I
n
t
e
r
n
at
i
on
al
Jou
r
n
al
of
E
l
e
c
t
r
i
c
al
an
d
C
om
p
u
t
e
r
E
n
gi
n
e
e
r
i
n
g (
I
JE
C
E
)
V
ol
.
11
, N
o.
1
,
F
e
br
ua
r
y
2021
, pp.
879~
891
I
S
S
N
:
2088
-
8708
,
D
O
I
:
10.11591/
i
j
e
c
e
.
v
11
i
1
.
pp
879
-
891
879
Jou
r
n
al
h
om
e
page
:
ht
t
p:
/
/
i
j
e
c
e
.i
ae
s
c
or
e
.c
om
S
i
m
i
l
ar
i
t
y
-
p
r
e
se
r
vi
n
g
h
a
s
h
f
or
c
o
n
t
e
n
t
-
b
ase
d
a
u
d
i
o r
e
t
r
i
e
val
u
si
n
g
u
n
s
u
p
e
r
v
i
se
d
d
e
e
p
n
e
u
r
a
l
n
e
t
w
or
k
s
P
e
t
c
h
ar
at
P
an
ya
p
a
n
u
w
at
, S
u
w
at
c
h
ai
K
am
on
s
a
n
t
i
r
oj
, L
u
e
p
ol
P
i
p
an
m
ae
k
ap
or
n
De
p
a
r
tm
e
n
t
o
f
C
o
m
p
u
ter
a
n
d
I
n
f
o
r
m
a
ti
o
n
S
c
i
e
n
c
e
,
K
i
n
g
M
o
n
g
k
u
t’
s
U
n
iv
e
r
s
i
ty
o
f
T
e
c
h
n
o
l
o
g
y
No
r
th
B
a
n
g
k
o
k
,
T
h
a
il
a
n
d
A
r
t
i
c
l
e
I
n
f
o
A
B
S
T
R
A
C
T
A
r
t
i
c
l
e
h
i
s
t
o
r
y
:
R
e
c
e
i
v
e
d J
a
n 1, 2020
R
e
v
i
s
e
d J
un 8, 2020
A
c
c
e
pt
e
d
A
u
g
18, 2020
Du
e
to
i
ts
e
f
f
i
c
i
e
n
c
y
i
n
s
to
r
a
g
e
a
n
d
s
e
a
r
c
h
s
p
e
e
d
,
b
i
n
a
r
y
h
a
s
h
i
n
g
h
a
s
b
e
c
o
m
e
a
n
a
tt
r
a
c
ti
v
e
a
p
p
r
o
a
c
h
f
o
r
a
l
a
r
g
e
a
u
d
i
o
d
a
tab
a
s
e
s
e
a
r
c
h
.
Ho
we
v
e
r
,
m
o
s
t
e
x
i
s
ti
n
g
h
a
s
h
i
n
g
-
b
a
s
e
d
m
e
th
o
d
s
f
o
c
u
s
o
n
d
a
ta
-
i
n
d
e
p
e
n
d
e
n
t
s
c
h
e
m
e
wh
e
r
e
r
a
n
d
o
m
li
n
e
a
r
p
r
o
jec
ti
o
n
s
o
r
s
o
m
e
a
r
i
th
m
e
ti
c
e
x
p
r
e
s
s
i
o
n
a
r
e
u
s
e
d
to
c
o
n
s
tr
u
c
t
h
a
s
h
f
u
n
c
ti
o
n
s
.
He
n
c
e
,
th
e
b
i
n
a
r
y
c
o
d
e
s
d
o
n
o
t
p
r
e
s
e
r
v
e
th
e
s
i
m
il
a
r
i
ty
a
n
d
m
a
y
d
e
g
r
a
d
e
th
e
s
e
a
r
c
h
p
e
r
f
o
r
m
a
n
c
e
.
I
n
th
i
s
p
a
p
e
r
,
a
n
u
n
s
u
p
e
r
v
i
s
e
d
s
i
m
il
a
r
i
ty
-
p
r
e
s
e
r
v
i
n
g
h
a
s
h
i
n
g
m
e
th
o
d
f
o
r
c
o
n
te
n
t
-
b
a
s
e
d
a
u
d
i
o
r
e
tr
i
e
v
a
l
i
s
p
r
o
p
o
s
e
d
.
Di
f
f
e
r
e
n
t
f
r
o
m
d
a
ta
-
i
n
d
e
p
e
n
d
e
n
t
h
a
s
h
i
n
g
m
e
th
o
d
s
,
we
d
e
v
e
l
o
p
a
d
e
e
p
n
e
two
r
k
to
l
e
a
r
n
c
o
m
p
a
c
t
b
i
n
a
r
y
c
o
d
e
s
f
r
o
m
m
u
l
ti
p
l
e
h
i
e
r
a
r
c
h
i
c
a
l
l
a
y
e
r
s
o
f
n
o
n
l
i
n
e
a
r
a
n
d
li
n
e
a
r
tr
a
n
s
f
o
r
m
a
ti
o
n
s
s
u
c
h
th
a
t
t
h
e
s
i
m
il
a
r
i
ty
b
e
twe
e
n
s
a
m
p
l
e
s
i
s
p
r
e
s
e
r
v
e
d
.
T
h
e
i
n
d
e
p
e
n
d
e
n
c
e
a
n
d
b
a
l
a
n
c
e
p
r
o
p
e
r
ti
e
s
a
r
e
i
n
c
l
u
d
e
d
an
d
o
p
ti
m
i
z
e
d
i
n
th
e
o
b
jec
ti
v
e
f
u
n
c
ti
o
n
to
i
m
p
r
o
v
e
th
e
c
o
d
e
s
.
E
x
p
e
r
i
m
e
n
tal
r
e
s
u
l
ts
o
n
th
e
E
x
te
n
d
e
d
B
a
ll
r
o
o
m
d
a
tas
e
t
wi
th
8
g
e
n
r
e
s
o
f
3
,
0
0
0
m
u
s
i
c
a
l
e
x
c
e
r
p
ts
s
h
o
w
th
a
t
o
u
r
p
r
o
p
o
s
e
d
m
e
t
h
o
d
s
i
g
n
i
f
i
c
a
n
t
l
y
o
u
tp
e
r
f
o
r
m
s
s
tate
-
of
-
th
e
-
a
r
t
d
a
ta
-
i
n
d
e
p
e
n
d
e
n
t
m
e
th
o
d
i
n
b
o
t
h
e
f
f
e
c
ti
v
e
n
e
s
s
a
n
d
e
f
f
i
c
i
e
n
c
y
.
K
e
y
w
or
d
s
:
C
ont
e
nt
-
ba
s
e
d a
udi
o r
e
t
r
i
e
v
a
l
D
e
e
p l
e
a
r
ni
n
g
D
e
e
p ne
ur
a
l
ne
t
w
or
ks
S
i
m
i
l
a
r
i
t
y
-
pr
e
s
e
r
v
i
n
g
ha
s
h
U
ns
upe
r
v
i
s
e
d l
e
a
r
ni
n
g
T
h
i
s
i
s
a
n
o
p
e
n
a
c
c
e
s
s
a
r
ti
c
l
e
u
n
d
e
r
th
e
C
C
B
Y
-
SA
l
i
c
e
n
s
e
.
C
or
r
e
s
pon
di
n
g A
u
t
h
or
:
P
e
t
c
ha
r
a
t
P
a
ny
a
pa
nu
w
a
t
,
D
e
pa
r
t
m
e
nt
of
C
o
m
put
e
r
a
nd
I
n
f
or
m
a
t
i
on S
c
i
e
nc
e
,
K
i
n
g
M
ong
kut
’
s
U
ni
v
e
r
s
i
t
y
of
T
e
c
hnol
ogy
N
or
t
h B
a
n
g
kok,
B
a
ng
kok, T
ha
i
l
a
nd
.
E
m
a
i
l
:
pa
ny
a
p
e
t
c
h@
hot
m
a
i
l
.c
o
m
1.
I
N
T
R
O
D
U
C
T
I
O
N
W
i
t
h
r
a
pi
dl
y
g
r
o
w
i
n
g
da
t
a
ba
s
e
of
di
g
i
t
a
l
a
udi
o
r
e
c
or
di
ng
s
,
t
he
no
v
e
l
r
e
t
r
i
e
v
a
l
s
t
r
a
t
e
gi
e
s
ha
v
e
r
e
c
e
i
v
e
d
g
r
e
a
t
a
t
t
e
nt
i
on.
E
a
r
l
y
r
e
t
r
i
e
v
a
l
a
ppr
oa
c
h
us
e
s
t
e
xt
ua
l
m
e
t
a
da
t
a
de
s
c
r
i
bi
ng
t
he
c
ont
e
nt
of
m
us
i
c
a
udi
o
(
e
.g
.,
a
r
t
i
s
t
na
m
e
,
s
ong
t
i
t
l
e
,
a
l
bum
na
m
e
,
g
e
nr
e
,
or
r
e
l
e
a
s
e
y
e
a
r
of
m
us
i
c
)
.
I
n
c
a
s
e
s
uc
h
de
s
c
r
i
pt
i
ons
a
r
e
not
a
v
a
i
l
a
b
l
e
,
i
t
i
s
r
e
qu
i
r
e
d
c
o
nt
e
n
t
-
ba
s
e
d
r
e
t
r
i
e
v
a
l
s
t
r
a
t
e
g
y
t
h
a
t
t
he
pe
r
c
e
p
t
ua
l
a
s
pe
c
t
s
of
t
h
e
a
u
d
i
o
a
r
e
ut
i
l
i
z
e
d
.
[
1]
.
C
ont
e
nt
-
ba
s
e
d
a
udi
o
r
e
t
r
i
e
v
a
l
a
ppr
o
a
c
h
i
s
g
e
ne
r
a
l
l
y
s
ol
v
e
d
w
i
t
h
t
w
o
s
t
e
ps
:
f
i
r
s
t
,
f
e
a
t
ur
e
s
a
r
e
e
xt
r
a
c
t
e
d
f
r
o
m
t
he
a
udi
o
f
i
l
e
a
nd
t
he
n
us
e
d
t
o
bui
l
d
i
nde
xe
s
f
or
s
e
a
r
c
hi
ng
.
T
w
o
m
a
i
n
i
s
s
ue
s
of
pe
r
f
or
m
i
n
g
a
s
e
a
r
c
h
ov
e
r
a
l
a
r
g
e
d
a
t
a
b
a
s
e
a
r
e
s
e
a
r
c
h
s
pe
e
d
a
n
d
e
f
f
i
c
i
e
n
t
s
t
o
r
a
g
e
.
T
he
m
o
s
t
i
n
t
e
r
e
s
t
i
ng
a
ppr
oa
c
h
f
o
r
ha
n
d
l
i
ng
t
h
e
s
e
pr
o
bl
e
m
s
i
s
b
i
n
a
r
y
ha
s
hi
ng
,
w
he
r
e
t
he
h
i
g
h
-
di
m
e
n
s
i
ona
l
f
e
a
t
ur
e
s
a
r
e
e
n
c
o
de
d
i
n
t
o
c
om
pa
c
t
b
i
na
r
y
c
ode
s
.
T
he
r
e
ha
v
e
be
e
n
s
e
v
e
r
a
l
ha
s
hi
ng
m
e
t
hods
pr
opos
e
d
i
n
t
he
l
i
t
e
r
a
t
ur
e
.
T
he
y
c
a
n
be
de
v
i
de
d
i
nt
o
t
w
o
c
a
t
e
g
or
i
e
s
,
da
t
a
-
i
n
de
pe
n
de
n
t
m
e
t
h
od
s
a
nd
da
t
a
-
de
pe
nd
e
n
t
m
e
t
h
o
ds
.
