
d
s
.
T
h
ey
s
tu
d
ied
d
if
f
er
en
t
c
o
d
e
s
m
ell
d
etec
tio
n
to
o
ls
an
d
ap
p
lied
th
e
m
to
f
i
n
d
co
d
e
s
m
e
lls
a
g
ain
s
t
ap
p
licatio
n
s
.
Neu
r
al
n
e
t
w
o
r
k
m
o
d
el
s
h
a
v
e
b
ee
n
u
s
ed
in
v
ar
io
u
s
d
o
m
ai
n
s
s
u
c
h
as
n
at
u
r
al
la
n
g
u
a
g
e
p
r
o
ce
s
s
in
g
,
in
f
o
r
m
atio
n
r
etr
ie
v
al,
co
m
p
u
ter
v
is
io
n
,
an
d
g
en
e
p
r
ed
icti
o
n
[
1
1
,
1
2
,
1
3
,
1
4
]
.
T
h
e
s
o
f
t
w
ar
e
en
g
i
n
ee
r
i
n
g
co
m
m
u
n
it
y
u
s
es
t
h
e
r
esear
ch
o
u
tco
m
es
o
f
n
e
u
r
al
n
et
w
o
r
k
s
to
ad
d
r
ess
in
tr
ac
tab
lean
d
p
r
ac
ticall
y
i
n
f
ea
s
ib
le
p
r
o
b
lem
s
.
I
n
f
ac
t,
p
o
p
u
lar
m
o
d
els
f
r
eq
u
e
n
tl
y
u
s
ed
in
co
m
p
u
ter
v
i
s
io
n
co
u
ld
b
e
ap
p
lied
f
o
r
p
r
o
g
r
a
m
m
i
n
g
ap
p
licatio
n
s
s
u
c
h
as
s
tati
s
ti
ca
l
p
r
o
g
r
a
m
s
y
n
th
e
s
is
an
d
co
d
e
r
ef
ac
to
r
in
g
[
1
5
,
1
6
,
1
7
]
.
W
e
also
u
s
e
th
e
n
et
w
o
r
k
m
o
d
el
to
d
etec
t
a
n
d
p
r
ed
ict
b
ad
co
d
e
s
m
ell
s
i
n
o
b
j
ec
t
-
o
r
ien
ted
s
o
f
t
w
ar
e
s
y
s
te
m
s
.
J
asp
r
ee
t
k
au
r
a
n
d
et
al.
[
1
8
]
h
av
e
p
r
esen
ted
a
b
ad
s
m
ell
d
etec
tio
n
m
e
th
o
d
o
lo
g
y
w
it
h
a
d
esig
n
o
f
n
e
u
r
al
n
et
w
o
r
k
m
o
d
el
u
s
in
g
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ics.
T
h
e
n
eu
r
al
n
et
w
o
r
k
m
o
d
el
w
as
ap
p
lied
t
o
f
in
d
t
w
e
lv
e
b
ad
s
m
ell
s
ag
a
in
s
t
t
w
o
d
if
f
er
en
t
v
er
s
io
n
s
o
f
a
J
av
aScr
ip
t
en
g
in
e
ca
lled
R
h
i
n
o
.
Fo
r
th
e
b
etter
ac
cu
r
ac
y
,
th
e
s
a
m
e
m
o
d
el
w
a
s
tr
ain
ed
an
d
test
ed
u
s
i
n
g
d
if
f
er
en
t e
p
o
ch
s
.
T
h
e
ex
p
er
im
e
n
tal
r
e
s
u
l
ts
d
e
m
o
n
s
tr
ated
th
e
r
elatio
n
s
h
ip
b
et
w
ee
n
b
ad
s
m
el
ls
a
n
d
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ics.
Ho
w
e
v
er
,
th
eir
n
e
u
r
al
n
et
w
o
r
k
m
o
d
el
is
to
o
s
i
m
p
le
to
co
v
er
m
u
c
h
la
r
g
er
J
av
a
p
r
o
g
r
am
s
s
in
ce
it
h
as
o
n
l
y
o
n
e
h
id
d
en
la
y
er
.
I
n
ad
d
itio
n
,
th
e
m
o
d
el
w
a
s
test
ed
a
g
ai
n
s
t
o
n
l
y
s
i
n
g
le
J
av
a
ap
p
licatio
n
.
I
t
is
a
r
ea
s
o
n
ab
le
d
o
u
b
t th
at
th
eir
a
p
p
r
o
ac
h
m
a
y
h
a
v
e
a
li
m
itatio
n
o
n
th
e
s
ca
lab
il
it
y
.
5.
