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.
15
,
No
.
2
,
A
p
r
il 2
0
2
6
,
p
p
.
1
7
0
1
~
1
7
0
8
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
15
.i
2
.
p
p
1
7
0
1
-
1
7
0
8
1701
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
A
deep
lea
rning
-
b
a
sed
a
ppro
a
ch f
o
r hearing
los
s de
t
ection
Dee
pa
,
M
a
njula
G
urura
j
Ra
o
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
m
a
t
i
o
n
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
N
i
t
t
e
(
D
e
e
m
e
d
t
o
b
e
U
n
i
v
e
r
si
t
y
)
,
N
M
A
M
I
n
st
i
t
u
t
e
o
f
T
e
c
h
n
o
l
o
g
y
,
K
a
r
k
a
l
a
Ta
l
u
k
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
n
1
9
,
2
0
2
5
R
ev
is
ed
Dec
2
9
,
2
0
2
5
Acc
ep
ted
Feb
6
,
2
0
2
6
M
il
li
o
n
s
o
f
p
e
o
p
le
a
c
ro
ss
th
e
wo
rld
a
re
a
ffe
c
ted
b
y
h
e
a
rin
g
lo
ss
a
n
d
e
a
rly
d
e
tec
ti
o
n
is
v
e
ry
imp
o
rtan
t
fo
r
e
f
fe
c
ti
v
e
in
terv
e
n
ti
o
n
.
Th
e
tra
d
it
i
o
n
a
l
h
e
a
rin
g
sc
re
e
n
in
g
m
e
th
o
d
s
a
re
e
ffe
c
ti
v
e
b
u
t
th
e
y
o
ften
re
l
y
o
n
sp
e
c
ialize
d
e
q
u
i
p
m
e
n
t
a
n
d
c
li
n
ica
l
re
so
u
rc
e
s
,
m
a
k
in
g
th
e
m
les
s
a
c
c
e
ss
ib
le
to
c
o
m
m
o
n
p
e
o
p
le.
He
a
rin
g
lo
ss
is
a
sta
te
th
a
t
a
ff
e
c
ts
th
e
a
b
il
it
y
to
c
o
m
m
u
n
ica
te,
so
c
iall
y
in
tera
c
t
a
n
d
o
v
e
ra
ll
q
u
a
li
t
y
o
f
l
if
e
.
Th
e
a
d
v
a
n
c
e
m
e
n
ts
i
n
re
c
e
n
t
y
e
a
rs
h
a
v
e
a
ime
d
to
e
n
h
a
n
c
e
t
h
e
a
c
c
e
ss
ib
il
it
y
a
n
d
e
fficie
n
c
y
o
f
h
e
a
rin
g
tes
ts
,
m
a
in
ly
in
re
m
o
te
a
re
a
s.
Th
e
a
c
c
u
ra
te
c
las
sifica
ti
o
n
o
f
h
e
a
rin
g
lo
ss
is
e
ss
e
n
ti
a
l
f
o
r
e
ffe
c
ti
v
e
d
e
tec
ti
o
n
a
n
d
trea
tme
n
t
in
a
u
d
io
l
o
g
y
.
Th
is
stu
d
y
p
re
se
n
ts
a
d
e
e
p
lea
rn
in
g
(DL)
-
b
a
se
d
a
p
p
r
o
a
c
h
b
a
se
d
o
n
a
fe
e
d
fo
rwa
rd
n
e
u
ra
l
n
e
two
r
k
(F
NN
)
.
Th
is
p
a
p
e
r
fo
c
u
se
s
o
n
c
o
m
m
o
n
c
a
u
se
s
li
k
e
c
e
ru
m
e
n
imp
a
c
ti
o
n
,
o
ti
ti
s
m
e
d
ia,
a
n
d
o
to
sc
lero
sis.
T
h
e
stu
d
y
tr
ies
to
e
x
p
l
o
re
wa
y
s
to
imp
ro
v
e
th
e
d
iag
n
o
sis
o
f
h
e
a
rin
g
l
o
ss
.
T
h
e
g
o
a
l
is
to
d
e
v
e
l
o
p
s
o
lu
ti
o
n
s
t
h
a
t
m
a
k
e
h
e
a
rin
g
sc
re
e
n
in
g
s
m
o
re
a
c
c
e
ss
ib
le
a
n
d
c
o
st
-
e
ffe
c
ti
v
e
fo
r
p
o
p
u
lati
o
n
s
wi
th
li
m
it
e
d
a
c
c
e
s
s
to
h
e
a
lt
h
c
a
re
re
so
u
rc
e
s.
Th
e
re
su
lt
s
sh
o
w
th
e
a
d
v
a
n
tag
e
s
o
f
DL
m
o
d
e
ls
in
su
p
p
o
rti
n
g
a
u
t
o
m
a
ted
a
c
c
u
ra
te
c
la
ss
ifi
c
a
ti
o
n
o
f
h
e
a
rin
g
lo
ss
fo
r
in
telli
g
e
n
t
d
iag
n
o
stic sy
ste
m
s in
a
u
d
i
o
lo
g
ica
l
h
e
a
lt
h
c
a
re
.
K
ey
w
o
r
d
s
:
Au
d
io
m
etr
ic
d
ata
C
las
s
if
icatio
n
Dee
p
lear
n
in
g
Diag
n
o
s
is
Feed
f
o
r
war
d
n
eu
r
al
n
etwo
r
k
Hea
lth
ca
r
e
Hea
r
in
g
lo
s
s
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
:
Ma
n
ju
la
Gu
r
u
r
aj
R
ao
Dep
ar
tm
en
t o
f
I
n
f
o
r
m
atio
n
Scien
ce
an
d
E
n
g
in
ee
r
i
n
g
,
Nitte
(
Dee
m
ed
to
b
e
U
n
iv
er
s
ity
)
NM
AM
I
n
s
titu
te
o
f
T
ec
h
n
o
l
o
g
y
Kar
k
ala
T
alu
k
,
I
n
d
ia
E
m
ail:
d
r
.
m
an
ju
la
g
u
r
u
r
ajh
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Hea
r
in
g
d
is
o
r
d
er
is
a
co
m
m
o
n
h
ea
lth
co
n
d
itio
n
th
at
af
f
e
cts
m
an
y
in
d
iv
id
u
als
ac
r
o
s
s
th
e
wo
r
ld
ir
r
esp
ec
tiv
e
o
f
th
eir
a
g
e,
g
en
d
er
,
a
n
d
s
o
cio
ec
o
n
o
m
ic
b
o
u
n
d
ar
ies.
I
t
ca
n
s
ig
n
if
ican
tly
af
f
ec
t
o
n
e'
s
ab
ilit
y
to
co
m
m
u
n
icate
,
en
g
ag
e
in
s
o
cia
l
in
ter
ac
tio
n
s
a
n
d
m
ain
tain
o
v
er
all
q
u
ality
o
f
life
.
T
h
er
e
co
u
ld
b
e
s
ev
er
al
ca
u
s
es
f
o
r
h
ea
r
in
g
lo
s
s
.
I
t
is
b
r
o
a
d
l
y
ca
teg
o
r
ized
i
n
to
two
ty
p
es
n
am
ely
s
en
s
o
r
in
eu
r
al
o
r
co
ch
lear
h
ea
r
in
g
an
d
co
n
d
u
ctiv
e
h
ea
r
i
n
g
l
o
s
s
.
Sen
s
o
r
in
eu
r
al
h
ea
r
in
g
d
is
o
r
d
er
o
cc
u
r
s
b
ec
au
s
e
o
f
i
n
ju
r
y
to
t
h
e
i
n
n
er
p
ar
t
o
f
t
h
e
ea
r
an
d
co
n
d
u
ctiv
e
h
ea
r
in
g
lo
s
s
r
esu
lts
d
u
e
to
o
b
s
tr
u
ctio
n
s
o
r
a
b
n
o
r
m
alities
in
th
e
o
u
ter
p
ar
t
o
r
m
id
d
le
ea
r
.
T
h
is
h
ea
r
in
g
l
o
s
s
is
ca
u
s
ed
as
a
r
esu
lt
o
f
ea
r
in
f
ec
tio
n
s
,
th
e
b
u
ild
u
p
o
f
ea
r
wax
,
o
to
s
cler
o
s
is
an
d
o
th
er
f
o
r
m
s
o
f
ea
r
o
b
s
tr
u
ctio
n
.
T
h
ese
co
n
d
itio
n
s
ar
e
tr
ea
tab
le
,
b
u
t th
ey
s
till
co
n
tr
ib
u
te
s
ig
n
if
ican
tly
to
h
ea
r
i
n
g
im
p
air
m
en
t a
cr
o
s
s
th
e
wo
r
ld
.
T
h
e
g
lo
b
al
b
u
r
d
en
o
f
h
ea
r
in
g
l
o
s
s
is
h
u
g
e
with
n
ea
r
ly
4
6
6
m
illi
o
n
p
eo
p
le
wo
r
l
d
wid
e
ex
p
e
r
ien
cin
g
s
o
m
e
f
o
r
m
o
f
d
is
ab
lin
g
h
e
ar
in
g
lo
s
s
.
Of
th
ese,
co
n
d
u
ctiv
e
h
ea
r
in
g
lo
s
s
,
wh
ich
i
s
less
s
ev
er
e
th
an
s
en
s
o
r
in
eu
r
al
h
ea
r
in
g
lo
s
s
af
f
ec
ts
a
lar
g
e
p
ar
t
o
f
in
d
iv
id
u
als.
I
t
is
r
esp
o
n
s
ib
le
f
o
r
im
p
air
in
g
th
e
ab
ilit
y
to
h
ea
r
s
o
u
n
d
s
clea
r
ly
th
u
s
d
is
r
u
p
tin
g
d
aily
co
m
m
u
n
icatio
n
a
n
d
s
o
ci
al
in
ter
ac
tio
n
s
.
T
h
e
q
u
ality
o
f
a
p
er
s
o
n
'
s
life
ca
n
b
e
g
r
ea
tly
im
p
ac
ted
b
y
h
ea
r
in
g
im
p
air
m
en
ts
.
E
f
f
ec
ti
v
e
in
ter
v
en
tio
n
an
d
r
eh
a
b
ilit
atio
n
o
f
h
e
ar
in
g
l
o
s
s
d
ep
en
d
o
n
ea
r
ly
d
etec
tio
n
,
esp
ec
ially
in
p
o
p
u
latio
n
s
th
at
ar
e
at
r
is
k
,
s
u
ch
as
in
f
an
ts
,
th
e
eld
er
ly
,
a
n
d
p
e
o
p
l
e
wh
o
ar
e
ex
p
o
s
ed
to
n
o
is
e
at
wo
r
k
.
T
r
a
d
itio
n
al
au
d
i
o
m
etr
i
c
test
in
g
tech
n
iq
u
es
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il 2
0
2
6
:
1
7
0
1
-
1
7
0
8
1702
f
r
eq
u
e
n
tly
u
s
e
s
u
b
jectiv
e
ass
e
s
s
m
en
t
an
d
co
u
ld
n
o
t
alwa
y
s
o
f
f
er
th
e
ac
cu
r
ac
y
n
ee
d
ed
to
i
d
en
tify
m
ild
h
ea
r
in
g
lo
s
s
at
its
ea
r
lie
s
t
s
tag
es.
T
y
m
p
an
o
m
etr
y
an
d
au
d
i
o
m
etr
y
test
in
g
ar
e
two
ex
am
p
les
o
f
t
h
e
s
p
ec
ialized
to
o
ls
th
at
h
av
e
h
is
to
r
ically
b
ee
n
u
s
ed
.
T
h
ese
tech
n
i
q
u
es
f
r
e
q
u
en
t
ly
n
ee
d
e
x
p
en
s
iv
e
e
q
u
ip
m
e
n
t,
s
k
illed
p
er
s
o
n
n
el,
an
d
ac
ce
s
s
to
m
ed
ical
f
ac
ilit
ies wh
ich
m
ay
n
o
t b
e
av
ailab
le
i
n
r
u
r
al
l
o
ca
tio
n
s
.
Acc
o
r
d
in
g
t
o
B
ar
b
o
u
r
et
a
l.