M
e
t
h
o
ds
i
n
da
t
a
-
i
nd
e
p
e
n
de
n
t
c
a
t
e
g
or
y
[2
-
7
]
us
e
r
a
ndo
m
l
i
ne
a
r
pr
oj
e
c
t
i
ons
or
s
o
m
e
a
r
i
t
h
m
e
t
i
c
e
xpr
e
s
s
i
on t
o c
ons
t
r
uc
t
ha
s
h f
unc
t
i
ons
. W
i
t
hout
t
he
t
r
a
i
ni
ng
pr
oc
e
s
s
,
t
he
y
a
r
e
r
obus
t
t
o
da
t
a
v
a
r
i
a
t
i
on.
H
o
w
e
v
e
r
,
s
uc
h
m
e
t
hods
r
e
qui
r
e
l
ong
ha
s
h
c
ode
s
t
o
a
c
hi
e
v
e
hi
g
h
pr
e
c
i
s
i
on. T
hi
s
i
nc
r
e
a
s
e
s
t
he
s
t
or
a
ge
c
os
t
a
nd de
g
r
a
de
s
t
he
s
e
a
r
c
h e
f
f
i
c
i
e
nc
y
[
8]
.
M
e
t
hods
i
n da
t
a
-
de
pe
nde
nt
c
a
t
e
g
or
y
,
a
l
s
o c
a
l
l
e
d l
e
a
r
ni
ng t
o ha
s
h
m
e
t
hods
, a
i
m
t
o l
e
a
r
n a
s
e
t
of
ha
s
h
f
u
nc
t
i
o
n
s
f
r
om
a
v
a
i
l
a
b
l
e
t
r
a
i
n
i
ng
da
t
a
t
ha
t
y
i
e
l
d
c
om
pa
c
t
c
o
de
s
t
o
a
c
h
i
e
v
e
s
a
t
i
s
f
a
c
t
or
y
s
e
a
r
c
h
pe
r
f
or
m
a
nc
e
[
9]
.
E
xi
s
t
i
ng
da
t
a
-
de
pe
nde
nt
m
e
t
hods
c
a
n
b
e
c
l
a
s
s
i
f
i
e
d
i
nt
o
uns
upe
r
v
i
s
e
d,
s
upe
r
v
i
s
e
d,
a
n
d
s
e
m
i
-
s
upe
r
v
i
s
e
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
l
e
c
&
C
om
p E
n
g
,
V
ol
. 11,
N
o. 1, F
e
br
ua
r
y
2021 :
879
-
891
880
l
e
a
r
ni
ng
a
ppr
oa
c
h.
U
ns
upe
r
v
i
s
e
d
ha
s
hi
n
g
m
e
t
hods
[
10
-
12]
us
e
unl
a
be
l
e
d
da
t
a
t
o
bui
l
d
t
he
ha
s
h
f
unc
t
i
ons
w
he
r
e
t
he
ne
i
g
hbor
di
s
t
a
nc
e
(
e
.
g
.,
L
2
nor
m
)
a
m
on
g
t
he
t
r
a
i
ni
ng
da
t
a
i
s
pr
e
s
e
r
v
e
d.
S
upe
r
v
i
s
e
d
or
s
e
m
i
-
s
upe
r
v
i
s
e
d
ha
s
hi
ng
m
e
t
hods
[
13
-
17]
a
t
t
e
m
p
t
o
i
m
pr
o
v
e
t
he
qua
l
i
t
y
of
ha
s
hi
ng
by
l
e
v
e
r
a
g
i
ng
t
he
s
e
m
a
nt
i
c
l
a
be
l
s
i
nt
o
t
he
l
e
a
r
ni
ng
pr
oc
e
s
s
.
C
om
pa
r
e
d
w
i
t
h
da
t
a
-
i
nde
pe
nde
nt
m
e
t
hods
,
i
t
a
ppe
a
r
s
t
h
a
t
da
t
a
-
de
pe
nde
nt
m
e
t
hods
c
a
n
a
c
hi
e
v
e
be
t
t
e
r
a
c
c
ur
a
c
y
w
i
t
h
s
hor
t
e
r
c
ode
s
[
12,
14,
17]
.
H
ow
e
v
e
r
,
da
t
a
-
de
pe
n
de
nt
m
e
t
hods
m
a
y
be
t
oo de
pe
nde
nt
on t
he
t
r
a
i
ni
ng
da
t
a
[
18]
.
T
he
r
e
a
r
e
bot
h
a
d
v
a
nt
a
g
e
s
a
nd
s
hor
t
c
om
i
n
g
s
of
us
i
ng
da
t
a
-
i
nde
pe
n
de
nt
a
nd
da
t
a
-
d
e
pe
nde
nt
m
e
t
hods
.
H
o
w
e
v
e
r
,
t
he
pr
e
v
i
ous
w
or
ks
of
t
he
t
w
o
c
a
t
e
g
or
i
e
s
do
not
f
ul
l
y
t
a
ke
i
nt
o
c
ons
i
de
r
a
t
i
on
t
he
s
i
m
i
l
a
r
i
t
y
p
r
e
s
e
r
v
i
n
g
a
nd
t
hi
s
m
a
y
d
e
gr
a
de
t
he
r
e
t
r
i
e
v
a
l
pe
r
f
or
m
a
nc
e
.
I
n
t
hi
s
w
or
k,
a
n
uns
upe
r
v
i
s
e
d
s
i
m
i
l
a
r
i
t
y
-
pr
e
s
e
r
v
i
n
g
ha
s
hi
n
g
m
e
t
hod
f
or
c
ont
e
nt
-
ba
s
e
d
a
udi
o
r
e
t
r
i
e
v
a
l
i
s
pr
opos
e
d.
W
e
de
v
e
l
op
a
de
e
p
ne
t
w
or
k
w
i
t
h
s
e
v
e
r
a
l
hi
e
r
a
r
c
hi
c
a
l
l
a
y
e
r
s
of
nonl
i
ne
a
r
a
nd
l
i
ne
a
r
t
r
a
ns
f
or
m
a
t
i
ons
t
o
l
e
a
r
n
c
o
m
pa
c
t
bi
na
r
y
c
ode
s
w
he
r
e
t
he
s
i
m
i
l
a
r
i
t
y
be
t
w
e
e
n
s
a
m
pl
e
s
i
s
pr
e
s
e
r
v
e
d.
F
ur
t
he
r
m
or
e
,
t
he
i
nde
pe
n
de
nc
e
a
nd
ba
l
a
nc
e
pr
ope
r
t
i
e
s
a
r
e
i
nc
l
ude
d
i
n
t
he
obj
e
c
t
i
v
e
f
unc
t
i
on
t
o
i
m
pr
o
v
e
t
he
c
od
e
s
.
T
he
pr
opos
e
d
m
e
t
hod
i
s
c
om
p
a
r
e
d
w
i
t
h
t
he
S
ha
z
a
m
a
l
g
or
i
t
h
m
[
3]
,
t
he
da
t
a
-
i
nde
pe
nd
e
nt
ha
s
hi
ng
m
e
t
hod,
i
n
t
e
r
m
s
of
a
c
c
ur
a
c
y,
pr
e
c
i
s
i
on,
r
e
c
a
l
l
,
f
a
l
s
e
pos
i
t
i
v
e
r
a
t
e
, a
nd
t
he
s
t
or
a
g
e
c
os
t
.
2.
B
A
C
K
G
R
O
U
N
D
2.1.
L
e
ar
n
i
n
g t
o
h
as
h
L
e
a
r
ni
ng
t
o
ha
s
h
a
t
t
e
m
pt
s
t
o
l
e
a
r
n
a
ha
s
h
f
unc
t
i
on
=
ℎ
(
)
t
ha
t
m
a
p
s
a
hi
g
h
-
di
m
e
ns
i
ona
l
i
nput
i
t
e
m
∈
t
o
a
c
o
m
pa
c
t
c
ode
,
a
i
m
i
ng
t
o
i
m
pr
o
v
e
t
he
s
e
a
r
c
h
pe
r
f
or
m
a
nc
e
[
1
9
]
.
T
he
r
e
a
r
e
4
t
opi
c
s
t
o
c
ons
i
de
r
f
or
t
he
l
e
a
r
ni
ng
t
o
ha
s
h:
(
1)
h
a
s
h
f
unc
t
i
on,
(
2)
s
i
m
i
l
a
r
i
t
y
-
pr
e
s
e
r
v
i
ng
,
(
3)
l
os
s
f
unc
t
i
on
,
a
nd
(
4)
d
e
e
p
l
e
a
r
ni
ng
to
ha
s
h.
2.1.1.
H
as
h
f
u
n
c
t
i
on
T
he
r
e
a
r
e
s
e
v
e
r
a
l
w
a
y
s
t
o
de
s
i
g
n
ha
s
h
f
unc
t
i
ons
.
T
he
m
os
t
w
i
de
l
y
us
e
d
ha
s
h
f
unc
t
i
ons
a
r
e
g
e
ne
r
a
l
i
z
e
d b
y
l
i
ne
a
r
pr
oj
e
c
t
i
on a
s
s
how
n i
n (
1)
.
=
ℎ
(
)
=
sg
n
(
(
+
)
)
w
he
r
e
∈
{
0
,
1
}
or
{
−
1
,
1
}
,
=
{
x
}
=
1
∈
x
i
s
t
he
t
r
a
i
ni
ng
s
e
t
w
hi
c
h
c
ont
a
i
ns
s
a
m
pl
e
s
,
i
s
t
he
di
m
e
ns
i
on
of
i
nput
v
e
c
t
or
,
=
{
}
=
1
∈
x
i
s
t
he
pr
oj
e
c
t
i
on
v
e
c
t
or
,
i
s
n
um
be
r
of
ha
s
h
bi
t
s
,
i
s
t
he
bi
a
s
v
a
r
i
a
bl
e
,
sg
n
(
)
=
−
1
or
0
i
f
<
0
a
nd
sg
n
(
)
=
1
ot
he
r
w
i
s
e
,
(
∙
)
i
s
a
pr
e
de
f
i
ne
d
f
unc
t
i
on
w
hi
c
h
c
a
n
pos
s
i
bl
y
be
ne
ur
a
l
ne
t
w
or
ks
or
nonl
i
ne
a
r
f
unc
t
i
on.
H
o
w
e
v
e
r
,
us
i
n
g
di
f
f
e
r
e
nt
(
∙
)
y
i
e
l
ds
di
f
f
e
r
e
nt
ha
s
h
f
unc
t
i
on pr
ope
r
t
i
e
s
.
2.1.2. S
i
m
i
l
ar
i
t
y
-
p
r
e
s
e
r
vi
n
g
T
he
di
s
t
a
nc
e
be
t
w
e
e
n
t
w
o
i
t
e
m
s
a
nd
c
a
n
be
de
f
i
ne
d
by
t
he
s
t
a
nda
r
di
z
e
d
E
uc
l
i
de
a
n
di
s
t
a
nc
e
‖
−
‖
2
or
ot
he
r
s
.
T
he
s
i
m
i
l
a
r
i
t
y
be
t
w
e
e
n
t
hos
e
i
t
e
m
s
i
s
of
t
e
n
de
f
i
ne
d
a
s
a
f
unc
t
i
on
of
t
he
di
s
t
a
nc
e
(
e
.g
.,
G
a
us
s
i
a
n
f
unc
t
i
on,
c
os
i
ne
s
i
m
i
l
a
r
i
t
y
,
a
nd
s
o
on)
.
I
n
a
ddi
t
i
o
n,
t
he
s
e
m
a
nt
i
c
s
i
m
i
l
a
r
i
t
y
a
ppr
oa
c
h
i
s
g
e
ne
r
a
l
l
y
us
e
d
i
n
s
i
m
i
l
a
r
i
t
y
s
e
a
r
c
h
a
ppl
i
c
a
t
i
on.
We
c
a
n
a
ppl
y
a
n
y
di
s
t
a
nc
e
t
o
t
he
ha
s
h
i
ng
a
l
g
or
i
t
h
m
f
o
r
s
e
m
a
nt
i
c
s
i
m
i
l
a
r
i
t
y
,
s
uc
h
a
s
E
uc
l
i
de
a
n
di
s
t
a
nc
e
,
by
de
f
i
ni
ng
s
e
m
a
nt
i
c
s
i
m
i
l
a
r
i
t
y
=
1
f
or
a
dj
a
c
e
nt
poi
nt
s
a
nd
=
0
or
−
1
f
or
f
a
r
t
he
r
poi
nt
s
.
