I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
1
4
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
5
,
p
p
.
62
~
72
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
1
4
.i
1
.
p
p
62
-
72
62
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
O
v
ercom
ing
imb
a
la
nced rice
se
ed
g
ermina
tion cla
ss
ificatio
n:
enha
ncing
accura
cy
f
o
r effec
tive se
edling
iden
tif
ica
ti
o
n
M
uh
la
s
a
h No
v
it
a
s
a
ri
M
a
ra
1
,
Sid
iq Sy
a
m
s
ul H
ida
y
a
t
1
,
F
a
rik
a
T
o
no
P
utr
i
2
,
Dwi
Ra
hm
a
wa
t
i
3
,
Sri Wa
hy
un
i
4
,
M
uh
a
m
a
d Ca
hy
o
Ardi
P
ra
bo
wo
1
,
No
er
Ni
'm
a
t
Sy
a
m
s
ul K
a
bir
5
,
Ra
g
il
T
ri
I
n
dra
wa
t
i
2
1
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
En
g
i
n
e
e
r
i
n
g
,
P
o
l
i
t
e
k
n
i
k
N
e
g
e
r
i
S
e
m
a
r
a
n
g
,
S
e
m
a
r
a
n
g
,
I
n
d
o
n
e
si
a
2
D
e
p
a
r
t
me
n
t
o
f
M
e
c
h
a
n
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
,
P
o
l
i
t
e
k
n
i
k
N
e
g
e
r
i
S
e
mara
n
g
,
S
e
m
a
r
a
n
g
,
I
n
d
o
n
e
si
a
3
D
e
p
a
r
t
me
n
t
o
f
A
g
r
i
c
u
l
t
u
r
a
l
P
r
o
d
u
c
t
i
o
n
,
P
o
l
i
t
e
k
n
i
k
N
e
g
e
r
i
J
e
mb
e
r
,
J
e
m
b
e
r
,
I
n
d
o
n
e
si
a
4
R
e
s
e
a
r
c
h
C
e
n
t
r
e
f
o
r
F
o
o
d
C
o
r
p
s
,
N
a
t
i
o
n
a
l
R
e
se
a
r
c
h
a
n
d
I
n
n
o
v
a
t
i
o
n
A
g
e
n
c
y
(
B
R
I
N
)
,
B
o
g
o
r
,
I
n
d
o
n
e
si
a
5
D
e
p
a
r
t
me
n
t
o
f
R
e
sea
r
c
h
a
n
d
D
e
v
e
l
o
p
me
n
t
,
M
o
n
e
v
I
n
d
o
n
e
s
i
a
,
M
a
k
a
ss
a
r
,
I
n
d
o
n
e
s
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
No
v
23
,
2
0
2
3
R
ev
is
ed
J
u
l
17
,
2
0
2
4
Acc
ep
ted
J
u
l
26
,
2
0
2
4
Th
is
st
u
d
y
a
ime
d
to
a
u
t
o
m
a
ti
c
a
ll
y
c
las
sify
rice
se
e
d
li
n
g
g
e
rm
in
a
ti
o
n
o
n
d
a
y
se
v
e
n
u
si
n
g
ima
g
e
a
n
a
ly
sis.
Th
e
c
a
teg
o
ries
in
c
l
u
d
e
d
n
o
rm
a
l,
a
b
n
o
rm
a
l,
a
n
d
d
e
a
d
se
e
d
s.
D
u
e
t
o
t
h
e
ra
rit
y
o
f
a
b
n
o
rm
a
l
se
e
d
li
n
g
s,
c
a
p
t
u
rin
g
th
e
ir
ima
g
e
s
re
su
lt
e
d
i
n
imb
a
lan
c
e
d
d
a
ta.
T
o
a
d
d
re
ss
t
h
is,
a
b
n
o
rm
a
l
c
a
teg
o
ries
we
re
c
o
m
b
in
e
d
i
n
to
a
si
n
g
le
c
las
s.
We
c
o
m
p
a
re
d
lo
g
isti
c
re
g
re
ss
io
n
,
ra
n
d
o
m
fo
re
st,
a
n
d
d
e
e
p
lea
rn
in
g
m
o
d
e
ls
(VG
G
1
9
,
VG
G
1
6
,
Ale
x
Ne
t)
fo
r
c
las
sifica
ti
o
n
.
S
u
rp
risi
n
g
ly
,
lo
g
isti
c
re
g
re
ss
io
n
a
c
h
iev
e
d
t
h
e
h
i
g
h
e
st
a
c
c
u
ra
c
y
(9
3
.
8
9
%
)
a
n
d
F1
-
sc
o
re
s
(0
.
9
6
n
o
rm
a
l,
0
.
8
1
a
b
n
o
rm
a
l)
d
e
sp
it
e
t
h
e
imb
a
lan
c
e
d
d
a
ta
a
n
d
c
o
m
p
lex
tas
k
.
T
h
e
e
ffe
c
ti
v
e
n
e
ss
o
f
lo
g
isti
c
re
g
re
ss
io
n
f
o
r
ric
e
se
e
d
li
n
g
c
las
sifica
ti
o
n
with
imb
a
lan
c
e
d
d
a
ta
h
a
s
b
e
e
n
d
e
m
o
n
stra
ted
i
n
th
is
n
o
v
e
l
re
se
a
rc
h
.
Histo
rica
ll
y
,
d
e
e
p
l
e
a
rn
i
n
g
m
o
d
e
ls d
o
m
i
n
a
te i
m
a
g
e
re
c
o
g
n
it
io
n
,
b
u
t
o
u
r
fin
d
i
n
g
s
su
g
g
e
st
sim
p
ler
m
o
d
e
ls
c
a
n
e
x
c
e
l
in
sp
e
c
ifi
c
s
c
e
n
a
rio
s,
e
sp
e
c
ially
with
li
m
it
e
d
d
a
ta
a
v
a
il
a
b
il
it
y
.
T
h
is
h
i
g
h
li
g
h
ts
t
h
e
imp
o
rta
n
c
e
o
f
se
lec
ti
n
g
m
o
d
e
ls
b
a
se
d
o
n
d
a
ta
c
h
a
ra
c
teristi
c
s.
Th
e
u
r
g
e
n
c
y
fo
r
th
is
re
se
a
rc
h
ste
m
s
fro
m
th
e
n
e
e
d
f
o
r
e
fficie
n
t
a
n
d
a
c
c
u
ra
te
rice
se
e
d
li
n
g
e
v
a
lu
a
ti
o
n
.
Im
p
ro
v
e
d
c
las
sifica
ti
o
n
c
a
n
e
n
h
a
n
c
e
a
g
ri
c
u
lt
u
ra
l
p
ra
c
ti
c
e
s
a
n
d
o
p
ti
m
ize
re
so
u
rc
e
a
ll
o
c
a
ti
o
n
.
K
ey
w
o
r
d
s
:
I
m
ag
e
class
if
icatio
n
L
o
g
is
tic
r
eg
r
ess
io
n
Ma
ch
in
e
lear
n
in
g
Mo
d
el
p
er
f
o
r
m
a
n
ce
Un
b
alan
ce
d
d
ata
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Sid
iq
Sy
am
s
u
l H
id
ay
at
Dep
ar
tm
en
t o
f
E
lectr
ical
E
n
g
i
n
ee
r
in
g
,
Po
litek
n
ik
Neg
er
i Se
m
ar
an
g
Sem
ar
an
g
5
0
2
7
5
,
C
en
tr
al
J
av
a
,
I
n
d
o
n
esia
E
m
ail:
s
id
iq
s
h
@
p
o
lin
es.a
c.
id
1.
I
NT
RO
D
UCT
I
O
N
R
ice
is
o
n
e
o
f
th
e
im
p
o
r
ta
n
t
f
o
o
d
co
m
m
o
d
ities
th
at
s
er
v
e
as
a
p
r
im
ar
y
f
o
o
d
s
o
u
r
ce
wo
r
ld
wid
e,
in
clu
d
in
g
in
I
n
d
o
n
esia
[
1
]
.
Me
etin
g
f
o
o
d
n
ee
d
s
s
u
s
tain
ab
ly
is
cr
u
cial
in
lin
e
w
ith
th
e
co
n
tin
u
o
u
s
in
cr
ea
s
e
in
p
o
p
u
latio
n
g
r
o
wth
.
T
h
e
av
aila
b
ilit
y
o
f
r
ice
in
th
e
co
m
m
u
n
ity
d
ep
en
d
s
o
n
cu
ltiv
atio
n
ef
f
o
r
ts
in
th
e
f
ield
.
Seed
s
g
er
m
in
atio
n
ar
e
o
n
e
o
f
th
e
ess
en
tial
elem
en
ts
in
p
lan
t
c
u
ltiv
atio
n
th
at
d
eter
m
in
e
p
r
o
d
u
ctio
n
ef
f
o
r
ts
an
d
y
ield
q
u
ality
[
2
]
.
Q
u
ality
s
ee
d
s
,
as
i
n
d
icate
d
b
y
ce
r
tific
atio
n
o
f
s
e
ed
s
g
er
m
in
atio
n
,
a
r
e
n
ec
ess
ar
y
.
Seed
ce
r
tific
atio
n
aim
s
to
m
ain
tain
th
e
g
en
etic,
p
h
y
s
ical,
an
d
p
h
y
s
io
lo
g
ical
p
u
r
ity
o
f
s
ee
d
s
u
s
ed
b
y
f
ar
m
er
s
t
h
r
o
u
g
h
m
o
n
ito
r
in
g
th
e
s
ee
d
p
r
o
d
u
ctio
n
p
r
o
ce
s
s
in
th
e
f
ield
,
s
ee
d
p
r
o
ce
s
s
in
g
,
l
ab
o
r
ato
r
y
s
ee
d
test
in
g
,
an
d
s
e
ed
lab
ellin
g
[
3
]
.
T
o
de
v
elo
p
s
u
p
er
io
r
s
ee
d
s
an
d
v
ar
ieties,
ef
f
o
r
ts
ar
e
n
ee
d
ed
to
d
et
ec
t
th
e
d
ev
elo
p
m
en
t
o
f
n
o
r
m
al
s
p
r
o
u
ts
ac
cu
r
atel
y
,
th
u
s
m
ak
in
g
th
e
p
r
o
ce
s
s
o
f
test
in
g
th
e
q
u
ality
o
f
r
ice
s
ee
d
s
a
n
d
s
ee
d
an
aly
s
is
f
aster
,
m
o
r
e
ef
f
ec
tiv
e,
a
n
d
m
o
r
e
ac
cu
r
ate.
As
o
f
th
e
cu
r
r
en
t
m
o
m
en
t,
th
e
p
r
ac
tice
o
f
o
b
s
er
v
in
g
an
d
tally
in
g
th
e
q
u
a
n
tity
o
f
n
o
r
m
al
s
p
r
o
u
ts
h
as
b
ee
n
co
n
d
u
cted
m
an
u
ally
b
y
lab
o
r
ato
r
y
an
aly
s
ts
.
Ho
wev
e
r
,
th
is
m
an
u
al
p
r
o
ce
d
u
r
e
is
s
u
s
ce
p
tib
le
to
er
r
o
r
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Ove
r
co
min
g
imb
a
la
n
ce
d
r
ice
s
ee
d
g
ermin
a
tio
n
cla
s
s
ifica
tio
n
…
(
Mu
h
la
s
a
h
N
o
vita
s
a
r
i Ma
r
a
)
63
ar
is
in
g
f
r
o
m
f
atig
u
e
f
ac
to
r
s
,
t
h
u
s
p
o
te
n
tially
co
m
p
r
o
m
is
in
g
th
e
in
te
g
r
ity
o
f
t
h
e
test
in
g
p
r
o
ce
s
s
.
Fu
r
th
er
m
o
r
e,
th
e
m
eth
o
d
is
tim
e
-
co
n
s
u
m
in
g
an
d
v
u
ln
er
ab
le
to
s
u
b
jectiv
ity
.
C
o
n
s
eq
u
en
tly
,
th
er
e
ar
is
es
a
n
ec
ess
ity
to
id
en
tify
a
m
o
r
e
e
x
p
ed
itio
u
s
,
p
r
ec
is
e,
an
d
s
tr
aig
h
tf
o
r
war
d
ap
p
r
o
ac
h
to
ascer
tain
n
o
r
m
al
s
p
r
o
u
t
s
.
T
h
er
ef
o
r
e,
it
is
n
ec
e
s
s
ar
y
to
f
in
d
a
m
eth
o
d
o
f
o
b
s
er
v
in
g
n
o
r
m
al
s
p
r
o
u
ts
th
at
is
f
aster
,
m
o
r
e
ac
cu
r
ate,
an
d
s
i
m
p
ler
.
Gen
ze
et
a
l
.
[
4
]
i
n
d
icate
th
at
t
h
e
u
s
e
o
f
im
ag
e
a
n
aly
s
i
s
in
s
ee
d
q
u
ality
test
in
g
s
h
o
ws
a
lo
we
r
er
r
o
r
r
ate
an
d
b
etter
p
er
f
o
r
m
a
n
ce
th
an
m
an
u
al
an
d
c
o
n
v
e
n
tio
n
al
m
et
h
o
d
s
.
T
h
e
class
if
icatio
n
o
f
r
ic
e
s
ee
d
lin
g
s
was b
ased
o
n
th
ei
r
g
er
m
in
atio
n
lev
els,
ca
teg
o
r
izin
g
th
em
as
n
o
r
m
al,
ab
n
o
r
m
al,
n
o
r
m
al
s
ee
d
lin
g
s
with
d
ec
ay
,
d
ea
d
s
ee
d
s
,
o
r
f
r
esh
s
ee
d
s
th
at
f
ailed
to
g
er
m
in
ate.
Du
r
in
g
th
is
r
esear
ch
,
th
e
team
en
co
u
n
ter
e
d
ch
allen
g
es
in
ca
p
tu
r
in
g
im
ag
es
o
f
th
e
r
ar
e
ca
te
g
o
r
ies,
i
n
clu
d
in
g
ab
n
o
r
m
al
s
ee
d
lin
g
s
,
d
ea
d
s
ee
d
s
,
n
o
r
m
al
s
ee
d
lin
g
s
with
d
ec
ay
,
a
n
d
f
r
esh
s
ee
d
lin
g
s
th
at
f
ailed
to
g
er
m
in
ate.
Han
d
lin
g
im
b
ala
n
ce
d
class
r
eq
u
ir
es
ap
p
r
o
p
r
iate
ap
p
r
o
ac
h
es
s
u
ch
as
o
v
er
s
am
p
lin
g
,
u
n
d
er
s
am
p
lin
g
,
to
alg
o
r
ith
m
ic
tec
h
n
iq
u
es
s
u
ch
as
s
in
g
le
class
lear
n
in
g
an
d
en
s
em
b
le
lear
n
in
g
.
T
o
o
v
er
co
m
e
th
e
p
r
o
b
lem
o
f
class
im
b
alan
ce
in
co
m
p
le
x
d
atasets
,
tech
n
iq
u
es
s
u
ch
as
d
e
ep
lear
n
i
n
g
ar
e
also
u
s
ed
[
5
]
.
T
h
er
ef
o
r
e,
th
is
s
tu
d
y
aim
s
to
ad
d
r
ess
th
e
is
s
u
e
o
f
class
im
b
alan
ce
in
th
e
class
if
icatio
n
o
f
r
ice
s
ee
d
g
er
m
in
atio
n
an
d
im
p
r
o
v
e
th
e
m
o
d
el'
s
p
er
f
o
r
m
a
n
ce
b
y
m
er
g
in
g
f
o
u
r
m
in
o
r
ity
class
es
in
to
a
s
in
g
le
ca
teg
o
r
y
ca
lled
'
Po
o
r
'
.
T
h
e
s
tu
d
y
u
tili
ze
d
n
in
e
class
if
icatio
n
alg
o
r
ith
m
s
,
in
clu
d
in
g
lo
g
is
tic
r
eg
r
ess
io
n
,
r
an
d
o
m
f
o
r
est,
k
-
n
ea
r
est
n
eig
h
b
o
u
r
s
(
KNN)
,
n
aïv
e
b
ay
esian
,
d
ec
is
io
n
tr
ee
,
ar
tific
ial
n
eu
r
al
n
etwo
r
k
(
AN
N)
,
an
d
c
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
(
C
NN
)
with
V
GG1
6
,
VGG1
9
,
an
d
Alex
n
et
ar
ch
itectu
r
es.
P
r
e
v
i
o
u
s
r
es
e
a
r
c
h
i
n
t
h
e
d
o
m
ain
o
f
s
e
e
d
c
l
as
s
i
f
i
c
a
ti
o
n
u
s
i
n
g
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
a
n
d
d
e
e
p
l
e
a
r
n
in
g
m
e
t
h
o
d
s
h
a
s
b
e
e
n
c
o
n
d
u
c
t
e
d
e
x
te
n
s
i
v
el
y
b
y
r
e
s
e
a
r
c
h
e
r
s
g
l
o
b
a
ll
y
.
T
h
e
s
e
l
ec
t
i
o
n
o
f
t
h
es
e
m
et
h
o
d
s
i
n
o
u
r
s
t
u
d
y
w
as
i
n
f
l
u
e
n
c
e
d
b
y
t
h
ei
r
s
u
c
c
e
s
s
f
u
l a
p
p
l
i
c
a
t
i
o
n
i
n
r
el
a
t
e
d
r
e
s
e
a
r
c
h
e
n
d
e
a
v
o
u
r
s
.
F
o
r
i
n
s
ta
n
c
e
,
P
r
a
b
o
w
o
a
n
d
H
i
d
a
y
at
[
6
]
p
r
o
p
o
s
e
a
s
o
lu
tio
n
to
ass
e
s
s
th
e
q
u
ality
o
f
r
ice
g
r
ain
s
b
y
a
n
al
y
zin
g
th
eir
s
h
ap
e
th
r
o
u
g
h
d
i
g
ital
im
ag
e
p
r
o
ce
s
s
in
g
tech
n
iq
u
es.
Hid
a
y
at
et
a
l
.
[
7
]
u
tili
ze
C
NN
alg
o
r
it
h
m
s
to
cl
ass
if
y
two
th
o
u
s
an
d
im
ag
es
i
n
to
two
ca
teg
o
r
ies:
s
u
p
er
io
r
s
ee
d
s
an
d
n
o
n
-
s
u
p
er
io
r
s
ee
d
s
.
T
h
ese
ex
p
er
im
en
tal
r
esu
lts
d
em
o
n
s
tr
ate
th
at
th
e
s
y
s
tem
ac
h
iev
es
a
p
r
ec
is
io
n
o
f
9
3
%
an
d
a
r
ec
all
o
f
9
5
%
in
class
if
y
in
g
s
u
p
e
r
io
r
an
d
n
o
n
-
s
u
p
er
i
o
r
s
ee
d
s
,
r
esp
e
ctiv
ely
[
8
]
.
C
r
ea
ted
th
e
r
ice
s
ee
d
g
e
r
m
in
atio
n
ev
al
u
atio
n
s
y
s
tem
(
R
SGES)
,
wh
ich
was
s
p
ec
if
ically
d
esig
n
ed
to
f
o
r
ec
ast
r
ice
s
ee
d
g
er
m
in
atio
n
u
s
in
g
d
ig
ital
im
ag
e
p
r
o
ce
s
s
in
g
an
d
ANN
tech
n
i
q
u
es.