[
1
]
,
an
o
n
lin
e
m
ac
h
in
e
lear
n
in
g
(
ML
)
b
ased
a
u
d
io
m
etr
y
tec
h
n
iq
u
e
was
p
r
esen
ted
th
at
esti
m
ates h
ea
r
in
g
th
r
esh
o
l
d
s
th
r
o
u
g
h
o
u
t th
e
e
n
tire
f
r
eq
u
en
cy
r
an
g
e.
T
h
e
ap
p
r
o
ac
h
im
p
r
o
v
es th
e
ass
es
s
m
en
t's
ef
f
icien
cy
b
y
p
r
o
v
id
in
g
m
o
r
e
a
u
d
io
g
r
am
d
et
ails
th
at
co
n
v
en
tio
n
al
m
et
h
o
d
s
d
o
n
o
t.
Dig
ital
m
eth
o
d
s
f
o
r
au
to
m
ate
d
au
d
io
m
etr
ic
test
in
g
with
ML
alg
o
r
ith
m
s
ar
e
d
is
cu
s
s
ed
in
[
2
]
.
T
h
e
au
th
o
r
s
ex
am
in
ed
2
7
au
t
o
m
ated
h
ea
r
in
g
test
te
ch
n
iq
u
es
a
n
d
h
ig
h
lig
h
ted
h
o
w
ML
lo
wer
s
th
e
n
u
m
b
er
o
f
tr
ials
n
ee
d
e
d
f
o
r
p
u
r
e
-
to
n
e
au
d
i
o
m
etr
y
(
PTA)
,
r
esu
ltin
g
in
r
ea
s
o
n
a
b
ly
p
r
ice
d
au
d
io
m
et
r
ic
ev
alu
atio
n
s
.
T
h
e
s
tu
d
y
in
[
3
]
,
[
4
]
cr
ea
ted
a
d
ee
p
lear
n
in
g
(
DL
)
m
o
d
el
to
p
r
ed
ict
co
n
d
u
ctiv
e
h
ea
r
in
g
to
p
r
o
p
er
l
y
d
iag
n
o
s
e
co
n
d
u
ctiv
e
h
ea
r
in
g
lo
s
s
b
r
o
u
g
h
t
o
n
b
y
o
titi
s
m
ed
ia,
an
d
o
to
s
co
p
ic
p
ictu
r
es
wer
e
an
aly
ze
d
.
T
h
e
m
o
d
el
s
h
o
wed
an
ac
cu
r
ac
y
o
f
81%
,
wh
ich
is
tr
ain
ed
o
n
a
d
at
aset o
f
m
o
r
e
th
a
n
2
,
0
0
0
o
to
s
co
p
ic
p
ictu
r
es.
Sch
litt
en
lach
er
et
a
l.
[
5
]
d
e
m
o
n
s
tr
ated
h
o
w
ML
tech
n
iq
u
es,
lik
e
B
ay
esian
ac
tiv
e
lea
r
n
in
g
,
ca
n
m
ax
im
ize
PTA'
s
ef
f
ec
tiv
en
ess
in
th
e
m
o
s
t
u
n
ce
r
tain
r
e
g
io
n
s
.
Gath
m
an
et
a
l.
[
6
]
u
s
ed
d
ec
is
io
n
tr
ee
alg
o
r
ith
m
s
,
n
am
ely
lig
h
t
g
r
ad
i
en
t
b
o
o
s
tin
g
m
ac
h
in
e
(
L
GB
M)
,
to
p
r
e
d
ict
au
d
io
m
etr
y
-
b
ased
h
ea
r
in
g
le
v
els
f
o
r
p
r
ed
ictin
g
h
ea
r
in
g
lo
s
s
u
s
in
g
d
em
o
g
r
a
p
h
ic
an
d
clin
ical
d
ata
.
T
h
e
ac
cu
r
ac
y
o
b
tain
e
d
b
y
th
e
m
o
d
el
was
v
er
y
h
ig
h
.
AlSam
h
o
r
i
et
a
l.
[
7
]
d
is
cu
s
s
ed
h
o
w
AI
a
n
d
ML
m
ig
h
t
h
elp
to
p
r
ev
e
n
t
h
ea
r
i
n
g
lo
s
s
.
T
h
e
s
tu
d
y
d
em
o
n
s
tr
ates
th
at
AI
alg
o
r
ith
m
s
ca
n
b
e
in
cl
u
d
ed
i
n
to
au
d
i
o
m
etr
ic
d
ev
ices
to
im
p
r
o
v
e
t
h
e
ef
f
icien
cy
an
d
in
d
iag
n
o
s
in
g
h
ea
r
in
g
lo
s
s
.
Sh
in
et
a
l.
[
8
]
cr
ea
ted
an
d
co
m
p
ar
ed
s
ev
er
al
ML
m
o
d
els
lik
e
g
r
ad
ie
n
t
b
o
o
s
tin
g
,
r
ec
u
r
r
en
t
n
eu
r
al
n
etwo
r
k
s
,
a
n
d
m
u
ltil
ay
er
p
er
ce
p
tr
o
n
s
.
W
en
g
et
a
l.
[
9
]
ex
am
in
e
d
th
e
ap
p
licatio
n
o
f
ML
m
o
d
els
to
p
r
e
d
ict
s
o
u
n
d
p
e
r
ce
p
tio
n
.
I
t
an
aly
ze
d
1
6
r
esear
ch
s
tu
d
ies
with
5
,
0
5
8
co
ch
lear
im
p
la
n
t
u
s
er
s
,
b
o
th
ad
u
lts
an
d
ch
ild
r
en
.
T
o
class
if
y
p
u
r
e
-
to
n
e
au
d
io
g
r
a
m
d
ata,
Do
u
et
a
l.
[
1
0
]
s
u
g
g
ested
a
h
y
b
r
id
ML
a
p
p
r
o
ac
h
th
at
b
le
n
d
s
d
ec
is
io
n
tr
ee
s
with
DL
m
eth
o
d
s
.
T
h
e
m
o
d
el
class
if
ied
v
ar
i
o
u
s
ty
p
es o
f
h
ea
r
in
g
lo
s
s
with
an
a
cc
u
r
ac
y
r
a
n
g
e
o
f
9
6
.
7
5
% to
9
9
.
8
5
%.
ML
m
eth
o
d
s
wer
e
u
s
ed
in
[
1
1
]
to
id
en
tify
m
id
d
le
-
ea
r
d
is
o
r
d
er
s
s
u
ch
as
o
to
s
cler
o
s
is
an
d
o
s
s
icu
lar
d
is
ar
ticu
latio
n
.
Sev
er
al
ML
m
o
d
els
wer
e
u
s
e
d
[
1
2
]
.
T
h
e
u
s
e
o
f
DL
m
o
d
el
s
f
o
r
ca
teg
o
r
izin
g
au
d
i
o
m
etr
i
c
d
ata
an
d
id
en
tify
in
g
h
ea
r
in
g
lo
s
s
is
d
is
cu
s
s
ed
[
1
3
]
.
C
r
o
wso
n
et
a
l.
[
1
4
]
in
v
esti
g
ated
th
e
ap
p
licatio
n
o
f
DL
to
au
to
m
ate
au
d
io
m
etr
ic
ex
am
i
n
atio
n
s
b
y
in
cr
ea
s
in
g
th
e
ac
ce
s
s
ib
ilit
y
an
d
af
f
o
r
d
ab
ilit
y
o
f
h
ea
r
in
g
lo
s
s
id
en
tific
atio
n
in
en
v
ir
o
n
m
en
ts
with
lim
ited
r
eso
u
r
ce
s
.
C
alab
r
ese
et
a
l.
[
1
5
]
p
r
esen
ted
d
ee
p
n
eu
r
al
n
etwo
r
k
s
f
o
r
au
d
io
g
r
am
esti
m
atio
n
.
Hep
s
ib
a
an
d
J
u
s
tin
[
1
6
]
s
h
o
wed
th
at
d
ee
p
n
e
u
r
al
n
etwo
r
k
s
ca
n
b
e
u
s
ed
t
o
im
p
r
o
v
e
s
p
ee
ch
co
m
p
r
eh
e
n
s
io
n
f
o
r
p
eo
p
le
w
h
o
m
ak
e
u
s
e
o
f
h
ea
r
in
g
aid
s
.
Sh
in
et
a
l.
[
1
7
]
ex
p
lain
e
d
th
at
h
ea
r
in
g
lo
s
s
d
etec
tio
n
ca
n
b
e
im
p
r
o
v
e
d
wh
en
p
u
r
e
-
to
n
e
au
d
io
g
r
a
m
s
ar
e
co
m
b
in
e
d
with
s
p
ee
ch
r
ec
o
g
n
itio
n
d
ata
.
An
ad
v
an
ce
m
e
n
t
f
o
r
telem
ed
ici
n
e
an
d
m
o
b
ile
h
ea
lth
ap
p
li
ca
tio
n
s
is
p
r
o
p
o
s
ed
in
[
1
8
]
f
o
r
th
e
r
ea
l
-
tim
e
class
if
icatio
n
o
f
h
ea
r
in
g
l
o
s
s
u
s
in
g
DL
m
o
d
els.
T
h
e
ap
p
licatio
n
o
f
c
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NNs
)
in
o
r
d
e
r
to
a
u
to
m
ate
th
e
class
i
f
icatio
n
o
f
au
d
io
m
etr
ic
r
esu
lts
is
d
is
cu
s
s
ed
in
[
1
9
]
.
Acc
o
r
d
in
g
to
th
e
ir
r
esear
ch
,
C
NN
-
b
ased
m
o
d
els
ar
e
ca
p
ab
le
o
f
ac
cu
r
ately
class
if
y
in
g
au
d
io
g
r
am
s
with
h
ig
h
p
r
ec
is
io
n
as
clo
s
e
to
p
r
o
f
ess
io
n
al
au
d
io
lo
g
is
ts
.
T
h
e
au
th
o
r
s
in
[
2
0
]
–
[
2
2
]
d
em
o
n
s
tr
ated
h
o
w
a
r
tific
ial
n
eu
r
al
n
etwo
r
k
s
(
AN
N)
ar
e
u
s
ed
i
n
th
e
d
etec
tio
n
o
f
h
ea
r
in
g
lo
s
s
.
T
h
eir
s
tu
d
y
s
h
o
ws
th
at
b
y
an
aly
zin
g
th
e
au
d
io
m
etr
ic
d
ata
in
a
v
er
y
p
r
ec
is
e
an
d
s
ca
lab
le
m
an
n
er
,
ANNs
ca
n
an
ticip
ate
th
e
s
ev
er
ity
o
f
im
p
air
m
en
ts
an
d
id
e
n
tify
d
if
f
er
en
t
f
o
r
m
s
o
f
h
ea
r
i
n
g
lo
s
s
.
T
h
e
s
tu
d
y
in
[
2
3
]
–
[
2
5
]
h
ig
h
lig
h
ted
th
at
ML
an
d
s
ig
n
a
l
p
r
o
ce
s
s
in
g
m
eth
o
d
s
m
ig
h
t
b
e
co
m
b
i
n
ed
t
o
p
r
ed
ict
h
ea
r
in
g
l
o
s
s
.
B
y
co
m
b
in
i
n
g
th
e
ad
v
an
tag
es
o
f
DL
with
s
ig
n
al
p
r
o
ce
s
s
in
g
,
th
is
h
y
b
r
id
tech
n
iq
u
e
ca
n
d
etec
t
h
ea
r
in
g
l
o
s
s
th
at
ar
e
m
o
r
e
p
r
ec
is
e
an
d
tr
u
s
two
r
th
y
.
Dete
ctin
g
th
e
h
ea
r
in
g
lo
s
s
is
d
if
f
icu
lt
,
esp
ec
ially
in
en
v
ir
o
n
m
en
ts
with
lim
ited
r
eso
u
r
ce
s
.
Ar
tific
ial
in
tellig
en
ce
h
as
tr
an
s
f
o
r
m
e
d
s
ev
er
al
ar
ea
s
o
f
h
ea
lth
ca
r
e
in
t
h
e
last
ten
y
ea
r
s
b
y
m
ea
n
s
o
f
p
r
ec
is
io
n
m
ed
icin
e,
au
to
m
atin
g
d
iag
n
o
s
tic
p
r
o
ce
d
u
r
es
,
an
d
im
p
r
o
v
in
g
p
atien
t
o
u
tco
m
es.