CO
NCLU
SI
O
N
S A
ND
F
UT
URE WO
RK
A
co
d
e
s
m
el
l
d
etec
tio
n
s
y
s
te
m
h
as
b
ee
n
p
r
esen
ted
to
f
in
d
m
aj
o
r
b
ad
co
d
e
s
m
ell
s
e
f
f
ec
ti
v
el
y
w
it
h
a
n
ar
tif
icial
n
eu
r
al
n
et
w
o
r
k
m
o
d
el
th
at
h
as
b
ee
n
tr
ai
n
ed
an
d
t
ested
w
i
th
t
h
e
t
w
e
n
t
y
p
o
p
u
lar
J
av
a
p
r
o
j
ec
t
f
r
o
m
GitHu
b
.
Fo
r
th
e
b
etter
p
er
f
o
r
m
an
ce
a
n
d
o
p
ti
m
izat
io
n
,
t
h
e
p
r
o
p
o
s
ed
n
et
w
o
r
k
m
o
d
el
h
as
b
ee
n
e
v
al
u
ated
o
n
m
u
ltip
le
ep
o
ch
s
a
n
d
h
id
d
en
l
a
y
er
s
.
T
h
e
p
r
esen
ted
s
o
l
u
tio
n
ap
p
lies
th
e
n
eu
r
al
n
et
w
o
r
k
m
o
d
el
f
o
r
co
d
e
s
m
e
ll
d
etec
tio
n
w
h
ic
h
is
o
n
e
o
f
t
h
e
ty
p
ical
s
o
f
t
w
ar
e
en
g
i
n
ee
r
in
g
p
r
o
b
lem
s
b
y
v
alid
ati
n
g
th
e
m
o
d
el
w
it
h
m
ass
iv
e
J
av
a
p
r
o
g
r
a
m
s
.
T
h
e
p
r
o
m
i
s
in
g
r
es
u
lt
s
f
r
o
m
t
h
e
ca
s
e
s
t
u
d
ie
s
s
u
g
g
e
s
t
t
h
at
n
e
u
r
al
n
e
t
w
o
r
k
m
o
d
els
ca
n
b
e
a
n
ess
e
n
tial
p
ar
t
o
f
co
d
e
s
m
ell
d
etec
tio
n
to
o
ls
b
y
id
en
ti
f
y
in
g
r
ef
ac
to
r
in
g
ar
ea
s
w
it
h
o
b
j
e
ct
-
o
r
ien
ted
m
etr
ics
.
Fu
r
t
h
er
m
o
r
e,
an
o
t
h
er
f
i
n
d
in
g
o
f
th
e
c
a
s
e
s
t
u
d
ies
i
s
t
h
at
o
b
j
e
ct
-
o
r
ien
ted
m
etr
ics
h
a
v
e
a
clo
s
e
r
elatio
n
s
h
ip
w
it
h
b
ad
co
d
e
s
m
ell
s
.
As
a
f
u
t
u
r
e
w
o
r
k
,
m
o
r
e
co
d
e
s
m
ell
s
w
i
l
l
b
e
ex
p
lo
r
ed
to
e
x
te
n
d
t
h
e
f
u
n
ct
io
n
alit
y
o
f
th
e
p
r
o
p
o
s
ed
d
etec
tio
n
s
y
s
te
m
w
it
h
m
o
r
e
p
r
ec
is
e
th
r
es
h
o
ld
s
o
f
t
h
e
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ic
s
.
B
ased
o
n
th
e
f
i
n
d
in
g
s
o
f
th
i
s
s
t
u
d
y
,
t
h
e
co
d
e
s
m
e
ll
d
etec
tio
n
s
y
s
te
m
w
ill
b
e
ap
p
lied
f
o
r
o
th
er
o
b
j
ec
t
-
o
r
ie
n
ted
p
r
o
g
r
a
m
m
in
g
lan
g
u
a
g
es
s
u
c
h
as
C
++
an
d
C
#
.
T
h
e
tr
an
s
itio
n
to
o
th
er
o
b
j
ec
t
-
o
r
ien
ted
s
y
s
te
m
s
m
a
y
b
e
s
tr
aig
h
t
f
o
r
w
ar
d
b
ec
au
s
e
th
e
y
ca
n
b
e
ex
p
o
s
ed
to
th
e
s
a
m
e
co
d
e
s
m
ell
s
.
T
h
e
p
r
o
p
o
s
ed
d
etec
tio
n
s
y
s
te
m
w
il
l
b
e
i
m
p
r
o
v
ed
b
y
u
s
i
n
g
m
o
r
e
o
b
j
ec
t
-
o
r
ien
ted
m
etr
ics
in
ter
m
s
o
f
ac
cu
r
ac
y
an
d
ef
f
ec
ti
v
e
n
ess
.