R
SGES
ac
h
iev
ed
im
p
r
ess
iv
e
p
r
ec
is
io
n
r
ates,
with
o
n
ly
7
.
6
6
%
f
alse
ac
ce
p
tan
ce
an
d
5
.
4
2
%
f
alse
r
ejec
tio
n
,
wh
ile
p
r
o
ce
s
s
in
g
e
ac
h
im
ag
e
in
ju
s
t
8
.
3
1
s
ec
o
n
d
s
.
Gu
lzar
et
a
l.
[
9
]
d
is
cu
s
s
ed
th
e
d
ev
elo
p
m
e
n
t o
f
an
i
d
en
tific
atio
n
an
d
class
if
icatio
n
s
y
s
tem
f
o
r
f
o
u
r
teen
ty
p
es o
f
s
ee
d
s
u
s
in
g
th
e
VGG
1
6
ar
ch
it
ec
tu
r
e.
T
h
e
class
if
icatio
n
ac
c
u
r
ac
y
o
f
th
e
m
o
d
el
is
ex
ce
ed
i
n
g
ly
h
ig
h
,
r
ea
c
h
in
g
0
.
9
9
.
SVM
u
s
ed
fo
r
th
e
class
if
icatio
n
o
f
t
h
r
ee
ty
p
es
o
f
r
ice
g
r
ain
s
,
n
am
ely
b
asm
ati,
b
an
g
s
a
n
d
b
r
o
wn
r
ice
[
1
0
]
.
T
h
e
r
esu
lts
s
h
o
wed
a
h
ig
h
SV
M
m
o
d
el
ac
c
u
r
ac
y
o
f
0
.
9
2
.
Ki
r
atir
atan
ap
r
u
k
et
a
l.
[
1
1
]
class
i
f
ied
r
ice
s
ee
d
u
s
in
g
f
iv
e
m
ac
h
in
e
lear
n
in
g
tech
n
iq
u
es,
i.e
.
,
lo
g
is
tic
r
eg
r
ess
io
n
,
li
n
ea
r
d
is
cr
im
in
atio
n
an
aly
s
is
,
KNN,
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM)
,
an
d
C
NN.
T
h
o
s
e
s
tu
d
y
r
esu
lted
in
b
est
ac
c
u
r
ac
y
v
alu
e
at
9
5
.
1
5
%
f
o
r
C
NN
m
eth
o
d
.
H
o
wev
er
,
all
th
o
s
e
p
r
ev
io
u
s
s
tu
d
ies
[
8
]
–
[
1
1
]
d
etec
ted
s
ee
d
q
u
ality
u
s
in
g
o
n
ly
th
e
s
ee
d
’
s
p
r
o
f
ile,
i.e
.
,
co
lo
u
r
,
tex
tu
r
e
,
an
d
s
h
a
p
e.
T
h
e
p
r
esen
t
s
tu
d
y
class
if
ied
s
ee
d
q
u
ality
b
ased
o
n
its
g
er
m
in
atio
n
d
e
v
elo
p
m
e
n
t.
I
n
co
n
tr
ast,
o
u
r
s
tu
d
y
d
is
tin
g
u
is
h
es
its
elf
b
y
class
if
y
in
g
s
ee
d
q
u
ality
b
ased
o
n
g
er
m
in
atio
n
d
ev
el
o
p
m
e
n
t,
ad
d
r
ess
in
g
a
s
ig
n
if
ica
n
t
g
ap
in
p
r
io
r
r
esear
ch
.
Siv
ak
u
m
ar
et
a
l.
[
1
2
]
co
n
d
u
cted
a
co
m
p
r
eh
en
s
iv
e
an
aly
s
is
o
f
s
ee
d
g
e
r
m
in
atio
n
r
ates
b
y
u
tili
zin
g
n
e
u
r
al
n
etwo
r
k
s
(
NN)
an
d
co
m
p
ar
in
g
th
e
m
to
e
x
ten
s
iv
e
d
atasets
co
n
tain
i
n
g
g
er
m
in
ated
an
d
n
o
n
-
g
er
m
in
ated
s
am
p
les.
T
h
eir
a
p
p
r
o
ac
h
f
o
cu
s
ed
o
n
an
aly
s
in
g
th
e
g
er
m
i
n
ated
r
o
o
ts
o
f
s
e
ed
s
,
lead
in
g
to
th
e
d
ev
elo
p
m
e
n
t
o
f
th
e
m
u
lti
-
lev
e
l
r
o
o
t
m
et
r
ic
r
atio
(
ML
R
M)
m
eth
o
d
.
T
h
is
m
eth
o
d
h
as
s
h
o
w
n
p
r
o
m
is
in
g
r
esu
lts
in
ac
cu
r
ately
ass
ess
in
g
s
ee
d
g
er
m
in
atio
n
r
ates.
On
th
e
o
th
er
h
an
d
,
Hu
an
g
et
a
l
.
[
1
3
]
ex
p
lo
r
ed
th
e
ap
p
licatio
n
o
f
C
NN
an
d
tr
a
n
s
f
er
lear
n
in
g
f
o
r
s
ee
d
q
u
ality
class
if
icatio
n
.
T
h
eir
s
tu
d
y
also
in
v
o
lv
ed
a
co
m
p
a
r
is
o
n
with
tr
ad
itio
n
al
m
ac
h
in
e
lea
r
n
in
g
alg
o
r
ith
m
s
.
T
h
e
e
x
p
er
im
e
n
tal
f
in
d
i
n
g
s
r
e
v
ea
led
t
h
at
d
ee
p
l
ea
r
n
in
g
alg
o
r
it
h
m
s
,
p
ar
ticu
lar
ly
Go
o
g
le
Net,
o
u
tp
er
f
o
r
m
e
d
m
ac
h
i
n
e
lear
n
i
n
g
al
g
o
r
ith
m
s
s
ig
n
if
ican
tly
.
Go
o
g
l
e
Net
ac
h
iev
ed
an
im
p
r
ess
iv
e
ac
cu
r
ac
y
o
f
9
5
%,
wh
ile
th
e
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
(
s
p
ee
d
ed
u
p
r
o
b
u
s
t
f
ea
tu
r
es
(
SUR
F
)
+S
VM
)
ac
h
iev
ed
a
n
ac
c
u
r
ac
y
o
f
7
9
.
2
%.
T
h
is
s
h
o
wca
s
es
th
e
s
u
p
e
r
io
r
ca
p
a
b
ilit
ies
o
f
d
ee
p
lear
n
i
n
g
in
s
ee
d
q
u
ality
class
if
icatio
n
task
s
.
Mo
h
an
an
d
R
aj
[
1
4
]
in
tr
o
d
u
ce
d
a
n
au
to
m
ated
s
y
s
tem
to
id
en
tify
an
d
class
if
y
r
ice
s
ee
d
s
b
ased
o
n
d
ig
ital
im
ag
e
p
r
o
ce
s
s
in
g
tech
n
iq
u
es
with
SVM
an
d
NN
class
if
icatio
n
m
eth
o
d
s
.
T
h
e
r
esu
lts
s
h
o
wed
th
at
SVM
p
r
o
v
id
ed
a
h
ig
h
er
a
cc
u
r
ac
y
o
f
9
1
%
co
m
p
ar
ed
to
NN
with
an
ac
cu
r
ac
y
o
f
8
3
%.
M
ac
h
in
e
lear
n
in
g
m
eth
o
d
s
s
u
ch
as
l
o
g
is
tic
r
eg
r
ess
io
n
,
lin
ea
r
d
is
cr
im
in
an
t
a
n
aly
s
is
(
L
DA
)
,
K
NN,
SVM
,
an
d
d
ee
p
lear
n
in
g
tech
n
iq
u
es
VGG1
6
,
VGG1
9
,
Xce
p
tio
n
,
I
n
ce
p
tio
n
V3
,
I
n
ce
p
t
io
n
R
esNetV2
u
s
ed
to
class
if
y
f
o
u
r
teen
v
ar
ieties
o
f
Oryz
a
S
a
tiva
r
ice
[
1
1
]
.
T
h
e
r
es
u
lts
s
h
o
wed
th
e
I
n
ce
p
tio
n
R
esNetV2
m
o
d
el
g
av
e
t
h
e
h
i
g
h
est
ac
cu
r
ac
y
o
f
9
5
.
1
5
%.
Ah
m
ed
et
a
l.
[
1
5
]
d
iv
id
e
alg
o
r
ith
m
s
an
d
tech
n
iq
u
es f
o
r
im
a
g
e
-
b
ased
r
ice
class
if
icatio
n
an
d
g
r
ad
atio
n
i
n
to
f
iv
e
d
if
f
er
en
t
ap
p
r
o
ac
h
es,
n
am
ely
:
g
e
o
m
etr
ic,
s
tatis
tical,
tr
ain
ed
,
u
n
s
u
p
er
v
is
ed
,
an
d
d
ee
p
lear
n
in
g
.
Dee
p
lear
n
in
g
ap
p
r
o
ac
h
es
h
av
e
d
e
m
o
n
s
tr
ate
d
m
o
r
e
p
r
o
m
is
in
g
o
u
tco
m
es
an
d
ar
e
g
ai
n
in
g
p
o
p
u
lar
ity
f
o
r
f
u
tu
r
e
r
esear
ch
.
M
ac
h
in
e
lear
n
i
n
g
m
et
h
o
d
s
s
u
c
h
as
lo
g
is
tic
r
eg
r
ess
io
n
,
d
ec
is
i
o
n
tr
ee
,
SVM,
r
an
d
o
m
f
o
r
est
,
an
d
n
aï
v
e
b
a
y
esian
to
class
if
y
f
iv
e
r
ice
v
ar
ieties
n
am
ely
B
asm
ati,
Ar
b
o
r
io
,
J
asm
in
e,
I
p
s
ala
,
an
d
Kar
ac
ad
a
g
[
1
6
]
.
T
h
e
ex
p
er
im
en
tal
r
esu
lts
s
h
o
wed
th
at
r
an
d
o
m
f
o
r
est
h
ad
th
e
h
ig
h
est
ac
cu
r
ac
y
r
ea
ch
in
g
9
9
.
8
5
%,
n
o
t
m
u
c
h
d
if
f
er
en
t
f
r
o
m
th
e
ac
cu
r
ac
y
o
f
d
ec
is
io
n
tr
ee
s
wh
i
ch
was
9
9
.
6
8
%.
R
u
s
lan
et
a
l.
[
1
7
]
u
s
ed
im
ag
e
p
r
o
ce
s
s
in
g
an
d
m
ac
h
in
e
lear
n
in
g
tech
n
iq
u
es
to
class
if
y
wild
r
i
ce
s
ee
d
s
an
d
cu
ltiv
ated
r
ice
b
ased
o
n
m
o
r
p
h
o
lo
g
y
,
c
o
lo
u
r
,
an
d
tex
t
u
r
e.
T
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
5
:
62
-
72
64
m
ac
h
in
e
lear
n
in
g
m
et
h
o
d
s
u
s
ed
ar
e
d
ec
is
io
n
tr
ee
,
d
is
cr
im
in
an
t
an
aly
s
is
,
lo
g
is
tic
r
eg
r
ess
i
o
n
,
n
aïv
e
b
a
y
esian
,
SVM,
KNN
,
an
d
ess
em
b
ler
c
lass
if
ier
s
.
T
h
e
r
esu
lts
s
h
o
wed
th
at
co
lo
u
r
-
b
ased
class
if
icati
o
n
with
th
e
lo
g
is
tic
r
eg
r
ess
io
n
m
o
d
el
p
r
o
v
id
ed
t
h
e
h
ig
h
est ac
cu
r
ac
y
an
d
s
en
s
itiv
ity
,
n
am
ely
0
.
9
8
an
d
0
.
8
5
r
esp
e
ctiv
ely
.
R
esear
ch
r
elate
d
to
o
v
er
c
o
m
in
g
u
n
b
alan
ce
d
class
es
in
class
if
icatio
n
h
as
also
b
ee
n
wid
ely
c
o
n
d
u
cte
d
.
W
u
[
1
8
]
ex
p
l
o
r
es
th
e
co
n
ce
p
t
o
f
class
im
b
alan
ce
an
d
h
o
w
i
t
ca
n
lead
to
m
is
lead
in
g
r
esu
lt
s
lik
e
h
ig
h
o
v
er
all
ac
cu
r
ac
y
d
esp
ite
p
o
o
r
p
e
r
f
o
r
m
an
ce
o
n
m
in
o
r
ity
class
es.
Go
s
ain
an
d
Sar
d
an
a
[
1
9
]
co
n
d
u
cted
a
co
m
p
ar
is
o
n
o
f
d
if
f
er
en
t
o
v
er
s
am
p
lin
g
m
eth
o
d
s
,
in
clu
d
in
g
s
y
n
th
etic
m
in
o
r
ity
o
v
er
-
s
am
p
lin
g
tech
n
iq
u
e
(
SMOT
E
)
,
ad
ap
tiv
e
s
y
n
th
etic
s
am
p
lin
g
(
ADASY
N
)
,
B
o
r
d
er
lin
e
-
SMOT
E
,
an
d
s
af
e
-
lev
el
SMOT
E
,
u
s
in
g
v
ar
io
u
s
class
if
ier
s
.
T
h
eir
r
esear
ch
r
ev
ea
le
d
th
at
s
af
e
-
lev
el
SMOT
E
,
wh
ich
g
en
er
ates
m
in
o
r
ity
i
n
s
tan
ce
s
m
o
r
e
p
r
o
m
in
en
tly
ar
o
u
n
d
lar
g
er
s
af
e
lev
els,
o
u
tp
er
f
o
r
m
s
SMOT
E
,
ADASYN,
an
d
B
o
r
d
er
lin
e
SMOT
E
in
ter
m
s
o
f
ac
cu
r
ac
y
p
er
f
o
r
m
an
ce
.
A
n
o
v
er
v
iew
o
f
th
e
liter
atu
r
e
[
2
0
]
s
p
an
n
in
g
th
e
y
ea
r
s
2
0
0
0
to
2
0
1
6
,
f
o
cu
s
in
g
o
n
d
if
f
er
en
t
ap
p
r
o
ac
h
es
ass
o
ciate
d
with
r
an
d
o
m
f
o
r
est
class
if
icati
o
n
f
o
r
ad
d
r
ess
in
g
class
im
b
alan
ce
.
T
h
e
s
ch
o
lar
ly
r
e
v
iew
d
eli
n
ea
tes
th
at
th
er
e
ar
e
two
ef
f
ec
tiv
e
m
eth
o
d
s
,
b
alan
c
ed
r
an
d
o
m
f
o
r
est
an
d
weig
h
te
d
r
an
d
o
m
f
o
r
est
,
f
o
r
ad
d
r
ess
in
g
th
e
is
s
u
e
o
f
class
im
b
alan
ce
.
A
liter
atu
r
e
r
ev
iew
co
n
d
u
cte
d
[
2
1
]
o
n
t
h
e
u
s
e
o
f
d
ee
p
lear
n
in
g
in
h
a
n
d
lin
g
u
n
b
alan
ce
d
cl
ass
r
o
o
m
d
ata.
T
h
e
r
esu
lts
s
h
o
w
th
at
d
e
ep
lear
n
in
g
m
eth
o
d
s
,
s
u
ch
as
C
NN
an
d
co
s
t
-
s
en
s
itiv
e
lear
n
in
g
,
ar
e
ef
f
ec
tiv
e
in
ad
d
r
ess
in
g
class
r
o
o
m
im
b
alan
ce
is
s
u
es.
Ma
h
m
u
d
ah
et
a
l
.
[
2
2
]
to
o
k
a
p
r
e
-
p
r
o
ce
s
s
in
g
a
p
p
r
o
ac
h
b
y
co
n
v
er
tin
g
b
in
ar
y
f
ea
tu
r
es
in
t
o
n
u
m
er
ical
f
ea
tu
r
es
u
s
in
g
th
e
f
ea
tu
r
e
ex
tr
ac
tio
n
m
eth
o
d
to
o
v
e
r
co
m
e
th
e
lim
itatio
n
s
o
f
th
e
ove
r
s
am
p
lin
g
m
eth
o
d
i
n
g
en
e
r
atin
g
d
iv
er
s
e
n
ew
s
am
p
les
o
f
b
in
ar
y
f
ea
tu
r
es.
T
h
e
r
esu
lts
in
d
icate
d
th
at
th
e
co
m
b
in
atio
n
o
f
r
an
d
o
m
f
o
r
e
s
t
an
d
o
v
er
s
am
p
lin
g
with
th
e
r
elo
ca
tin
g
-
s
af
e
-
lev
el
SMOT
E
(
R
SLS)
m
eth
o
d
p
r
o
v
id
e
d
t
h
e
h
i
g
h
est
p
e
r
f
o
r
m
an
ce
o
n
all
d
atasets
test
ed
.
Ho
wev
er
,
th
er
e
a
r
e
v
ar
iatio
n
s
in
t
h
e
b
est
f
ea
tu
r
e
ex
tr
ac
tio
n
m
eth
o
d
f
o
r
ea
ch
d
a
taset,
th
is
is
e
s
tim
ated
b
ec
au
s
e
it
is
af
f
ec
ted
b
y
th
e
im
b
alan
ce
r
atio
.
Alth
o
u
g
h
m
o
s
t
p
er
f
o
r
m
a
n
ce
im
p
r
o
v
e
m
e
n
ts
ar
e
n
o
t
s
tatis
tically
s
ig
n
if
ic
an
t,
th
er
e
is
a
s
ig
n
if
ican
t
im
p
r
o
v
em
en
t
i
n
ac
c
u
r
ac
y
with
th
e
u
s
e
o
f
t
-
d
is
tr
ib
u
ted
s
to
ch
asti
c
n
eig
h
b
o
r
em
b
ed
d
in
g
(
t
-
SNE)
f
ea
tu
r
e
ex
tr
ac
tio
n
.
K
u
m
ar
et
a
l
.
[
2
3
]
o
f
v
ar
io
u
s
ap
p
r
o
ac
h
es
p
r
esen
ted
to
class
if
y
in
g
u
n
b
alan
ce
d
d
a
tasets
an
d
th
eir
ap
p
licatio
n
s
.
T
h
e
r
esu
lts
o
f
th
e
r
ev
iew
s
h
o
w
th
at
th
e
s
am
p
lin
g
m
eth
o
d
is
ea
s
y
to
im
p
lem
en
t,
b
u
t
in
r
ea
l
-
wo
r
ld
ap
p
licatio
n
s
in
v
o
lv
in
g
b
iased
d
ata
d
is
tr
ib
u
tio
n
,
a
h
y
b
r
i
d
a
p
p
r
o
ac
h
is
also
r
eq
u
ir
ed
.
U
n
b
ala
n
ce
d
d
ata
class
if
icatio
n
is
th
e
s
u
b
ject
o
f
ex
ten
s
iv
e
r
esear
ch
in
m
ac
h
in
e
lear
n
in
g
.
Fen
g
et
a
l.