C
o
n
v
e
n
tio
n
al
a
u
d
io
m
etr
y
d
ep
e
n
d
s
h
ea
v
ily
o
n
tr
ai
n
ed
a
u
d
io
l
o
g
is
ts
to
in
ter
p
r
et
r
esu
lts
.
T
h
is
p
o
s
es
ch
allen
g
es
in
lar
g
e
-
s
ca
le
s
c
r
ee
n
in
g
s
.
T
h
is
h
as
m
ad
e
it n
ec
ess
ar
y
to
cr
ea
te
D
L
-
d
r
iv
en
m
o
d
els th
at
ca
n
im
p
r
o
v
e
th
e
p
er
f
o
r
m
an
ce
.
T
h
e
k
e
y
c
o
n
t
r
i
b
u
t
i
o
n
o
f
t
h
i
s
r
e
s
e
a
r
c
h
p
a
p
e
r
i
n
c
l
u
d
e
s
a
f
e
e
d
f
o
r
w
a
r
d
n
e
u
r
a
l
n
e
t
w
o
r
k
(
F
N
N
)
w
a
s
d
e
s
i
g
n
e
d
t
o
a
u
t
o
m
a
t
i
c
a
l
l
y
c
l
a
s
s
i
f
y
c
o
m
m
o
n
t
y
p
e
s
o
f
h
e
a
r
i
n
g
l
o
s
s
.
T
h
e
p
r
o
p
o
s
e
d
D
L
m
o
d
e
l
h
e
l
p
s
i
n
e
a
r
l
y
d
e
t
e
c
t
i
o
n
b
y
s
h
o
w
i
n
g
g
r
e
a
t
a
c
c
u
r
a
c
y
i
n
c
l
a
s
s
i
f
y
i
n
g
h
e
a
r
i
n
g
l
o
s
s
.
T
h
i
s
p
a
p
e
r
c
h
e
c
k
s
t
h
e
f
e
a
s
i
b
i
l
i
t
y
o
f
a
u
d
i
o
m
e
t
r
i
c
e
v
a
l
u
a
t
i
o
n
t
h
r
o
u
g
h
DL
t
e
c
h
n
i
q
u
e
s
f
o
r
t
h
e
e
a
r
l
y
i
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
h
e
a
r
i
n
g
i
m
p
a
i
r
m
e
n
t
s
.
I
n
c
l
u
d
i
n
g
D
L
t
o
a
u
d
i
o
m
e
t
r
i
c
t
e
s
t
i
n
g
m
a
k
e
s
t
h
e
d
i
a
g
n
o
s
i
s
p
r
o
c
e
s
s
b
e
t
t
e
r
.
I
t
a
l
s
o
l
e
a
d
s
t
o
s
c
a
l
a
b
l
e
a
n
d
c
o
s
t
-
e
f
f
e
c
t
i
v
e
s
o
l
u
t
i
o
n
s
i
n
b
o
t
h
c
l
i
n
i
c
a
l
a
n
d
r
e
m
o
t
e
s
e
t
t
i
n
g
s
.
T
h
i
s
p
a
p
e
r
a
i
m
s
t
o
i
n
v
e
s
t
i
g
a
t
e
n
o
v
e
l
a
p
p
r
o
a
c
h
e
s
t
o
i
m
p
r
o
v
e
t
h
e
d
i
a
g
n
o
s
i
s
o
f
h
e
a
r
i
n
g
l
o
s
s
t
o
e
n
h
a
n
c
e
a
c
c
e
s
s
i
b
i
l
i
t
y
f
o
r
i
n
d
i
v
i
d
u
a
l
s
a
f
f
e
c
t
e
d
.
I
t
a
i
m
s
t
o
i
d
e
n
t
i
f
y
s
o
l
u
t
i
o
n
s
t
h
a
t
c
a
n
b
e
a
p
p
l
i
e
d
i
n
l
o
w
-
r
e
s
o
u
r
c
e
e
n
v
i
r
o
n
m
e
n
t
s
b
y
i
n
v
e
s
t
i
g
a
t
i
n
g
c
u
t
t
i
n
g
-
e
d
g
e
t
e
c
h
n
i
q
u
e
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
A
d
ee
p
lea
r
n
in
g
b
a
s
ed
a
p
p
r
o
a
ch
fo
r
h
ea
r
in
g
lo
s
s
d
etec
tio
n
(
Dee
p
a
)
1703
2.
M
E
T
H
O
D
T
h
is
s
e
cti
o
n
o
u
tli
n
es
t
h
e
m
e
th
o
d
o
lo
g
y
f
o
ll
o
we
d
in
t
h
e
p
r
o
p
o
s
ed
DL
m
o
d
el
f
o
r
h
e
ar
i
n
g
lo
s
s
.
T
h
e
s
t
ep
s
p
r
o
p
o
s
e
d
i
n
cl
u
d
e
d
at
a
c
o
ll
ec
ti
o
n
,
d
at
a
p
r
e
p
r
o
ce
s
s
in
g
,
a
n
d
m
o
d
el
d
e
v
e
lo
p
m
e
n
t
.
E
a
c
h
s
te
p
i
s
d
esi
g
n
e
d
t
o
e
n
s
u
r
e
r
o
b
u
s
t
f
ea
tu
r
e
e
x
t
r
a
cti
o
n
,
ac
c
u
r
ate
le
ar
n
i
n
g
,
an
d
r
eli
ab
le
p
r
ed
i
cti
o
n
p
e
r
f
o
r
m
a
n
c
e
o
f
t
h
e
p
r
o
p
o
s
e
d
m
o
d
el
.
2
.
1
.
Da
t
a
c
o
llect
io
n
C
o
llectio
n
o
f
th
e
au
d
io
m
etr
ic
d
etails
is
th
e
in
itial
s
tag
e
f
o
r
d
ev
elo
p
in
g
th
e
p
r
o
p
o
s
ed
DL
m
o
d
el.
T
h
e
d
ata
was
co
llected
f
r
o
m
th
e
UC
I
Au
d
io
lo
g
y
d
ataset.
T
h
is
d
ataset
in
clu
d
es
b
o
th
ca
teg
o
r
ical
an
d
n
u
m
er
ical
in
f
o
r
m
atio
n
,
wh
ich
is
m
o
s
tly
u
tili
ze
d
f
o
r
th
e
class
if
icat
io
n
o
f
h
ea
r
in
g
lo
s
s
.
An
F
NN
f
o
r
m
u
lti
-
class
class
if
icatio
n
o
f
h
ea
r
in
g
lo
s
s
is
p
r
esen
ted
in
th
is
r
e
s
ea
r
ch
.
T
h
e
d
ataset
is
m
ad
e
u
p
o
f
au
d
io
lo
g
ical
test
d
ata
th
at
h
as
ch
ar
ac
ter
is
tics
wh
ich
h
elp
in
th
e
d
etec
tio
n
o
f
h
ea
r
in
g
lo
s
s
in
v
ar
io
u
s
co
n
te
x
ts
.
T
h
e
v
ar
io
u
s
f
ea
tu
r
es
in
th
e
d
ataset
wer
e
p
atien
t
attr
ib
u
tes
s
u
ch
as
ag
e,
s
ex
,
an
d
a
u
d
io
lo
g
ical
test
r
esu
lts
.
T
h
e
ca
teg
o
r
ical
f
ea
tu
r
es
in
clu
d
e
test
ty
p
es,
ea
r
s
id
e
(
le
f
t
o
r
r
ig
h
t)
a
n
d
ce
r
tain
test
r
es
u
lts
s
u
ch
as
lo
s
s
o
r
co
n
d
itio
n
.
Nu
m
er
ical
f
ea
tu
r
es
in
clu
d
ed
v
alu
es
s
u
ch
as
th
e
p
atien
t's
ag
e,
au
d
io
m
etr
ic
r
esu
l
ts
,
an
d
v
ar
io
u
s
m
ea
s
u
r
em
e
n
ts
r
elate
d
to
h
ea
r
in
g
th
r
esh
o
ld
s
an
d
im
p
ed
a
n
ce
s
.
T
h
is
d
ataset
p
r
esen
ts
a
ch
al
len
g
e
b
ec
au
s
e
o
f
th
e
v
ar
iety
o
f
f
ea
t
u
r
e
ty
p
es.
Pre
p
r
o
ce
s
s
in
g
ap
p
r
o
ac
h
es
ar
e
n
ee
d
ed
to
n
o
r
m
alize
n
u
m
er
i
ca
l
f
ea
tu
r
es
an
d
tr
an
s
f
o
r
m
ca
teg
o
r
ical
v
ar
iab
les
in
to
a
f
o
r
m
at
th
at
ca
n
b
e
u
s
ed
.
T
h
e
g
o
al
was
to
p
r
ed
ict
o
n
e
o
f
f
iv
e
p
o
s
s
ib
le
class
es
o
f
h
e
ar
in
g
l
o
s
s
:
n
o
r
m
al,
co
n
d
u
ctiv
e
,
m
ix
ed
,
co
ch
lea
r
,
a
n
d
r
etr
o
co
ch
le
ar
.
2
.
2
.
Da
t
a
prepro
ce
s
s
ing
T
h
e
p
r
e
p
r
o
ce
s
s
in
g
o
f
d
ata
is
p
er
f
o
r
m
ed
in
o
r
d
er
to
p
r
ep
a
r
e
th
e
d
ataset
th
at
is
ap
p
licab
le
f
o
r
DL
m
o
d
el
tr
ain
in
g
.
T
h
is
s
tep
en
s
u
r
es
d
ata
co
n
s
is
ten
cy
,
r
ed
u
ce
s
n
o
is
e,
an
d
im
p
r
o
v
es
o
v
e
r
all
d
ata
q
u
ality
.
T
h
er
e
wer
e
m
u
ltip
le
s
tep
s
in
th
is
p
h
a
s
e
,
wh
ich
ar
e
lis
ted
as f
o
llo
ws
:
i)
C
ateg
o
r
ical
en
co
d
in
g
: th
e
o
n
e
-
h
o
t e
n
co
d
i
n
g
was u
s
ed
to
tr
an
s
f
o
r
m
th
e
ca
teg
o
r
ical
f
ea
tu
r
es
,
lik
e
test
k
in
d
an
d
ea
r
s
id
e
,
in
to
b
in
ar
y
v
ec
to
r
s
.
E
ac
h
ca
teg
o
r
y
in
th
e
f
ea
tu
r
e
h
as
its
o
wn
b
in
ar
y
c
o
lu
m
n
c
r
ea
ted
in
th
is
s
tag
e.
Fo
r
ex
am
p
le,
th
e
ea
r
s
i
d
e
was
r
ep
r
esen
ted
as
b
in
a
r
y
co
lu
m
n
s
f
o
r
th
e
lef
t
p
ar
t
o
f
ea
r
s
an
d
r
ig
h
t
ea
r
s
,
an
d
th
e
test
ty
p
e
f
ea
tu
r
e
was e
n
co
d
ed
in
to
s
ep
ar
ate
co
l
u
m
n
s
f
o
r
ea
ch
test
ty
p
e.
ii)
Featu
r
e
s
ca
lin
g
:
s
tan
d
ar
d
izatio
n
was
u
s
ed
to
s
tan
d
ar
d
ize
n
u
m
er
ical
f
ea
tu
r
es
wh
ich
s
ca
les
th
e
f
ea
tu
r
es
t
o
h
av
e
u
n
it
v
ar
ian
ce
an
d
ze
r
o
m
ea
n
v
al
u
e.
T
h
is
h
elp
s
in
m
o
r
e
ef
f
icien
t
tr
ain
i
n
g
b
y
p
r
ev
e
n
tin
g
ch
ar
ac
ter
is
tics
f
r
o
m
h
av
i
n
g
an
u
n
ev
e
n
ef
f
ec
t
o
n
th
e
m
o
d
el'
s
lear
n
in
g
.
iii)
Data
s
p
litt
in
g
:
th
e
d
ata
is
r
an
d
o
m
ly
g
r
o
u
p
e
d
in
to
tr
ain
in
g
s
et
an
d
v
alid
atio
n
s
et
as
an
8
0
/2
0
s
p
lit.