I
n
t
he
ha
s
h
c
odi
ng
s
pa
c
e
,
t
he
H
a
m
m
i
n
g
di
s
t
a
nc
e
be
t
w
e
e
n
t
he
c
ode
a
nd
c
a
n
be
de
f
i
ne
d
a
s
‖
−
‖
1
=
∑
‖
ℎ
(
)
−
ℎ
(
)
‖
=
1
.
I
t
i
s
t
he
num
be
r
of
bi
na
r
y
di
g
i
t
s
w
he
r
e
t
he
v
a
l
ue
s
a
r
e
di
f
f
e
r
e
nt
.
H
a
m
m
i
n
g
s
i
m
i
l
a
r
i
t
y
i
s
de
f
i
ne
d
a
s
=
−
f
or
t
he
c
ode
s
v
a
l
ue
d
by
1
a
nd
0.
F
or
t
he
c
ode
s
v
a
l
u
e
d
by
1
a
nd
-
1, t
he
i
nne
r
pr
oduc
t
=
i
s
de
f
i
ne
d a
s
t
he
s
i
m
i
l
a
r
i
t
y
.
L
e
t
’
s
f
oc
us
on
t
he
t
e
r
m
of
s
i
m
i
l
a
r
i
t
y
pr
e
s
e
r
v
i
n
g
.
I
n
F
i
g
ur
e
1(
a
)
,
t
he
r
e
i
s
a
s
e
t
of
t
hr
e
e
poi
nt
s
(
1
,
2
,
a
nd
3
)
i
n
a
n
i
nput
s
pa
c
e
.
B
y
m
e
a
s
ur
i
ng
t
he
E
uc
l
i
de
a
n
di
s
t
a
nc
e
be
t
w
e
e
n
t
he
poi
nt
s
,
w
e
c
a
n
f
i
nd
t
ha
t
1
i
s
c
l
os
e
r
t
o
2
t
ha
n
t
o
3
,
i
.e
.,
1
i
s
m
or
e
s
i
m
i
l
a
r
t
o
2
t
ha
n
3
.
T
h
e
ℎ
(
1
)
,
ℎ
(
2
)
,
a
nd
ℎ
(
3
)
a
r
e
t
he
r
e
pr
e
s
e
nt
a
t
i
ons
of
1
,
2
,
a
nd
3
i
n
t
he
ha
s
h
c
odi
ng
s
pa
c
e
(
or
H
a
m
m
i
n
g
s
pa
c
e
)
,
r
e
s
pe
c
t
i
v
e
l
y
.
F
r
om
t
he
F
i
g
ur
e
1(
b)
,
w
e
c
a
n
s
e
e
ℎ
(
1
)
i
s
c
l
os
e
r
t
o
ℎ
(
3
)
w
hi
l
e
ℎ
(
2
)
i
s
f
a
r
a
w
a
y
.
I
n
t
hi
s
c
a
s
e
,
i
t
s
how
s
t
ha
t
t
he
s
i
m
i
l
a
r
i
t
i
e
s
a
r
e
not
pr
e
s
e
r
v
e
d. F
i
g
ur
e
1(
c
)
, on t
he
ot
he
r
h
a
nd, s
how
s
a
n
e
xa
m
pl
e
of
t
he
s
i
m
i
l
a
r
i
t
i
e
s
t
ha
t
a
r
e
w
e
l
l
pr
e
s
e
r
v
e
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
l
e
c
&
C
om
p E
n
g
I
S
S
N
:
2088
-
8708
Si
m
i
l
ar
i
t
y
-
pr
e
s
e
r
v
i
ng has
h f
o
r
c
ont
e
nt
-
b
as
e
d
au
di
o r
e
t
r
i
e
v
al
us
i
ng
…
(
P
e
t
c
har
at
P
any
apanuw
at
)
881
S
i
m
i
l
a
r
x
1
x
2
x
3
D
i
ss
i
m
i
l
a
r
H
a
m
m
i
n
g
S
p
a
c
e
(
2
-
d
i
m
)
h
(
x
2
)
h
(
x
3
)
0
1
1
h
(
x
1
)
H
a
m
m
i
n
g
S
p
a
c
e
(
2
-
d
i
m
)
h
(
x
1
)
h
(
x
2
)
h
(
x
3
)
0
1
1
S
i
m
i
l
a
r
i
t
i
e
s
a
r
e
n
o
t
p
r
e
se
r
v
e
d
S
i
m
i
l
a
r
i
t
i
e
s
a
r
e
w
e
l
l
p
r
e
se
r
v
e
d
H
a
sh
i
n
g
H
a
sh
i
n
g
(
a
)
(
b
)
(
c
)
F
i
g
ur
e
1. S
i
m
i
l
a
r
i
t
y
-
pr
e
s
e
r
v
i
n
g
ha
s
hi
ng
2.1.3.
L
os
s
f
u
n
c
t
i
o
n
T
he
l
os
s
f
unc
t
i
on
i
s
i
n
t
e
nde
d
t
o
pr
e
s
e
r
v
e
t
he
s
i
m
i
l
a
r
i
t
y
or
de
r
,
i
.e
.,
m
i
ni
m
i
z
e
t
he
di
f
f
e
r
e
nc
e
be
t
w
e
e
n
t
he
ne
a
r
e
s
t
ne
i
g
hbor
s
e
a
r
c
h
r
e
s
ul
t
i
n
t
he
ha
s
h
c
odi
n
g
s
pa
c
e
a
nd
t
he
s
e
a
r
c
h
r
e
s
ul
t
i
n
t
he
i
nput
s
pa
c
e
.
T
he
l
os
s
f
unc
t
i
on
(
,
)
i
s
de
f
i
ne
d a
s
f
ol
l
ow
s:
(
,
)
=
∑
‖
−
‖
2
,
∈
w
he
r
e
i
s
t
he
i
np
ut
da
t
a
,
a
nd
i
s
t
he
pr
oj
e
c
t
i
on
v
e
c
t
or
.
S
pe
c
i
f
i
c
a
l
l
y
,
=
ℎ
(
)
ne
e
ds
t
o
be
bi
na
r
y
.
T
hi
s
bi
na
r
y
c
ons
t
r
a
i
nt
l
e
a
ds
t
o
a
di
f
f
i
c
ul
t
opt
i
m
i
z
a
t
i
on
pr
obl
e
m
.
T
o
s
ol
v
e
t
he
pr
obl
e
m
,
w
e
dr
op
t
he
bi
na
r
y
c
ons
t
r
a
i
nt
a
nd
l
e
t
t
he
c
ode
s
be
c
ont
i
nu
ous
.
T
he
c
ode
s
a
r
e
t
he
n
bi
na
r
i
z
e
d
w
i
t
h
t
hr
e
s
hol
di
ng
.
F
or
bi
na
r
y
c
ons
t
r
a
i
nt
r
e
l
a
xa
t
i
on,
v
a
r
i
ous
s
t
a
nda
r
d
opt
i
m
i
z
a
t
i
on
t
e
c
hni
que
s
c
a
n be
a
ppl
i
e
d.
2.1.4.
D
e
e
p
l
e
ar
n
i
n
g t
o h
a
s
h
T
he
g
o
a
l
of
l
e
a
r
ni
ng
t
o
ha
s
h
i
s
t
o
l
e
a
r
n
t
he
s
pe
c
i
f
i
c
ha
s
h
f
unc
t
i
ons
t
ha
t
m
a
p
hi
g
h
di
m
e
ns
i
o
na
l
i
nput
v
e
c
t
or
t
o
a
c
om
p
a
c
t
bi
na
r
y
v
e
c
t
or
t
ha
t
y
i
e
l
ds
a
g
ood
qua
l
i
t
y
of
r
e
t
r
i
e
v
a
l
a
nd
s
e
a
r
c
h
s
pe
e
d
[
20]
.
F
or
unl
a
be
l
e
d
da
t
a
,
a
n
i
l
l
us
t
r
a
t
i
on
of
uns
upe
r
v
i
s
e
d
de
e
p
l
e
a
r
ni
ng
t
o
h
a
s
h
m
ode
l
t
ha
t
m
a
p
t
he
i
nput
v
e
c
t
or
∈
t
o
c
o
m
pa
c
t
bi
na
r
y
c
ode
s
i
s
s
how
n i
n F
i
g
ur
e
2
.
w
(
1
)
w
(
L
-
1)
w
(
L
)
...
...
...
Lay
er
1
La
y
er
2
Lay
er
L
-
1
Lay
er
L
Lay
er
L
+1
(Inp
ut
L
ayer)
(Rec
ons
truc
tion
Lay
er)
D
xR
ˆ
D
xR
F
i
g
ur
e
2. U
ns
upe
r
v
i
s
e
d de
e
p l
e
a
r
ni
ng
m
ode
l
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
l
e
c
&
C
om
p E
n
g
,
V
ol
. 11,
N
o. 1, F
e
br
ua
r
y
2021 :
879
-
891
882
A
s
s
u
m
e
t
ha
t
a
n
uns
upe
r
v
i
s
e
d
de
e
p
ne
t
w
or
k
c
ons
i
s
t
s
of
L
+
1
l
a
y
e
r
s
.
A
bi
na
r
y
v
e
c
t
o
y
i
s
g
e
ne
r
a
t
e
d
by
pa
s
s
i
ng
t
he
i
nput
v
e
c
t
or
t
hr
oug
h
t
he
ne
t
w
or
k
t
ha
t
c
ont
a
i
ns
m
ul
t
i
pl
e
hi
e
r
a
r
c
hi
c
a
l
l
a
ye
r
s
of
nonl
i
ne
a
r
f
unc
t
i
ons
. T
he
bi
na
r
y
c
ode
of
a
t
L
t
h
l
a
y
e
r
c
a
n be
c
a
l
c
ul
a
t
e
d
a
s
f
ol
l
ow
s
:
=
ℎ
(
)
=
(
(
,
)
)
w
he
r
e
(
,
)
i
s
a
c
om
pos
i
t
i
on of
nonl
i
ne
a
r
t
r
a
ns
f
or
m
a
t
i
ons
de
f
i
ne
d a
s
f
ol
l
ow
s
:
(
,
)
=
(
∙
∙
∙
2
(
1
(
,
(
1
)
)
,
(
2
)
)
∙
∙
∙
(
)
)
w
he
r
e
t
he
v
e
c
t
or
a
nd t
he
w
e
i
g
ht
v
e
c
t
or
(
)
a
r
e
us
e
d
a
s
i
nput
, t
he
pr
oj
e
c
t
i
on
+
1
i
s
pr
oduc
e
d b
y
(
∙
)
.
T
h
e
l
e
a
r
ni
ng
a
l
g
o
r
i
t
hm
a
i
m
s
t
o
l
e
a
r
n
a
s
e
t
of
nonl
i
ne
a
r
w
e
i
g
ht
v
e
c
t
or
s
=
{
(
1
)
,
…
,
(
)
}
w
he
r
e
t
he
i
nf
or
m
a
t
i
on
f
r
o
m
t
he
i
nput
s
pa
c
e
i
s
pr
e
s
e
r
v
e
d.
2
.
2
.
S
e
ar
c
h
w
i
t
h
h
as
h
i
n
g
T
he
r
e
a
r
e
t
w
o
s
t
r
a
t
e
g
i
e
s
t
o
p
e
r
f
or
m
a
s
e
a
r
c
h
w
i
t
h
ha
s
h
i
ng
,
h
a
s
h
c
od
e
r
a
n
ki
ng
a
nd
ha
s
h t
a
b
l
e
l
ook
u
p
[
19]
.
F
or
t
he
ha
s
h
c
ode
r
a
nki
ng
,
a
n
e
xha
us
t
i
v
e
s
e
a
r
c
h
i
s
pe
r
f
or
m
e
d
b
y
c
o
m
p
a
r
i
ng
t
he
di
s
t
a
nc
e
(
e
.
g
.,
H
a
m
m
i
n
g
di
s
t
a
nc
e
)
be
t
w
e
e
n
t
he
que
r
y
a
nd
t
he
r
e
f
e
r
e
nc
e
i
t
e
m
s
.