[
2
4
]
p
r
o
v
id
e
d
g
u
id
an
ce
in
ch
o
o
s
in
g
th
e
r
ig
h
t
r
es
am
p
lin
g
tech
n
i
q
u
e
an
d
class
if
icatio
n
m
eth
o
d
f
o
r
class
if
icatio
n
p
r
o
b
lem
s
with
cl
ass
im
b
alan
ce
.
T
h
e
r
an
d
o
m
f
o
r
est
alg
o
r
ith
m
u
s
ed
o
n
th
e
Go
o
g
le
E
ar
th
E
n
g
in
e
p
latf
o
r
m
to
class
if
y
r
ice
p
h
en
o
lo
g
y
f
r
o
m
l
an
d
s
at
-
8
s
atel
lite
im
ag
er
y
[
2
5
]
.
Ov
er
s
am
p
lin
g
tech
n
iq
u
es
ar
e
u
s
ed
to
o
v
e
r
co
m
e
th
e
p
r
o
b
lem
o
f
d
ata
im
b
ala
n
ce
.
T
h
e
r
esu
lts
s
h
o
wed
th
e
ac
cu
r
ac
y
o
f
r
a
n
d
o
m
f
o
r
est
with
o
v
er
s
am
p
lin
g
is
b
etter
t
h
an
r
an
d
o
m
f
o
r
est
with
o
u
t
o
v
er
s
am
p
lin
g
.
Z
h
en
g
et
a
l
.
[
2
6
]
p
r
o
v
id
es
g
u
id
an
ce
f
o
r
s
elec
tin
g
ap
p
r
o
p
r
iate
m
eth
o
d
s
in
o
v
er
co
m
i
n
g
th
e
p
r
o
b
lem
o
f
u
n
b
alan
ce
d
d
ata
class
if
icatio
n
.
A
co
m
p
r
eh
en
s
iv
e
in
v
esti
g
atio
n
co
n
d
u
cted
in
to
th
e
ch
allen
g
es
ass
o
ciate
d
with
a
d
d
r
ess
in
g
im
b
alan
ce
is
s
u
es
in
th
e
f
ield
o
f
o
b
ject
d
etec
tio
n
[
2
7
]
.
T
h
eir
r
esear
ch
r
ev
ea
ls
th
at
at
p
r
esen
t,
th
er
e
is
n
o
s
in
g
le
m
eth
o
d
th
at
ca
n
b
e
u
n
iv
er
s
ally
ap
p
lied
to
all
ar
ch
itectu
r
al
s
etu
p
s
to
ef
f
ec
tiv
ely
m
itig
ate
o
b
j
ec
tiv
e
im
b
alan
ce
s
.
T
h
e
af
o
r
e
m
en
tio
n
e
d
liter
atu
r
e
p
r
esen
ts
v
alu
ab
le
in
s
ig
h
ts
i
n
to
m
eth
o
d
o
lo
g
ies
f
o
r
a
d
d
r
es
s
in
g
class
im
b
alan
ce
in
d
if
f
e
r
e
n
t
d
o
m
ain
s
.
I
t
is
n
o
tab
le
t
h
at
th
er
e
is
a
s
ca
r
city
o
f
s
tu
d
ies
ad
d
r
ess
in
g
t
h
e
s
p
ec
if
ic
ch
allen
g
es
p
o
s
ed
b
y
im
b
alan
ce
d
d
atasets
in
r
ice
s
ee
d
g
er
m
in
atio
n
clas
s
if
icatio
n
.
Desp
ite
th
e
ab
u
n
d
a
n
ce
o
f
r
esear
ch
o
n
class
im
b
alan
ce
with
in
th
e
b
r
o
ad
er
f
ield
s
o
f
m
ac
h
i
n
e
lear
n
i
n
g
an
d
d
e
ep
lear
n
in
g
,
th
er
e
h
as
b
ee
n
n
o
p
r
e
v
io
u
s
in
v
esti
g
atio
n
in
to
a
d
d
r
ess
in
g
class
im
b
alan
ce
s
p
ec
if
ically
in
th
e
r
ea
lm
o
f
r
ice
s
ee
d
g
er
m
in
atio
n
-
b
ased
class
if
icatio
n
.
T
h
is
h
ig
h
lig
h
ts
a
s
ig
n
if
ican
t r
esear
ch
g
a
p
th
at
o
u
r
s
tu
d
y
s
ee
k
s
to
b
r
id
g
e.
2.
ME
T
H
O
D
I
n
th
is
s
tu
d
y
,
we
clas
s
if
ied
r
ice
s
ee
d
lin
g
s
b
ased
o
n
th
eir
g
er
m
in
atio
n
lev
els
in
to
f
iv
e
ca
teg
o
r
ies:
n
o
r
m
al,
a
b
n
o
r
m
al,
d
ea
d
s
ee
d
s
,
n
o
r
m
al
s
ee
d
lin
g
s
with
d
ec
a
y
,
an
d
f
r
esh
s
ee
d
s
th
at
f
ailed
t
o
g
er
m
in
ate
.
Fo
r
t
h
e
"No
r
m
al"
class
if
icatio
n
,
we
id
en
tifie
d
s
ev
er
al
cr
iter
ia
th
at
m
u
s
t
b
e
m
et
b
y
th
e
r
ice
s
ee
d
lin
g
s
.
Firstl
y
,
th
e
p
r
im
ar
y
r
o
o
t
m
u
s
t
g
r
o
w
s
tr
o
n
g
,
ac
c
o
m
p
an
ied
b
y
well
-
d
ev
elo
p
ed
s
ec
o
n
d
ar
y
r
o
o
ts
,
with
a
m
in
im
u
m
o
f
2
t
o
3
well
-
d
ev
elo
p
ed
s
ec
o
n
d
ar
y
s
em
in
al
r
o
o
ts
.
Ad
d
itio
n
ally
,
th
e
p
r
im
ar
y
leaf
m
u
s
t
g
r
o
w
alo
n
g
th
e
co
leo
p
tile
an
d
em
er
g
e
f
r
o
m
it
in
a
h
ea
lth
y
co
n
d
itio
n
.
T
h
e
p
lu
m
u
le,
th
e
p
a
r
t
o
f
th
e
p
lan
t
th
at
will
d
ev
elo
p
i
n
to
th
e
s
tem
,
m
u
s
t
also
b
e
p
r
esen
t
with
o
u
t
an
y
s
ig
n
s
o
f
d
ec
ay
.
Fu
r
th
er
m
o
r
e,
th
e
d
ev
elo
p
m
e
n
t
o
f
th
e
h
y
p
o
co
ty
l,
th
e
p
ar
t
o
f
th
e
s
tem
b
elo
w
th
e
c
o
t
y
led
o
n
s
,
m
u
s
t
b
e
p
er
f
ec
t
with
o
u
t
an
y
d
am
a
g
e
to
its
tis
s
u
es.
"No
r
m
al"
s
ee
d
lin
g
s
m
u
s
t
also
h
av
e
o
n
e
co
ty
le
d
o
n
f
o
r
m
o
n
o
co
t
s
ee
d
lin
g
s
an
d
two
co
t
y
led
o
n
s
f
o
r
d
ico
t
s
ee
d
lin
g
s
.
Fo
r
s
ee
d
s
with
ep
ig
ea
l
g
er
m
in
atio
n
ty
p
e,
th
e
len
g
th
o
f
th
e
r
o
o
t
s
h
o
u
ld
b
e
at
least
f
o
u
r
tim
es
th
e
len
g
th
o
f
th
e
s
ee
d
an
d
h
av
e
a
n
o
r
m
al
s
tr
u
ctu
r
al
d
ev
elo
p
m
en
t.
L
astl
y
,
s
ee
d
lin
g
s
th
at
h
av
e
b
e
en
d
a
m
ag
ed
d
u
e
to
in
f
ec
tio
n
b
y
o
th
er
s
ee
d
lin
g
s
ar
e
s
till
co
n
s
id
er
ed
n
o
r
m
al
as
lo
n
g
as
th
e
ess
en
tial
p
ar
ts
o
f
th
e
s
ee
d
lin
g
s
r
em
ai
n
in
tact.
Fo
r
th
e
"Ab
n
o
r
m
al”
class
if
icatio
n
,
s
ee
d
lin
g
s
m
u
s
t
m
ee
t
th
e
f
o
llo
win
g
cr
iter
ia.
F
ir
s
tly
,
n
o
p
r
im
ar
y
o
r
s
ec
o
n
d
ar
y
r
o
o
ts
g
r
o
w,
o
r
if
th
ey
g
r
o
w,
th
e
r
o
o
ts
ar
e
wea
k
an
d
s
h
o
r
t.
Ad
d
itio
n
ally
,
th
e
s
ee
d
lin
g
s
d
o
n
o
t
g
r
o
w
th
eir
f
ir
s
t
leaf
,
an
d
th
eir
co
leo
p
tile
is
co
lo
r
less
.
So
m
et
im
es,
th
e
p
lu
m
u
le
m
a
y
g
r
o
w
wh
ite
o
r
ex
p
e
r
ien
ce
d
ec
ay
.
Seed
lin
g
s
th
at
ar
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Ove
r
co
min
g
imb
a
la
n
ce
d
r
ice
s
ee
d
g
ermin
a
tio
n
cla
s
s
ifica
tio
n
…
(
Mu
h
la
s
a
h
N
o
vita
s
a
r
i Ma
r
a
)
65
d
am
ag
ed
,
lack
co
ty
le
d
o
n
s
,
em
b
r
y
o
s
,
a
n
d
s
h
o
r
t
p
r
im
ar
y
r
o
o
ts
ar
e
also
in
clu
d
ed
in
th
e
"A
b
n
o
r
m
al”
class
if
icatio
n
.
Mo
r
eo
v
er
,
s
ee
d
lin
g
s
s
h
o
win
g
d
ef
o
r
m
ities
,
wea
k
d
e
v
elo
p
m
en
t,
o
r
im
b
alan
ce
in
ess
en
tial
p
ar
ts
will
also
b
e
ca
teg
o
r
ized
as
"Ab
n
o
r
m
al”.
Seed
s
th
at
d
ec
ay
b
ef
o
r
e
g
er
m
in
atio
n
o
r
f
ail
to
g
r
o
w
with
in
th
e
s
p
ec
if
ied
test
in
g
p
er
io
d
,
b
u
t
ar
e
n
o
t
in
a
d
o
r
m
a
n
t
s
tate,
will
b
e
class
if
ied
as
"
Dea
d
s
ee
d
s
”.
T
h
e
"No
r
m
al
s
e
ed
lin
g
s
with
d
ec
ay
”
class
i
f
icatio
n
is
u
s
ed
f
o
r
s
ee
d
s
th
at
r
em
ain
h
ar
d
at
th
e
en
d
o
f
th
e
g
er
m
in
ati
o
n
test
d
u
e
to
n
o
t
ab
s
o
r
b
in
g
wate
r
,
ca
u
s
ed
b
y
an
im
p
e
r
m
ea
b
le
s
ee
d
co
at
th
at
h
in
d
er
s
th
e
g
e
r
m
in
atio
n
p
r
o
ce
s
s
.
L
astl
y
,
th
e
class
if
icatio
n
o
f
"Fr
esh
s
ee
d
s
th
at
f
ailed
to
g
er
m
in
ate
”
is
g
iv
e
n
t
o
s
ee
d
s
th
at
d
o
n
o
t
g
er
m
in
ate
b
y
th
e
e
n
d
o
f
t
h
e
test
in
g
p
er
i
o
d
b
u
t
s
till
h
av
e
th
e
p
o
ten
tial
to
g
r
o
w
n
o
r
m
ally
.
T
h
is
ty
p
e
o
f
s
ee
d
c
an
ab
s
o
r
b
wate
r
d
u
r
in
g
th
e
t
esti
n
g
p
r
o
ce
s
s
b
u
t
ex
p
er
ien
ce
s
o
b
s
tacle
s
in
f
u
r
th
er
d
ev
elo
p
m
en
t.
I
n
th
is
s
tu
d
y
,
we
test
ed
th
e
g
e
r
m
in
atio
n
o
f
r
ice
s
ee
d
s
f
o
r
7
d
ay
s
o
n
b
lo
ttin
g
p
a
p
er
.
Su
b
s
eq
u
en
tly
,
we
ca
p
tu
r
ed
im
ag
es
o
f
ea
c
h
s
ee
d
lin
g
.
Fig
u
r
e
1
s
h
o
ws
th
e
d
ata
ac
q
u
is
itio
n
p
r
o
ce
d
u
r
e
in
th
is
s
tu
d
y
.
T
o
ca
p
tu
r
e
h
ig
h
-
q
u
ality
im
ag
es
o
f
g
e
r
m
i
n
atin
g
r
ice
s
ee
d
lin
g
s
,
a
Nik
o
n
D5
6
0
0
d
ig
ital
ca
m
er
a
with
a
r
eso
lu
tio
n
o
f
2
4
.
2
m
eg
ap
ix
els
was
u
s
ed
f
o
r
d
ata
ac
q
u
is
itio
n
t
o
o
l.
T
h
is
ca
m
er
a
p
r
o
v
id
es
ex
ce
p
tio
n
al
im
a
g
e
q
u
ality
an
d
d
etail.
I
m
ag
es
tak
en
in
a
6
0
×
6
0
cm
Min
ib
o
x
s
tu
d
i
o
.
T
h
e
Min
ib
o
x
s
tu
d
io
p
r
o
v
id
es
co
n
tr
o
lled
lig
h
tin
g
co
n
d
itio
n
s
an
d
co
n
s
is
ten
t
b
ac
k
g
r
o
u
n
d
s
,
en
s
u
r
in
g
u
n
if
o
r
m
ity
ac
r
o
s
s
d
atasets
.
Dif
f
u
s
ed
an
d
u
n
i
f
o
r
m
lig
h
tin
g
was
m
ain
tain
ed
in
a
s
tu
d
io
s
ettin
g
to
elim
i
n
ate
s
h
ad
o
ws
an
d
g
u
ar
an
tee
c
o
n
s
is
ten
t
lig
h
tin
g
ac
r
o
s
s
all
s
ee
d
lin
g
s
.
Neu
tr
al,
non
-
r
e
f
lectiv
e
b
ac
k
g
r
o
u
n
d
s
ar
e
u
s
ed
in
th
e
s
tu
d
io
to
m
in
im
i
z
e
d
is
tr
ac
tio
n
s
an
d
m
ain
tain
f
o
c
u
s
o
n
th
e
s
ee
d
lin
g
s
.
E
ac
h
g
er
m
in
atin
g
r
ice
s
ee
d
lin
g
was
p
h
o
to
g
r
ap
h
e
d
in
d
iv
id
u
ally
f
r
o
m
ab
o
v
e
th
e
s
tu
d
io
b
o
x
with
a
d
is
tan
ce
b
etwe
en
th
e
ca
m
er
a
an
d
t
h
e
o
b
ject
as f
ar
as 6
0
cm
(
as h
ig
h
as th
e
s
tu
d
io
b
o
x
)
.
Fig
u
r
e
1
.
Data
ac
q
u
is
itio
n
p
r
o
ce
d
u
r
e
B
ased
o
n
th
eir
g
er
m
in
atio
n
lev
els,
r
ice
s
ee
d
lin
g
s
wer
e
c
lass
if
ied
as
n
o
r
m
al,
ab
n
o
r
m
a
l,
n
o
r
m
al
s
ee
d
lin
g
s
with
d
ec
a
y
,
d
ea
d
s
ee
d
s
,
o
r
f
r
esh
s
ee
d
s
th
at
f
ailed
to
g
er
m
i
n
ate.
T
h
e
im
a
g
e
d
ata
w
as
en
s
u
r
ed
to
b
e
o
f
h
ig
h
q
u
ality
to
e
n
s
u
r
e
ac
cu
r
at
e
class
if
icatio
n
.
T
h
e
r
esu
lts
o
f
th
e
im
ag
e
ca
p
tu
r
e
f
o
r
ea
ch
cl
ass
if
icatio
n
ar
e
as
f
o
llo
ws
n
o
r
m
al
s
ee
d
lin
g
s
7
5
3
im
ag
es,
ab
n
o
r
m
al
s
ee
d
lin
g
s
:
6
8
im
ag
es,
d
ea
d
s
ee
d
s
:
1
0
im
ag
e
s
,
n
o
r
m
al
s
e
e
d
l
i
n
g
s
w
i
t
h
d
e
c
a
y
:
5
0
i
m
a
g
e
s
,
a
n
d
f
r
e
s
h
s
e
e
d
s
t
h
a
t
f
a
i
l
e
d
t
o
g
e
r
m
i
n
a
t
e
:
1
3
i
m
a
g
e
s
.
T
h
u
s
,
t
h
e
t
o
t
a
l
n
u
m
b
e
r
o
f
i
m
a
g
e
s
s
u
c
c
e
s
s
f
u
l
l
y
c
o
l
l
e
c
t
e
d
f
o
r
t
h
i
s
s
t
u
d
y
i
s
8
9
4
i
m
a
g
e
s
,
w
h
i
c
h
w
i
l
l
b
e
u
s
e
d
f
o
r
t
h
e
c
l
a
s
s
i
f
i
c
a
t
i
o
n
a
n
a
l
y
s
i
s
.
T
h
e
r
esear
ch
co
n
d
u
cted
in
t
h
is
in
v
esti
g
atio
n
f
ac
ed
d
if
f
icu
lties
in
ca
p
tu
r
in
g
im
ag
es
o
f
ab
n
o
r
m
al
s
ee
d
lin
g
s
,
d
ea
d
s
ee
d
s
,
n
o
r
m
al
s
ee
d
lin
g
s
with
d
ec
ay
,
an
d
f
r
e
s
h
s
ee
d
lin
g
s
th
at
f
ailed
to
g
er
m
in
ate
d
u
e
to
th
ei
r
r
ar
ity
.
As
a
r
esu
lt,
th
e
n
u
m
b
e
r
o
f
im
ag
es
f
o
r
th
ese
ca
teg
o
r
i
es
was
co
n
s
id
er
ab
ly
lo
wer
th
an
th
at
f
o
r
n
o
r
m
al
s
ee
d
lin
g
s
.
T
o
a
d
d
r
ess
th
is
i
s
s
u
e,
th
e
f
o
u
r
c
r
iter
ia
wer
e
m
er
g
ed
in
to
a
s
in
g
le
ca
te
g
o
r
y
lab
eled
"Po
o
r
.
"
C
o
n
s
eq
u
en
tly
,
th
e
d
ataset
co
n
s
is
ted
o
f
7
5
3
im
ag
e
d
ata
o
f
n
o
r
m
al
r
ice
s
ee
d
lin
g
s
class
if
ied
as
"Go
o
d
"
an
d
1
4
1
d
ata
ca
teg
o
r
ize
d
as "Po
o
r
.
"
Fi
g
u
r
e
2
illu
s
tr
ates th
e
class
if
icatio
n
o
f
r
ice
s
ee
d
lin
g
s
,
d
e
p
ictin
g
th
e
d
is
tr
ib
u
tio
n
o
f
im
ag
es
b
etwe
en
th
e
"Go
o
d
"
an
d
"Po
o
r
"
ca
teg
o
r
ies.