T
h
is
d
iv
is
io
n
allo
ws th
e
m
o
d
el
t
o
b
e
ev
alu
ated
o
n
n
ew
d
ata
,
wh
ic
h
h
elp
s
to
m
in
im
ize
t
h
e
r
is
k
o
f
o
v
er
f
itti
n
g
.
iv
)
T
en
s
o
r
co
n
v
er
s
io
n
:
af
ter
p
r
e
p
r
o
ce
s
s
in
g
th
e
d
ataset
was
c
o
n
v
er
ted
in
to
Py
T
o
r
c
h
ten
s
o
r
s
,
wh
ich
is
a
n
ec
ess
ar
y
f
o
r
m
at
f
o
r
tr
ain
in
g
DL
m
o
d
els
in
Py
T
o
r
ch
.
T
h
e
d
ata
was
th
en
p
ac
k
ag
ed
i
n
to
d
ata
lo
ad
er
o
b
jects
f
o
r
m
i
n
i
-
b
atch
t
r
ain
in
g
th
at
allo
ws
f
o
r
m
o
r
e
ef
f
icie
n
t
p
r
o
ce
s
s
in
g
a
n
d
h
elp
s
to
m
a
n
ag
e
m
em
o
r
y
co
n
s
u
m
p
tio
n
d
u
r
i
n
g
m
o
d
el
tr
a
in
in
g
.
2
.
3
.
M
o
del dev
elo
pm
ent
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
is
d
esig
n
ed
to
ca
p
tu
r
e
r
elev
an
t
p
atter
n
s
f
r
o
m
th
e
in
p
u
t
d
ata.
T
h
e
m
o
d
el
is
m
ad
e
u
p
o
f
m
u
ltip
le
i
n
ter
co
n
n
ec
te
d
lay
er
s
th
at
e
n
ab
le
h
ie
r
ar
c
h
ical
f
ea
tu
r
e
lear
n
in
g
an
d
im
p
r
o
v
e
d
p
r
ed
ictio
n
ac
cu
r
ac
y
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
is
ex
p
lain
ed
as f
o
llo
ws
:
i)
I
n
p
u
t
la
y
er
:
th
e
in
p
u
t
lay
e
r
co
r
r
esp
o
n
d
s
to
t
h
e
f
ea
tu
r
es
in
th
e
p
r
ep
r
o
ce
s
s
ed
d
ataset.
E
ac
h
n
o
d
e
r
e
p
r
esen
ts
o
n
e
f
ea
tu
r
e,
i
n
clu
d
in
g
b
o
th
ca
teg
o
r
ical
an
d
n
u
m
er
ical
f
e
atu
r
es.
T
h
is
lay
er
s
er
v
es
as
th
e
in
ter
f
ac
e
b
etwe
en
th
e
r
aw
in
p
u
t d
ata
a
n
d
th
e
DL
m
o
d
el.
ii)
Hid
d
en
lay
er
s
:
th
e
m
o
d
el
h
as
two
h
id
d
en
lay
er
s
.
T
h
e
f
ir
s
t
lay
er
h
as
1
2
8
n
eu
r
o
n
s
,
wh
ich
is
ac
tiv
ated
b
y
th
e
r
ec
tifie
d
li
n
ea
r
u
n
it
(
R
eL
U
)
ac
tiv
atio
n
f
u
n
ctio
n
th
at
h
elp
s
to
ad
d
n
o
n
-
lin
ea
r
ity
to
th
e
m
o
d
el.
A
d
r
o
p
o
u
t
r
ate
o
f
ab
o
u
t
0
.
3
was
ap
p
lied
to
th
is
lay
er
to
m
in
i
m
ize
an
y
o
v
e
r
f
itti
n
g
.
T
h
e
n
ex
t
lay
er
is
m
ad
e
u
p
o
f
6
4
n
eu
r
o
n
s
wh
ich
also
u
s
es
th
e
R
eL
U
ac
tiv
atio
n
f
u
n
cti
o
n
a
n
d
a
0
.
3
d
r
o
p
o
u
t
r
ate
to
f
u
r
th
er
p
r
ev
en
t
o
v
er
f
itti
n
g
a
n
d
e
n
h
an
ce
g
en
er
aliza
tio
n
.
iii)
Ou
tp
u
t
lay
er
:
th
is
lay
er
h
as
5
n
eu
r
o
n
s
r
ep
r
esen
tin
g
o
n
e
o
f
th
e
f
iv
e
p
o
ten
tial
class
es
o
f
h
ea
r
in
g
lo
s
s
.
A
So
f
tMa
x
a
ctiv
atio
n
is
ap
p
lied
to
th
is
lay
er
to
tr
a
n
s
f
o
r
m
th
e
r
aw
v
alu
es
to
a
p
r
o
b
ab
ilit
y
d
is
tr
ib
u
tio
n
.
T
h
e
h
ig
h
est p
r
o
b
ab
ilit
y
class
is
ch
o
s
en
as th
e
p
r
ed
icted
o
u
tp
u
t.
2
.
4
.
M
o
del t
ra
ini
ng
T
h
e
m
o
d
el
tr
ain
in
g
p
r
o
ce
s
s
is
a
cr
itical
p
h
ase
th
at
d
eter
m
in
e
s
th
e
lear
n
in
g
ef
f
icien
cy
o
f
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
.
I
t
f
o
c
u
s
es
o
n
m
in
im
izin
g
p
r
e
d
ictio
n
er
r
o
r
s
.
T
h
e
m
o
d
el
tr
ain
in
g
p
r
o
ce
s
s
in
v
o
lv
es
s
elec
tin
g
ap
p
r
o
p
r
iate
o
p
tim
izatio
n
tech
n
iq
u
es,
lo
s
s
f
u
n
ctio
n
s
,
an
d
ev
al
u
atio
n
m
etr
ics:
i)
L
o
s
s
f
u
n
ctio
n
:
th
e
cr
o
s
s
-
en
tr
o
p
y
lo
s
s
was
ap
p
lied
as
th
e
lo
s
s
f
u
n
ctio
n
.
T
h
is
is
id
ea
l
f
o
r
th
e
task
o
f
m
u
lti
-
class
clas
s
if
icatio
n
.
T
h
is
f
u
n
ctio
n
co
m
b
in
es
So
f
tMa
x
a
ctiv
atio
n
with
a
lo
g
l
o
s
s
to
co
m
p
u
te
t
h
e
d
if
f
er
en
ce
b
etwe
en
th
e
p
r
o
b
ab
ilit
y
o
f
th
e
p
r
ed
icted
class
an
d
th
e
ac
tu
al
lab
els.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il 2
0
2
6
:
1
7
0
1
-
1
7
0
8
1704
ii)
Op
tim
izer
:
Ad
am
o
p
tim
izer
w
as
u
s
ed
to
tr
ain
th
e
m
o
d
el.
T
h
is
o
p
tim
izer
is
r
ec
o
g
n
ized
f
o
r
its
ef
f
icien
cy
an
d
ca
p
ac
ity
t
o
a
d
ju
s
t
lear
n
in
g
r
ates
d
u
r
in
g
t
h
e
tr
ain
i
n
g
p
r
o
ce
s
s
.
Ad
am
o
p
tim
izer
u
s
es
th
e
ad
v
a
n
tag
es
o
f
b
o
th
m
o
m
en
tu
m
an
d
a
d
ap
tiv
e
lear
n
in
g
r
ates f
o
r
f
aster
co
n
v
er
g
en
ce
.
iii)
T
r
ain
in
g
p
r
o
ce
s
s
:
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
was
tr
ain
ed
f
o
r
5
0
ep
o
ch
s
.
I
n
ea
ch
ep
o
c
h
th
e
en
tire
tr
ain
in
g
d
ataset
was f
ed
in
to
th
e
m
o
d
el
an
d
th
e
weig
h
ts
wer
e
ad
ju
s
te
d
b
ased
o
n
th
e
ca
lcu
lated
g
r
ad
ien
ts
.
Dr
o
p
o
u
t
was u
s
ed
d
u
r
in
g
th
e
tr
ain
in
g
p
r
o
ce
s
s
s
o
as
to
p
r
ev
en
t a
n
y
o
v
er
f
itti
n
g
.
T
h
e
p
er
f
o
r
m
a
n
ce
was e
v
alu
ated
o
n
th
e
v
alid
atio
n
s
et
af
ter
ea
ch
ep
o
ch
t
o
m
o
n
ito
r
d
e
v
elo
p
m
en
t
an
d
g
u
ar
a
n
tee
s
tr
o
n
g
g
e
n
er
aliza
tio
n
to
u
n
k
n
o
wn
d
ata.
T
h
e
h
y
p
er
p
ar
am
eter
s
lik
e
th
e
lear
n
in
g
an
d
d
r
o
p
o
u
t
r
ate
wer
e
ad
ju
s
ted
in
r
esp
o
n
s
e
to
v
alid
atio
n
r
esu
lts
s
o
as to
m
ax
im
ize
th
e
p
er
f
o
r
m
an
ce
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
T
h
e
r
esu
lts
o
f
th
e
p
r
o
p
o
s
ed
D
L
m
o
d
el
f
o
r
m
u
lti
-
class
class
i
f
icatio
n
o
f
h
ea
r
in
g
lo
s
s
ar
e
p
r
esen
ted
in
th
is
s
ec
tio
n
.
T
h
e
d
ataset
u
s
ed
was
th
e
UC
I
Au
d
io
lo
g
y
d
atas
et.
T
h
e
ef
f
ec
tiv
en
ess
o
f
t
h
e
m
o
d
el
was
ev
alu
ated
b
ased
o
n
its
ab
ilit
y
to
class
if
y
h
ea
r
in
g
lo
s
s
in
to
f
iv
e
ca
teg
o
r
ies:
n
o
r
m
al,
co
n
d
u
ctiv
e,
m
i
x
ed
,
co
ch
lear
,
an
d
r
etr
o
co
ch
le
ar
.
T
h
e
m
etr
ics
lik
e
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all
an
d
F1
-
s
co
r
e
wer
e
u
s
ed
to
d
e
ter
m
in
e
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
.
T
h
e
tr
ain
-
v
alid
at
io
n
s
p
lit o
f
8
0
/2
0
was u
s
ed
to
tr
ain
th
e
m
o
d
el
f
o
r
5
0
ep
o
c
h
s
.
C
r
o
s
s
en
tr
o
p
y
lo
s
s
was
u
s
ed
as
th
e
lo
s
s
f
u
n
ctio
n
in
th
e
tr
ain
in
g
p
h
ase,
an
d
th
e
Ad
am
o
p
tim
izer
was
u
s
ed
.
T
h
e
lear
n
in
g
r
ate
was
s
et
to
0
.
0
0
1
.
T
o
a
v
o
id
o
v
er
f
itti
n
g
,
d
r
o
p
o
u
t r
eg
u
lar
izatio
n
was
d
o
n
e
t
o
b
o
t
h
h
id
d
en
lay
er
s
at
a
r
ate
o
f
0
.
3
.
Fig
u
r
e
1
s
h
o
ws
t
h
e
g
r
ap
h
o
f
tr
ain
in
g
a
n
d
v
alid
atio
n
lo
s
s
(
Fig
u
r
e
1
(
a)
)
as
well
tr
ai
n
in
g
an
d
v
alid
atio
n
ac
cu
r
ac
y
(
Fig
u
r
e
1
(
b
)
)
.
T
h
e
t
r
ain
in
g
lo
s
s
d
ec
r
ea
s
ed
c
o
n
s
is
ten
tly
o
v
er
th
e
5
0
ep
o
ch
s
.
T
h
e
ac
cu
r
ac
y
o
n
th
e
tr
ain
in
g
s
et
in
c
r
ea
s
ed
s
tead
ily
.
T
h
e
v
alid
atio
n
lo
s
s
s
h
o
wed
a
s
im
ilar
tr
en
d
.
T
h
e
m
o
d
el
d
em
o
n
s
tr
ated
ef
f
e
ctiv
e
co
n
v
er
g
en
ce
,
as in
d
icate
d
b
y
th
e
s
tab
le
lo
s
s
an
d
h
ig
h
ac
cu
r
a
cy
to
war
d
s
th
e
e
n
d
o
f
tr
ain
in
g
.
(
a)
(
b
)
Fig
u
r
e
1
.