I
n
ad
d
itio
n
,
f
u
tu
r
e
r
esear
ch
m
ig
h
t
co
v
er
m
at
h
e
m
a
tical
f
o
r
m
alis
m
s
to
r
ep
r
esen
t th
e
r
elatio
n
s
h
ip
b
et
w
ee
n
m
etr
ic
s
an
d
co
d
e
s
m
e
lls
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
C
E
I
SS
N:
2
0
8
8
-
8708
F
in
d
in
g
B
a
d
C
o
d
e
S
mells w
ith
N
eu
r
a
l Netw
o
r
k
Mo
d
els
(
Do
n
g
K
w
a
n
K
im
)
3621
RE
F
E
R
E
NC
E
S
[1
]
S
o
f
tw
a
r
e
En
g
in
e
e
rin
g
S
tan
d
a
r
d
s
Co
m
m
it
tee
o
f
th
e
IEE
E,
IE
EE
S
ta
n
d
a
r
d
f
o
r S
o
ft
w
a
re
M
a
in
ten
a
n
c
e
,
Ju
n
e
1
9
9
8
.
[2
]
M
.
F
o
w
ler,
e
t
a
l
,
Re
f
a
c
to
rin
g
:
Im
p
ro
v
in
g
t
h
e
De
sig
n
o
f
Ex
isti
n
g
Co
d
e
.
A
d
d
iso
n
-
W
e
sle
y
,
1
9
9
9
.
[3
]
M
.
A
ll
a
m
a
n
is,
e
t
a
l.
,
“
S
u
g
g
e
stin
g
A
c
c
u
ra
te
M
e
th
o
d
a
n
d
Clas
s
Na
m
e
s
”
,
Pro
c
e
e
d
in
g
s o
f
t
h
e
2
0
1
5
1
0
th
J
o
in
t
M
e
e
ti
n
g
o
n
F
o
u
n
d
a
ti
o
n
s
o
f
S
o
f
twa
re
En
g
i
n
e
e
rin
g
,
2
0
1
5
,
p
p
.
38
-
4
9
.
[4
]
R.
Ku
m
a
r
,
e
t
a
l
.
,
“
A
n
E
m
p
iri
c
a
l
S
tu
d
y
o
f
Ba
d
S
m
e
ll
in
Co
d
e
o
n
M
a
in
ten
a
n
c
e
Eff
o
rt”
,
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
Co
mp
u
ter
S
c
ien
c
e
En
g
in
e
e
rin
g
,
v
o
l.
5
,
2
0
1
6
.
[5
]
M.
T
u
f
a
n
o
,
e
t
a
l
.
,
“
W
h
e
n
a
n
d
W
h
y
Yo
u
r
C
o
d
e
S
tarts
t
o
S
m
e
ll
Ba
d
”
,
Pr
o
c
e
e
d
in
g
s
o
f
t
h
e
3
7
t
h
I
n
ter
n
a
t
io
n
a
l
Co
n
fer
e
n
c
e
o
n
S
o
ft
w
a
re
En
g
in
e
e
rin
g
,
M
a
y
2
0
1
5
.
[6
]
T
.
G
.
S
.
F
il
ó
,
e
t
a
l
.
,
“
A
Ca
talo
g
u
e
o
f
T
h
re
sh
o
ld
s
f
o
r
Ob
jec
t
-
Orie
n
ted
S
o
f
tw
a
re
M
e
tri
c
s”
,
T
h
e
Fi
rs
t
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
Ad
v
a
n
c
e
s a
n
d
T
re
n
d
s i
n
S
o
ft
w
a
re
En
g
in
e
e
rin
g
,
A
p
r
il
2
0
1
5
.
[7
]
N.
Ka
y
a
r
v
izh
y
,
“
S
y
ste
m
a
ti
c
Re
v
ie
w
o
f
Ob
jec
t
Ori
e
n
ted
M
e
tri
c
T
o
o
ls
”
,
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Co
mp
u
ter
Ap
p
li
c
a
ti
o
n
s,
v
o
l
.
1
3
5
,
F
e
b
2
0
1
6
.
[8
]
F.
A
.
F
o
n
ta
n
a
,
e
t
a
l
.
,
“
A
u
to
m
a
ti
c
m
e
tri
c
th
re
sh
o
l
d
s
d
e
r
iv
a
ti
o
n
f
o
rc
o
d
e
sm
e
ll
d
e
tec
ti
o
n
”
,
In
ter
n
a
ti
o
n
a
l
W
o
rk
sh
o
p
o
n
Eme
rg
in
g
T
re
n
d
s i
n
S
o
f
twa
re
M
e
trics
,
M
a
y
2
0
1
5
.
[9
]
A
.
Ha
m
id
,
e
t
a
l
.
,
“
A
Co
m
p
a
ra
ti
v
e
S
tu
d
y
o
n
Co
d
e
S
m
e
ll
De
te
c
ti
o
n
T
o
o
ls”
,
In
t’l
J
o
u
r
n
a
l
o
f
A
d
v
a
n
c
e
d
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l.