T
h
is
s
tr
u
ctu
r
ed
v
is
u
aliza
tio
n
a
llo
ws
f
o
r
a
clea
r
u
n
d
er
s
tan
d
i
n
g
o
f
th
e
im
b
alan
ce
in
th
e
d
ataset,
with
a
lar
g
e
r
p
r
o
p
o
r
tio
n
o
f
im
ag
es
f
allin
g
u
n
d
e
r
th
e
"Go
o
d
"
ca
teg
o
r
y
c
o
m
p
ar
e
d
to
th
e
"Po
o
r
"
ca
teg
o
r
y
.
B
y
m
e
r
g
in
g
th
e
r
ar
e
ca
teg
o
r
ies
in
to
a
s
in
g
le
g
r
o
u
p
,
we
aim
ed
to
en
h
an
ce
th
e
r
o
b
u
s
tn
ess
o
f
o
u
r
class
if
icatio
n
m
o
d
el
wh
ile
ad
d
r
ess
in
g
th
e
im
b
alan
ce
in
t
h
e
d
ataset.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
5
:
62
-
72
66
Fig
u
r
e
2
.
C
lass
if
icatio
n
o
f
r
ice
s
ee
d
lin
g
s
T
h
e
n
in
e
class
if
icatio
n
alg
o
r
it
h
m
s
u
s
ed
in
th
is
s
tu
d
y
w
er
e
l
o
g
is
tic
r
eg
r
ess
io
n
,
r
an
d
o
m
f
o
r
est,
KNN,
n
aïv
e
B
ay
esian
,
d
ec
is
io
n
t
r
ee
,
ANN,
VGG1
6
,
VGG1
9
,
an
d
Alex
Net
wer
e
test
ed
to
class
if
y
r
ice
s
ee
d
lin
g
g
er
m
in
atio
n
in
t
o
g
o
o
d
o
r
b
a
d
.
T
h
e
r
esu
lts
o
f
ea
ch
alg
o
r
ith
m
a
r
e
g
iv
en
in
th
e
f
o
r
m
o
f
ac
c
u
r
ac
y
,
co
n
f
u
s
io
n
m
atr
ix
,
an
d
class
if
icatio
n
r
ep
o
r
t.
B
ef
o
r
e
th
e
im
ag
es
wer
e
f
e
d
in
t
o
th
e
m
ac
h
in
e
lear
n
i
n
g
an
d
d
ee
p
l
e
ar
n
in
g
m
o
d
els
f
o
r
class
if
icatio
n
,
a
s
er
ies
o
f
p
r
ep
r
o
ce
s
s
in
g
s
tep
s
wer
e
a
p
p
lied
to
en
h
a
n
ce
t
h
e
q
u
ality
a
n
d
co
n
s
i
s
ten
cy
o
f
th
e
d
ata.
All
im
ag
es
wer
e
r
esized
to
a
u
n
if
o
r
m
d
im
e
n
s
io
n
o
f
2
2
7
×
2
2
7
p
ix
els.
T
h
is
s
tan
d
a
r
d
izatio
n
wa
s
ess
en
tial
to
en
s
u
r
e
th
at
th
e
m
o
d
el
s
r
ec
eiv
ed
co
n
s
is
ten
t
in
p
u
t
s
izes.
E
ac
h
im
ag
e
was
n
o
r
m
alize
d
b
y
r
escalin
g
p
ix
el
v
alu
es
to
th
e
r
an
g
e
[
0
,
1
]
.
No
r
m
aliza
tio
n
h
e
lp
s
in
s
p
ee
d
in
g
u
p
th
e
tr
ain
in
g
p
r
o
ce
s
s
an
d
allo
ws
th
e
m
o
d
els
to
co
n
v
er
g
e
f
aster
.
T
h
e
im
ag
es we
r
e
o
r
g
an
ized
in
to
b
atch
es t
o
ef
f
icien
tly
tr
ain
t
h
e
m
ac
h
i
n
e
lear
n
in
g
an
d
d
ee
p
lear
n
in
g
m
o
d
els.
I
n
th
is
ca
s
e,
a
b
atch
s
ize
o
f
3
2
was
u
s
ed
.
T
h
e
d
ataset
was
th
en
d
iv
id
ed
in
to
two
p
ar
ts
,
with
8
0
%
o
f
th
e
d
ata
u
s
ed
f
o
r
tr
ai
n
in
g
an
d
2
0
% f
o
r
test
in
g
.
So
t
h
at
th
e
tr
ain
in
g
d
ata
f
o
r
th
e
g
o
o
d
class
was
6
0
3
im
a
g
e
s
d
ata
an
d
f
o
r
p
o
o
r
d
ata
as m
u
ch
as 1
1
1
im
ag
es
d
ata.
W
h
ile
1
5
0
g
o
o
d
class
im
ag
e
d
ata
a
n
d
3
0
p
o
o
r
class
im
ag
e
d
ata
wer
e
u
s
ed
as
test
in
g
d
ata.
T
h
is
s
p
litt
in
g
en
s
u
r
ed
th
at
th
e
m
o
d
els
wer
e
tr
ain
ed
o
n
a
r
ep
r
esen
tativ
e
p
o
r
t
io
n
o
f
th
e
d
ata
an
d
ev
alu
ated
o
n
u
n
s
ee
n
s
am
p
les
to
ass
ess
th
eir
g
en
er
aliza
tio
n
p
er
f
o
r
m
an
ce
.
T
h
e
I
m
ag
eDa
ta
Gen
er
ato
r
f
u
n
ctio
n
f
r
o
m
th
e
Ker
as lib
r
ar
y
was u
ti
lized
to
ef
f
icien
tly
lo
a
d
,
p
r
ep
r
o
ce
s
s
,
an
d
au
g
m
en
t im
ag
e
d
at
a.
Acc
u
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
a
n
d
F1
s
co
r
e
u
tili
zin
g
a
co
n
f
u
s
io
n
m
atr
ix
wer
e
em
p
lo
y
ed
a
s
in
tu
itiv
e
p
er
f
o
r
m
an
ce
i
n
d
icato
r
s
to
ass
ess
th
e
p
er
f
o
r
m
an
ce
o
f
th
e
clas
s
if
icatio
n
m
o
d
el.
T
o
g
en
er
ate
a
co
n
f
u
s
io
n
m
atr
i
x
,
th
e
n
u
m
b
er
o
f
tr
u
e
p
o
s
itiv
es,
tr
u
e
n
eg
ativ
es,
f
alse
p
o
s
itiv
es,
an
d
f
alse
n
eg
ativ
es
m
u
s
t
b
e
ca
lcu
lated
.
T
h
e
f
o
llo
win
g
m
ath
e
m
atica
l m
eth
o
d
is
u
s
ed
to
ca
lcu
late
th
e
p
er
f
o
r
m
an
ce
in
d
icato
r
[
2
8
]
.
Acc
u
r
ac
y
=(
T
P+T
N)
/(
T
P+T
N
+FP+FN)
Pre
cisi
o
n
=T
P/(T
P+FP
)
R
ec
all=T
P/(T
P+
FN)
F1
-
s
co
r
e=
(
2
×Pr
ec
is
io
n
×Reca
ll Pr
ec
is
io
n
)
/(
R
ec
all+Reca
ll)
T
h
e
co
n
f
u
s
io
n
m
atr
ix
illu
s
tr
a
tes
th
e
p
r
e
d
ictio
n
s
p
r
o
v
i
d
ed
b
y
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
an
d
th
e
ac
tu
al
o
cc
u
r
r
e
n
ce
s
.
Acc
u
r
ac
y
is
th
e
r
atio
o
f
tr
u
e
p
o
s
itiv
es
to
tr
u
e
n
eg
ativ
es
f
o
r
ea
ch
test
in
g
d
ata
s
et.
T
h
e
ac
cu
r
ac
y
o
f
th
e
alg
o
r
ith
m
af
f
ec
ts
h
o
w
well
it
p
r
ed
icts
.
Pre
cisi
o
n
is
d
ef
in
ed
as
th
e
p
r
o
p
o
r
tio
n
o
f
ac
c
u
r
ately
p
r
ed
icted
p
o
s
itiv
e
test
f
in
d
in
g
s
to
all
p
r
o
jecte
d
p
o
s
itiv
e
test
r
esu
lts
.
W
h
en
c
o
m
p
ar
e
d
t
o
a
ll
test
d
ata,
r
ec
all
i
n
d
i
c
a
t
e
s
h
o
w
o
f
t
e
n
f
o
r
e
c
a
s
t
s
t
u
r
n
o
u
t
t
o
b
e
c
o
r
r
e
c
t
.
T
h
e
F
1
S
c
o
r
e
i
s
c
o
m
p
u
t
e
d
b
y
a
v
e
r
a
g
i
n
g
p
r
e
c
i
s
i
o
n
a
n
d
r
e
c
a
l
l
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
R
esu
lts
f
r
o
m
th
e
lo
g
is
tic
r
eg
r
ess
io
n
,
r
an
d
o
m
f
o
r
est
,
KNN,
n
aïv
e
B
ay
esian
,
d
ec
is
io
n
tr
ee
,
VGG
1
6
,
VGG
19
,
an
d
Alex
Net
alg
o
r
it
h
m
s
s
h
o
wed
d
if
f
er
e
n
t
ac
cu
r
ac
y
an
d
p
e
r
f
o
r
m
an
ce
.
I
n
th
e
r
es
u
lts
o
f
th
e
an
aly
s
is
u
s
in
g
lo
g
is
tic
r
eg
r
ess
io
n
,
an
ac
cu
r
ac
y
o
f
0
.
9
4
was
o
b
tain
e
d
,
in
d
icatin
g
th
at
th
is
m
o
d
el
s
u
cc
ee
d
ed
well
in
class
if
y
in
g
th
e
g
er
m
in
atio
n
o
f
r
ice
s
ee
d
lin
g
s
in
to
g
o
o
d
o
r
p
o
o
r
ca
teg
o
r
ies.
I
n
th
e
c
o
n
f
u
s
io
n
m
atr
ix
,
o
u
t
o
f
1
5
0
r
ice
s
ee
d
d
ata
th
at
wer
e
g
o
o
d
,
1
4
6
o
f
th
e
m
wer
e
co
r
r
ec
tly
cl
ass
if
ied
as
g
o
o
d
(
tr
u
e
p
o
s
itiv
e
)
,
wh
ile
4
s
ee
d
lin
g
s
wer
e
in
co
r
r
ec
tly
class
if
ied
as
p
o
o
r
(
f
alse
n
eg
ativ
e
)
.
Of
th
e
3
0
p
o
o
r
r
ice
s
ee
d
d
ata
,
2
3
o
f
t
h
em
wer
e
co
r
r
ec
tly
class
if
ied
as
p
o
o
r
(
tr
u
e
n
eg
ativ
e
)
,
b
u
t
7
p
o
o
r
s
ee
d
lin
g
s
wer
e
i
n
co
r
r
ec
tly
class
if
ied
as
g
o
o
d
(
f
alse
p
o
s
itiv
e
)
.
T
h
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Ove
r
co
min
g
imb
a
la
n
ce
d
r
ice
s
ee
d
g
ermin
a
tio
n
cla
s
s
ifica
tio
n
…
(
Mu
h
la
s
a
h
N
o
vita
s
a
r
i Ma
r
a
)
67
class
if
icatio
n
r
ep
o
r
t
s
h
o
ws
th
at
th
e
m
o
d
el
p
r
ec
is
io
n
is
0
.
7
7
f
o
r
th
e
p
o
o
r
class
,
an
d
0
.
9
7
f
o
r
t
h
e
g
o
o
d
class
.
W
h
i
le
th
e
r
ec
all
f
o
r
g
o
o
d
an
d
p
o
o
r
d
class
es
is
0
.
9
5
a
n
d
0
.
8
5
r
esp
ec
tiv
ely
.
I
n
th
e
F1
-
Sco
r
e
,
wh
ich
co
m
b
i
n
es
in
f
o
r
m
atio
n
a
b
o
u
t
p
r
ec
is
io
n
an
d
r
ec
all,
th
e
g
o
o
d
class
h
as
a
s
co
r
e
o
f
0
.
9
6
an
d
th
e
p
o
o
r
class
h
as
a
s
co
r
e
o
f
0
.
8
1
.
Hig
h
F1
-
s
co
r
e
in
b
o
th
class
es
in
d
ica
te
th
at
t
h
e
m
o
d
el
p
er
f
o
r
m
s
well
in
class
if
y
in
g
th
e
m
.
T
h
e
ex
ec
u
tio
n
tim
e
o
f
th
e
lo
g
is
tic
r
eg
r
ess
io
n
m
o
d
el
is
ab
o
u
t
1
6
.
7
s
ec
o
n
d
s
,
wh
ich
s
h
o
ws
th
at
th
is
m
o
d
el
h
as
a
r
elativ
ely
f
ast
p
r
o
ce
s
s
in
g
tim
e
in
class
if
y
in
g
r
ice
s
ee
d
d
ata.
T
h
is
s
u
g
g
ests
its
ef
f
ec
tiv
en
ess
f
o
r
t
h
is
class
if
icatio
n
task
.
Fo
r
in
s
tan
ce
,
R
ah
m
an
an
d
Yap
[
2
9
]
in
v
esti
g
ated
th
e
ef
f
ec
t
o
f
im
b
alan
c
ed
r
atio
s
o
n
lo
g
is
tic
r
eg
r
ess
i
o
n
,
co
n
f
i
r
m
in
g
th
at
im
b
alan
ce
im
p
ac
ts
p
ar
am
eter
esti
m
ates
an
d
clas
s
if
icatio
n
.
R
ah
m
an
et
a
l.
[
3
0
]
p
r
o
v
id
e
e
v
id
en
ce
th
at
lo
g
is
tic
r
eg
r
ess
io
n
ca
n
s
till
p
er
f
o
r
m
we
ll
with
im
b
alan
ce
d
d
ata,
alth
o
u
g
h
th
e
im
b
alan
ce
r
atio
af
f
ec
ts
p
ar
am
eter
esti
m
ates
an
d
p
er
f
o
r
m
a
n
ce
m
etr
ics lik
e
p
r
ec
is
io
n
an
d
r
ec
all,
wh
ich
is
s
im
ilar
to
o
u
r
r
e
p
o
r
ted
r
esu
lts
.
T
h
e
r
esu
lts
o
f
th
e
an
aly
s
is
u
s
i
n
g
th
e
r
an
d
o
m
f
o
r
est
alg
o
r
ith
m
s
h
o
wed
an
ac
cu
r
ac
y
o
f
0
.
9
1
.
Ho
wev
er
,
th
is
ac
cu
r
ac
y
v
alu
e
is
d
o
m
in
ated
b
y
th
e
ac
cu
r
ac
y
o
f
th
e
m
o
d
el
p
r
ed
ictin
g
th
e
m
ajo
r
ity
class
.
T
h
is
ca
n
b
e
s
ee
n
f
r
o
m
th
e
c
o
n
f
u
s
io
n
m
at
r
ix
wh
ich
s
h
o
ws
th
at
o
u
t
o
f
a
t
o
tal
o
f
1
5
0
d
ata
o
n
r
ice
s
ee
d
s
th
at
ar
e
ac
tu
ally
g
o
o
d
,
all
o
f
t
h
em
ar
e
class
if
ied
co
r
r
ec
tly
as
g
o
o
d
(
tr
u
e
p
o
s
itiv
e
)
,
s
o
th
er
e
is
n
o
er
r
o
r
in
id
en
tify
i
n
g
g
o
o
d
r
ice
s
ee
d
s
Ho
wev
er
,
f
r
o
m
3
0
p
o
o
r
r
ice
s
e
ed
d
ata,
o
n
ly
1
4
o
f
th
e
m
ar
e
co
r
r
ec
tly
class
if
ied
as
p
o
o
r
,
wh
il
e
1
6
p
o
o
r
s
ee
d
s
ar
e
in
co
r
r
ec
tly
class
if
ied
as
g
o
o
d
(
f
alse
p
o
s
itiv
e
)
.
I
n
th
e
F1
-
Sco
r
e,
wh
ich
c
o
m
b
in
es
in
f
o
r
m
ati
o
n
ab
o
u
t
p
r
ec
is
io
n
an
d
r
ec
all,
t
h
e
g
o
o
d
class
h
as a
v
er
y
h
ig
h
s
co
r
e,
wh
ic
h
is
0
.
9
5
,
wh
ile
th
e
p
o
o
r
class
h
as
a
s
co
r
e
o
f
0
.
6
4
.
T
h
ese
r
esu
lts
in
d
icate
th
at
th
e
r
an
d
o
m
f
o
r
est
m
o
d
el
h
as
d
if
f
ic
u
lty
in
id
en
tify
in
g
p
o
o
r
r
ice
s
ee
d
lin
g
s
.
T
h
e
e
x
ec
u
tio
n
tim
e
o
f
th
e
r
an
d
o
m
f
o
r
est
m
o
d
el
is
ab
o
u
t
1
1
.
9
s
ec
o
n
d
s
,
wh
ich
in
d
icate
s
th
at
th
is
m
o
d
el
h
as
a
r
elativ
ely
f
ast
p
r
o
ce
s
s
in
g
tim
e
in
class
if
y
in
g
r
ice
s
ee
d
d
ata.
Sh
u
et
a
l
.
[
3
1
]
h
av
e
id
en
tifie
d
s
im
il
ar
co
n
s
tr
ain
ts
o
f
r
an
d
o
m
f
o
r
est
alg
o
r
ith
m
s
in
m
an
ag
in
g
im
b
alan
ce
d
d
ata,
r
ec
o
m
m
e
n
d
in
g
th
e
g
en
er
atio
n
o
f
m
in
o
r
ity
class
s
am
p
les
u
s
in
g
g
en
er
ativ
e
ad
v
er
s
ar
ial
n
etwo
r
k
(
GAN)
to
en
h
a
n
ce
class
if
icatio
n
p
er
f
o
r
m
an
ce
.
Mo
r
e
a
n
d
R
an
a
[
3
2
]
r
e
v
iew
v
ar
io
u
s
tech
n
iq
u
es
lik
e
weig
h
ted
r
a
n
d
o
m
f
o
r
est
an
d
b
alan
ce
d
r
an
d
o
m
f
o
r
est
to
tack
le
t
h
e
class
im
b
alan
ce
p
r
o
b
lem
,
ac
k
n
o
wled
g
in
g
t
h
e
lim
itatio
n
s
o
f
s
tan
d
ar
d
r
an
d
o
m
f
o
r
ests
in
im
b
a
lan
ce
d
s
ce
n
ar
io
s
.
Kh
o
s
h
g
o
f
taar
et
a
l.
[
3
3
]
p
r
o
v
id
e
an
ex
te
n
s
iv
e
em
p
i
r
ical
ev
a
lu
atio
n
o
f
r
an
d
o
m
f
o
r
ests
b
u
ilt
f
r
o
m
im
b
alan
ce
d
d
ata,
h
ig
h
lig
h
ti
n
g
th
e
ch
allen
g
es
an
d
lim
itatio
n
s
f
ac
ed
b
y
th
e
alg
o
r
ith
m
,
p
ar
ticu
lar
l
y
its
te
n
d
en
cy
to
f
a
v
o
r
th
e
m
ajo
r
ity
class
.
T
h
e
r
esu
lts
o
f
th
e
a
n
aly
s
is
u
s
in
g
th
e
KNN
alg
o
r
ith
m
s
h
o
we
d
a
lo
w
ac
cu
r
ac
y
o
f
0
.