T
h
e
g
r
a
p
h
o
f
(
a)
tr
ai
n
in
g
lo
s
s
v
s
v
alid
atio
n
l
o
s
s
an
d
(
b
)
t
r
ain
in
g
ac
c
u
r
ac
y
v
s
v
ali
d
atio
n
ac
cu
r
ac
y
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
A
d
ee
p
lea
r
n
in
g
b
a
s
ed
a
p
p
r
o
a
ch
fo
r
h
ea
r
in
g
lo
s
s
d
etec
tio
n
(
Dee
p
a
)
1705
I
n
th
is
s
tu
d
y
,
t
h
e
p
r
o
p
o
s
ed
m
o
d
el
was
ev
alu
ated
ag
ai
n
s
t
s
u
p
p
o
r
t
v
ec
t
o
r
m
ac
h
in
e
(
SVM)
a
n
d
r
an
d
o
m
f
o
r
est
(
R
F).
T
h
e
d
etailed
s
tu
d
y
o
f
th
e
p
er
f
o
r
m
an
ce
o
f
th
ese
m
o
d
els
is
p
r
esen
ted
as
f
o
llo
w
.
T
h
e
f
o
cu
s
is
o
n
th
eir
ef
f
ec
tiv
en
ess
in
p
r
ed
icti
n
g
th
e
f
iv
e
h
ea
r
in
g
lo
s
s
ca
teg
o
r
ies:
n
o
r
m
al,
c
o
n
d
u
ctiv
e,
m
ix
ed
,
co
ch
lea
r
,
an
d
r
etr
o
c
o
ch
lear
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
d
em
o
n
s
tr
ated
v
er
y
g
o
o
d
o
v
er
all
p
e
r
f
o
r
m
an
ce
o
b
tain
i
n
g
a
n
ac
cu
r
ac
y
o
f
8
5
% a
cr
o
s
s
all
ca
teg
o
r
ies.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el’
s
class
if
icati
o
n
r
ep
o
r
t is su
m
m
ar
ized
i
n
T
a
b
le
1
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
p
er
f
o
r
m
ed
well
f
o
r
p
r
ed
ictin
g
co
c
h
le
ar
h
ea
r
i
n
g
l
o
s
s
with
a
0
.
9
3
p
r
ec
is
io
n
,
0
.
9
3
r
ec
all
,
an
d
an
F1
-
s
co
r
e
o
f
0
.
9
3
.
T
h
e
m
ix
ed
class
d
em
o
n
s
tr
ated
a
b
alan
ce
d
r
esu
lt
with
a
p
r
ec
is
io
n
o
f
0
.
6
0
,
r
ec
all
o
f
0
.
7
5
,
a
n
d
an
F1
-
s
co
r
e
o
f
0
.
6
7
.
Alth
o
u
g
h
th
e
m
o
d
el
d
em
o
n
s
tr
ated
s
tr
o
n
g
r
ec
all
f
o
r
th
is
class
th
er
e
is
a
s
co
p
e
f
o
r
im
p
r
o
v
in
g
p
r
ec
is
io
n
.
T
h
e
r
etr
o
co
c
h
lear
class
also
p
er
f
o
r
m
e
d
well,
with
a
p
r
ec
is
io
n
o
f
1
.
0
0
an
d
a
r
ec
all
o
f
0
.
6
7
.
T
h
is
h
ig
h
lig
h
t
s
g
o
o
d
p
r
ec
is
io
n
b
u
t
s
u
g
g
ests
th
at
th
e
m
o
d
el
co
u
ld
im
p
r
o
v
e
in
f
in
d
in
g
all
in
s
tan
ce
s
o
f
r
etr
o
co
ch
lear
h
e
ar
in
g
lo
s
s
.
Fo
r
th
e
co
n
d
u
ctiv
e
class
,
th
e
m
o
d
el
g
a
v
e
a
g
o
o
d
p
er
f
o
r
m
an
ce
with
a
p
r
ec
is
io
n
o
f
0
.
6
7
,
r
ec
all
o
f
0
.
6
7
,
an
d
a
n
F1
-
s
co
r
e
o
f
0
.
6
7
s
h
o
win
g
g
o
o
d
p
er
f
o
r
m
an
ce
.
T
h
e
n
o
r
m
al
class
,
with
o
n
ly
2
s
am
p
les
in
th
e
d
ataset,
was
d
if
f
icu
lt
to
class
if
y
.
T
h
e
m
o
d
el
was
ab
le
to
g
en
er
ate
a
p
r
ec
is
io
n
o
f
0
.
5
0
,
r
ec
all
o
f
0
.
5
0
,
an
d
F1
-
s
co
r
e
o
f
0
.
5
0
,
r
ef
lectin
g
th
e
d
if
f
ic
u
l
ty
in
ac
cu
r
ately
class
if
y
in
g
th
is
s
m
all
cla
s
s
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
p
er
f
o
r
m
e
d
w
ell
f
o
r
co
c
h
lear
a
n
d
m
i
x
ed
cla
s
s
es.
I
t
ac
h
iev
ed
a
g
o
o
d
ac
cu
r
ac
y
an
d
F1
-
s
co
r
es.
T
h
e
r
esu
lts
wer
e
m
o
r
e
b
alan
ce
d
ac
r
o
s
s
th
e
v
ar
i
o
u
s
class
es a
s
co
m
p
ar
ed
to
SVM
an
d
R
F.
T
ab
le
1
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
class
if
icatio
n
r
ep
o
r
t
C
l
a
s
s
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
N
o
r
mal
0
.
5
0
0
.
5
0
0
.
5
0
C
o
n
d
u
c
t
i
v
e
0
.
6
7
0
.
6
7
0
.
6
7
M
i
x
e
d
0
.
6
0
0
.
7
5
0
.
6
7
C
o
c
h
l
e
a
r
0
.
9
3
0
.
9
3
0
.
9
3
R
e
t
r
o
c
o
c
h
l
e
a
r
1
.
0
0
0
.
6
7
0
.
8
0
SVM
is
a
wid
ely
u
s
ed
s
u
p
er
v
is
ed
lear
n
in
g
m
eth
o
d
wh
ich
is
ef
f
ec
tiv
e
in
h
an
d
lin
g
h
ig
h
-
d
i
m
en
s
io
n
al
d
ata.
I
t
aim
s
at
f
in
d
i
n
g
an
o
p
ti
m
al
h
y
p
er
p
lan
e
th
at
m
ax
im
ize
s
th
e
s
ep
ar
atio
n
m
ar
g
in
am
o
n
g
d
if
f
er
e
n
t
class
es.
SVM
clas
s
if
icatio
n
r
ep
o
r
t
is
p
r
esen
ted
in
T
ab
le
2
.
T
h
e
SVM
g
en
er
ated
an
o
v
er
all
ac
cu
r
a
cy
o
f
8
2
%.
L
ik
e
th
e
p
r
o
p
o
s
ed
m
o
d
el,
th
e
SVM
m
o
d
el
p
er
f
o
r
m
ed
well
o
n
th
e
c
o
ch
lear
class
,
ac
h
iev
in
g
a
p
r
e
cisi
o
n
v
alu
e
o
f
0
.
8
5
an
d
r
ec
all
o
f
1
.
0
0
r
esu
ltin
g
in
an
F1
-
s
co
r
e
o
f
0
.
9
2
.
T
h
is
r
ep
r
esen
ts
th
at
SVM
was
h
ig
h
ly
ef
f
ec
tiv
e
in
class
if
y
in
g
co
ch
lear
h
ea
r
in
g
l
o
s
s
.
C
o
n
d
u
ctiv
e
class
s
h
o
wed
an
im
p
r
ess
iv
e
p
r
ec
is
io
n
o
f
1
.
0
0
b
u
t
with
a
r
ec
all
o
f
0
.
6
7
an
d
F1
-
s
co
r
e
o
f
0
.
8
0
.
T
h
e
m
ix
e
d
class
s
h
o
wed
p
o
o
r
r
esu
lt
with
an
F1
-
s
co
r
e
o
f
0
.
3
3
.
T
h
e
p
r
ec
is
io
n
o
f
0
.
5
0
an
d
r
ec
all
o
f
0
.
2
5
in
d
icat
e
th
at
th
e
SVM
m
o
d
el
s
tr
u
g
g
led
s
ig
n
if
ican
tly
with
th
is
clas
s
.
L
ik
e
th
e
p
r
o
p
o
s
ed
m
o
d
el,
th
e
R
etr
o
co
c
h
lear
cla
s
s
was
an
o
th
er
s
tr
en
g
th
o
f
th
e
SVM
m
o
d
el
wh
e
r
e
th
e
p
r
ec
i
s
io
n
s
co
r
e
was
1
.
0
,
r
ec
all
was 0
.
6
7
an
d
F1
-
s
co
r
e
was 0
.
8
0
.
T
h
e
n
o
r
m
al
class
was n
o
t c
lass
if
ied
co
r
r
ec
tly
b
y
th
e
SVM
m
o
d
el,
with
all
s
co
r
es
b
ein
g
0
.
0
0
s
h
o
win
g
th
at
th
e
SVM
s
tr
u
g
g
led
t
o
r
ec
o
g
n
ize
in
s
tan
ce
s
o
f
n
o
r
m
a
l
h
ea
r
in
g
lo
s
s
.
T
h
e
SVM
m
o
d
el
s
u
b
s
tan
tially
f
ail
ed
with
t
h
e
n
o
r
m
al
an
d
m
ix
e
d
ca
teg
o
r
ies
b
u
t
d
id
well
o
v
e
r
all
o
n
th
e
co
ch
lear
an
d
r
etr
o
co
ch
lear
class
es.
T
ab
le
2
.
T
h
e
class
if
icatio
n
r
ep
o
r
t o
f
SVM
C
l
a
s
s
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
N
o
r
mal
0
.
0
0
0
.
0
0
0
.
0
0
C
o
n
d
u
c
t
i
v
e
1
.
0
0
0
.
6
7
0
.
8
0
M
i
x
e
d
0
.
5
0
0
.
2
5
0
.
3
3
C
o
c
h
l
e
a
r
0
.
8
5
1
.
0
0
0
.
9
2
R
e
t
r
o
c
o
c
h
l
e
a
r
1
.
0
0
0
.
6
7
0
.
8
0
R
F
i
s
a
tech
n
iq
u
e
th
at
m
ak
es
u
s
e
o
f
m
u
ltip
le
d
ec
is
io
n
tr
ee
s
.
T
h
e
R
F
cla
s
s
if
ier
a
ch
iev
ed
a
n
ac
cu
r
ac
y
o
f
8
2
%
with
p
e
r
f
o
r
m
an
ce
v
ar
iatio
n
s
s
im
ilar
to
th
o
s
e
o
f
SVM.
RF
class
if
icatio
n
r
ep
o
r
t
is
s
h
o
wn
in
T
ab
le
3
.
T
h
e
co
ch
lear
class
was
well
-
p
r
ed
icted
b
y
th
e
R
F
m
o
d
el
s
co
r
in
g
a
p
r
ec
is
io
n
o
f
0
.
9
0
,
r
ec
al
l
o
f
0
.
9
6
,
a
n
d
0
.
9
3
F1
-
s
co
r
e.
T
h
is
co
n
f
ir
m
s
m
o
d
el’
s
ef
f
ec
tiv
en
ess
in
class
if
y
in
g
in
s
tan
ce
s
o
f
co
ch
lea
r
h
ea
r
in
g
lo
s
s
.
T
h
e
co
n
d
u
ctiv
e
class
ac
h
iev
ed
d
ec
en
t
r
esu
lts
with
a
p
r
ec
is
io
n
s
c
o
r
e
0
.
6
7
,
a
r
ec
all
o
f
s
co
r
e
0
.
6
7
,
an
d
F1
-
s
co
r
e
o
f
0
.
6
7
.
T
h
e
p
e
r
f
o
r
m
an
ce
was
b
alan
ce
d
b
u
t
d
id
n
o
t
h
av
e
i
d
ea
l
ac
cu
r
ac
y
.
T
h
e
m
ix
e
d
class
s
h
o
wed
a
m
o
d
er
ate
p
er
f
o
r
m
an
ce
with
an
F1
-
s
co
r
e
o
f
0
.
5
7
,
wh
e
r
e
th
e
p
r
ec
is
io
n
was
0
.
6
7
an
d
r
ec
all
was
0
.
5
0
.