T
he
i
t
e
m
s
w
i
t
h
t
he
s
m
a
l
l
e
s
t
di
s
t
a
nc
e
s
,
c
a
l
l
e
d
n
e
a
r
e
s
t
ne
i
g
hbor
s
,
a
r
e
r
e
t
r
i
e
v
e
d.
H
o
w
e
v
e
r
,
t
he
c
os
t
of
c
om
put
i
ng
t
he
di
s
t
a
nc
e
r
e
s
ul
t
s
i
n
pe
r
f
or
m
a
nc
e
de
g
r
a
da
t
i
on.
T
he
a
l
t
e
r
na
t
i
v
e
a
pp
r
oa
c
h,
ha
s
h
t
a
bl
e
l
ookup
a
i
m
s
t
o
a
c
c
e
l
e
r
a
t
e
t
he
s
e
a
r
c
h
b
y
r
e
du
c
i
ng
t
he
di
s
t
a
nc
e
c
o
m
put
a
t
i
ons
.
T
he
i
n
v
e
r
s
e
l
ookup
da
t
a
ba
s
e
,
c
a
l
l
e
d
h
a
s
h
t
a
bl
e
,
i
s
c
o
m
pos
e
d
of
bu
c
ke
t
s
w
hi
c
h
a
r
e
i
nde
xe
d
b
y
t
he
ha
s
h c
ode
s
. G
i
v
e
n t
he
que
r
y
, t
he
m
a
t
c
hi
ng
i
t
e
m
s
s
t
or
i
ng
i
n t
he
buc
ke
t
a
r
e
r
e
t
r
i
e
v
e
d.
2
.
3
.
A
u
d
i
o
f
i
n
ge
r
p
r
i
n
t
i
n
g
A
udi
o
f
i
ng
e
r
p
r
i
nt
i
ng
i
s
be
s
t
know
n
f
or
i
t
s
a
bi
l
i
t
y
t
o
i
de
nt
i
f
y
a
n
uknow
n
a
udi
o
r
e
c
or
di
ng
by
us
i
ng
i
t
s
c
o
m
pa
c
t
c
ont
e
nt
-
ba
s
e
d
s
i
g
n
a
t
ur
e
s
o
-
c
a
l
l
e
d
f
i
nge
r
pr
i
nt
[
21]
.
I
t
doe
s
t
hi
s
by
c
on
v
e
r
t
i
ng
t
he
a
udi
o
f
e
a
t
ur
e
s
i
nt
o
ha
s
h
c
ode
s
,
a
i
m
i
n
g
t
o
uni
que
l
y
i
de
nt
i
f
y
a
n
a
udi
o
r
e
c
or
di
ng
.
T
h
e
a
d
v
a
nt
a
ge
of
f
i
nge
r
pr
i
nt
i
s
t
ha
t
,
i
t
r
e
duc
e
s
s
t
or
a
g
e
c
os
t
s
a
s
f
i
nge
r
pr
i
nt
i
s
r
e
l
a
t
i
v
e
l
y
s
m
a
l
l
.
M
o
r
e
o
v
e
r
,
t
he
pe
r
c
e
pt
ua
l
i
r
r
e
l
e
v
a
n
c
i
e
s
ha
v
e
be
e
n
r
e
m
o
v
e
d
f
r
o
m
f
i
n
g
e
r
pr
i
nt
, r
e
s
ul
t
i
ng
i
n e
f
f
i
c
i
e
nt
c
o
m
p
a
r
i
s
on a
nd s
e
a
r
c
hi
n
g
.
3.
M
E
T
H
O
D
T
he
a
i
m
of
t
hi
s
pa
pe
r
i
s
t
o
pr
o
v
i
de
a
n
e
f
f
i
c
i
e
nt
t
e
c
hni
que
t
ha
t
y
i
e
l
ds
a
g
ood
qu
a
l
i
t
y
of
r
e
t
r
i
e
v
a
l
a
nd
c
o
m
put
a
t
i
ona
l
e
f
f
i
c
i
e
n
c
y
.
I
n
t
hi
s
w
or
k,
c
o
m
pa
c
t
bi
na
r
y
c
ode
s
a
r
e
l
e
a
r
ne
d
f
or
f
i
n
g
e
r
p
r
i
nt
i
nde
xi
ng
w
i
t
h
uns
upe
r
v
i
s
e
d
de
e
p
ne
t
w
or
k
i
n
a
w
a
y
t
ha
t
t
he
s
i
m
i
l
a
r
i
t
y
be
t
w
e
e
n
s
a
m
pl
e
s
i
s
pr
e
s
e
r
v
e
d.
O
nc
e
,
a
s
hor
t
a
udi
o
s
a
m
pl
e
i
s
t
a
ke
n
t
o
our
c
ont
e
nt
-
ba
s
e
d
a
udi
o
r
e
t
r
i
e
v
a
l
s
y
s
t
e
m
,
t
he
s
y
s
t
e
m
pe
r
f
or
m
s
d
a
t
a
ba
s
e
l
ookup
f
or
m
a
t
c
hi
ng
t
r
a
c
k
a
nd
t
he
n
r
e
t
ur
ns
t
he
s
on
g
I
D
t
ha
t
t
he
que
r
y
i
s
t
a
ke
n.
A
s
s
how
n
i
n
F
i
g
ur
e
3,
t
he
s
y
s
t
e
m
i
s
de
s
i
g
ne
d
w
i
t
h
t
hr
e
e
s
t
e
ps
:
(
1)
F
i
ng
e
r
pr
i
nt
f
e
a
t
ur
e
e
xt
r
a
c
t
i
on,
(
2)
U
ns
upe
r
v
i
s
e
d
s
i
m
i
l
a
r
i
t
y
-
p
r
e
s
e
r
v
i
ng
h
a
s
hi
ng
,
a
nd (
3)
S
e
que
nc
e
m
a
t
c
hi
ng
.
R
e
f
e
r
e
nc
e
Au
dio
Un
s
upe
r
vis
e
d
S
im
il
a
r
it
y
-
P
r
e
s
e
r
ving
Ha
s
hing
De
ep
L
ea
r
n
in
g
to
Ha
s
h
Ha
s
h
Fu
n
ctio
n
s
16
-
bi
t
Ha
s
h
X
1
,X
3
X
2
...
X
n
I
t
e
m
s
Da
tab
as
e
16
-
b
it
Ha
s
h
S
e
quenc
e
M
a
tch
ing
S
o
n
g
I
D
R
etu
r
n
ed
I
tem
Qu
e
r
y
A
udi
o
0
20
40
60
100
80
120
140
160
180
F
r
e
q
u
e
n
c
y
500
1000
1500
2500
2000
3000
3500
4000
(t
2
,f
2
)
(t
1
,f
1
)
Δ
f
=
f
2
-
f
1
,
Δ
t
=
t
2
-
t
1
F
e
at
ur
e
s
=
[
f
1
,
Δ
f
,
Δ
t]
T
i
m
e
0
20
40
60
100
80
120
140
160
180
F
r
e
q
u
e
n
c
y
500
1000
1500
2500
2000
3000
3500
4000
(t
2
,f
2
)
(t
1
,f
1
)
Ca
ndi
da
t
e
P
oi
nt
Δ
f
=
f
2
-
f
1
,
Δ
t
=
t
2
-
t
1
F
e
atur
e
s
=
[
f
1
,
Δ
f
,
Δ
t]
F
e
a
tur
e
E
xtr
a
c
ti
on
F
e
a
tur
e
E
xtr
a
c
ti
on
h
1
2
h
1
3
h
1
19
h
2
2
h
2
18
h
3
16
x
3
x
4
x
2
x
1
h
1
1
h
2
1
h
3
1
.
.
.
.
.
.
.
.
.
...
.
.
.
1
ˆ
x
2
ˆ
x
3
ˆ
x
4
ˆ
x
20
ˆ
x
x
20
Ca
ndi
da
t
e
P
oi
nt
L
a
y
e
r
1
L
a
y
e
r
2
L
a
y
e
r
3
L
a
y
e
r
4
L
a
y
e
r
5
I
n
p
u
t
L
a
y
e
r
R
ec
o
n
s
tr
u
ctio
n
L
a
y
e
r
T
a
r
g
e
t
R
e
g
i
o
n
T
a
r
g
e
t
R
e
g
i
o
n
F
i
g
ur
e
3. T
he
c
ons
t
r
uc
t
i
on of
pr
opos
e
d m
e
t
hod f
or
our
c
ont
e
nt
-
ba
s
e
d a
udi
o r
e
t
r
i
e
v
a
l
s
y
s
t
e
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
l
e
c
&
C
om
p E
n
g
I
S
S
N
:
2088
-
8708
Si
m
i
l
ar
i
t
y
-
pr
e
s
e
r
v
i
ng has
h f
o
r
c
ont
e
nt
-
b
as
e
d
au
di
o r
e
t
r
i
e
v
al
us
i
ng
…
(
P
e
t
c
har
at
P
any
apanuw
at
)
883
3.1.
F
i
n
ge
r
p
r
i
n
t
f
e
at
u
r
e
e
xt
r
ac
t
i
on
B
e
f
or
e
f
i
nge
r
pr
i
nt
f
e
a
t
ur
e
e
xt
r
a
c
t
i
on
i
s
pe
r
f
or
m
e
d,
t
he
a
udi
o
s
i
g
na
l
i
s
c
on
v
e
r
t
e
d
i
nt
o
a
c
o
m
m
o
n
f
or
m
a
t
f
or
a
na
l
y
s
i
s
.
N
e
xt
,
t
he
t
i
m
e
-
s
e
r
i
e
s
a
udi
o
s
i
g
na
l
i
s
c
on
v
e
r
t
e
d
i
nt
o
t
i
m
e
-
f
r
e
que
nc
y
d
om
a
i
n
f
r
o
m
w
hi
c
h
m
o
r
e
m
e
a
ni
n
g
f
ul
i
nf
or
m
a
t
i
on c
a
n be
e
xt
r
a
c
t
e
d. B
e
l
ow
e
a
c
h a
s
pe
c
t
i
s
de
t
a
i
l
e
d.
3.1.1.
P
r
e
p
r
oc
e
s
s
i
n
g a
n
d
t
r
an
s
f
or
m
I
n
t
hi
s
pa
pe
r
,
t
he
f
i
ng
e
r
pr
i
nt
e
xt
r
a
c
t
i
on
pr
e
s
e
nt
e
d
i
n
[
3]
i
s
a
pp
l
i
e
d.
F
i
r
s
t
l
y
,
w
e
c
on
v
e
r
t
t
he
i
nput
a
udi
o
t
o
m
ono
s
i
g
na
l
a
nd
dow
ns
a
m
pl
e
d
f
r
o
m
t
he
s
t
a
nda
r
d
di
g
i
t
a
l
a
udi
o
of
44.1K
H
z
,
t
o
8K
H
z
,
t
o
m
a
k
e
t
he
da
t
a
e
a
s
i
e
r
t
o
ha
ndl
e
,
r
e
duc
i
ng
da
t
a
ba
s
e
s
i
z
e
,
a
nd
i
nc
r
e
a
s
i
ng
s
p
e
e
d
of
t
he
a
l
g
or
i
t
h
m
.
T
he
a
udi
o
s
i
g
na
l
i
s
t
he
n
c
on
v
e
r
t
e
d
i
nt
o
t
he
t
i
m
e
-
f
r
e
qu
e
nc
y
r
e
pr
e
s
e
nt
a
t
i
on.
W
e
pe
r
f
or
m
a
s
hor
t
-
t
i
m
e
f
our
i
e
r
t
r
a
ns
f
or
m
(
S
T
F
T
)
w
i
t
h
a
w
i
ndo
w
s
i
z
e
of
64
m
s
.
f
or
g
ood
s
pe
c
t
r
u
m
r
e
s
ol
ut
i
on
[
22]
a
nd
a
hop
s
i
z
e
of
32
m
s
.