3
6
,
in
d
i
ca
tin
g
th
at
th
is
m
o
d
el
h
ad
d
if
f
ic
u
lty
in
c
lass
if
y
in
g
r
ic
e
s
ee
d
g
er
m
in
atio
n
in
to
g
o
o
d
o
r
p
o
o
r
ca
teg
o
r
ies.
T
h
e
co
n
f
u
s
io
n
m
atr
ix
s
h
o
ws
th
at
o
u
t o
f
a
to
ta
l
o
f
1
5
0
r
ice
s
ee
d
d
ata
t
h
at
ar
e
ac
tu
ally
g
o
o
d
,
o
n
ly
3
6
ar
e
c
o
r
r
ec
tly
class
if
ied
as
g
o
o
d
(
tr
u
e
p
o
s
itiv
e
)
,
wh
ile
1
1
4
g
o
o
d
s
ee
d
lin
g
s
ar
e
in
co
r
r
ec
tly
class
if
ied
as
p
o
o
r
(
f
alse
n
eg
ati
v
e
)
.
Of
th
e
3
0
p
o
o
r
r
ice
s
ee
d
d
ata,
o
n
l
y
2
9
o
f
th
e
m
wer
e
co
r
r
ec
tly
class
if
ied
as
p
o
o
r
(
tr
u
e
n
eg
ativ
e
)
,
w
h
ile
o
n
ly
1
p
o
o
r
s
ee
d
was
in
co
r
r
ec
tly
class
if
ied
as g
o
o
d
(
f
alse p
o
s
itiv
e
)
.
T
h
e
class
if
ica
tio
n
r
ep
o
r
t sh
o
ws th
at
th
e
m
o
d
el
h
as a
h
ig
h
r
e
ca
ll,
wh
ich
is
0
.
9
7
,
f
o
r
th
e
g
o
o
d
c
lass
b
u
t
f
o
r
th
e
p
o
o
r
class
,
th
e
lo
w
p
r
ec
is
io
n
is
0
.
2
0
.
I
n
th
e
F
1
-
s
co
r
e,
wh
ich
co
m
b
in
es in
f
o
r
m
atio
n
ab
o
u
t p
r
ec
is
io
n
an
d
r
ec
all,
th
e
g
o
o
d
cl
ass
h
as a
s
co
r
e
o
f
0
.
3
9
,
w
h
ile
th
e
p
o
o
r
class
h
as
a
s
co
r
e
o
f
0
.
3
4
.
T
h
e
e
x
ec
u
tio
n
t
im
e
o
f
th
e
KNN
m
o
d
el
is
v
er
y
f
ast,
wh
ich
is
ab
o
u
t
0
.
2
1
s
e
co
n
d
s
,
b
u
t
th
e
lo
w
p
er
f
o
r
m
an
ce
in
class
if
y
in
g
r
ic
e
s
ee
d
lin
g
s
r
eq
u
ir
es
im
p
r
o
v
e
d
m
o
d
el
p
e
r
f
o
r
m
an
ce
to
p
r
o
d
u
ce
b
etter
an
d
m
o
r
e
ac
cu
r
ate
class
if
icatio
n
r
esu
lts
.
R
elate
d
to
th
is
r
esu
lt,
Su
n
an
d
C
h
en
[
3
4
]
p
r
o
v
id
e
a
co
m
p
r
eh
en
s
iv
e
s
u
r
v
ey
o
f
KNN
alg
o
r
ith
m
s
an
d
h
ig
h
lig
h
t
th
e
d
if
f
icu
lty
KNN
f
ac
es
with
im
b
alan
ce
d
d
ata,
wh
ich
o
f
ten
lead
s
to
p
o
o
r
p
er
f
o
r
m
an
ce
.
I
n
th
eir
th
o
r
o
u
g
h
an
aly
s
is
o
f
KNN
alg
o
r
ith
m
s
,
[
3
4
]
p
o
in
t
o
u
t
th
at
KNN
f
r
eq
u
en
tly
p
er
f
o
r
m
s
p
o
o
r
ly
wh
en
f
ac
ed
with
u
n
b
al
an
ce
d
d
ata.
T
h
e
r
esu
lts
o
f
th
e
an
aly
s
is
u
s
in
g
th
e
n
aïv
e
B
ay
esian
alg
o
r
ith
m
s
h
o
wed
an
ac
cu
r
ac
y
o
f
0
.
5
4
,
in
d
icatin
g
th
at
th
is
m
o
d
el
h
as
a
lo
w
p
e
r
f
o
r
m
an
ce
in
class
if
y
in
g
r
ice
s
ee
d
g
er
m
in
atio
n
in
to
g
o
o
d
o
r
p
o
o
r
ca
teg
o
r
ies.
T
h
e
co
n
f
u
s
io
n
m
atr
ix
s
h
o
ws
th
at
o
u
t
o
f
a
to
tal
o
f
1
5
0
r
ice
s
ee
d
d
ata
th
at
a
r
e
ac
t
u
ally
g
o
o
d
,
o
n
ly
7
2
a
r
e
c
o
r
r
ec
tly
class
if
ied
as
g
o
o
d
(
tr
u
e
p
o
s
itiv
e
)
,
w
h
ile
7
8
g
o
o
d
s
ee
d
lin
g
s
ar
e
in
co
r
r
ec
tly
class
if
ied
as
p
o
o
r
(
f
alse
n
e
g
ativ
e
)
.
Of
th
e
3
0
p
o
o
r
r
ice
s
ee
d
d
ata,
o
n
ly
2
6
o
f
t
h
em
wer
e
co
r
r
ec
t
ly
class
if
ied
as
p
o
o
r
(
tr
u
e
n
eg
ativ
e
)
,
wh
ile
4
p
o
o
r
s
ee
d
lin
g
s
wer
e
in
c
o
r
r
ec
tly
cla
s
s
if
ied
as
g
o
o
d
(
f
alse
p
o
s
itiv
e
)
.
T
h
e
class
if
icatio
n
r
e
p
o
r
t
s
h
o
ws
th
at
th
e
m
o
d
el
h
as a
h
ig
h
r
ec
al
l r
ate,
wh
ich
is
0
.
9
5
,
f
o
r
th
e
g
o
o
d
class
h
o
we
v
er
,
f
o
r
t
h
e
p
o
o
r
class
th
e
r
ec
all
v
alu
e
is
v
er
y
lo
w
at
0
.
2
5
.
I
n
th
e
F
1
-
s
co
r
e,
wh
ic
h
co
m
b
in
es
in
f
o
r
m
atio
n
ab
o
u
t
p
r
ec
is
io
n
an
d
r
ec
all,
th
e
g
o
o
d
class
h
as
a
s
co
r
e
o
f
0
.
6
4
,
wh
ile
th
e
p
o
o
r
class
h
as
a
s
co
r
e
o
f
0
.
3
9
.
T
h
is
in
d
icat
es
th
at
th
er
e
is
d
if
f
icu
lty
in
id
en
tify
in
g
p
o
o
r
r
ice
s
ee
d
lin
g
s
.
T
h
e
ex
ec
u
tio
n
tim
e
o
f
th
e
n
aïv
e
B
ay
esi
an
m
o
d
el
is
ab
o
u
t
0
.
7
8
s
ec
o
n
d
s
,
wh
ich
in
d
icate
s
th
at
th
is
m
o
d
el
h
as
a
f
air
ly
f
ast
p
r
o
ce
s
s
in
g
tim
e.
Alth
o
u
g
h
th
e
ac
c
u
r
ac
y
is
s
till
ab
o
v
e
0
.
5
,
th
e
clas
s
if
icatio
n
r
esu
lts
in
p
o
o
r
g
r
ad
es
n
ee
d
to
b
e
im
p
r
o
v
ed
to
im
p
r
o
v
e
th
e
m
o
d
el’
s
p
er
f
o
r
m
an
ce
in
b
etter
id
en
tify
i
n
g
p
o
o
r
r
ice
s
ee
d
lin
g
s
.
H
ig
h
lig
h
t
s
im
ilar
d
i
f
f
icu
lties
o
f
n
aï
v
e
B
ay
es
f
ac
es
i
n
im
b
a
lan
ce
d
d
atasets
an
d
th
e
im
p
a
ct
o
n
class
if
icatio
n
ac
cu
r
ac
y
[
3
5
]
.
T
h
e
r
esu
lts
o
f
th
e
an
aly
s
is
u
s
in
g
th
e
d
ec
is
io
n
tr
ee
alg
o
r
it
h
m
s
h
o
wed
an
ac
cu
r
ac
y
o
f
0
.
7
4
,
wh
ich
in
d
icate
s
th
at
th
is
m
o
d
el
h
as
a
f
air
ly
g
o
o
d
le
v
el
o
f
ac
c
u
r
ac
y
in
class
if
y
in
g
r
ice
s
ee
d
g
er
m
i
n
atio
n
in
t
o
g
o
o
d
o
r
p
o
o
r
ca
teg
o
r
ies.
T
h
e
co
n
f
u
s
io
n
m
atr
ix
s
h
o
ws
th
at
o
u
t
o
f
a
t
o
tal
o
f
1
5
0
d
ata
o
n
r
ice
s
ee
d
s
th
at
ar
e
ac
tu
ally
g
o
o
d
,
1
1
2
o
f
th
em
ar
e
c
o
r
r
e
ctly
class
if
ied
as
g
o
o
d
(
tr
u
e
p
o
s
itiv
e
)
,
wh
ile
3
8
g
o
o
d
s
ee
d
lin
g
s
ar
e
in
co
r
r
ec
tly
class
if
ied
as
p
o
o
r
(
f
alse
n
eg
ativ
e
)
.
Of
t
h
e
3
0
p
o
o
r
r
ice
s
ee
d
d
ata,
2
1
o
f
th
em
wer
e
co
r
r
ec
tly
class
if
ied
as
p
o
o
r
(
tr
u
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
5
:
62
-
72
68
n
eg
ativ
e
)
,
b
u
t
9
p
o
o
r
s
ee
d
lin
g
s
wer
e
in
co
r
r
ec
tly
class
if
ied
a
s
g
o
o
d
(
f
alse
p
o
s
itiv
e
)
.
T
h
e
cl
ass
if
icatio
n
r
ep
o
r
t
s
h
o
ws
th
at
th
e
m
o
d
el
h
as
a
h
ig
h
r
ec
all
r
ate,
wh
ich
is
0
.
9
3
,
f
o
r
th
e
g
o
o
d
class
b
u
t
f
o
r
th
e
p
o
o
r
class
,
th
e
r
ec
all
v
alu
e
is
lo
w
at
0
.
3
6
.
I
n
th
e
F
1
-
s
co
r
e,
wh
ich
co
m
b
in
es
in
f
o
r
m
atio
n
ab
o
u
t
p
r
ec
is
io
n
an
d
r
e
ca
ll,
th
e
g
o
o
d
class
h
as
a
s
co
r
e
o
f
0
.
8
3
,
wh
ile
th
e
p
o
o
r
class
h
as
a
s
co
r
e
o
f
0
.
4
7
s
ig
n
alin
g
th
e
d
if
f
ic
u
lty
in
id
en
tify
in
g
p
o
o
r
r
ic
e
s
ee
d
lin
g
s
.
T
h
e
ex
ec
u
tio
n
tim
e
o
f
th
e
d
ec
is
io
n
tr
ee
m
o
d
el
is
a
p
p
r
o
x
im
ately
7
7
.
7
3
s
ec
o
n
d
s
,
wh
ich
in
d
icate
s
th
at
th
is
m
o
d
el
tak
es
a
co
n
s
id
er
a
b
le
am
o
u
n
t
o
f
tim
e
f
o
r
p
r
o
ce
s
s
in
g
.
Alth
o
u
g
h
th
e
ac
c
u
r
ac
y
is
q
u
ite
g
o
o
d
,
th
e
class
if
icatio
n
r
esu
lts
in
p
o
o
r
g
r
ad
es
n
ee
d
to
b
e
im
p
r
o
v
ed
to
im
p
r
o
v
e
th
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
m
o
d
el
in
b
etter
id
en
tify
in
g
p
o
o
r
r
ice
s
ee
d
lin
g
s
.
I
n
a
d
d
itio
n
,
m
o
d
els
al
s
o
n
e
ed
to
b
e
o
p
tim
ized
to
r
eq
u
ir
e
s
h
o
r
ter
p
r
o
ce
s
s
in
g
tim
es.
T
r
u
icǎ
an
d
L
eo
r
d
ea
n
u
[
3
6
]
d
is
cu
s
s
v
ar
io
u
s
d
ec
is
io
n
tr
ee
alg
o
r
ith
m
s
ap
p
lied
to
im
b
al
an
ce
d
d
atasets
an
d
th
eir
r
esp
ec
tiv
e
p
er
f
o
r
m
a
n
ce
i
m
p
r
o
v
e
m
en
ts
.
T
h
e
ANN,
VGG1
9
,
an
d
Alex
Net
ar
ch
itectu
r
es
p
r
o
v
id
e
th
e
s
am
e
test
r
esu
lts
.
T
h
e
ANN
ar
ch
itectu
r
e
co
n
s
is
ts
o
f
a
s
eq
u
en
tial
m
o
d
el
with
a
f
latten
in
g
lay
er
to
co
n
v
er
t
in
p
u
t
im
a
g
es
in
to
a
o
n
e
-
d
im
en
s
io
n
al
v
ec
to
r
,
a
h
id
d
en
lay
e
r
with
1
2
8
n
eu
r
o
n
s
an
d
r
ec
t
if
ied
lin
ea
r
u
n
it
(
R
eL
U)
ac
tiv
atio
n
,
an
d
a
n
o
u
t
p
u
t
la
y
er
with
a
s
ig
m
o
id
ac
tiv
atio
n
f
o
r
b
in
ar
y
class
if
icatio
n
.
T
h
e
m
o
d
el
is
co
m
p
iled
w
ith
th
e
A
d
am
o
p
tim
izer
an
d
b
i
n
ar
y
cr
o
s
s
-
en
tr
o
p
y
lo
s
s
.
T
h
e
Alex
Net
a
r
ch
itectu
r
e
co
m
m
en
ce
s
with
c
o
n
v
o
lu
tio
n
l
ay
er
1
,
f
ea
tu
r
in
g
9
6
f
ilter
s
with
an
1
1
×
11
r
ec
ep
tiv
e
f
ield
an
d
R
eL
U
ac
tiv
atio
n
.
A
4
×
4
s
tr
id
e
is
ap
p
lied
t
o
ac
co
m
m
o
d
ate
2
2
7
×
2
2
7
p
ix
el
R
GB
in
p
u
t
im
ag
es.
Su
b
s
eq
u
en
tly
,
m
ax
p
o
o
lin
g
lay
er
1
with
a
3
×
3
win
d
o
w
a
n
d
2
×
2
s
tr
id
e
r
ed
u
ce
s
s
p
atial
d
im
e
n
s
io
n
s
.
C
o
n
v
o
l
u
tio
n
l
ay
er
2
f
o
llo
ws,
co
m
p
r
is
in
g
2
5
6
5
×
5
f
ilter
s
with
R
eL
U
ac
ti
v
atio
n
,
an
d
is
s
u
cc
ee
d
ed
b
y
m
ax
p
o
o
lin
g
lay
er
2
.
C
o
n
v
o
lu
tio
n
l
ay
e
r
s
3
,
4
,
an
d
5
in
tr
o
d
u
ce
3
8
4
,
3
8
4
,
an
d
2
5
6
3
×
3
f
ilter
s
with
R
eL
U
ac
tiv
atio
n
,
p
r
o
g
r
ess
iv
ely
ca
p
tu
r
in
g
h
ier
ar
ch
ical
f
ea
tu
r
e
s
.
Ma
x
p
o
o
lin
g
lay
er
3
co
n
tin
u
es
s
p
atial
r
ed
u
ctio
n
.
T
h
e
d
ata
is
f
latten
ed
an
d
p
r
o
ce
s
s
ed
th
r
o
u
g
h
two
f
u
lly
c
o
n
n
ec
ted
lay
er
s
with
5
1
2
u
n
its
,
ap
p
ly
in
g
R
eL
U
ac
tiv
atio
n
an
d
d
r
o
p
o
u
t
o
f
0
.
5
f
o
r
r
eg
u
lar
izatio
n
.
T
h
e
ar
c
h
itectu
r
e
co
n
clu
d
es
with
a
s
in
g
le
-
u
n
it
o
u
tp
u
t
L
a
y
er
with
s
ig
m
o
id
ac
tiv
atio
n
f
o
r
b
in
ar
y
class
if
icatio
n
.
Mo
d
el
tr
ain
in
g
em
p
lo
y
s
th
e
Ad
am
o
p
tim
izer
an
d
b
in
a
r
y
cr
o
s
s
-
en
tr
o
p
y
lo
s
s
.
T
h
e
f
o
u
n
d
atio
n
al
elem
en
t
with
in
th
e
a
r
ch
itectu
r
e
o
f
t
h
e
V
GG1
9
b
ase
m
o
d
el
is
th
e
VG
G1
9
C
NN
,
r
en
o
wn
ed
f
o
r
its
ad
ep
tn
ess
in
ex
tr
ac
tin
g
in
tr
icate
h
ier
ar
ch
ical
f
ea
tu
r
es
f
r
o
m
im
a
g
es.
I
n
th
is
s
p
ec
if
ic
in
s
tan
tiatio
n
,
its
r
o
le
is
th
at
o
f
a
f
ea
tu
r
e
ex
tr
ac
t
o
r
.
Su
b
s
eq
u
en
t
to
th
e
VGG1
9
co
n
v
o
lu
tio
n
al
lay
er
s
,
a
f
latten
lay
er
is
in
tr
o
d
u
ce
d
,
s
er
v
in
g
th
e
p
u
r
p
o
s
e
o
f
r
esh
ap
in
g
th
e
m
u
lti
-
d
im
en
s
io
n
al
o
u
tp
u
t
in
to
a
s
in
g
u
la
r
one
-
d
im
e
n
s
io
n
al
v
ec
to
r
.
T
wo
d
en
s
ely
co
n
n
ec
ted
la
y
er
s
,
ea
c
h
co
m
p
r
is
in
g
4
0
9
6
u
n
its
,
a
r
e
i
n
co
r
p
o
r
ated
i
n
to
th
e
ar
ch
itectu
r
e.
T
h
ese
lay
e
r
s
estab
lis
h
co
m
p
r
eh
en
s
iv
e
in
ter
co
n
n
ec
tio
n
s
with
u
n
its
f
r
o
m
t
h
e
p
r
ec
ed
in
g
lay
er
,
an
d
R
eL
U
ac
tiv
atio
n
f
u
n
ctio
n
s
ar
e
ap
p
lied
to
m
o
d
u
late
th
eir
o
u
t
p
u
ts
.
T
h
e
ter
m
in
al
lay
er
co
n
s
i
s
ts
o
f
a
s
o
li
tar
y
u
n
it
an
d
em
p
lo
y
s
a
s
ig
m
o
id
ac
tiv
atio
n
f
u
n
ctio
n
,
co
n
f
o
r
m
in
g
to
th
e
cu
s
to
m
ar
y
co
n
f
ig
u
r
atio
n
f
o
r
b
in
ar
y
class
if
icatio
n
task
s
.