T
h
is
in
d
icate
s
th
a
t
th
e
m
o
d
el
was
s
o
m
ewh
at
ef
f
e
ctiv
e
b
u
t
f
ailed
t
o
co
r
r
ec
tly
cl
ass
if
y
in
s
tan
ce
s
o
f
m
ix
ed
h
ea
r
in
g
lo
s
s
.
As
in
th
e
SVM
m
o
d
el
th
e
r
etr
o
co
ch
lea
r
p
er
f
o
r
m
ed
well
with
a
p
r
ec
is
io
n
o
f
1
.
0
0
an
d
r
ec
all
o
f
0
.
6
7
an
d
0
.
8
0
F1
-
s
co
r
e.
Similar
to
SVM
th
e
n
o
r
m
al
cl
ass
p
o
s
ed
a
ch
allen
g
e
f
o
r
th
e
R
F m
o
d
el
with
all
s
co
r
es a
t 0
.
0
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il 2
0
2
6
:
1
7
0
1
-
1
7
0
8
1706
T
ab
le
3
.
T
h
e
class
if
icatio
n
r
ep
o
r
t o
f
R
F
C
l
a
s
s
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
N
o
r
mal
0
.
0
0
0
.
0
0
0
.
0
0
C
o
n
d
u
c
t
i
v
e
0
.
6
7
0
.
6
7
0
.
6
7
M
i
x
e
d
0
.
6
7
0
.
5
0
0
.
5
7
C
o
c
h
l
e
a
r
0
.
9
0
0
.
9
6
0
.
9
3
R
e
t
r
o
c
o
c
h
l
e
a
r
1
.
0
0
0
.
6
7
0
.
8
0
All
m
o
d
els
p
er
f
o
r
m
ed
well
f
o
r
th
e
co
ch
lear
class
s
co
r
in
g
a
F1
-
s
co
r
es
th
at
ar
e
clo
s
e
to
o
r
ab
o
v
e
0
.
9
0
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
a
n
d
SVM
m
o
d
els
e
x
h
ib
ited
alm
o
s
t
p
er
f
ec
t
r
ec
all
w
h
ile
R
F
s
h
o
we
d
a
s
lig
h
tly
h
ig
h
er
p
r
ec
is
io
n
(
0
.
9
0
)
th
an
th
e
o
th
e
r
two
m
o
d
els.
T
h
e
SVM
m
o
d
el
h
ad
th
e
h
ig
h
est
p
r
ec
is
io
n
f
o
r
th
e
co
n
d
u
ctiv
e
class
(
1
.
0
0
)
b
u
t th
e
p
r
o
p
o
s
ed
m
o
d
el
an
d
R
F
h
ad
m
o
r
e
b
alan
ce
d
p
er
f
o
r
m
an
ce
with
F1
-
s
co
r
es o
f
0
.
6
7
ea
ch
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
p
er
f
o
r
m
ed
t
h
e
b
est
o
n
th
e
m
ix
ed
class
with
0
.
6
7
F1
-
s
co
r
e.
T
h
e
r
etr
o
co
ch
lear
class
was
class
if
ied
p
r
o
p
er
ly
b
y
all
th
e
m
o
d
els
s
co
r
i
n
g
F1
-
s
co
r
es
a
b
o
v
e
0
.
8
0
an
d
alm
o
s
t
p
er
f
ec
t
p
r
ec
is
io
n
f
o
r
th
e
p
r
o
p
o
s
ed
m
o
d
el
a
n
d
SVM.
T
h
e
n
o
r
m
al
class
was
n
o
t
class
if
ied
p
r
o
p
er
l
y
b
y
all
m
o
d
els
b
ec
au
s
e
o
f
its
s
m
all
s
am
p
le
s
ize
(
o
n
ly
2
in
s
tan
ce
s
)
an
d
all
m
o
d
els
s
tr
u
g
g
le
d
to
class
if
y
th
is
clas
s
co
r
r
ec
tly
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
was a
ls
o
co
m
p
ar
ed
with
ex
is
ti
n
g
m
eth
o
d
o
lo
g
ies
[
3
]
,
[
2
5
]
.
T
h
e
co
m
p
a
r
ativ
e
an
aly
s
is
is
p
r
e
s
en
ted
in
T
ab
le
4
.
T
h
e
m
o
d
el
p
r
o
p
o
s
ed
in
th
is
p
a
p
er
s
h
o
wed
b
etter
r
esu
lts
th
an
all
th
e
o
th
er
co
m
p
ar
ed
m
o
d
el
s
in
ter
m
s
o
f
o
v
er
all
a
cc
u
r
ac
y
(
8
5
%).
I
t
d
em
o
n
s
tr
ates
a
g
o
o
d
p
er
f
o
r
m
an
ce
ac
r
o
s
s
m
o
s
t
class
es
esp
ec
ially
in
p
r
ed
ictin
g
co
ch
lear
an
d
m
ix
ed
h
ea
r
in
g
lo
s
s
.
B
a
s
ed
o
n
th
e
r
esu
lts
,
th
e
p
r
o
p
o
s
ed
m
o
d
el
p
r
o
v
id
es
an
ef
f
ec
tiv
e
s
o
lu
tio
n
f
o
r
th
is
class
if
icatio
n
task
.
T
h
er
e
is
a
p
o
ten
tial
f
o
r
f
u
r
t
h
er
im
p
r
o
v
em
en
t
b
y
m
o
d
el
tu
n
i
n
g
,
d
a
ta
au
g
m
e
n
tatio
n
o
r
u
s
in
g
m
o
r
e
a
d
v
an
ce
d
ar
ch
ite
ctu
r
es.
T
h
e
r
esu
lts
d
ep
ict
th
at
th
e
p
r
o
p
o
s
ed
FNN
m
o
d
el
ca
n
ac
h
iev
e
h
ig
h
class
if
icatio
n
ac
cu
r
ac
y
f
o
r
h
ea
r
in
g
lo
s
s
p
r
e
d
ictio
n
.
Fu
t
u
r
e
im
p
r
o
v
em
e
n
ts
co
u
ld
in
v
o
lv
e
f
in
e
-
tu
n
in
g
th
e
m
o
d
el
,
ex
p
lo
r
in
g
d
ee
p
e
r
ar
ch
itectu
r
es a
n
d
in
clu
d
in
g
a
d
d
itio
n
al
f
ea
tu
r
es to
f
u
r
th
e
r
im
p
r
o
v
e
class
if
icatio
n
ac
cu
r
ac
y
.
T
ab
le
4
.
C
o
m
p
a
r
ativ
e
an
aly
s
is
with
ex
itin
g
m
eth
o
d
M
o
d
e
l
A
c
c
u
r
a
c
y
(
%)
D
L
[
3
]
81
M
L
[
2
5
]
78
P
r
o
p
o
se
d
m
o
d
e
l
85
4.
CO
NCLU
SI
O
N
T
h
e
d
iag
n
o
s
is
an
d
m
an
ag
em
e
n
t
o
f
h
ea
r
in
g
lo
s
s
h
av
e
m
ad
e
a
s
ig
n
if
ican
t
d
if
f
er
en
ce
in
r
ec
en
t
y
ea
r
s
.
E
ar
ly
d
etec
tio
n
o
f
h
ea
r
in
g
lo
s
s
an
d
in
ter
v
en
tio
n
ar
e
v
er
y
cr
i
tical
f
o
r
p
r
ev
e
n
tin
g
an
y
lo
n
g
-
t
er
m
d
is
ab
ilit
y
.
T
h
e
co
n
tin
u
o
u
s
r
esear
ch
in
th
e
f
i
eld
o
f
n
ew
d
iag
n
o
s
tic
tech
n
o
lo
g
ies
,
as
well
as
th
e
d
ev
elo
p
m
en
t
o
f
a
f
f
o
r
d
ab
le
tr
ea
tm
en
t
is
ess
en
tial
to
e
n
s
u
r
e
th
at
in
d
iv
id
u
als
ca
n
b
en
e
f
it
f
r
o
m
tim
ely
in
ter
v
e
n
tio
n
s
.
T
h
e
m
o
d
el
t
h
at
is
p
r
o
p
o
s
ed
in
t
h
is
p
ap
er
s
h
o
we
d
p
r
o
m
is
in
g
r
esu
lts
in
class
if
y
in
g
h
ea
r
in
g
lo
s
s
o
f
d
if
f
er
en
t
ty
p
es
,
ac
h
iev
in
g
an
o
v
er
all
ac
cu
r
ac
y
o
f
8
5
%.
Ho
wev
er
,
th
er
e
ar
e
s
till
ch
allen
g
es
with
m
in
o
r
ity
class
es
wh
ich
ca
n
b
e
d
ea
lt with
b
y
u
s
in
g
m
eth
o
d
s
s
u
ch
as
d
ata
a
u
g
m
en
tatio
n
o
r
ad
ju
s
tin
g
th
e
weig
h
ts
o
f
class
.
T
h
e
ex
p
er
im
en
tal
f
in
d
in
g
s
s
h
o
w
th
at
th
e
p
r
o
p
o
s
ed
DL
m
o
d
els
ar
e
ef
f
ec
tiv
e
to
o
ls
f
o
r
th
e
au
to
m
ated
class
if
icatio
n
task
s
in
h
ea
lth
ca
r
e.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
Au
th
o
r
s
s
tate
n
o
f
u
n
d
in
g
in
v
o
lv
ed
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
Dee
p
a
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Ma
n
ju
la
Gu
r
u
r
aj
R
ao
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
C
:
C
o
n
c
e
p
t
u
a
l
i
z
a
t
i
o
n
M
:
M
e
t
h
o
d
o
l
o
g
y
So
:
So
f
t
w
a
r
e
Va
:
Va
l
i
d
a
t
i
o
n
Fo
:
Fo
r
mal
a
n
a
l
y
s
i
s
I
:
I
n
v
e
s
t
i
g
a
t
i
o
n
R
:
R
e
so
u
r
c
e
s
D
:
D
a
t
a
C
u
r
a
t
i
o
n
O
:
W
r
i
t
i
n
g
-
O
r
i
g
i
n
a
l
D
r
a
f
t
E
:
W
r
i
t
i
n
g
-
R
e
v
i
e
w
&
E
d
i
t
i
n
g
Vi
:
Vi
su
a
l
i
z
a
t
i
o
n
Su
:
Su
p
e
r
v
i
s
i
o
n
P
:
P
r
o
j
e
c
t
a
d
mi
n
i
st
r
a
t
i
o
n
Fu
:
Fu
n
d
i
n
g
a
c
q
u
i
si
t
i
o
n
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
A
d
ee
p
lea
r
n
in
g
b
a
s
ed
a
p
p
r
o
a
ch
fo
r
h
ea
r
in
g
lo
s
s
d
etec
tio
n
(
Dee
p
a
)
1707
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est.
I
NF
O
RM
E
D
CO
NS
E
N
T
No
t
ap
p
licab
le.
T
h
is
s
tu
d
y
u
t
ilizes
a
p
u
b
licly
a
v
ailab
le
d
at
aset
f
r
o
m
UC
I
r
v
in
e
Ma
c
h
in
e
L
ea
r
n
in
g
R
ep
o
s
ito
r
y
,
with
n
o
id
e
n
tific
atio
n
o
f
p
er
s
o
n
al
in
f
o
r
m
a
tio
n
o
r
in
v
o
l
v
em
en
t
o
f
i
n
d
iv
id
u
als
r
eq
u
ir
i
n
g
in
f
o
r
m
e
d
co
n
s
en
t.
E
T
H
I
CAL AP
P
RO
V
AL
No
t
ap
p
licab
le.