F
i
g
ur
e
4
s
how
s
t
i
m
e
-
fr
e
qu
e
nc
y
gr
a
ph s
o
-
c
a
l
l
e
d a
s
pe
c
t
r
ogr
a
m
. O
n t
he
hor
i
z
ont
a
l
a
xi
s
i
s
t
i
m
e
, on t
he
v
e
r
t
i
c
a
l
i
s
f
r
e
que
nc
y
,
a
n
d
o
n
t
h
e
t
h
i
r
d
i
s
i
n
t
e
n
s
i
t
y
.
E
a
c
h
p
o
i
n
t
o
n
t
h
e
g
r
a
p
h
r
e
p
r
e
s
e
n
t
s
t
he
i
n
t
e
n
s
i
t
y
o
f
a
g
i
v
e
n
f
r
e
q
ue
n
c
y
a
t
s
p
e
c
i
f
i
c
t
i
m
e
.
F
i
g
ur
e
4. S
pe
c
t
r
og
r
a
m
w
i
t
h pe
a
k
i
nt
e
ns
i
t
i
e
s
3.1.2.
F
e
at
u
r
e
e
xt
r
ac
t
i
on
A
f
t
e
r
c
on
v
e
r
t
i
ng
t
he
s
i
g
na
l
i
nt
o
t
he
t
i
m
e
-
f
r
e
qu
e
nc
y
do
m
a
i
n,
t
he
f
e
a
t
ur
e
s
a
r
e
t
he
n
e
xt
r
a
c
t
e
d
f
r
o
m
t
he
s
pe
c
t
r
um
.
D
ue
t
o
t
he
i
r
r
obus
t
ne
s
s
t
o
noi
s
e
a
nd
di
s
t
or
t
i
o
ns
, t
he
a
m
pl
i
t
ude
pe
a
ks
i
n
e
a
c
h f
r
a
m
e
a
r
e
s
e
l
e
c
t
e
d
a
s
c
a
ndi
da
t
e
poi
nt
s
.
E
a
c
h
c
a
ndi
da
t
e
poi
n
t
i
s
pa
i
r
e
d
w
i
t
h
t
he
a
dj
a
c
e
nt
pe
a
ks
.
T
he
c
ons
t
e
l
l
a
t
i
on
m
a
p
of
pa
i
r
e
d
poi
nt
s
w
i
t
h
c
oor
di
na
t
e
l
i
s
t
i
s
s
how
n
i
n
F
i
g
ur
e
5.
I
n
t
hi
s
w
or
k,
e
a
c
h
c
a
ndi
da
t
e
poi
nt
i
s
pa
i
r
e
d
w
i
t
hi
n
31
f
r
e
que
nc
y
bi
ns
a
nd
63
t
i
m
e
f
r
a
m
e
s
.
O
nl
y
t
he
c
l
os
e
s
t
3
pe
a
ks
i
n
t
i
m
e
t
o
e
a
c
h
ot
he
r
a
r
e
s
e
l
e
c
t
e
d.
F
i
g
ur
e
6
s
h
ow
s
t
h
e
c
om
bi
na
t
or
i
a
l
a
s
s
oc
i
a
t
i
o
n
o
f
a
p
a
i
r
of
t
w
o p
oi
n
t
s
w
h
i
c
h
i
s
c
a
l
l
e
d
a
‘
l
a
ndm
a
r
k’
. F
or
e
a
c
h
pa
i
r
,
i
t
c
o
n
s
i
s
t
s
o
f
f
ou
r
c
om
po
ne
n
t
s
,
t
he
s
t
a
r
t
i
ng
f
r
e
q
ue
nc
y
1
,
t
he
s
t
a
r
t
i
ng
t
i
m
e
1
,
t
he
e
nd
f
r
e
q
ue
nc
y
2
,
a
nd
t
h
e
e
n
d
t
i
m
e
2
.
F
i
g
ur
e
5. A
c
ons
t
e
l
l
a
t
i
on m
a
p of
pa
i
r
e
d poi
nt
s
(t
2
,f
2
)
T
i
m
e
Freq
u
en
cy
(t
1
,f
1
)
C
a
n
d
i
d
a
t
e
P
o
i
n
t
3
1
F
r
e
q
u
e
n
c
y
B
i
n
6
3
T
i
m
e
F
r
a
m
e
Δ
f
=
f
2
-
f
1
,
Δ
t
=
t
2
-
t
1
Feat
u
re
=
[f
1
,
Δ
f,
Δ
t]
3
1
F
r
e
q
u
e
n
c
y
B
in
F
i
g
ur
e
6. T
he
c
o
m
bi
na
t
or
i
a
l
a
s
s
oc
i
a
t
i
on of
a
pa
i
r
of
t
w
o poi
nt
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
l
e
c
&
C
om
p E
n
g
,
V
ol
. 11,
N
o. 1, F
e
br
ua
r
y
2021 :
879
-
891
884
3.1.3. A
u
d
i
o f
i
n
ge
r
p
r
i
n
t
F
or
t
he
l
a
ndm
a
r
k
a
s
m
e
nt
i
one
d a
bo
v
e
, t
he
a
udi
o f
i
ng
e
r
pr
i
nt
c
a
n be
de
f
i
ne
d a
s
f
ol
l
ow
s
:
=
[
1
,
Δ
,
Δ
]
w
he
r
e
t
he
f
r
e
que
nc
y
di
f
f
e
r
e
nc
e
∆
=
2
−
1
,
a
nd
t
he
t
i
m
e
di
f
f
e
r
e
nc
e
be
t
w
e
e
n
t
he
t
w
o
poi
nt
s
∆
=
2
−
1
.
T
he
f
i
ng
e
r
pr
i
n
t
i
s
a
l
s
o
a
s
s
oc
i
a
t
e
d
w
i
t
h
t
h
e
o
f
f
s
e
t
t
i
m
e
f
r
om
t
h
e
be
g
i
n
n
i
ng
of
t
h
e
a
u
d
i
o
f
i
l
e
t
o
t
h
e
s
t
a
r
t
i
n
g
t
i
m
e
1
.
T
hi
s
f
i
ng
e
r
p
r
i
nt
f
e
a
t
ur
e
[
1
,
Δ
,
Δ
]
i
s
us
e
d
t
o
g
e
ne
r
a
t
e
ha
s
h
c
ode
i
n
[
3]
.
T
h
e
ha
s
h
m
ode
l
c
a
n
b
e
de
f
i
ne
d a
s
s
how
n i
n (
6)
.
1
×
2
12
+
∆
×
2
6
+
∆
w
he
r
e
a
f
i
n
g
e
r
pr
i
nt
ha
s
h
i
s
c
o
m
pos
e
d
of
8
-
bi
t
f
r
e
que
n
c
y
1
,
6
-
bi
t
f
r
e
que
nc
y
di
f
f
e
r
e
nc
e
∆
,
a
nd
6
-
bi
t
t
i
m
e
di
f
f
e
r
e
nc
e
∆
. F
i
g
ur
e
7 s
how
s
a
n
e
xa
m
pl
e
of
20
-
bi
t
ha
s
h
a
ddr
e
s
s
c
a
l
c
ul
a
t
e
d f
r
o
m
(
6)
.
[
2
1
6
, 1
8
,
3
6
]
H
a
sh
F
u
n
c
t
i
o
n
8
8
5
9
2
4
=
H
a
sh
A
d
d
r
e
ss
12
6
1
22
f
f
t
1
1
0
1
1
0
0
0
0
1
0
0
1
0
1
0
0
1
0
0
1
,
[
,
]
f
f
t
F
i
g
ur
e
7. A
n
e
x
a
m
pl
e
of
20
-
bi
t
ha
s
h a
ddr
e
s
s
F
or
a
16
-
bi
t
f
i
ng
e
r
pr
i
nt
ha
s
h,
i
t
i
s
c
om
pos
e
d
of
6
-
bi
t
f
r
e
que
nc
y
1
,
5
-
bi
t
f
r
e
que
nc
y
di
f
f
e
r
e
nc
e
∆
,
a
nd 5
-
bi
t
t
i
m
e
di
f
f
e
r
e
nc
e
∆
. T
he
ha
s
h
m
ode
l
c
a
n be
de
f
i
ne
d a
s
s
how
n i
n (
7)
.
1
×
2
10
+
∆
×
2
5
+
∆
(
7)
A
f
t
e
r
t
he
ha
s
h
c
ode
i
s
c
a
l
c
ul
a
t
e
d,
t
he
s
y
s
t
e
m
t
he
n
us
e
s
t
hi
s
c
ode
a
s
a
n
i
nde
x
f
or
s
e
a
r
c
hi
ng
i
n
t
he
da
t
a
ba
s
e
.
A
n e
xa
c
t
m
a
t
c
hi
ng
a
l
g
or
i
t
h
m
i
s
a
ppl
i
e
d i
n [
3]
.
U
nl
i
ke
t
he
S
ha
z
a
m
a
l
g
or
i
t
h
m
,
w
e
de
v
e
l
op a
de
e
p
ne
ur
a
l
ne
t
w
o
r
k
w
i
t
h
m
ul
t
i
pl
e
hi
e
r
a
r
c
hi
c
a
l
l
a
y
e
r
s
of
nonl
i
ne
a
r
a
nd
l
i
ne
a
r
t
r
a
ns
f
or
m
a
t
i
ons
t
o
l
e
a
r
n
c
o
m
pa
c
t
c
ode
s
f
r
o
m
t
he
s
e
f
i
nge
r
pr
i
nt
f
e
a
t
ur
e
s
s
uc
h
t
ha
t
t
he
s
i
m
i
l
a
r
i
t
y
be
t
w
e
e
n
s
a
m
pl
e
s
i
s
pr
e
s
e
r
v
e
d.
T
he
de
t
a
i
l
s
a
r
e
de
s
c
r
i
be
d f
ur
t
he
r
i
n t
he
ne
xt
s
e
c
t
i
on.
3
.
2
.
U
n
s
u
p
e
r
vi
s
e
d
s
i
m
i
l
ar
i
t
y
-
p
r
e
s
e
r
vi
n
g
h
as
h
i
n
g (
U
S
H
)
I
n
t
hi
s
pa
pe
r
,
t
he
ha
s
h
t
r
a
ns
f
or
m
a
t
i
ons
a
r
e
c
r
e
a
t
e
d
b
y
a
n
uns
upe
r
v
i
s
e
d
de
e
p
n
e
ur
a
l
ne
t
w
or
k
.
A
s
s
ho
w
n i
n F
i
g
ur
e
8, t
he
r
e
a
r
e
5
l
a
y
e
r
s
i
n our
de
e
p n
e
t
w
o
r
k:
t
he
i
nput
l
a
ye
r
c
ons
i
s
t
s
of
20
node
s
of
i
nput
,
t
he
t
hr
e
e
hi
dde
n
l
a
y
e
r
s
c
ons
i
s
t
of
19,
18,
a
nd
1
6
node
s
r
e
s
pe
c
t
i
v
e
l
y
,
a
nd
t
he
r
e
a
r
e
2
0
node
s
of
̂
i
n
t
he
o
u
t
p
u
t
l
a
y
e
r
.
.
.
.
.
.
.
.
.
.
.
.
.
L
ay
er
1
L
ay
er
2
L
ay
er
3
L
ay
er
4
L
ay
er
5
.
.
.