Su
b
s
eq
u
en
tly
,
th
e
m
o
d
el
is
m
eticu
lo
u
s
ly
ass
em
b
led
,
in
teg
r
a
tin
g
th
e
A
d
am
o
p
tim
izer
an
d
b
in
a
r
y
cr
o
s
s
-
en
tr
o
p
y
lo
s
s
f
u
n
ctio
n
.
T
h
e
ac
cu
r
ac
y
v
alu
e
o
f
ANN,
VGG1
9
,
an
d
Alex
Net
m
o
d
els
is
0
.
8
3
o
r
8
3
%.
Ho
wev
er
,
wh
en
lo
o
k
in
g
at
th
e
co
n
f
u
s
io
n
m
atr
ix
,
th
e
m
o
d
el
ca
n
n
o
t
d
is
tin
g
u
is
h
p
o
o
r
c
lass
es
(
m
in
o
r
ity
class
es)
p
er
f
ec
tly
b
ec
au
s
e
it
h
as
a
v
er
y
lo
w
r
ec
all
v
alu
e
a
n
d
F1
-
s
co
r
e
f
o
r
t
h
at
class
.
T
h
e
co
n
f
u
s
io
n
m
atr
ix
r
esu
lts
s
h
o
wed
th
at
o
u
t o
f
a
to
tal
o
f
3
0
d
ata
th
at
s
h
o
u
ld
b
elo
n
g
to
class
1
,
th
e
m
o
d
el
f
ailed
in
clas
s
if
y
in
g
an
y
o
f
th
em
,
r
esu
ltin
g
in
3
0
f
alse
n
eg
ativ
es.
I
n
co
n
t
r
ast,
f
o
r
eith
e
r
class
(
m
ajo
r
ity
class
)
,
th
e
m
o
d
el
m
an
a
g
ed
to
class
if
y
all
d
ata
c
o
r
r
ec
t
ly
,
r
esu
ltin
g
in
1
5
0
tr
u
e
p
o
s
itiv
es.
T
h
is
ca
u
s
es th
e
p
r
ec
is
io
n
an
d
r
ec
all
f
o
r
class
1
to
ze
r
o
.
Fro
m
th
e
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
e
g
en
er
ated
,
it
ca
n
b
e
s
ee
n
th
at
th
e
m
o
d
el
h
as
p
r
o
b
lem
s
in
r
ec
o
g
n
izin
g
m
in
o
r
ity
class
es
well,
an
d
th
is
ca
n
b
e
ca
u
s
ed
b
y
an
im
b
alan
ce
in
th
e
am
o
u
n
t
o
f
d
ata
b
etwe
en
th
e
m
ajo
r
ity
class
an
d
th
e
m
in
o
r
ity
c
lass
.
ANN,
VG
G1
9
,
an
d
Alex
Net
m
o
d
els
ac
h
iev
ed
s
im
ilar
ac
cu
r
ac
y
(
8
3
%)
b
u
t
s
u
f
f
er
ed
f
r
o
m
d
ata
im
b
ala
n
ce
,
as
s
h
o
wn
b
y
th
eir
h
ig
h
F1
-
s
co
r
es
f
o
r
th
e
m
aj
o
r
it
y
class
an
d
an
F1
-
Sco
r
e
o
f
ze
r
o
f
o
r
th
e
m
in
o
r
ity
class
,
h
ig
h
lig
h
tin
g
th
is
lim
itatio
n
as seen
in
[
2
1
]
,
[
3
7
]
.
T
h
e
ar
c
h
itectu
r
e
f
o
r
VGG1
6
clo
s
ely
p
ar
allels
th
e
s
tr
u
ctu
r
e
an
d
r
atio
n
ale
o
f
VGG1
9
,
as
p
r
ev
io
u
s
ly
d
escr
ib
ed
.
I
n
th
e
co
n
tex
t
o
f
te
s
tin
g
,
th
e
VGG1
6
ar
ch
itectu
r
e
y
ield
ed
an
ac
cu
r
ac
y
o
f
0
.
6
5
o
r
6
5
%.
Ho
we
v
er
,
wh
en
lo
o
k
in
g
at
th
e
co
n
f
u
s
io
n
m
atr
ix
,
t
h
e
m
o
d
el
h
as
p
r
o
b
l
em
s
class
if
y
in
g
b
o
th
class
es.
T
h
e
m
o
d
el
co
r
r
ec
tly
class
if
ied
1
1
3
d
ata
i
n
to
g
o
o
d
class
b
u
t
m
is
class
if
ied
3
7
d
ata
f
r
o
m
g
o
o
d
class
in
to
f
alse
p
o
s
itiv
es.
I
n
ad
d
itio
n
,
th
e
m
o
d
el
also
m
an
a
g
ed
to
cla
s
s
if
y
4
d
ata
f
r
o
m
class
1
co
r
r
e
ctly
b
u
t
f
ailed
to
clas
s
if
y
2
6
d
ata
f
r
o
m
class
1
as
class
g
o
o
d
(
f
alse
n
e
g
ativ
es).
Fro
m
th
e
p
r
ec
is
io
n
,
r
ec
all,
a
n
d
F1
-
s
co
r
e
g
e
n
er
ated
,
th
e
m
o
d
el
h
as
p
r
o
b
lem
s
r
ec
o
g
n
izin
g
b
o
th
class
es
well.
Pre
cisi
o
n
f
o
r
g
o
o
d
class
es
is
q
u
ite
h
ig
h
(
0
.
7
5
)
,
b
u
t
p
r
ec
is
io
n
f
o
r
p
o
o
r
class
es
i
s
v
er
y
lo
w
(
0
.
1
3
)
.
T
h
is
in
d
icate
s
th
at
th
e
m
o
d
el
ten
d
s
to
class
if
y
d
ata
in
to
m
ajo
r
ity
class
es
well
b
u
t
h
as
p
r
o
b
lem
s
class
if
y
in
g
d
ata
in
t
o
m
in
o
r
ity
class
es.
Similar
ly
,
r
ec
all
f
o
r
g
o
o
d
class
es is
q
u
ite
h
ig
h
(
0
.
8
1
)
,
b
u
t r
ec
all
f
o
r
p
o
o
r
class
es i
s
also
lo
w
(
0
.
1
0
)
.
L
o
w
r
ec
all
in
d
icate
s
th
at
th
e
m
o
d
el
f
ailed
to
id
en
tif
y
m
o
s
t o
f
th
e
d
ata
f
r
o
m
m
in
o
r
ity
class
es.
Sh
ah
an
d
Ma
n
ju
la
[
3
8
]
em
p
h
asized
th
at
wh
ile
VGG1
6
is
e
f
f
ec
tiv
e
f
o
r
b
a
lan
ce
d
d
atasets
,
its
p
er
f
o
r
m
an
ce
d
r
o
p
s
in
im
b
alan
ce
d
s
ce
n
ar
io
s
.
T
o
en
h
an
ce
t
h
e
clar
ity
an
d
c
o
m
p
r
eh
en
s
iv
e
n
ess
o
f
o
u
r
ev
al
u
atio
n
,
we
h
av
e
th
o
u
g
h
t
f
u
lly
co
m
p
iled
th
e
k
ey
p
er
f
o
r
m
a
n
ce
m
etr
ics f
o
r
e
ac
h
m
eth
o
d
i
n
T
ab
le
1
a
n
d
r
ep
r
esen
ted
th
e
m
eth
o
d
'
s
p
er
f
o
r
m
an
ce
g
r
ap
h
ically
in
Fig
u
r
e
3
.
T
h
e
r
esu
lts
o
f
o
u
r
ass
es
s
m
en
ts
r
ev
ea
l
lo
g
is
tic
r
e
g
r
ess
io
n
as
th
e
lead
in
g
p
er
f
o
r
m
er
,
ac
h
iev
in
g
an
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Ove
r
co
min
g
imb
a
la
n
ce
d
r
ice
s
ee
d
g
ermin
a
tio
n
cla
s
s
ifica
tio
n
…
(
Mu
h
la
s
a
h
N
o
vita
s
a
r
i Ma
r
a
)
69
im
p
r
ess
iv
e
ac
cu
r
ac
y
o
f
0
.
9
4
(
9
4
%),
with
F1
-
s
co
r
es o
f
0
.
9
6
f
o
r
t
h
e
"Go
o
d
"
class
an
d
0
.
8
1
f
o
r
t
h
e
"Po
o
r
"
class
.
R
an
d
o
m
f
o
r
est
f
o
ll
o
ws
clo
s
ely
b
eh
in
d
with
an
ac
cu
r
ac
y
o
f
0
.
9
1
(
9
1
%).
Ho
wev
er
,
a
m
o
r
e
d
etailed
ex
am
in
atio
n
o
f
p
r
ec
is
io
n
an
d
F1
-
s
co
r
es,
p
ar
ticu
lar
ly
f
o
r
th
e
m
in
o
r
ity
cl
ass
es
(
th
e
"
Po
o
r
"
class
)
,
u
n
co
v
er
s
a
s
ig
n
if
ican
t
lim
itatio
n
in
r
an
d
o
m
f
o
r
est
,
c
h
ar
ac
ter
ized
b
y
lo
w
p
r
ec
is
io
n
a
n
d
F1
-
Sco
r
es,
in
d
icatin
g
ch
all
en
g
es in
ef
f
ec
tiv
ely
id
en
tify
in
g
th
ese
m
in
o
r
ity
cla
s
s
es.
I
m
p
o
r
tan
tly
,
th
is
lim
itatio
n
e
x
ten
d
s
b
ey
o
n
d
r
an
d
o
m
f
o
r
est
to
e
n
co
m
p
ass
m
o
r
e
co
m
p
lex
m
o
d
els
lik
e
ANN,
VGG1
9
,
an
d
Alex
N
et.
Desp
ite
th
eir
h
ig
h
o
v
er
a
ll
ac
cu
r
ac
y
,
t
h
ese
s
o
p
h
is
ticated
m
o
d
els
e
x
h
ib
it
a
n
F1
-
s
co
r
e
o
f
0
f
o
r
th
e
m
in
o
r
it
y
class
.
T
h
is
o
b
s
er
v
atio
n
h
ig
h
l
ig
h
ts
th
e
s
u
b
s
tan
tial
im
p
ac
t
o
f
d
ata
im
b
ala
n
ce
,
wh
e
r
e
th
e
m
in
o
r
ity
class
is
n
o
tab
ly
u
n
d
er
r
ep
r
esen
ted
c
o
m
p
ar
e
d
to
th
e
m
ajo
r
ity
class
.
Similar
f
in
d
in
g
s
wer
e
r
ep
o
r
ted
b
y
L
u
o
et
a
l
.
[
3
9
]
,
wh
o
c
o
m
p
a
r
ed
th
e
p
er
f
o
r
m
a
n
ce
o
f
r
a
n
d
o
m
f
o
r
est
an
d
lo
g
is
tic
r
eg
r
ess
io
n
o
n
im
b
alan
ce
d
d
ata
s
ets,
n
o
tin
g
t
h
at
lo
g
is
tic
r
eg
r
es
s
io
n
o
f
ten
o
u
tp
er
f
o
r
m
s
r
an
d
o
m
f
o
r
est
in
te
r
m
s
o
f
h
an
d
lin
g
im
b
alan
ce
d
d
ata
.
T
h
is
s
tu
d
y
u
n
d
e
r
s
co
r
es
th
at
wh
ile
co
m
p
lex
m
o
d
els
s
u
ch
a
s
ANN,
VGG1
9
,
an
d
Alex
N
et
h
av
e
th
e
ca
p
ab
ilit
y
to
d
is
ce
r
n
in
tr
icate
im
ag
e
f
ea
tu
r
es.
Ho
wev
er
,
th
e
y
m
ig
h
t
n
o
t
b
e
th
e
m
o
s
t
s
u
itab
le
ch
o
ice
f
o
r
o
u
r
r
ice
s
p
r
o
u
ts
im
ag
e
class
if
icatio
n
ta
s
k
s
in
v
o
lv
in
g
im
b
alan
ce
d
d
at
a.
I
n
s
u
c
h
s
ce
n
ar
io
s
,
s
im
p
ler
m
o
d
els
lik
e
lo
g
is
tic
r
eg
r
ess
io
n
n
o
t
o
n
ly
y
ield
s
u
p
er
io
r
r
esu
lts
b
u
t
also
o
f
f
er
th
e
ad
v
an
tag
e
o
f
r
elativ
ely
s
h
o
r
t
er
ex
ec
u
tio
n
tim
es,
r
en
d
er
in
g
th
em
a
m
o
r
e
ef
f
icien
t
an
d
ef
f
ec
tiv
e
o
p
tio
n
.
T
h
e
s
e
f
in
d
in
g
s
em
p
h
asize
th
e
cr
i
tical
im
p
o
r
tan
ce
o
f
ju
d
icio
u
s
ly
s
elec
tin
g
m
o
d
els b
ased
o
n
th
e
s
p
ec
if
ic
c
h
ar
ac
ter
i
s
tics
o
f
th
e
d
ataset
an
d
th
e
o
b
j
ec
tiv
es a
t h
an
d
.
T
ab
le
1
.
C
o
m
p
r
eh
en
s
iv
e
c
o
m
p
ar
is
o
n
o
f
class
if
icatio
n
r
esu
lts
M
o
d
e
l
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
(
G
o
o
d
)
P
r
e
c
i
s
i
o
n
(
P
o
o
r
)
R
e
c
a
l
l
(
G
o
o
d
)
R
e
c
a
l
l
(
P
o
o
r
)
F
1
s
c
o
r
e
(
G
o
o
d
)
F
1
s
c
o
r
e
(
P
o
o
r
)
Ex
e
c
u
t
i
o
n
t
i
m
e
(
s)
Lo
g
i
s
t
i
c
r
e
g
r
e
ssi
o
n
0
.
9
4
0
.
9
7
0
.
7
7
0
.
9
5
0
.
8
5
0
.
9
6
0
.
8
1
1
6
.
6
9
R
a
n
d
o
m
f
o
r
e
s
t
0
.
9
1
1
.
0
0
0
.
4
7
0
.
9
0
1
.
0
0
0
.
9
5
0
.
6
4
1
1
.
9
3
K
N
N
0
.
3
6
0
.
2
4
0
.
9
7
0
.
9
7
0
.
2
0
0
.
3
9
0
.
3
4
0
.
2
1
N
a
ï
v
e
B
a
y
e
s
i
a
n
0
.
5
4
0
.
4
8
0
.
8
7
0
.
9
5
0
.
2
5
0
.
6
4
0
.
3
9
0
.
7
8
D
e
c
i
s
i
o
n
t
r
e
e
0
.
7
4
0
.
7
5
0
.
7
0
0
.
9
3
0
.
3
6
0
.
8
3
0
.
4
7
7
7
.
7
3
ANN
0
.
8
3
1
.
0
0
0
.
0
0
0
.
8
3
0
.
0
0
0
.
9
1
0
.
0
0
1
1
0
1
3
.
7
0
V
G
G
1
9
0
.
8
3
1
.
0
0
0
.
0
0
0
.
8
3
0
.
0
0
0
.
9
1
0
.
0
0
8
1
2
4
.
8
9
A
l
e
x
N
e
t
0
.
8
3
1
.
0
0
0
.
0
0
0
.
8
3
0
.
0
0
0
.
9
1
0
.
0
0
8
0
2
1
.
2
4
V
G
G
1
6
0
.
6
5
0
.
7
5
0
.
1
3
0
.
8
1
0
.
1
0
0
.
7
8
0
.
1
1
4
7
2
0
.
6
9
Fig
u
r
e
3
.
Me
th
o
d
’
s
p
e
r
f
o
r
m
an
ce
4.
CO
NCLU
SI
O
N
I
n
co
n
clu
s
io
n
,
o
u
r
e
v
alu
atio
n
o
f
v
ar
io
u
s
class
if
icatio
n
m
eth
o
d
s
f
o
r
r
ice
s
p
r
o
u
t
im
a
g
e
class
if
icatio
n
h
as
p
r
o
v
id
e
d
v
alu
a
b
le
in
s
ig
h
ts
.
L
o
g
is
tic
r
eg
r
ess
io
n
em
er
g
ed
as
th
e
to
p
p
er
f
o
r
m
e
r
with
an
im
p
r
ess
iv
e
ac
cu
r
ac
y
o
f
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
4
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
5
:
62
-
72
70
9
4
%,
d
em
o
n
s
tr
atin
g
its
ef
f
ec
ti
v
en
ess
in
th
i
s
task
.
R
an
d
o
m
f
o
r
est
f
o
llo
wed
clo
s
ely
with
a
n
ac
cu
r
ac
y
o
f
9
1
%,
alth
o
u
g
h
it
ex
h
ib
ited
lim
itatio
n
s
in
r
ec
o
g
n
izin
g
m
i
n
o
r
ity
class
es
d
u
e
to
lo
w
p
r
ec
is
io
n
an
d
F1
-
s
co
r
es.
T
h
is
lim
itatio
n
ex
ten
d
ed
t
o
m
o
r
e
co
m
p
lex
m
o
d
els
lik
e
ANN,
VGG1
9
,
an
d
Alex
Net,
wh
ic
h
,
d
esp
ite
th
eir
h
ig
h
o
v
er
all
ac
c
u
r
ac
y
,
s
tr
u
g
g
led
with
th
e
m
i
n
o
r
ity
class
.
T
h
i
s
u
n
d
er
s
co
r
es
th
e
im
p
ac
t
o
f
d
ata
im
b
alan
ce
in
class
if
icatio
n
task
s
.
Ou
r
s
tu
d
y
h
ig
h
lig
h
ts
th
e
im
p
o
r
tan
ce
o
f
m
o
d
el
s
elec
tio
n
b
ased
o
n
d
at
aset
ch
ar
ac
ter
is
tics
an
d
class
if
icatio
n
g
o
als.
W
h
i
le
co
m
p
lex
m
o
d
els
h
av
e
th
e
ca
p
ac
ity
to
an
aly
ze
in
tr
icate
im
ag
e
f
ea
tu
r
es,
th
ey
m
a
y
n
o
t
b
e
th
e
m
o
s
t
s
u
itab
le
ch
o
ic
e
f
o
r
task
s
in
v
o
lv
in
g
im
b
alan
c
ed
d
ata,
s
u
ch
as
r
ice
s
p
r
o
u
t
cla
s
s
if
icatio
n
.
I
n
s
u
ch
s
ce
n
ar
io
s
,
s
im
p
ler
m
o
d
els
lik
e
lo
g
is
tic
r
eg
r
ess
io
n
o
f
f
er
s
u
p
er
io
r
r
esu
lts
an
d
s
h
o
r
ter
e
x
ec
u
tio
n
tim
es,
m
ak
in
g
th
em
a
m
o
r
e
ef
f
icien
t
o
p
tio
n
.
T
h
ese
f
in
d
in
g
s
em
p
h
asize
th
e
n
ee
d
f
o
r
a
t
h
o
u
g
h
tf
u
l
a
p
p
r
o
ac
h
to
m
o
d
el
s
elec
tio
n
,
co
n
s
id
er
in
g
b
o
th
ac
cu
r
ac
y
an
d
th
e
s
p
ec
if
ic
ch
allen
g
es
p
o
s
ed
b
y
im
b
alan
ce
d
d
ata.