T
h
is
s
tu
d
y
u
s
es
a
p
u
b
licly
av
ailab
le
d
ataset
f
r
o
m
UC
I
r
v
in
e
Ma
ch
in
e
L
ea
r
n
in
g
R
ep
o
s
ito
r
y
an
d
d
o
es
n
o
t
in
v
o
lv
e
h
u
m
an
o
r
a
n
im
al
s
u
b
jects;
th
u
s
,
n
o
in
s
titu
tio
n
al
r
e
v
iew
b
o
ar
d
ap
p
r
o
v
al
o
r
co
m
p
lian
ce
with
th
e
Helsin
k
i
Dec
lar
atio
n
is
r
eq
u
ir
ed
.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
a
t
a
t
h
at
s
u
p
p
o
r
t
t
h
e
f
i
n
d
i
n
g
s
o
f
t
h
is
s
t
u
d
y
w
il
l
b
e
av
a
i
l
a
b
l
e
i
n
UC
I
r
v
i
n
e
Ma
c
h
i
n
e
L
e
a
r
n
i
n
g
R
e
p
o
s
it
o
r
y
a
t
h
t
t
p
s
:
//
a
r
c
h
i
v
e
.
ics
.
u
c
i
.
e
d
u
/
m
l/
m
a
c
h
i
n
e
-
le
a
r
n
i
n
g
-
d
a
t
a
b
a
s
es
/
a
u
d
i
o
l
o
g
y
/
a
u
d
i
o
l
o
g
y
.
s
t
a
n
d
a
r
d
i
z
e
d
.
d
a
ta
.
RE
F
E
R
E
NC
E
S
[
1
]
D
.
L.
B
a
r
b
o
u
r
e
t
a
l
.
,
“
O
n
l
i
n
e
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
a
u
d
i
o
m
e
t
r
y
,
”
Ea
r
&
H
e
a
r
i
n
g
,
v
o
l
.
4
0
,
n
o
.
4
,
p
p
.
9
1
8
–
9
2
6
,
J
u
l
.
2
0
1
9
,
d
o
i
:
1
0
.
1
0
9
7
/
A
U
D
.
0
0
0
0
0
0
0
0
0
0
0
0
0
6
6
9
.
[
2
]
J.
-
W
.
W
a
sma
n
n
,
L
.
P
r
a
g
t
,
R
.
Ei
k
e
l
b
o
o
m,
a
n
d
D
.
W
.
S
w
a
n
e
p
o
e
l
,
“
D
i
g
i
t
a
l
a
p
p
r
o
a
c
h
e
s
t
o
a
u
t
o
ma
t
e
d
a
n
d
mac
h
i
n
e
l
e
a
r
n
i
n
g
a
ssessm
e
n
t
s
o
f
h
e
a
r
i
n
g
:
sc
o
p
i
n
g
r
e
v
i
e
w
,
”
J
o
u
rn
a
l
o
f
Me
d
i
c
a
l
I
n
t
e
rn
e
t
Re
s
e
a
rc
h
,
v
o
l
.
2
4
,
n
o
.
2
,
F
e
b
.
2
0
2
2
,
d
o
i
:
1
0
.
2
1
9
6
/
3
2
5
8
1
.
[
3
]
J.
Ze
n
g
e
t
a
l
.
,
“
A
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
t
o
p
r
e
d
i
c
t
c
o
n
d
u
c
t
i
v
e
h
e
a
r
i
n
g
l
o
ss
i
n
p
a
t
i
e
n
t
s
w
i
t
h
o
t
i
t
i
s
med
i
a
w
i
t
h
e
f
f
u
si
o
n
u
si
n
g
o
t
o
sco
p
i
c
i
mag
e
s,”
J
AM
A
O
t
o
l
a
ryn
g
o
l
o
g
y
–
H
e
a
d
& N
e
c
k
S
u
rg
e
ry
,
v
o
l
.
1
4
8
,
n
o
.
7
,
Ju
l
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
1
/
j
a
ma
o
t
o
.
2
0
2
2
.
0
9
0
0
.
[
4
]
H
.
B
y
u
n
,
C
.
J.
P
a
r
k
,
S
.
J.
O
h
,
M
.
J
.
C
h
u
n
g
,
B
.
H
.
C
h
o
,
a
n
d
Y
.
-
S
.
C
h
o
,
“
A
u
t
o
ma
t
i
c
p
r
e
d
i
c
t
i
o
n
o
f
c
o
n
d
u
c
t
i
v
e
h
e
a
r
i
n
g
l
o
ss
u
si
n
g
v
i
d
e
o
p
n
e
u
m
a
t
i
c
o
t
o
s
c
o
p
y
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
m,
”
E
a
r
&
H
e
a
ri
n
g
,
v
o
l
.
4
3
,
n
o
.
5
,
p
p
.
1
5
6
3
–
1
5
7
3
,
S
e
p
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
9
7
/
A
U
D
.
0
0
0
0
0
0
0
0
0
0
0
0
1
2
1
7
.
[
5
]
J.
S
c
h
l
i
t
t
e
n
l
a
c
h
e
r
,
R
.
E.
Tu
r
n
e
r
,
a
n
d
B
.
C
.
J.
M
o
o
r
e
,
“
A
u
d
i
o
g
r
a
m
e
st
i
mat
i
o
n
u
si
n
g
B
a
y
e
si
a
n
a
c
t
i
v
e
l
e
a
r
n
i
n
g
,
”
T
h
e
J
o
u
r
n
a
l
o
f
t
h
e
Ac
o
u
s
t
i
c
a
l
S
o
c
i
e
t
y
o
f
Am
e
ri
c
a
,
v
o
l
.
1
4
4
,
n
o
.
1
,
p
p
.
4
2
1
–
4
3
0
,
J
u
l
.
2
0
1
8
,
d
o
i
:
1
0
.
1
1
2
1
/
1
.
5
0
4
7
4
3
6
.
[
6
]
T.
J.
G
a
t
h
m
a
n
,
J
.
S
.
C
h
o
i
,
R
.
M
.
S
.
V
a
sd
e
v
,
J.
A
.
S
c
h
o
e
p
h
o
e
r
st
e
r
,
a
n
d
M
.
E.
A
d
a
ms,
“
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
p
r
e
d
i
c
t
i
o
n
o
f
o
b
j
e
c
t
i
v
e
h
e
a
r
i
n
g
l
o
s
s
w
i
t
h
d
e
mo
g
r
a
p
h
i
c
s
,
c
l
i
n
i
c
a
l
f
a
c
t
o
r
s
,
a
n
d
su
b
j
e
c
t
i
v
e
h
e
a
r
i
n
g
st
a
t
u
s,
”
O
t
o
l
a
r
y
n
g
o
l
o
g
y
–
H
e
a
d
a
n
d
N
e
c
k
S
u
r
g
e
r
y
,
v
o
l
.
1
6
9
,
n
o
.
3
,
p
p
.
5
0
4
–
5
1
3
,
S
e
p
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
2
/
o
h
n
.
2
8
8
.
[
7
]
J.
F
.
A
l
S
a
m
h
o
r
i
e
t
a
l
.
,
“
A
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
f
o
r
h
e
a
r
i
n
g
l
o
ss
p
r
e
v
e
n
t
i
o
n
,
d
i
a
g
n
o
s
i
s,
a
n
d
ma
n
a
g
e
m
e
n
t
,
”
J
o
u
rn
a
l
o
f
M
e
d
i
c
i
n
e
,
S
u
r
g
e
ry,
a
n
d
P
u
b
l
i
c
H
e
a
l
t
h
,
v
o
l
.
3
,
A
u
g
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
g
l
me
d
i
.
2
0
2
4
.
1
0
0
1
3
3
.
[
8
]
J.
S
.
S
h
i
n
e
t
a
l
.
,
“
D
e
v
e
l
o
p
me
n
t
a
n
d
c
o
mp
a
r
i
s
o
n
o
f
ma
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
i
n
g
m
o
d
e
l
s
f
o
r
sp
e
e
c
h
a
u
d
i
o
met
r
y
p
r
e
d
i
c
t
i
o
n
,
”
Ap
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
1
5
,
n
o
.
6
,
M
a
r
.
2
0
2
5
,
d
o
i
:
1
0
.
3
3
9
0
/
a
p
p
1
5
0
6
3
0
7
1
.
[
9
]
J.
W
e
n
g
e
t
a
l
.
,
“
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
-
b
a
s
e
d
p
r
e
d
i
c
t
i
o
n
o
f
t
h
e
o
u
t
c
o
mes
o
f
c
o
c
h
l
e
a
r
i
mp
l
a
n
t
a
t
i
o
n
i
n
p
a
t
i
e
n
t
s
w
i
t
h
i
n
n
e
r
e
a
r
mal
f
o
r
m
a
t
i
o
n
,
”
Eu
r
o
p
e
a
n
Arc
h
i
v
e
s
o
f
O
t
o
-
R
h
i
n
o
-
L
a
ry
n
g
o
l
o
g
y
,
v
o
l
.
2
8
1
,
n
o
.
7
,
p
p
.
3
5
3
5
–
3
5
4
5
,
Ju
l
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s
0
0
4
0
5
-
0
2
4
-
0
8
4
6
3
-
w.
[
1
0
]
Z.
D
o
u
e
t
a
l
.
,
“
P
u
r
e
t
o
n
e
a
u
d
i
o
g
r
a
m
c
l
a
ss
i
f
i
c
a
t
i
o
n
u
s
i
n
g
d
e
e
p
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
,
”
C
l
i
n
i
c
a
l
O
t
o
l
a
ry
n
g
o
l
o
g
y
,
v
o
l
.
4
9
,
n
o
.
5
,
p
p
.
5
9
5
–
6
0
3
,
S
e
p
.
2
0
2
4
,
d
o
i
:
1
0
.
1
1
1
1
/
c
o
a
.
1
4
1
7
0
.
[
1
1
]
S
.
T.
W
i
n
k
l
e
r
,
B
.
S
a
c
k
ma
n
n
,
B
.
P
r
i
w
i
t
z
e
r
,
a
n
d
M
.
L
a
u
x
m
a
n
n
,
“
E
x
p
l
o
r
i
n
g
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
d
i
a
g
n
o
s
t
i
c
d
e
c
i
si
o
n
s
b
a
se
d
o
n
w
i
d
e
b
a
n
d
i
mm
i
t
t
a
n
c
e
m
e
a
s
u
r
e
m
e
n
t
s
f
o
r
o
t
o
scl
e
r
o
s
i
s
a
n
d
d
i
s
a
r
t
i
c
u
l
a
t
i
o
n
,
”
J
o
u
r
n
a
l
o
f
M
e
d
i
c
a
l
Art
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
v
o
l
.
7
,
p
p
.
9
–
9
,
Ju
n
.
2
0
2
4
,
d
o
i
:
1
0
.
2
1
0
3
7
/
j
ma
i
-
24
-
1
0
.
[
1
2
]
E.
S
o
y
l
e
mez
e
t
a
l
.
,
“
P
r
e
d
i
c
t
i
n
g
n
o
i
se
-
i
n
d
u
c
e
d
h
e
a
r
i
n
g
l
o
ss
w
i
t
h
mac
h
i
n
e
l
e
a
r
n
i
n
g
:
t
h
e
i
n
f
l
u
e
n
c
e
o
f
t
i
n
n
i
t
u
s
a
s
a
p
r
e
d
i
c
t
i
v
e
f
a
c
t
o
r
,
”
T
h
e
J
o
u
r
n
a
l
o
f
L
a
r
y
n
g
o
l
o
g
y
&
O
t
o
l
o
g
y
,
v
o
l
.
1
3
8
,
n
o
.
1
0
,
p
p
.
1
0
3
0
–
1
0
3
5
,
O
c
t
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
7
/
S
0
0
2
2
2
1
5
1
2
4
0
0
0
9
4
X
.
[
1
3
]
M
.
K
a
s
sj
a
ń
s
k
i
e
t
a
l
.
,
“
E
f
f
i
c
i
e
n
c
y
o
f
a
r
t
i
f
i
c
i
a
l
i
n
t
e
l
l
i
g
e
n
c
e
me
t
h
o
d
s
f
o
r
h
e
a
r
i
n
g
l
o
ss
t
y
p
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
:
a
n
e
v
a
l
u
a
t
i
o
n
,
”
J
o
u
rn
a
l
o
f
Au
t
o
m
a
t
i
o
n
,
Mo
b
i
l
e
R
o
b
o
t
i
c
s
a
n
d
I
n
t
e
l
l
i
g
e
n
t
S
y
s
t
e
m
s
,
v
o
l
.
1
8
,
n
o
.
3
,
p
p
.
2
8
–
3
8
,
S
e
p
.
2
0
2
4
,
d
o
i
:
1
0
.
1
4
3
1
3
/
J
A
M
R
I
S
/
3
-
2
0
2
4
/
1
9
.
[
1
4
]
M
.
G
.
C
r
o
w
s
o
n
e
t
a
l
.