20
20
18
19
16
(
1
)
,
(
1
)
,
ii
fx
(
1
)
,
(
1
)
,
ii
af
(
2
)
,
(
2
)
,
1
1
i
i
H
f
e
(
2
)
,
(
2
)
,
ii
af
(
3
)
,
(
3
)
,
1
1
i
i
H
f
e
(
3
)
,
(
3
)
,
ii
af
(
4
)
,
(
4
)
,
(
4
)
,
(
4
)
,
(
4
)
,
ii
ii
i
H
HH
H
ee
f
ee
(
4
)
,
(
4
)
,
ii
af
(
5
)
,
(
5
)
,
ii
fH
(
5
)
,
(
5
)
,
ii
af
F
i
g
ur
e
8. O
ur
p
r
opos
e
d uns
upe
r
v
i
s
e
d s
i
m
i
l
a
r
i
t
y
-
pr
e
s
e
r
v
i
n
g
ha
s
hi
ng
ne
t
w
or
k (
U
S
H
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
l
e
c
&
C
om
p E
n
g
I
S
S
N
:
2088
-
8708
Si
m
i
l
ar
i
t
y
-
pr
e
s
e
r
v
i
ng has
h f
o
r
c
ont
e
nt
-
b
as
e
d
au
di
o r
e
t
r
i
e
v
al
us
i
ng
…
(
P
e
t
c
har
at
P
any
apanuw
at
)
885
O
u
r
de
e
p
n
e
t
w
or
k
i
s
l
e
a
r
n
e
d
s
o
t
ha
t
t
h
e
o
u
t
p
u
t
of
t
he
f
o
u
r
t
h
l
a
y
e
r
c
a
n
b
e
u
s
e
d
a
s
t
he
b
i
na
r
y
ha
s
h
c
od
e
s
.
F
o
r
t
h
e
ne
t
w
or
k
de
s
i
g
n,
e
a
c
h
no
de
i
s
c
om
po
s
e
d
of
o
ne
i
n
p
ut
s
um
m
a
t
i
o
n
f
u
nc
t
i
on
a
n
d
o
ne
o
u
t
p
u
t
t
r
a
ns
f
or
m
a
t
i
on
f
unc
t
i
on. T
he
f
unc
t
i
on f
(
∙)
i
s
us
e
d t
o c
om
bi
ne
i
nf
or
m
a
t
i
on by
t
he
l
i
nks
f
r
o
m
ot
he
r
node
s
, a
s
s
how
n i
n (
8)
.
=
(
(
)
,
1
,
(
)
,
2
,
…
,
(
)
,
;
(
)
,
1
,
(
)
,
2
,
…
,
(
)
,
)
(
8)
w
he
r
e
(
)
,
1
,
(
)
,
2
,
…
,
(
)
,
a
r
e
t
he
i
nput
s
t
o
t
he
node
,
(
)
,
1
,
(
)
,
2
,
…
,
(
)
,
a
r
e
t
he
a
s
s
oc
i
a
t
e
d
w
e
i
g
ht
s
,
i
ndi
c
a
t
e
s
t
he
l
a
y
e
r
nu
m
be
r
,
a
nd
i
s
t
he
num
b
e
r
of
i
nput
node
s
.
T
he
out
put
a
c
t
i
v
a
t
i
on
f
unc
t
i
on
(
(
∙
)
)
i
s
s
how
n i
n (
9)
.
=
(
)
(
)
=
(
)
(
∙
)
(
9)
L
e
t
be
t
he
s
um
of
pr
oduc
t
of
i
nput
a
nd
w
e
i
g
ht
,
=
∑
,
t
he
f
unc
t
i
ons
of
t
he
node
s
f
r
o
m
l
a
y
e
r
1 t
o l
a
y
e
r
5 of
our
pr
opos
e
d ne
t
w
or
k a
r
e
de
f
i
ne
d a
s
f
ol
l
ow
s
:
L
a
ye
r
1:
I
n
t
hi
s
l
a
y
e
r
,
t
he
node
s
onl
y
c
on
v
e
y
i
nput
s
t
o
t
he
node
s
of
t
he
ne
xt
c
ons
e
c
ut
i
v
e
l
a
y
e
r
.
T
he
f
unc
t
i
ons
of
t
he
t
h node
a
r
e
s
how
n
a
s
;
(
1
)
,
=
(
1
)
,
a
nd
(
1
)
,
=
(
1
)
,
L
a
ye
r
2:
F
or
t
hi
s
l
a
y
e
r
,
t
he
s
i
gm
oi
d
f
unc
t
i
on
i
s
us
e
d
a
s
a
c
t
i
v
a
t
i
on
f
unc
t
i
on.
L
e
t
(
2
)
,
=
(
1
)
,
a
nd
(
2
)
,
=
∑
(
2
)
,
(
2
)
,
. T
hus
, t
he
f
unc
t
i
ons
of
t
he
t
h node
a
r
e
de
f
i
ne
d a
s
;
(
2
)
,
=
1
1
+
−
(
2
)
,
a
nd
(
2
)
,
=
(
2
)
,
L
a
ye
r
3:
I
n
t
hi
s
l
a
y
e
r
,
t
he
s
i
gm
oi
d
f
unc
t
i
on
i
s
a
ppl
i
e
d
f
or
a
c
t
i
v
a
t
i
on
f
unc
t
i
on.
L
e
t
(
3
)
,
=
(
2
)
,
a
nd
(
3
)
,
=
∑
(
3
)
,
(
3
)
,
. T
he
f
unc
t
i
ons
of
t
he
t
h node
a
r
e
de
f
i
ne
d a
s
;
(
3
)
,
=
1
1
+
−
(
3
)
,
a
nd
(
3
)
,
=
(
3
)
,
L
a
ye
r
4:
T
he
out
put
of
e
a
c
h
node
i
n
t
hi
s
l
a
ye
r
w
i
l
l
be
us
e
d
a
s
t
he
bi
na
r
y
c
ode
s
.
D
ur
i
n
g
t
r
a
i
ni
ng
,
t
he
s
e
c
ode
s
a
r
e
us
e
d
t
o
r
e
c
ons
t
r
uc
t
t
he
i
nput
da
t
a
a
t
t
he
out
put
l
a
y
e
r
.
T
he
h
y
pe
r
bol
i
c
t
a
ng
e
nt
f
unc
t
i
on
i
s
pa
r
t
i
c
ul
a
r
l
y
us
e
d
a
s
a
c
t
i
v
a
t
i
on
f
unc
t
i
on
i
n
t
hi
s
l
a
ye
r
.
L
e
t
(
4
)
,
=
(
3
)
,
a
nd
(
4
)
,
=
∑
(
4
)
,
(
4
)
,
.
T
he
f
unc
t
i
ons
of
t
he
t
h node
a
r
e
de
f
i
ne
d a
s
;
(
4
)
,
=
(
4
)
,
−
−
(
4
)
,
(
4
)
,
+
−
(
4
)
,
a
nd
(
4
)
,
=
(
4
)
,
L
a
ye
r
5:
T
hi
s
l
a
y
e
r
i
s
t
he
out
put
or
r
e
c
ons
t
r
uc
t
i
on l
a
y
e
r
.
T
o p
r
e
s
e
r
v
e
t
he
s
i
m
i
l
a
r
i
t
y
b
e
t
w
e
e
n
s
a
m
pl
e
s
,
t
hus
,
t
he
t
a
r
g
e
t
out
put
s
a
r
e
g
i
v
e
n
t
he
s
a
m
e
a
s
t
he
i
nput
s
of
l
a
y
e
r
1.
L
e
t
(
5
)
,
=
(
4
)
,
a
nd
(
5
)
,
=
∑
(
5
)
,
(
5
)
,
. T
he
f
unc
t
i
ons
of
t
he
t
h out
put
no
de
a
r
e
de
f
i
ne
d a
s
;
(
5
)
,
=
(
5
)
,
a
nd
(
5
)
,
=
(
5
)
,
T
o
a
c
hi
e
v
e
t
he
e
f
f
i
c
i
e
nt
bi
na
r
y
c
ode
s
,
w
e
i
nc
l
ude
c
ons
t
r
a
i
nt
s
i
n
t
he
obj
e
c
t
i
v
e
f
unc
t
i
o
n
s
o
t
ha
t
t
he
c
ode
s
ha
v
e
4
pr
ope
r
t
i
e
s
:
(
1)
be
l
ong
i
ng
t
o
{
1,
-
1}
,
(
2)
s
i
m
i
l
a
r
i
t
y
-
pr
e
s
e
r
v
i
n
g
,
(
3)
i
nde
pe
nde
nt
,
a
nd
(
4)
ba
l
a
nc
i
ng
.
I
n
t
hi
s
pa
pe
r
,
t
he
m
e
t
hod
pr
e
s
e
nt
e
d
i
n
t
he
U
H
-
B
D
N
N
[
23]
i
s
a
pp
l
i
e
d
t
o
opt
i
m
i
z
e
t
he
obj
e
c
t
i
v
e
f
unc
t
i
on w
hi
c
h i
s
de
f
i
ne
d a
s
f
ol
l
ow
s
:
,
=
1
2
‖
−
(
(
−
1
)
+
(
−
1
)
×
[
1
]
1
×
)
‖
2
+
1
2
∑
‖
(
)
‖
2
−
1
=
1
+
2
2
‖
(
−
1
)
−
‖
2
+
3
2
‖
1
(
−
1
)
(
−
1
)
−
‖
2
+
4
2
‖
(
−
1
)
[
1
]
×
1
‖
2
(
15)
.
.
∈
{
1
,
−
1
}
×
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
l
e
c
&
C
om
p E
n
g
,
V
ol
. 11,
N
o. 1, F
e
br
ua
r
y
2021 :
879
-
891
886
w
he
r
e
∈
×
i
s
a
s
e
t
of
t
r
a
i
ni
ng
da
t
a
w
i
t
h
di
m
e
ns
i
on,
∈
{
1
,
−
1
}
×
i
s
out
put
bi
na
r
y
c
ode
of
,
i
s
num
be
r
o
f
bi
t
s
,
i
s
num
be
r
of
l
a
y
e
r
s
,
(
)
i
s
w
e
i
g
ht
m
a
t
r
i
x be
t
w
e
e
n l
a
y
e
r
+
1
a
nd l
a
ye
r
,
i
s
bi
a
s
v
e
c
t
or
f
or
n
o
de
s
i
n
l
a
y
e
r
+
1
,
(
)
=
(
)
(
(
−
1
)
(
−
1
)
+
(
−
1
)
[
1
]
1
×
)
i
s
t
he
o
ut
p
u
t
v
a
l
u
e
s
of
l
a
y
e
r
,
(
1
)
=
,
(
)
i
s
a
c
t
i
v
a
t
i
o
n
f
un
c
t
i
o
n
of
l
a
y
e
r
,
a
nd
1
-
4
a
r
e
t
h
e
pa
r
a
m
e
t
e
r
s
f
or
o
p
t
i
m
i
z
i
ng
t
he
o
b
j
e
c
t
i
v
e
f
u
nc
t
i
on
.
T
he
f
i
r
s
t
t
e
r
m
of
(
15)
m
a
ke
s
s
ur
e
t
ha
t
t
he
bi
na
r
y
c
ode
a
l
l
ow
s
a
g
ood r
e
c
ons
t
r
uc
t
i
on of
. T
h
e
s
e
c
ond
t
e
r
m
i
s
a
w
e
i
g
ht
r
e
g
ul
a
r
i
z
a
t
i
on
t
ha
t
e
nc
our
a
g
e
s
t
he
ne
t
w
or
k
t
o
ke
e
p
t
he
w
e
i
g
ht
s
s
m
a
l
l
i
n
or
de
r
t
o
r
e
duc
e
o
v
e
r
f
i
t
t
i
ng
.
T
he
t
hi
r
d
t
e
r
m
m
e
a
s
ur
e
s
t
he
e
qua
l
i
t
y
c
ons
t
r
a
i
nt
v
i
ol
a
t
i
on.
T
he
f
our
t
h
t
e
r
m
i
s
t
he
i
nde
pe
nde
nc
e
,
a
nd
t
he
f
i
f
t
h
t
e
r
m
i
s
ba
l
a
nc
e
of
t
he
bi
na
r
y
c
ode
s
.
A
s
s
how
n
i
n
(
16)
i
s
t
o
e
ns
ur
e
t
ha
t
e
a
c
h
bi
t
of
t
he
b
i
na
r
y
c
ode
s
be
l
ong
s
t
o
{
1,
-
1}
.
Af
t
e
r
t
he
e
f
f
i
c
i
e
nt
c
ode
s
a
r
e
pr
oduc
e
d
f
r
o
m
de
e
p
-
l
e
a
r
ni
ng
n
e
t
w
or
k,
t
he
s
e
c
ode
s
a
r
e
us
e
d
a
s
s
e
a
r
c
h
i
nde
x
i
n our
c
ont
e
nt
-
ba
s
e
d
a
ud
i
o
r
e
t
r
i
e
v
a
l
s
y
s
t
e
m
.