Ultim
ately
,
t
h
e
ch
o
ice
o
f
m
o
d
el
s
h
o
u
ld
alig
n
with
th
e
o
b
jectiv
es
o
f
th
e
clas
s
if
icatio
n
ta
s
k
an
d
th
e
ch
ar
ac
ter
is
tics
o
f
th
e
d
ataset
to
ac
h
iev
e
o
p
tim
al
r
esu
lts
.
Mo
v
in
g
f
o
r
w
ar
d
,
f
u
tu
r
e
r
esear
ch
s
h
o
u
ld
f
o
cu
s
o
n
d
ev
elo
p
in
g
n
o
v
el
tec
h
n
iq
u
es
to
ad
d
r
ess
im
b
alan
ce
d
d
ata
ch
allen
g
es
ef
f
ec
tiv
ely
.
Ad
d
itio
n
ally
,
e
x
p
lo
r
in
g
en
s
em
b
le
m
eth
o
d
s
th
at
c
o
m
b
in
e
th
e
s
tr
en
g
th
s
o
f
d
if
f
e
r
en
t
m
o
d
els
co
u
ld
o
f
f
er
im
p
r
o
v
ed
p
e
r
f
o
r
m
an
ce
in
h
an
d
lin
g
im
b
alan
ce
d
d
atasets
.
Mo
r
eo
v
er
,
in
v
esti
g
atin
g
tech
n
iq
u
es
f
o
r
g
en
er
atin
g
s
y
n
th
etic
d
ata
t
o
b
alan
ce
class
d
is
tr
ib
u
tio
n
s
a
n
d
r
ef
in
in
g
f
ea
tu
r
e
en
g
in
ee
r
in
g
p
r
o
ce
s
s
es
s
p
ec
if
ic
to
r
ice
s
p
r
o
u
t
im
a
g
es
co
u
ld
f
u
r
th
e
r
en
h
an
ce
class
if
icatio
n
ac
cu
r
ac
y
.
Ov
e
r
all,
co
n
tin
u
ed
ex
p
l
o
r
atio
n
i
n
th
e
s
e
d
ir
ec
tio
n
s
will
co
n
tr
ib
u
te
to
ad
v
an
cin
g
th
e
f
ield
o
f
r
ice
s
p
r
o
u
t
im
a
g
e
class
if
icatio
n
an
d
ad
d
r
ess
in
g
r
ea
l
-
wo
r
ld
ch
allen
g
es in
ag
r
icu
ltu
r
e
an
d
f
o
o
d
s
ec
u
r
ity
.
ACK
NO
WL
E
DG
E
M
E
NT
S
W
e
wo
u
ld
lik
e
to
ex
p
r
ess
o
u
r
d
ee
p
est
g
r
atitu
d
e
to
th
e
I
n
d
o
n
esian
Natio
n
al
R
e
s
ea
r
ch
an
d
I
n
n
o
v
atio
n
Ag
en
cy
f
o
r
th
e
f
u
n
d
in
g
p
r
o
v
i
d
ed
in
th
is
r
esear
ch
.
T
h
e
f
u
n
d
in
g
was
p
r
o
v
id
ed
in
c
o
llab
o
r
atio
n
with
th
e
E
d
u
ca
tio
n
Fu
n
d
Ma
n
ag
em
e
n
t
Ag
en
cy
o
f
th
e
Min
is
tr
y
o
f
Fin
an
ce
u
n
d
er
th
e
R
esear
ch
an
d
I
n
n
o
v
atio
n
f
o
r
Ad
v
an
ce
I
n
d
o
n
esia
f
u
n
d
i
n
g
s
ch
em
e
,
co
n
tr
ac
t
n
u
m
b
er
8
4
/I
V/KS/1
1
/2
0
2
2
.
W
e
also
ex
te
n
d
o
u
r
g
r
at
itu
d
e
to
Po
litek
n
ik
Neg
er
i
Sem
a
r
an
g
f
o
r
th
e
f
ac
i
litatio
n
p
r
o
v
id
e
d
.
I
n
ad
d
itio
n
,
we
wo
u
ld
also
lik
e
to
e
x
p
r
e
s
s
o
u
r
g
r
atitu
d
e
to
Po
litek
n
ik
Neg
er
i Jem
b
er
f
o
r
t
h
eir
ass
is
tan
ce
in
p
r
o
v
id
in
g
f
a
cilities
f
o
r
d
ata
co
llectio
n
o
f
r
i
ce
s
ee
d
s
p
r
o
u
ts
.
RE
F
E
R
E
NC
E
S
[
1
]
F
.
R
o
z
i
e
t
a
l
.
,
“
I
n
d
o
n
e
s
i
a
n
m
a
r
k
e
t
d
e
man
d
p
a
t
t
e
r
n
s
f
o
r
f
o
o
d
c
o
mm
o
d
i
t
y
so
u
r
c
e
s
o
f
c
a
r
b
o
h
y
d
r
a
t
e
s i
n
f
a
c
i
n
g
t
h
e
g
l
o
b
a
l
f
o
o
d
c
r
i
s
i
s,”
H
e
l
i
y
o
n
,
v
o
l
.
9
,
n
o
.
6
,
Ju
n
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
h
e
l
i
y
o
n
.
2
0
2
3
.
e
1
6
8
0
9
.
[
2
]
P
.
A
.
T
u
a
n
,
M
.
S
u
n
,
T.
N
.
N
g
u
y
e
n
,
S
.
P
a
r
k
,
a
n
d
B
.
T.
A
y
e
l
e
,
“
M
o
l
e
c
u
l
a
r
mec
h
a
n
i
sms
o
f
s
e
e
d
g
e
r
m
i
n
a
t
i
o
n
,
”
i
n
S
p
r
o
u
t
e
d
G
ra
i
n
s:
N
u
t
ri
t
i
o
n
a
l
V
a
l
u
e
,
Pr
o
d
u
c
t
i
o
n
,
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
El
s
e
v
i
e
r
,
2
0
1
8
,
p
p
.
1
–
24
,
d
o
i
:
1
0
.
1
0
1
6
/
B
9
7
8
-
0
-
12
-
8
1
1
5
2
5
-
1
.
0
0
0
0
1
-
4.
[
3
]
L.
O
.
C
o
p
e
l
a
n
d
a
n
d
M
.
B
.
M
c
D
o
n
a
l
d
,
“
S
e
e
d
E
n
h
a
n
c
e
m
e
n
t
s
,
”
i
n
Pr
i
n
c
i
p
l
e
s
o
f
S
e
e
d
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
B
o
s
t
o
n
,
M
A
:
S
p
r
i
n
g
e
r
U
S
,
2
0
0
1
,
p
p
.
2
7
7
–
2
9
6
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
1
-
4
6
1
5
-
1
6
1
9
-
4
_
1
3
.
[
4
]
N
.
G
e
n
z
e
,
R
.
B
h
a
r
t
i
,
M
.
G
r
i
e
b
,
S
.
J
.
S
c
h
u
l
t
h
e
i
s
s
,
a
n
d
D
.
G
.
G
r
i
m
m
,
“
A
c
c
u
r
a
t
e
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
-
b
a
s
e
d
g
e
r
m
i
n
a
t
i
o
n
d
e
t
e
c
t
i
o
n
,
p
r
e
d
i
c
t
i
o
n
a
n
d
q
u
a
l
i
t
y
a
s
s
e
s
s
m
e
n
t
o
f
t
h
r
e
e
g
r
a
i
n
c
r
o
p
s
,
”
P
l
a
n
t
M
e
t
h
o
d
s
,
v
o
l
.
1
6
,
n
o
.
1
,
D
e
c
.
2
0
2
0
,
d
o
i
:
1
0
.
1
1
8
6
/
s
1
3
0
0
7
-
020
-
0
0
6
9
9
-
x.
[
5
]
S
.
S
.
R
a
w
a
t
a
n
d
A
.
K
.
M
i
s
h
r
a
,
“
R
e
v
i
e
w
o
f
m
e
t
h
o
d
s
f
o
r
h
a
n
d
l
i
n
g
c
l
a
ss
-
i
m
b
a
l
a
n
c
e
d
i
n
c
l
a
ss
i
f
i
c
a
t
i
o
n
p
r
o
b
l
e
ms,”
a
rXi
v
-
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
p
p
.
1
-
1
5
,
2
0
2
2
,
d
o
i
:
1
0
.
4
8
5
5
0
/
a
r
X
i
v
.
2
2
1
1
.
0
5
4
5
6
.
[
6
]
M
.
C
.
A
.
P
r
a
b
o
w
o
a
n
d
S
.
S
.
H
i
d
a
y
a
t
,
“
E
d
g
e
D
e
t
e
c
t
i
o
n
t
e
c
h
n
i
q
u
e
f
o
r
r
i
c
e
q
u
a
l
i
t
y
a
n
a
l
y
s
i
s
u
s
i
n
g
d
i
g
i
t
a
l
i
ma
g
e
p
r
o
c
e
ssi
n
g
,
”
i
n
AI
P
C
o
n
f
e
re
n
c
e
Pr
o
c
e
e
d
i
n
g
s
,
2
0
2
3
,
v
o
l
.
2
4
3
1
,
n
o
.
1
,
d
o
i
:
1
0
.
1
0
6
3
/
5
.
0
1
1
7
5
1
0
.
[
7
]
S
.
S
.
H
i
d
a
y
a
t
,
D
.
R
a
h
m
a
w
a
t
i
,
M
.
C
.
A
.
P
r
a
b
o
w
o
,
L
.
T
r
i
y
o
n
o
,
a
n
d
F
.
T
.
P
u
t
r
i
,
“
D
e
t
e
r
m
i
n
i
n
g
t
h
e
r
i
c
e
s
e
e
d
s
q
u
a
l
i
t
y
u
s
i
n
g
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
n
I
n
f
o
r
m
a
t
i
c
s
V
i
s
u
a
l
i
z
a
t
i
o
n
,
v
o
l
.
7
,
n
o
.
2
,
p
p
.
5
2
7
–
5
3
4
,
J
u
n
.
2
0
2
3
,
d
o
i
:
1
0
.
3
0
6
3
0
/
j
o
i
v
.
7
.
2
.
1
1
7
5
.
[
8
]
B
.
Lu
r
st
w
u
t
a
n
d
C
.
P
o
r
n
p
a
n
o
mc
h
a
i
,
“
I
mag
e
a
n
a
l
y
s
i
s b
a
s
e
d
o
n
c
o
l
o
r
,
s
h
a
p
e
a
n
d
t
e
x
t
u
r
e
f
o
r
r
i
c
e
s
e
e
d
(
O
ryz
a
sa
t
i
v
a
L
.
)
g
e
r
mi
n
a
t
i
o
n
e
v
a
l
u
a
t
i
o
n
,
”
Ag
r
i
c
u
l
t
u
re
a
n
d
N
a
t
u
r
a
l
Re
so
u
r
c
e
s
,
v
o
l
.
5
1
,
n
o
.
5
,
p
p
.
3
8
3
–
3
8
9
,
O
c
t
.
2
0
1
7
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
a
n
r
e
s.
2
0
1
7
.
1
2
.
0
0
2
.
[
9
]
Y
.
G
u
l
z
a
r
,
Y
.
H
a
m
i
d
,
A
.
B
.
S
o
o
mr
o
,
A
.
A
.
A
l
w
a
n
,
a
n
d
L.
J
o
u
r
n
a
u
x
,
“
A
c
o
n
v
o
l
u
t
i
o
n
n
e
u
r
a
l
n
e
t
w
o
r
k
-
b
a
se
d
s
e
e
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
sy
st
e
m,
”
S
y
m
m
e
t
ry
,
v
o
l
.
1
2
,
n
o
.
1
2
,
p
p
.
1
–
1
8
,
D
e
c
.
2
0
2
0
,
d
o
i
:
1
0
.
3
3
9
0
/
sy
m
1
2
1
2
2
0
1
8
.
[
1
0
]
S
.
I
b
r
a
h
i
m
,
N
.
A
.
Z
u
l
k
i
f
l
i
,
N
.
S
a
b
r
i
,
A
.
A
.
S
h
a
r
i
,
a
n
d
M
.
R
.
M
.
N
o
o
r
d
i
n
,
“
R
i
c
e
g
r
a
i
n
c
l
a
ssi
f
i
c
a
t
i
o
n
u
si
n
g
mu
l
t
i
-
c
l
a
ss su
p
p
o
r
t
v
e
c
t
o
r
mac
h
i
n
e
(
S
V
M
)
,
”
I
AE
S
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
8
,
n
o
.
3
,
p
p
.
2
1
5
–
2
2
0
,
D
e
c
.
2
0
1
9
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
a
i
.
v
8
.
i
3
.
p
p
2
1
5
-
2
2
0
.
[
1
1
]
K
.
K
i
r
a
t
i
r
a
t
a
n
a
p
r
u
k
e
t
a
l
.
,
“
D
e
v
e
l
o
p
m
e
n
t
o
f
p
a
d
d
y
r
i
c
e
se
e
d
c
l
a
ssi
f
i
c
a
t
i
o
n
p
r
o
c
e
ss
u
s
i
n
g
mac
h
i
n
e
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
f
o
r
a
u
t
o
ma
t
i
c
g
r
a
d
i
n
g
ma
c
h
i
n
e
,
”
J
o
u
rn
a
l
o
f
S
e
n
s
o
rs
,
v
o
l
.
2
0
2
0
,
p
p
.
1
–
1
4
,
Ju
l
.
2
0
2
0
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
0
/
7
0
4
1
3
1
0
.
[
1
2
]
D
.
S
i
v
a
k
u
m
a
r
,
K
.
S
u
r
i
y
a
k
r
i
sh
n
a
a
n
,
P
.
A
k
s
h
a
y
a
,
G
.
V
.
A
n
u
j
a
,
a
n
d
G
.
T
.
D
e
v
a
d
h
a
r
sh
i
n
i
,
“
C
o
m
p
u
t
e
r
i
z
e
d
g
r
o
w
t
h
a
n
a
l
y
s
i
s
o
f
se
e
d
s
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
m
e
t
h
o
d
,
”
I
n
t
e
rn
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
Re
c
e
n
t
T
e
c
h
n
o
l
o
g
y
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
7
,
n
o
.
6
,
p
p
.
1
8
8
5
–
1
8
9
2
,
2
0
1
9
.
[
1
3
]
S
.
H
u
a
n
g
,
X
.
F
a
n
,
L.
S
u
n
,
Y
.
S
h
e
n
,
a
n
d
X
.
S
u
o
,
“
R
e
sea
r
c
h
o
n
c
l
a
ss
i
f
i
c
a
t
i
o
n
met
h
o
d
o
f
ma
i
z
e
s
e
e
d
d
e
f
e
c
t
b
a
s
e
d
o
n
mac
h
i
n
e
v
i
si
o
n
,
”
J
o
u
rn
a
l
o
f
S
e
n
s
o
rs
,
v
o
l
.
2
0
1
9
,
p
p
.
1
–
9
,
N
o
v
.
2
0
1
9
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
1
9
/
2
7
1
6
9
7
5
.
[
1
4
]
D
.
M
o
h
a
n
a
n
d
M
.
G
.
R
a
j
,
“
Q
u
a
l
i
t
y
a
n
a
l
y
s
i
s
o
f
r
i
c
e
g
r
a
i
n
s
u
s
i
n
g
A
N
N
a
n
d
S
V
M
,
”
J
o
u
rn
a
l
o
f
C
ri
t
i
c
a
l
R
e
v
i
e
w
s
,
v
o
l
.
7
,
n
o
.
1
,
p
p
.
3
9
5
–
4
0
2
,
J
a
n
.
2
0
2
0
,
d
o
i
:
1
0
.
3
1
8
3
8
/
j
c
r
.
0
7
.
0
1
.
7
9
.
[
1
5
]
S
.
B
.
A
h
me
d
,
S
.
F
.
A
l
i
,
a
n
d
A
.
Z.
K
h
a
n
,
“
O
n
t
h
e
f
r
o
n
t
i
e
r
s
o
f
r
i
c
e
g
r
a
i
n
a
n
a
l
y
si
s,
c
l
a
ss
i
f
i
c
a
t
i
o
n
a
n
d
q
u
a
l
i
t
y
g
r
a
d
i
n
g
:
a
r
e
v
i
e
w
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
9
,
p
p
.
1
6
0
7
7
9
–
1
6
0
7
9
6
,
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
1
.
3
1
3
0
4
7
2
.
[
1
6
]
P
.
S
a
x
e
n
a
,
K
.
P
r
i
y
a
,
S
.
G
o
e
l
,
P
.
K
.
A
g
g
a
r
w
a
l
,
A
.
S
i
n
h
a
,
a
n
d
P
.
J
a
i
n
,
“
R
i
c
e
v
a
r
i
e
t
i
e
s
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
mac
h
i
n
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
ms,
”
J
o
u
rn
a
l
o
f
Ph
a
rm
a
c
e
u
t
i
c
a
l
N
e
g
a
t
i
v
e
R
e
su
l
t
s
,
v
o
l
.
1
3
,
2
0
2
2
,
d
o
i
:
1
0
.
4
7
7
5
0
/
p
n
r
.
2
0
2
2
.
1
3
.
S
0
7
.
4
7
9
.
[
1
7
]
R
.
R
u
s
l
a
n
,
S
.
K
h
a
i
r
u
n
n
i
z
a
-
B
e
j
o
,
M
.
J
a
h
a
r
i
,
a
n
d
M
.
F
.
I
b
r
a
h
i
m
,
“
W
e
e
d
y
r
i
c
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
i
ma
g
e
p
r
o
c
e
ssi
n
g
a
n
d
a
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
,
”
A
g
ri
c
u
l
t
u
re
,
v
o
l
.
1
2
,
n
o
.
5
,
A
p
r
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
a
g
r
i
c
u
l
t
u
r
e
1
2
0
5
0
6
4
5
.
[
1
8
]
O
.
W
u
,
“
R
e
t
h
i
n
k
i
n
g
c
l
a
ss
i
mb
a
l
a
n
c
e
i
n
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
,”
a
rX
i
v
-
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
2
0
2
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
Ove
r
co
min
g
imb
a
la
n
ce
d
r
ice
s
ee
d
g
ermin
a
tio
n
cla
s
s
ifica
tio
n
…
(
Mu
h
la
s
a
h
N
o
vita
s
a
r
i Ma
r
a
)
71
[
1
9
]
A
.
G
o
s
a
i
n
a
n
d
S
.
S
a
r
d
a
n
a
,
“
H
a
n
d
l
i
n
g
c
l
a
s
s
i
m
b
a
l
a
n
c
e
p
r
o
b
l
e
m
u
s
i
n
g
o
v
e
r
s
a
m
p
l
i
n
g
t
e
c
h
n
i
q
u
e
s
:
A
r
e
v
i
e
w
,
”
i
n
2
0
1
7
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
A
d
v
a
n
c
e
s
i
n
C
o
m
p
u
t
i
n
g
,
C
o
m
m
u
n
i
c
a
t
i
o
n
s
a
n
d
I
n
f
o
r
m
a
t
i
c
s
,
S
e
p
.
2
0
1
7
,
p
p
.
7
9
–
85
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
C
C
I
.
2
0
1
7
.
8
1
2
5
8
2
0
.