,
“
A
u
t
o
A
u
d
i
o
:
d
e
e
p
l
e
a
r
n
i
n
g
f
o
r
a
u
t
o
ma
t
i
c
a
u
d
i
o
g
r
a
m
i
n
t
e
r
p
r
e
t
a
t
i
o
n
,
”
J
o
u
rn
a
l
o
f
Me
d
i
c
a
l
S
y
st
e
m
s
,
v
o
l
.
4
4
,
n
o
.
9
,
S
e
p
.
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
0
9
1
6
-
0
2
0
-
0
1
6
2
7
-
1.
[
1
5
]
S
.
C
a
l
a
b
r
e
se
,
E.
D
o
n
a
t
i
,
a
n
d
C
.
C
h
o
u
s
i
d
i
s,
“
P
r
e
d
i
c
t
i
o
n
o
f
h
e
a
r
i
n
g
l
o
ss
t
h
r
o
u
g
h
a
p
p
l
i
c
a
t
i
o
n
o
f
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
AES
C
o
n
v
e
n
t
i
o
n
P
a
p
e
rs
Fo
r
u
m
,
v
o
l
.
1
4
8
,
2
0
2
0
.
[
1
6
]
D
.
H
e
p
si
b
a
a
n
d
J.
Ju
s
t
i
n
,
“
R
o
l
e
o
f
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
i
n
s
p
e
e
c
h
e
n
h
a
n
c
e
men
t
:
a
r
e
v
i
e
w
,
”
i
n
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
(
S
L
AAI
-
I
C
AI
2
0
1
8
)
,
2
0
1
9
,
p
p
.
1
0
3
–
1
1
2
.
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
981
-
13
-
9
1
2
9
-
3
_
8
.
[
1
7
]
J.
S
.
S
h
i
n
,
J.
M
a
,
S
.
J
.
C
h
o
i
,
S
.
K
i
m
,
a
n
d
M
.
H
o
n
g
,
“
D
e
v
e
l
o
p
m
e
n
t
o
f
a
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
f
o
r
p
r
e
d
i
c
t
i
n
g
s
p
e
e
c
h
a
u
d
i
o
m
e
t
r
y
u
si
n
g
p
u
r
e
-
t
o
n
e
a
u
d
i
o
me
t
r
y
d
a
t
a
,
”
Ap
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
1
4
,
n
o
.
2
0
,
O
c
t
.
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
a
p
p
1
4
2
0
9
3
7
9
.
[
1
8
]
M
.
K
a
ssj
a
ń
s
k
i
e
t
a
l
.
,
“
A
u
t
o
ma
t
e
d
h
e
a
r
i
n
g
l
o
ss t
y
p
e
c
l
a
ssi
f
i
c
a
t
i
o
n
b
a
s
e
d
o
n
p
u
r
e
t
o
n
e
a
u
d
i
o
m
e
t
r
y
d
a
t
a
,
”
S
c
i
e
n
t
i
f
i
c
Re
p
o
r
t
s
,
v
o
l
.
1
4
,
n
o
.
1
,
J
u
n
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
3
8
/
s
4
1
5
9
8
-
0
2
4
-
6
4
3
1
0
-
2.
[
1
9
]
H.
-
C
.
C
h
u
,
Y
.
-
L.
Z
h
a
n
g
,
a
n
d
H
.
-
C
.
C
h
i
a
n
g
,
“
A
C
N
N
s
o
u
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
mec
h
a
n
i
sm
u
s
i
n
g
d
a
t
a
a
u
g
me
n
t
a
t
i
o
n
,
”
S
e
n
so
rs
,
v
o
l
.
2
3
,
n
o
.
1
5
,
A
u
g
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
s2
3
1
5
6
9
7
2
.
[
2
0
]
S.
-
C
.
Li
n
e
t
a
l
.
,
“
A
r
t
i
f
i
c
i
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
-
a
ss
i
st
e
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
o
f
h
e
a
r
i
n
g
p
r
o
g
n
o
s
i
s
o
f
s
u
d
d
e
n
se
n
s
o
r
i
n
e
u
r
a
l
h
e
a
r
i
n
g
l
o
ss
w
i
t
h
v
e
r
t
i
g
o
,
”
I
EEE
J
o
u
rn
a
l
o
f
T
r
a
n
sl
a
t
i
o
n
a
l
En
g
i
n
e
e
ri
n
g
i
n
H
e
a
l
t
h
a
n
d
Me
d
i
c
i
n
e
,
v
o
l
.
1
1
,
p
p
.
1
7
0
–
1
8
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
J
TEH
M
.
2
0
2
3
.
3
2
4
2
3
3
9
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
15
,
No
.
2
,
Ap
r
il 2
0
2
6
:
1
7
0
1
-
1
7
0
8
1708
[
2
1
]
T.
K
h
a
j
o
n
k
l
i
n
e
t
a
l
.
,
“
U
t
i
l
i
z
i
n
g
a
r
t
i
f
i
c
i
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s
f
o
r
e
s
t
a
b
l
i
sh
i
n
g
h
e
a
r
i
n
g
-
l
o
ss
p
r
e
d
i
c
t
i
n
g
m
o
d
e
l
s
b
a
se
d
o
n
a
l
o
n
g
i
t
u
d
i
n
a
l
d
a
t
a
se
t
a
n
d
t
h
e
i
r
i
m
p
l
i
c
a
t
i
o
n
s
f
o
r
m
a
n
a
g
i
n
g
t
h
e
h
e
a
r
i
n
g
c
o
n
s
e
r
v
a
t
i
o
n
p
r
o
g
r
a
m,”
S
a
f
e
t
y
a
n
d
H
e
a
l
t
h
a
t
W
o
rk
,
v
o
l
.
1
5
,
n
o
.
2
,
p
p
.
2
2
0
–
2
2
7
,
J
u
n
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
s
h
a
w
.
2
0
2
4
.
0
2
.
0
0
4
.
[
2
2
]
E.
M
.
C
á
r
d
e
n
a
s
,
J.
P
.
d
e
l
a
R
.
G
u
t
i
é
r
r
e
z
,
L
.
M
.
A
.
Lo
b
o
,
a
n
d
G
.
O
.
R
u
i
z
,
“
A
u
t
o
ma
t
i
c
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ssi
f
i
c
a
t
i
o
n
o
f
h
e
a
r
i
n
g
l
o
ss
c
o
n
d
i
t
i
o
n
s
u
s
i
n
g
a
n
a
r
t
i
f
i
c
i
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
a
p
p
r
o
a
c
h
,
”
i
n
P
a
t
t
e
r
n
Re
c
o
g
n
i
t
i
o
n
(
M
C
PR
2
0
1
9
)
,
2
0
1
9
,
p
p
.
2
2
7
–
237
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
030
-
2
1
0
7
7
-
9
_
2
1
.
[
2
3
]
B
.
W
i
l
so
n
,
“
D
i
g
i
t
a
l
si
g
n
a
l
p
r
o
c
e
ssi
n
g
a
p
p
l
i
c
a
t
i
o
n
s
f
o
r
h
e
a
r
i
n
g
a
c
c
e
ss
i
b
i
l
i
t
y
,
”
I
EEE
S
i
g
n
a
l
Pr
o
c
e
ss
i
n
g
Ma
g
a
zi
n
e
,
v
o
l
.
2
0
,
n
o
.
5
,
p
p
.
1
4
–
1
8
,
S
e
p
.
2
0
0
3
,
d
o
i
:
1
0
.
1
1
0
9
/
M
S
P
.
2
0
0
3
.
1
2
3
6
7
6
9
.
[
2
4
]
M
.
A
.
B
o
l
a
r
i
n
w
a
a
n
d
M
.
E
.
A
y
o
m
a
,
“
D
e
v
e
l
o
p
m
e
n
t
o
f
a
s
i
g
n
a
l
p
r
o
c
e
ssi
n
g
a
l
g
o
r
i
t
h
m
f
o
r
f
e
e
d
b
a
c
k
(
n
o
i
s
e
)
r
e
d
u
c
t
i
o
n
i
n
h
e
a
r
i
n
g
a
i
d
s
u
si
n
g
t
h
e
s
p
e
c
t
r
a
l
s
u
b
t
r
a
c
t
i
o
n
t
e
c
h
n
i
q
u
e
,
”
Eu
ro
p
e
a
n
J
o
u
r
n
a
l
o
f
El
e
c
t
ri
c
a
l
E
n
g
i
n
e
e
ri
n
g
a
n
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
8
,
n
o
.
3
,
p
p
.
2
1
–
2
8
,
J
u
n
.
2
0
2
4
,
d
o
i
:
1
0
.
2
4
0
1
8
/
e
j
e
c
e
.
2
0
2
4
.
8
.
3
.
6
2
4
.
[
2
5
]
S
.
Li
e
t
a
l
.
,
“
O
p
t
i
m
i
z
i
n
g
a
c
l
a
ssi
f
i
c
a
t
i
o
n
m
o
d
e
l
t
o
e
v
a
l
u
a
t
e
i
n
d
i
v
i
d
u
a
l
su
s
c
e
p
t
i
b
i
l
i
t
y
i
n
n
o
i
se
-
i
n
d
u
c
e
d
h
e
a
r
i
n
g
l
o
ss
:
c
r
o
ss
-
se
c
t
i
o
n
a
l
st
u
d
y
,
”
J
MI
R
P
u
b
l
i
c
H
e
a
l
t
h
a
n
d
S
u
rv
e
i
l
l
a
n
c
e
,
v
o
l
.
1
0
,
p
p
.
e
6
0
3
7
3
–
e
6
0
3
7
3
,
N
o
v
.
2
0
2
4
,
d
o
i
:
1
0
.
2
1
9
6
/
6
0
3
7
3
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
De
e
p
a
re
c
e
iv
e
d
th
e
P
h
.
D
.
in
Co
m
p
u
ter
S
c
ien
c
e
En
g
in
e
e
rin
g
fr
o
m
Visv
e
sv
a
ra
y
a
Tec
h
n
o
l
o
g
ica
l
Un
i
v
e
rsity
,
I
n
d
ia
.
S
h
e
is
c
u
rre
n
tl
y
w
o
rk
i
n
g
a
s
a
n
a
ss
istan
t
p
ro
fe
ss
o
r
a
t
In
fo
rm
a
ti
o
n
S
c
ien
c
e
an
d
En
g
in
e
e
rin
g
a
t
NMAM
I
n
stit
u
te
o
f
Te
c
h
n
o
l
o
g
y
,
Ni
tt
e
,
I
n
d
ia.
He
r
re
se
a
rc
h
in
tere
sts
a
re
in
th
e
field
s
o
f
c
o
m
p
u
ter
v
isio
n
a
n
d
d
i
g
it
a
l
ima
g
e
p
ro
c
e
ss
in
g
.
S
h
e
h
a
s
p
u
b
li
sh
e
d
se
v
e
ra
l
p
a
p
e
rs
i
n
i
n
ter
n
a
ti
o
n
a
l
jo
u
rn
a
ls
a
n
d
c
o
n
fe
re
n
c
e
s.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
d
e
e
p
a
sh
e
tt
y
1
7
@n
i
tt
e
.
e
d
u
.
i
n
.
Ma
n
ju
l
a
G
u
r
u
r
a
j
Ra
o
re
c
e
iv
e
d
t
h
e
d
e
g
re
e
o
f
P
h
.
D
.
in
Co
m
p
u
ter
S
c
ien
c
e
fro
m
Ja
in
Un
i
v
e
rsity
,
Ba
n
g
a
l
o
re
,
In
d
i
a
.
S
h
e
is
a
p
r
o
fe
ss
o
r
sc
ien
c
e
a
n
d
e
n
g
i
n
e
e
rin
g
a
t
NMA
M
In
stit
u
te
o
f
Tec
h
n
o
lo
g
y
,
Nitt
e
,
I
n
d
ia
.
He
r
re
se
a
r
c
h
in
tere
sts
a
re
in
ima
g
e
p
ro
c
e
ss
in
g
,
c
o
m
p
u
ter
v
isio
n
,
DL
,
a
n
d
NLP
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
d
r
.
m
a
n
ju
lag
u
r
u
ra
jh
@n
i
tt
e
.
e
d
u
.
in
.
Evaluation Warning : The document was created with Spire.PDF for Python.