6
0
,
2
0
1
3
.
[1
0
]
Di
W
u
,
e
t
a
l
.
,
“
A
m
e
tri
c
s
-
b
a
se
d
c
o
m
p
a
ra
ti
v
e
stu
d
y
o
n
o
b
jec
t
-
o
rien
ted
p
ro
g
ra
m
m
in
g
lan
g
u
a
g
e
s
”
,
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
S
o
ft
w
a
re
En
g
in
e
e
rin
g
a
n
d
K
n
o
wled
g
e
E
n
g
i
n
e
e
rin
g
,
Ju
ly
2
0
1
5
.
[1
1
]
Ch
a
n
d
ra
h
a
s M
ish
ra
,
e
t
a
l
.
,
“
De
e
p
M
a
c
h
in
e
L
e
a
rn
in
g
a
n
d
Ne
u
ra
l
N
e
tw
o
rk
s:
A
n
O
v
e
rv
i
e
w
”
,
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.
6
,
n
o
.
2
,
Ju
n
e
2
0
1
7
.
[1
2
]
S
a
li
su
M
u
sa
Bo
ro
d
o
,
e
t
a
l
.
,
“
Bi
g
Da
ta
P
latf
o
r
m
s
a
n
d
T
e
c
h
n
iq
u
e
s
”
,
In
d
o
n
e
sia
n
J
o
u
rn
a
l
o
f
El
e
c
trica
l
En
g
in
e
e
rin
g
and
Co
m
p
u
ter
S
c
ie
n
c
e
,
v
o
l.
1
,
n
o
.
1
,
Ja
n
u
a
ry
2
0
1
6
.
[1
3
]
P.
Bielik
,
e
t
a
l
.
,
“
P
ro
g
ra
m
m
in
g
w
it
h
Big
Co
d
e
:
L
e
ss
o
n
s,
T
e
c
h
n
iq
u
e
s
a
n
d
A
p
p
li
c
a
ti
o
n
s
”
,
T
h
e
In
a
u
g
u
r
a
l
S
u
mm
i
t
o
n
Ad
v
a
n
c
e
sin
Pro
g
ra
mm
in
g
L
a
n
g
u
a
g
e
s
,
2
0
1
5
[1
4
]
L
i
G
u
o
p
in
g
,
e
t
a
l
.
,
“
Co
m
b
in
a
ti
o
n
o
f
F
a
u
lt
T
re
e
a
n
d
Ne
u
ra
l
Ne
t
w
o
rk
s
in
Ex
c
a
v
a
to
r
Dia
g
n
o
sis
”
,
In
d
o
n
e
sia
n
J
o
u
r
n
a
l
o
f
El
e
c
trica
l
En
g
in
e
e
rin
g
a
n
d
Co
mp
u
ter
S
c
ien
c
e
,
v
o
l
.
1
1
,
n
o
.
4
,
A
p
ril
2
0
1
3
.
[1
5
]
P.
Bielik
,
e
t
a
l
.
,
“
P
HO
G
:
P
ro
b
a
b
il
isti
c
M
o
d
e
l
f
o
r
Co
d
e
”
,
I
n
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
M
a
c
h
in
e
L
e
a
r
n
in
g
,
2
0
1
6
.
[1
6
]
M
.
W
h
it
e
,
e
t
a
l
.
,
“
T
o
w
a
rd
D
e
e
p
L
e
a
rn
in
g
S
o
f
twa
re
Re
p
o
sito
ries
”
,
In
t’l
Co
n
fer
e
n
c
e
o
n
M
i
n
in
g
S
o
ft
w
a
re
Rep
o
sit
o
rie
s
,
2
0
1
5
.
[1
7
]
H.K.
Da
m
,
e
t
a
l
.,
“
A
d
e
e
p
l
a
n
g
u
a
g
e
m
o
d
e
l
f
o
r
so
f
t
w
a
re
c
o
d
e
”
,
W
o
r
k
sh
o
p
o
n
Na
t
u
ra
l
n
e
ss
o
f
S
o
ft
w
a
re
,
A
u
g
2
0
1
6
.
[1
8
]
J.Ka
u
r
,
e
t
a
l.
,
“
Ne
u
ra
l
Ne
t
w
o
rk
b
a
se
d
Re
fa
c
to
rin
g
A
re
a
Id
e
n
ti
f
ic
a
ti
o
n
in
S
o
f
twa
re
S
y
ste
m
w
it
h
Ob
jec
t
Orie
n
ted
M
e
tri
c
s”
,
In
d
i
a
n
J
o
u
r
n
a
l
o
f
S
c
ien
c
e
&
T
e
c
h
n
o
lo
g
y
,
v
o
l
.
9
,
M
a
rc
h
2
0
1
6
.
Evaluation Warning : The document was created with Spire.PDF for Python.