T
h
e
s
ong
I
D
,
1
,
1
,
∆
a
nd
∆
a
r
e
s
t
or
e
d
a
t
t
he
i
r
ha
s
h a
ddr
e
s
s
i
n t
he
da
t
a
ba
s
e
. T
a
bl
e
1 s
how
s
t
he
r
e
pr
e
s
e
nt
a
t
i
on of
i
nf
or
m
a
t
i
on da
t
a
.
T
a
bl
e
1. R
e
pr
e
s
e
nt
a
t
i
on of
i
nf
or
m
a
t
i
on da
t
a
M
e
t
h
od
I
nde
x
I
nf
or
m
a
t
i
o
n
da
t
a
U
S
H
16
bi
t
s
H
a
s
h
A
ddr
e
s
s
S
o
ng
I
D
,
1
,
1
,
∆
,
∆
S
h
a
z
a
m
[
3]
16
-
bi
t
/
20
-
bi
t
H
a
s
h
A
ddr
e
s
s
S
o
ng
I
D
,
1
3
.
3
.
S
e
q
u
e
n
c
e
m
at
c
h
i
n
g
F
or
t
he
que
r
y
s
t
e
p,
a
s
e
qu
e
nc
e
of
qu
e
r
y
f
e
a
t
ur
e
s
i
s
ge
ne
r
a
t
e
d
t
o
a
s
e
t
of
c
o
m
p
a
c
t
ha
s
h
c
ode
s
a
nd
us
e
d
f
or
s
e
a
r
c
hi
ng
i
n
t
he
i
nv
e
r
s
e
l
ookup
da
t
a
ba
s
e
.
L
e
t
r
e
pr
e
s
e
nt
a
s
e
t
of
s
e
que
nc
e
s
of
que
r
y
f
e
a
t
ur
e
s
,
=
{
1
,
2
,
…
,
}
,
w
he
r
e
i
s
a
que
r
y
a
t
or
e
de
r
,
=
1
,
2
,
…
,
,
a
nd
i
s
t
he
t
ot
a
l
num
be
r
of
s
e
que
nc
e
s
.
T
he
l
e
a
r
ni
ng
ha
s
h
f
unc
t
i
on
:
→
ℎ
(
)
,
i
s
us
e
d
t
o
m
a
p
t
he
que
r
y
f
e
a
t
ur
e
s
t
o
bi
na
r
y
ha
s
h
c
ode
s
. W
e
c
a
n de
f
i
ne
=
{
}
=
1
t
o t
he
c
or
r
e
s
pondi
ng
bi
na
r
y
c
ode
s
a
s
f
ol
l
ow
s
:
=
(
)
=
{
ℎ
(
1
)
,
ℎ
(
2
)
,
…
,
ℎ
(
)
}
w
he
r
e
i
s
t
he
ha
s
h
c
ode
s
of
,
∈
{
1
,
−
1
}
,
a
nd
i
s
t
he
num
be
r
of
bi
t
s
.
A
f
t
e
r
l
e
a
r
ni
n
g
de
e
p
ne
t
w
or
k
,
w
e
obt
a
i
n
a
s
e
t
of
i
t
e
m
s
t
ha
t
i
nde
xe
d
by
t
he
ha
s
h
a
ddr
e
s
s
.
L
e
t
=
{
1
,
2
,
…
,
}
be
a
s
e
t
of
i
t
e
m
s
of
,
∈
5x
1
be
t
he
i
nf
or
m
a
t
i
on
v
e
c
t
or
t
ha
t
a
r
e
s
t
or
e
d
i
n
t
he
da
t
a
ba
s
e
,
a
nd
is
t
he
num
be
r
of
i
t
e
m
s
of
.
G
i
v
e
n
=
{
}
=
1
i
s
a
s
e
t
of
w
he
r
e
=
1
,
2
,
…
,
,
t
h
e
s
e
q
u
e
n
c
e
m
a
t
c
h
i
ng
pr
o
c
e
s
s
i
s
s
h
ow
n
i
n
F
i
g
ur
e
9
.
A
S
e
q
u
e
n
c
e
o
f
Q
u
e
r
i
e
s
Q
=
{
q
1
, q
2
, q
3
}
q
1
q
2
q
3
L
e
a
r
n
i
n
g
H
a
sh
F
u
n
c
t
i
o
n
H
(
q
1
)
H
(
q
2
)
H
(
q
3
)
1
0
0
0
1
0
.
.
.
1
0
0
0
1
0
0
0
1
0
.
.
.
1
0
0
1
1
0
0
0
1
0
.
.
.
1
0
1
0
1
0
0
0
1
0
.
.
.
1
0
1
1
1
0
0
0
1
0
.
.
.
1
1
0
0
1
1
1
1
1
1
.
.
.
1
1
1
1
T
h
e
R
e
f
e
r
e
n
c
e
I
n
v
e
r
se
L
o
o
k
u
p
D
a
t
a
b
a
s
e
[
]
5x1
[
]
5x1
x
21
x
22
[
]
5x1
[
]
5x1
[
]
5x1
[
]
5x1
x
11
x
12
x
13
x
14
[
]
5x1
[
]
5x1
[
]
5x1
x
31
x
32
x
33
S
x
11
x
12
x
14
x
21
x
22
x
31
x
32
x
13
x
33
I
n
f
o
r
m
a
t
i
o
n
D
a
t
a
[
S
o
n
g
I
D
,
t
1
,
f
1,
,
Δ
f
,
Δ
t
]
Mo
s
t
fre
quen
tly
occu
rred
sim
ila
r
rel
ativ
e
offs
ets
t
ime
Mi
n
imu
m
Dis
ta
nce
S
o
n
g
I
D
R
e
t
u
r
n
e
d
I
t
e
m
D
a
t
a
C
o
l
l
e
c
t
i
o
n
F
i
g
ur
e
9. O
ur
pr
opos
e
d s
e
que
nc
e
m
a
t
c
hi
ng
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
l
e
c
&
C
om
p E
n
g
I
S
S
N
:
2088
-
8708
Si
m
i
l
ar
i
t
y
-
pr
e
s
e
r
v
i
ng has
h f
o
r
c
ont
e
nt
-
b
as
e
d
au
di
o r
e
t
r
i
e
v
al
us
i
ng
…
(
P
e
t
c
har
at
P
any
apanuw
at
)
887
A
s
c
a
n be
s
e
e
n i
n F
i
g
ur
e
9,
a
s
s
u
m
e
t
ha
t
=
{
1
,
2
,
3
}
, a
nd g
e
t
t
he
bi
na
r
y
c
ode
s
(
)
, w
e
obt
a
i
n a
s
e
t
of
,
1
=
{
11
,
12
,
13
,
14
}
,
2
=
{
21
,
22
}
,
3
=
{
31
,
32
,
33
}
,
a
nd
=
{
11
,
12
,
13
,
14
,
21
,
22
,
31
,
32
,
33
}
.
O
n
e
ne
a
r
e
s
t
ne
i
g
h
b
o
r
s
e
a
r
c
h
(
)
f
o
r
a
q
ue
r
y
i
t
e
m
a
t
o
r
de
r
f
r
om
i
s
d
e
f
i
ne
d
a
s
f
o
l
l
ow
s
:
(
)
=
∈
‖
−
‖
2
w
he
r
e
‖
−
‖
2
i
s
t
he
2
-
nor
m
b
e
t
w
e
e
n
a
nd
s
e
que
nc
e
s
.
G
i
v
e
n
=
{
(
)
}
=
1
i
s
a
c
a
ndi
da
t
e
i
t
e
m
s
s
e
t
of
.
W
e
a
l
s
o
a
ppl
y
a
t
i
m
e
of
f
s
e
t
c
ons
t
r
a
i
nt
f
or
i
m
pr
o
v
i
n
g
t
he
a
c
c
ur
a
c
y
of
t
he
s
e
que
nc
e
m
a
t
c
hi
ng
.
T
he
t
i
m
e
of
f
s
e
t
c
ons
t
r
a
i
nt
|
−
|
i
s
t
he
a
bs
ol
ut
e
d
i
f
f
e
r
e
nc
e
be
t
w
e
e
n
a
nd
.
I
t
c
a
n
b
e
de
f
i
ne
d a
s
f
ol
l
ow
s
:
|
1
−
1
|
=
⋯
=
|
−
|
w
he
r
e
a
nd
a
r
e
t
he
of
f
s
e
t
t
i
m
e
of
r
e
f
e
r
e
nc
e
f
i
l
e
a
nd
que
r
y
,
r
e
s
pe
c
t
i
v
e
l
y
.
T
he
c
ons
t
r
a
i
nt
of
of
f
s
e
t
t
i
m
e
c
a
n
b
e
a
na
l
y
z
e
d
t
ha
t
a
s
e
que
nc
e
of
c
a
ndi
da
t
e
i
t
e
m
s
s
houl
d
oc
c
ur
w
i
t
h
t
he
s
a
m
e
a
bs
ol
ut
e
di
f
f
e
r
e
nc
e
a
m
on
g
t
he
t
i
m
e
s
e
que
n
c
e
s
.
O
ur
pr
opos
e
d ha
s
t
he
f
ol
l
ow
i
ng
pr
o
c
e
dur
e
s
.
I
n
s
u
m
m
a
r
y
,
t
he
pr
opos
e
d
a
udi
o
r
e
t
r
i
e
v
a
l
a
l
g
or
i
t
h
m
i
s
ba
s
e
d
on
t
w
o
pa
r
t
s
,
t
he
s
i
m
i
l
a
r
i
t
y
(
m
i
ni
m
u
m
di
s
t
a
nc
e
)
be
t
w
e
e
n
t
he
a
udi
o
que
r
y
a
nd
t
he
s
ong
i
n
t
he
r
e
f
e
r
e
nc
e
da
t
a
ba
s
e
,
a
nd
t
he
a
bs
ol
ut
e
di
f
f
e
r
e
nc
e
a
m
on
g
t
he
t
i
m
e
s
e
que
n
c
e
s
.
Alg
o
r
it
h
m
:
S
i
m
il
a
r
it
y
-
p
r
e
s
e
r
v
in
g
h
a
s
h
f
o
r
c
o
n
t
e
n
t
-
b
a
s
e
d
a
u
d
io
r
e
t
r
ie
v
a
l
u
s
i
n
g
u
n
s
u
p
e
r
v
i
s
e
d
d
e
e
p
n
e
u
r
a
l
n
e
t
wo
r
k
s
I
n
p
u
t
:
=
{
}
=
1
r
e
f
e
r
e
n
c
e
s
e
t
;
q
u
e
r
y
s
e
t
;
Ou
tp
u
t
:
S
I
D
S
o
n
g
I
D
i
s
r
e
tu
r
n
e
d
b
y
p
r
o
p
o
s
e
d
a
l
g
o
r
it
h
m
.
S
tep
1
:
E
x
tr
a
c
t
i
n
g
f
i
n
g
e
r
p
r
i
n
t
f
e
a
tu
r
e
s
=
{
}
=
1
∈
×
f
r
o
m
th
e
r
e
f
e
r
e
n
c
e
d
a
tas
e
t
.
S
tep
2
:
L
e
a
r
n
i
n
g
h
a
s
h
wh
e
r
e
∈
i
s
a
n
i
n
p
u
t
v
e
c
to
r
.
T
h
e
o
b
jec
t
iv
e
f
u
n
c
t
i
o
n
d
e
f
i
n
e
d
a
s
e
q
u
a
ti
o
n
s
15
-
1
6
.
I
n
th
i
s
s
t
e
p
,
we
a
p
p
l
y
th
e
l
e
a
r
n
i
n
g
f
u
n
c
ti
o
n
ℎ
(
)
f
o
r
a
u
d
i
o
f
i
n
g
e
r
p
r
i
n
t
i
n
s
tep
3
.