[
2
0
]
S.
J.
B
a
sh
a
,
S
.
R
.
M
a
d
a
l
a
,
K
.
V
i
v
e
k
,
E
.
S
.
K
u
m
a
r
,
a
n
d
T.
A
mm
a
n
n
a
mm
a
,
“
A
r
e
v
i
e
w
o
n
i
mb
a
l
a
n
c
e
d
d
a
t
a
c
l
a
ss
i
f
i
c
a
t
i
o
n
t
e
c
h
n
i
q
u
e
s,
”
i
n
2
0
2
2
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Ad
v
a
n
c
e
d
C
o
m
p
u
t
i
n
g
T
e
c
h
n
o
l
o
g
i
e
s
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s,
I
C
AC
T
A
2
0
2
2
,
M
a
r
.
2
0
2
2
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
C
TA
5
4
4
8
8
.
2
0
2
2
.
9
7
5
3
3
9
2
.
[
2
1
]
J.
M
.
J
o
h
n
so
n
a
n
d
T.
M
.
K
h
o
s
h
g
o
f
t
a
a
r
,
“
S
u
r
v
e
y
o
n
d
e
e
p
l
e
a
r
n
i
n
g
w
i
t
h
c
l
a
ss
i
m
b
a
l
a
n
c
e
,
”
J
o
u
rn
a
l
o
f
Bi
g
D
a
t
a
,
v
o
l
.
6
,
n
o
.
1
,
D
e
c
.
2
0
1
9
,
d
o
i
:
1
0
.
1
1
8
6
/
s4
0
5
3
7
-
0
1
9
-
0
1
9
2
-
5.
[
2
2
]
K
.
R
.
M
a
h
mu
d
a
h
,
F
.
I
n
d
r
i
a
n
i
,
Y
.
Ta
k
e
mo
r
i
‐
sa
k
a
i
,
Y
.
I
w
a
t
a
,
T
.
W
a
d
a
,
a
n
d
K
.
S
a
t
o
u
,
“
C
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
i
mb
a
l
a
n
c
e
d
d
a
t
a
r
e
p
r
e
se
n
t
e
d
a
s b
i
n
a
r
y
f
e
a
t
u
r
e
s,
”
A
p
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
1
1
,
n
o
.
1
7
,
A
u
g
.
2
0
2
1
,
d
o
i
:
1
0
.
3
3
9
0
/
a
p
p
1
1
1
7
7
8
2
5
.
[
2
3
]
P
.
K
u
m
a
r
,
R
.
B
h
a
t
n
a
g
a
r
,
K
.
G
a
u
r
,
a
n
d
A
.
B
h
a
t
n
a
g
a
r
,
“
C
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
i
m
b
a
l
a
n
c
e
d
d
a
t
a
:
r
e
v
i
e
w
o
f
me
t
h
o
d
s
a
n
d
a
p
p
l
i
c
a
t
i
o
n
s
,
”
I
O
P
C
o
n
f
e
re
n
c
e
S
e
r
i
e
s:
Ma
t
e
r
i
a
l
s
S
c
i
e
n
c
e
a
n
d
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
1
0
9
9
,
n
o
.
1
,
M
a
r
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
8
8
/
1
7
5
7
-
8
9
9
x
/
1
0
9
9
/
1
/
0
1
2
0
7
7
.
[
2
4
]
Y
.
F
e
n
g
,
M
.
Z
h
o
u
,
a
n
d
X
.
T
o
n
g
,
“
I
mb
a
l
a
n
c
e
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
:
A
p
a
r
a
d
i
g
m
-
b
a
sed
r
e
v
i
e
w
,
”
S
t
a
t
i
st
i
c
a
l
An
a
l
y
si
s
a
n
d
D
a
t
a
M
i
n
i
n
g
,
v
o
l
.
1
4
,
n
o
.
5
,
p
p
.
3
8
3
–
4
0
6
,
O
c
t
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
0
2
/
sam
.
1
1
5
3
8
.
[
2
5
]
H
.
S
u
r
y
o
n
o
,
H
.
K
u
sw
a
n
t
o
,
a
n
d
N
.
I
r
i
a
w
a
n
,
“
R
i
c
e
p
h
e
n
o
l
o
g
y
c
l
a
ssi
f
i
c
a
t
i
o
n
b
a
se
d
o
n
r
a
n
d
o
m
f
o
r
e
s
t
a
l
g
o
r
i
t
h
m
f
o
r
d
a
t
a
i
mb
a
l
a
n
c
e
u
si
n
g
G
o
o
g
l
e
E
a
r
t
h
e
n
g
i
n
e
,
”
P
ro
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
1
9
7
,
p
p
.
6
6
8
–
6
7
6
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s.
2
0
2
1
.
1
2
.
2
0
1
.
[
2
6
]
M
.
Zh
e
n
g
,
F
.
W
a
n
g
,
X
.
H
u
,
Y
.
M
i
a
o
,
H
.
C
a
o
,
a
n
d
M
.
Ta
n
g
,
“
A
me
t
h
o
d
f
o
r
a
n
a
l
y
z
i
n
g
t
h
e
p
e
r
f
o
r
m
a
n
c
e
i
m
p
a
c
t
o
f
i
mb
a
l
a
n
c
e
d
b
i
n
a
r
y
d
a
t
a
o
n
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
m
o
d
e
l
s,”
A
x
i
o
m
s
,
v
o
l
.
1
1
,
n
o
.
1
1
,
N
o
v
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
a
x
i
o
ms
1
1
1
1
0
6
0
7
.
[
2
7
]
K
.
O
k
s
u
z
,
B
.
C
.
C
a
m,
S
.
K
a
l
k
a
n
,
a
n
d
E.
A
k
b
a
s,
“
I
mb
a
l
a
n
c
e
p
r
o
b
l
e
ms
i
n
o
b
j
e
c
t
d
e
t
e
c
t
i
o
n
:
a
r
e
v
i
e
w
,
”
I
E
EE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
P
a
t
t
e
r
n
An
a
l
y
si
s
a
n
d
M
a
c
h
i
n
e
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
4
3
,
n
o
.
1
0
,
p
p
.
3
3
8
8
–
3
4
1
5
,
O
c
t
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
TPA
M
I
.
2
0
2
0
.
2
9
8
1
8
9
0
.
[
2
8
]
J
.
H
a
,
M
.
K
a
m
b
e
,
a
n
d
J
.
P
e
,
D
a
t
a
m
i
n
i
n
g
:
c
o
n
c
e
p
t
s
a
n
d
t
e
c
h
n
i
q
u
e
s
,
B
u
r
l
i
n
g
t
o
n
:
M
o
r
g
a
n
K
a
u
f
m
a
n
n
,
2
0
1
1
,
d
o
i
:
1
0
.
1
0
1
6
/
C
2
0
0
9
-
0
-
61819
-
5.
[
2
9
]
H
.
A
.
A
.
R
a
h
m
a
n
a
n
d
B
.
W
.
Y
a
p
,
“
I
mb
a
l
a
n
c
e
e
f
f
e
c
t
s
o
n
c
l
a
ssi
f
i
c
a
t
i
o
n
u
s
i
n
g
b
i
n
a
r
y
l
o
g
i
s
t
i
c
r
e
g
r
e
ss
i
o
n
,
”
i
n
C
o
m
m
u
n
i
c
a
t
i
o
n
s
i
n
C
o
m
p
u
t
e
r
a
n
d
I
n
f
o
rm
a
t
i
o
n
S
c
i
e
n
c
e
,
v
o
l
.
6
5
2
,
2
0
1
6
,
p
p
.
1
3
6
–
1
4
7
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
981
-
10
-
2
7
7
7
-
2
_
1
2
.
[
3
0
]
H
.
A
.
A
.
R
a
h
ma
n
,
Y
.
B
.
W
a
h
,
a
n
d
O
.
S
.
H
u
a
t
,
“
P
r
e
d
i
c
t
i
v
e
p
e
r
f
o
r
ma
n
c
e
o
f
l
o
g
i
st
i
c
r
e
g
r
e
ssi
o
n
f
o
r
i
m
b
a
l
a
n
c
e
d
d
a
t
a
w
i
t
h
c
a
t
e
g
o
r
i
c
a
l
c
o
v
a
r
i
a
t
e
,
”
Pe
rt
a
n
i
k
a
J
o
u
r
n
a
l
o
f
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
2
8
,
n
o
.
4
,
p
p
.
1
1
4
1
–
1
1
6
1
,
O
c
t
.
2
0
2
0
,
d
o
i
:
1
0
.
4
7
8
3
6
/
p
j
s
t
.
2
8
.
4
.
0
2
.
[
3
1
]
Q
.
S
h
u
,
T
.
H
u
,
a
n
d
S
.
L
i
u
,
“
R
a
n
d
o
m
f
o
r
e
st
a
l
g
o
r
i
t
h
m
b
a
s
e
d
o
n
G
A
N
f
o
r
I
mb
a
l
a
n
c
e
d
d
a
t
a
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
J
o
u
r
n
a
l
o
f
P
h
y
s
i
c
s
:
C
o
n
f
e
re
n
c
e
S
e
r
i
e
s
,
v
o
l
.
1
5
4
4
,
n
o
.
1
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
8
8
/
1
7
4
2
-
6
5
9
6
/
1
5
4
4
/
1
/
0
1
2
0
1
4
.
[
3
2
]
A
.
S
.
M
o
r
e
a
n
d
D
.
P
.
R
a
n
a
,
“
R
e
v
i
e
w
o
f
r
a
n
d
o
m
f
o
r
e
s
t
c
l
a
ss
i
f
i
c
a
t
i
o
n
t
e
c
h
n
i
q
u
e
s
t
o
r
e
s
o
l
v
e
d
a
t
a
i
m
b
a
l
a
n
c
e
,
”
i
n
1
st
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
I
n
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
a
n
d
I
n
f
o
rm
a
t
i
o
n
M
a
n
a
g
e
m
e
n
t
,
O
c
t
.
2
0
1
7
,
p
p
.
7
2
–
78
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
I
S
I
M
.
2
0
1
7
.
8
1
2
2
1
5
1
.
[
3
3
]
T.
M
.
K
h
o
sh
g
o
f
t
a
a
r
,
M
.
G
o
l
a
w
a
l
a
,
a
n
d
J.
V
a
n
H
u
l
se,
“
A
n
e
m
p
i
r
i
c
a
l
s
t
u
d
y
o
f
l
e
a
r
n
i
n
g
f
r
o
m
i
m
b
a
l
a
n
c
e
d
d
a
t
a
u
s
i
n
g
r
a
n
d
o
m fo
r
e
st
,
”
i
n
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
T
o
o
l
s
w
i
t
h
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
O
c
t
.
2
0
0
7
,
v
o
l
.
2
,
p
p
.
3
1
0
–
317
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
TA
I
.
2
0
0
7
.
4
6
.
[
3
4
]
B
.
S
u
n
a
n
d
H
.
C
h
e
n
,
“
A
S
u
r
v
e
y
o
f
k
n
e
a
r
e
s
t
n
e
i
g
h
b
o
r
a
l
g
o
r
i
t
h
ms
f
o
r
so
l
v
i
n
g
t
h
e
c
l
a
ss
i
mb
a
l
a
n
c
e
d
p
r
o
b
l
e
m
,
”
Wi
r
e
l
e
ss
C
o
m
m
u
n
i
c
a
t
i
o
n
s
a
n
d
M
o
b
i
l
e
C
o
m
p
u
t
i
n
g
,
v
o
l
.
2
0
2
1
,
n
o
.
1
,
J
a
n
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
5
5
/
2
0
2
1
/
5
5
2
0
9
9
0
.
[
3
5
]
W
.
N
.
L
.
W
.
H
.
I
b
e
n
i
,
M
.
Z
.
M
.
S
a
l
i
k
o
n
,
A
.
M
u
st
a
p
h
a
,
S
.
A
.
D
a
u
d
,
a
n
d
M
.
N
.
M
.
S
a
l
l
e
h
,
“
C
o
mp
a
r
a
t
i
v
e
a
n
a
l
y
si
s
o
n
b
a
y
e
s
i
a
n
c
l
a
ss
i
f
i
c
a
t
i
o
n
f
o
r
b
r
e
a
st
c
a
n
c
e
r
p
r
o
b
l
e
m,
”
Bu
l
l
e
t
i
n
o
f
El
e
c
t
ri
c
a
l
E
n
g
i
n
e
e
ri
n
g
a
n
d
I
n
f
o
rm
a
t
i
c
s
,
v
o
l
.
8
,
n
o
.
4
,
D
e
c
.
2
0
1
9
,
d
o
i
:
1
0
.
1
1
5
9
1
/
e
e
i
.
v
8
i
4
.
1
6
2
8
.
[
3
6
]
C
.
O
.
Tr
u
i
c
ǎ
a
n
d
C
.
A
.
L
e
o
r
d
e
a
n
u
,
“
C
l
a
ssi
f
i
c
a
t
i
o
n
o
f
a
n
i
m
b
a
l
a
n
c
e
d
d
a
t
a
se
t
u
si
n
g
d
e
c
i
s
i
o
n
t
r
e
e
a
l
g
o
r
i
t
h
ms,
”
U
PB
S
c
i
e
n
t
i
f
i
c
B
u
l
l
e
t
i
n
,
S
e
r
i
e
s
C
:
E
l
e
c
t
r
i
c
a
l
En
g
i
n
e
e
r
i
n
g
a
n
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
7
9
,
n
o
.
4
,
p
p
.
6
9
–
8
4
,
2
0
1
7
.
[
3
7
]
S
.
W
a
n
g
,
W
.
L
i
u
,
J
.
W
u
,
L.
C
a
o
,
Q
.
M
e
n
g
,
a
n
d
P
.
J.
K
e
n
n
e
d
y
,
“
Tr
a
i
n
i
n
g
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
s
o
n
i
m
b
a
l
a
n
c
e
d
d
a
t
a
se
t
s,
”
i
n
Pro
c
e
e
d
i
n
g
s
o
f
t
h
e
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
i
n
t
C
o
n
f
e
r
e
n
c
e
o
n
N
e
u
r
a
l
N
e
t
w
o
rks
,
v
o
l
.
2
0
1
6
,
p
p
.
4
3
6
8
–
4
3
7
4
,
Ju
l
.
2
0
1
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
JC
N
N
.
2
0
1
6
.
7
7
2
7
7
7
0
.
[
3
8
]
H
.
S
h
a
h
a
n
d
V
.
M
a
n
j
u
l
a
,
“
A
p
e
r
f
o
r
m
a
n
c
e
a
n
a
l
y
s
i
s
o
f
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s
u
s
i
n
g
k
u
z
u
s
h
i
j
i
c
h
a
r
a
c
t
e
r
r
e
c
o
g
n
i
t
i
o
n
,
”
i
n
2
0
2
0
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
D
e
c
i
s
i
o
n
A
i
d
S
c
i
e
n
c
e
s
a
n
d
A
p
p
l
i
c
a
t
i
o
n
,
p
p
.
1
0
6
8
–
1
0
7
1
,
N
o
v
.
2
0
2
0
,
d
o
i
:
1
0
.
1
1
0
9
/
D
A
S
A
5
1
4
0
3
.
2
0
2
0
.
9
3
1
7
1
8
8
.
[
3
9
]
H
.
L
u
o
,
X
.
P
a
n
,
Q
.
W
a
n
g
,
S
.
Y
e
,
a
n
d
Y
.
Q
i
a
n
,
“
L
o
g
i
s
t
i
c
r
e
g
r
e
s
s
i
o
n
a
n
d
r
a
n
d
o
m
f
o
r
e
s
t
f
o
r
e
f
f
e
c
t
i
v
e
i
m
b
a
l
a
n
c
e
d
c
l
a
s
s
i
f
i
c
a
t
i
o
n
,
”
i
n
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
m
p
u
t
e
r
S
o
f
t
w
a
r
e
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
C
o
n
f
e
r
e
n
c
e
,
v
o
l
.
1
,
p
p
.
9
1
6
–
9
1
7
,
J
u
l
.
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
C
O
M
P
S
A
C
.
2
0
1
9
.
0
0
1
3
9
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Mu
h
la
s
a
h
No
v
ita
s
a
r
i
Ma
r
a
h
o
ld
s
b
o
th
b
a
c
h
e
l
o
r
a
n
d
M
a
ste
r
o
f
M
a
th
e
m
a
ti
c
s
d
e
g
re
e
fro
m
Un
i
v
e
rsitas
G
a
d
jah
M
a
d
a
,
In
d
o
n
e
sia
.
S
h
e
is
c
u
rre
n
tl
y
a
lec
tu
re
r
a
t
De
p
a
rtme
n
t
o
f
Tele
c
o
m
m
u
n
ica
ti
o
n
E
n
g
i
n
e
e
rin
g
S
c
ien
c
e
i
n
P
o
l
it
e
k
n
ik
Ne
g
e
ri
S
e
m
a
ra
n
g
,
In
d
o
n
e
sia
.
He
r
re
se
a
rc
h
p
ro
we
ss
e
x
ten
d
s
to
m
a
c
h
in
e
lea
rn
i
n
g
,
d
a
ta
m
i
n
in
g
,
a
n
d
ti
m
e
se
ries
fo
re
c
a
stin
g
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
m
u
h
las
a
h
@p
o
li
n
e
s.a
c
.
id
.
S
id
iq
S
y
a
m
sul
H
i
d
a
y
a
t
n
o
ta
b
le
c
o
n
tr
ib
u
ti
o
n
s
in
th
e
d
e
v
e
lo
p
m
e
n
t
o
f
a
ro
b
o
t
f
o
r
p
rima
ry
sc
h
o
o
l
lea
rn
i
n
g
,
th
e
d
e
sig
n
o
f
h
i
g
h
-
fre
q
u
e
n
c
y
tran
sm
issio
n
li
n
e
s
u
sin
g
m
icro
stri
p
tec
h
n
o
l
o
g
y
,
a
n
d
th
e
c
re
a
ti
o
n
o
f
a
sm
a
rt
E
-
He
a
lt
h
a
p
p
li
c
a
ti
o
n
f
o
r
m
o
n
i
to
ri
n
g
p
a
ti
e
n
ts'
d
a
il
y
a
c
ti
v
it
ies
.
His
b
a
c
h
e
lo
r’s
in
e
lec
tri
c
a
l
e
n
g
in
e
e
rin
g
fro
m
Un
i
v
e
rsitas
Dip
o
n
e
g
o
ro
,
I
n
d
o
n
e
sia
,
m
a
ste
r’s
in
e
lec
tri
c
a
l
e
n
g
in
e
e
rin
g
fro
m
In
stit
u
t
Te
k
n
o
lo
g
i
Ba
n
d
u
n
g
,
In
d
o
n
e
sia
,
a
n
d
Do
c
to
r
in
In
telli
g
e
n
t
In
tera
c
ti
o
n
Tec
h
n
o
lo
g
y
fro
m
th
e
Un
iv
e
rsit
y
o
f
Tsu
k
u
b
a
,
Ja
p
a
n
.
He
c
a
n
b
e
c
o
n
t
a
c
ted
a
t
e
m
a
il
:
sid
iq
sh
@
p
o
l
in
e
s.a
c
.
id
.
Evaluation Warning : The document was created with Spire.PDF for Python.