S
tep
3
:
S
e
q
u
e
n
c
e
m
a
tch
i
n
g
1
.
T
h
e
q
u
e
r
y
s
a
m
p
le
i
s
d
iv
i
d
e
d
i
n
to
f
i
n
g
e
r
p
r
i
n
ts
=
{
}
=
1
2.
=
{
}
,
=
{
}
,
=
{
}
,
=
{
}
f
o
r
=
1,
2,
,
do
=
ℎ
(
)
=
a
ll
i
tem
s
i
n
(
)
a
r
e
c
o
ll
e
c
ted
i
n
to
=
∪
e
n
d
f
o
r
=
1,
2,
,
do
Au
x
=
m
a
x
Va
l
u
e
f
o
r
=
1
,
2,
,
s
ize
Of
(
)
do
if
‖
−
‖
2
<
Au
x
Au
x
=
‖
−
‖
2
a
b
s
T
i
m
e
=
|
−
|
S
I
D
=
e
n
d
e
n
d
=
∪
{
,
}
e
n
d
3
.
F
i
n
d
i
n
g
m
a
x
f
r
e
q
u
e
n
c
y
o
f
e
a
c
h
S
o
n
g
I
D
o
f
wh
e
r
e
th
e
y
h
a
v
e
s
a
m
e
a
b
s
o
l
u
te
d
if
f
e
r
e
n
c
e
a
m
o
n
g
th
e
t
i
m
e
s
e
q
u
e
n
c
e
s
.
4.
E
X
P
E
R
I
M
E
N
T
A
L
A
N
D
P
E
R
F
O
R
M
A
N
C
E
A
N
A
L
Y
S
I
S
4.1.
D
at
ab
as
e
T
he
pe
r
f
or
m
a
nc
e
of
our
pr
opos
e
d
U
S
H
m
e
t
hod i
s
e
v
a
l
ua
t
e
d on t
he
E
xt
e
nde
d B
a
l
l
r
oo
m
da
t
a
s
e
t
f
r
e
e
l
y
a
v
a
i
l
a
bl
e
i
n
[
24, 25]
.
T
he
da
t
a
s
e
t
c
ons
i
s
t
s
of
4,18
0
m
us
i
c
a
l
e
xc
e
r
pt
s
of
13
g
e
nr
e
s
w
i
t
h
a
l
e
ng
t
h
of
30
s
e
c
onds
e
a
c
h.
T
he
a
ud
i
o
qua
l
i
t
y
of
t
hi
s
da
t
a
i
s
44.1kH
z
,
192
-
kbps
,
s
t
e
r
e
o,
m
p3
f
or
m
a
t
.
I
n
t
hi
s
w
or
k,
t
he
a
udi
o
s
i
g
na
l
is
dow
ns
a
m
pl
e
d
t
o
8K
H
z
.
t
o
m
a
k
e
t
he
da
t
a
e
a
s
i
e
r
t
o
ha
ndl
e
a
s
pr
e
v
i
ous
l
y
m
e
nt
i
one
d.
T
he
t
r
a
i
ni
ng
s
e
t
(
a
l
s
o
us
e
d
a
s
r
e
f
e
r
e
nc
e
d
a
t
a
ba
s
e
f
or
r
e
t
r
i
e
v
a
l
)
i
s
c
o
m
pos
e
d
of
3,000
t
r
a
c
ks
f
r
o
m
8
g
e
nr
e
s
,
t
he
s
a
m
e
r
h
y
t
h
m
c
l
a
s
s
a
s
our
pr
e
v
i
ous
w
or
ks
[
26,
27]
.
A
s
e
t
of
1,0
00
a
udi
o
que
r
i
e
s
w
i
t
h
a
l
e
n
g
t
h
of
10
s
e
c
onds
e
a
c
h
a
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
l
e
c
&
C
om
p E
n
g
,
V
ol
. 11,
N
o. 1, F
e
br
ua
r
y
2021 :
879
-
891
888
r
a
ndo
m
l
y
s
e
l
e
c
t
e
d
f
r
o
m
t
hos
e
3,000
t
r
a
c
ks
.
A
not
he
r
s
e
t
of
200
a
udi
o
que
r
i
e
s
c
om
e
s
f
r
om
a
udi
o
f
i
l
e
s
t
ha
t
do
not
a
ppe
a
r
i
n
t
he
da
t
a
ba
s
e
,
i
n or
de
r
t
o
a
na
l
y
z
e
t
he
f
a
l
s
e
pos
i
t
i
v
e
r
a
t
e
.
E
a
c
h
a
udi
o
s
a
m
pl
e
i
s
r
e
pr
e
s
e
nt
e
d
by
a
20
-
di
m
e
ns
i
ona
l
f
e
a
t
ur
e
v
e
c
t
or
e
xt
r
a
c
t
e
d
b
y
f
i
nge
r
pr
i
nt
a
l
g
or
i
t
h
m
.
T
a
bl
e
2
s
ho
w
s
t
he
nu
m
b
e
r
o
f
s
a
m
pl
e
s
i
n
t
he
da
t
a
ba
s
e
a
nd t
he
que
r
y
s
e
t
.
T
a
bl
e
2.
A
udi
o s
a
m
pl
e
s
i
n t
he
da
t
a
ba
s
e
a
nd que
r
y
s
e
t
S
e
t
N
um
be
r
of
t
r
a
c
ks
L
e
ngt
h
of
s
e
gm
e
nt
(
s
)
N
um
be
r
of
s
a
m
pl
e
s
D
a
t
a
ba
s
e
3,
000
30
441,
184
Q
ue
r
y
1,
200
10
67,
785
4.2.
P
e
r
f
or
m
an
c
e
e
val
u
at
i
on
4.2.1.
E
f
f
e
c
t
i
ve
n
e
s
s
of
r
e
t
r
i
e
val
O
n
a
t
ot
a
l
of
1,200
a
udi
o
que
r
i
e
s
,
t
he
r
e
t
r
i
e
v
a
l
r
e
s
ul
t
s
obt
a
i
ne
d
f
r
om
our
pr
opos
e
d
U
S
H
m
e
t
hod
a
nd
s
t
a
t
e
-
of
-
t
he
-
a
r
t
d
a
t
a
-
i
nde
pe
nde
nt
m
e
t
hod,
t
he
S
ha
z
a
m
a
l
g
or
i
t
h
m
,
a
r
e
s
ho
w
n
i
n
T
a
bl
e
3.
T
he
f
a
l
s
e
n
e
g
a
t
i
v
e
(
F
N
)
r
e
f
e
r
s
t
o
t
he
i
nc
or
r
e
c
t
i
de
nt
i
f
i
c
a
t
i
on
t
ha
t
t
he
q
ue
r
y
a
udi
o
doe
s
not
e
xi
s
t
i
n
t
he
da
t
a
b
a
s
e
w
h
e
n
i
t
doe
s
,
t
r
ue
pos
i
t
i
v
e
(
T
P
)
r
e
f
e
r
s
t
o t
he
c
or
r
e
c
t
i
de
nt
i
f
i
c
a
t
i
on of
t
he
a
udi
o r
e
c
or
di
n
g
f
r
o
m
t
he
que
r
y
,
f
a
l
s
e
pos
i
t
i
v
e
(
F
P
)
r
e
f
e
r
s
t
o
t
he
i
nc
or
r
e
c
t
i
de
nt
i
f
i
c
a
t
i
on
of
t
he
w
r
ong
r
e
c
or
di
n
g
w
he
n
t
he
c
or
r
e
c
t
r
e
c
or
di
ng
do
e
s
not
e
xi
s
t
i
n
t
he
da
t
a
ba
s
e
, a
nd t
r
ue
ne
ga
t
i
v
e
(
T
N
)
r
e
f
e
r
s
t
o t
he
c
o
r
r
e
c
t
i
de
nt
i
f
i
c
a
t
i
on t
ha
t
no a
udi
o r
e
c
o
r
di
ng
m
a
t
c
he
s
t
h
e
qu
e
r
y
.
A
c
c
or
di
ng
t
o
t
he
e
x
pe
r
i
m
e
n
t
a
l
r
e
s
u
l
t
s
,
w
e
o
b
t
a
i
n
h
i
g
he
r
pe
r
c
e
n
t
a
g
e
of
a
c
c
ur
a
c
y
(
8
8.
9
2%
)
f
or
t
h
e
pr
opos
e
d
U
S
H
t
ha
n
s
t
a
t
e
-
of
-
t
he
-
a
r
t
d
a
t
a
-
i
nde
pe
nde
nt
m
e
t
hod,
t
he
S
ha
z
a
m
a
l
g
or
i
t
h
m
(
71.67%
f
or
16
-
bi
t
ha
s
h
c
ode
,
8
7
.
4
2%
f
o
r
2
0
-
b
i
t
h
a
s
h
c
o
d
e
)
.
F
i
g
ur
e
1
0
s
h
ow
s
t
he
r
e
t
r
i
e
v
a
l
a
c
c
ur
a
c
y
c
om
pa
r
i
s
o
n
be
t
w
e
e
n
t
h
e
t
w
o
d
i
f
f
e
r
e
n
t
m
e
t
h
o
d
s
.
T
a
bl
e
3. R
e
t
r
i
e
v
a
l
r
e
s
ul
t
s
c
o
m
pa
r
i
s
on be
t
w
e
e
n
U
S
H
a
nd s
t
a
t
e
-
of
-
t
he
-
a
r
t
da
t
a
-
i
nde
pe
nde
nt
m
e
t
hod
M
e
t
h
od
FN
TP
FP
TN
U
S
H
16
-
bi
t
114
886
19
181
S
h
a
z
a
m
16
-
bi
t
290
710
50
150
S
h
a
z
a
m
20
-
bi
t
132
868
19
181
F
i
g
ur
e
10. T
he
r
e
t
r
i
e
v
a
l
a
c
c
ur
a
c
y
of
t
he
pr
opos
e
d
U
S
H
a
nd t
he
S
ha
z
a
m
a
l
g
or
i
t
h
m
T
he
e
f
f
e
c
t
i
v
e
n
e
s
s
of
t
he
U
S
H
i
s
e
v
a
l
u
a
t
e
d
t
hr
oug
h
t
he
e
xpe
r
i
m
e
nt
s
a
nd
c
o
m
pa
r
e
d
w
i
t
h
s
t
a
t
e
-
of
t
h
e
-
a
r
t
d
a
t
a
-
i
n
de
p
e
n
de
n
t
m
e
t
h
o
d
i
n
t
e
r
m
s
o
f
p
r
e
c
i
s
i
o
n
,
r
e
c
a
l
l
,
F
1
s
c
or
e
,
a
n
d
f
a
l
s
e
p
o
s
i
t
i
v
e
r
a
t
e
[
2
8
]
,
a
s
f
o
l
l
ow
s
:
=
+
×
100
(
20)
=
+
×
1
0
0
(
21)
F1
=
2
×
x
+
(
22)
=
+
×
1
0
0
(
23)
A
s
c
a
n
be
s
e
e
n
i
n
T
a
bl
e
4,
w
e
obt
a
i
n
h
i
g
he
r
pr
e
c
i
s
i
on
a
nd
r
e
c
a
l
l
v
a
l
ue
s
f
or
t
he
pr
opos
e
d
U
S
H
t
ha
n
s
t
a
t
e
-
of
-
t
he
-
a
r
t
da
t
a
-
i
nde
pe
nd
e
nt
m
e
t
hod
bot
h
i
n
16
-
b
i
t
a
nd
20
-
bi
t
ha
s
h
c
ode
.
T
he
F
1
s
c
or
e
(
i
n
t
he
f
our
t
h
c
ol
um
n
)
s
ho
w
s
t
he
o
v
e
r
a
l
l
e
f
f
e
c
t
i
v
e
ne
s
s
of
t
w
o
di
f
f
e
r
e
nt
m
e
t
hods
.
F
ur
t
he
r
m
or
e
,
t
he
U
S
H
ha
s
a
s
i
g
ni
f
i
c
a
nt
l
y
Evaluation Warning : The document was created with Spire.PDF for Python.