I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
pu
t
er
E
ng
ineering
(
I
J
E
CE
)
Vo
l.
1
6
,
No
.
2
,
A
p
r
il
20
2
6
,
p
p
.
895
~
904
I
SS
N:
2088
-
8
7
0
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijece.
v
1
6
i
2
.
pp
8
9
5
-
9
0
4
895
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
An int
erpre
table
deep learning
fra
mewo
rk f
o
r
early
det
ec
tion o
f
depress
io
n using
hy
brid archit
ectu
res
Cha
it
hra
I
nd
a
v
a
ra
Venk
a
t
es
ha
g
o
wda
1
,
Ro
o
pa
s
hree
H
ej
j
a
j
j
i R
a
ng
a
na
t
ha
s
ha
rm
a
2
,
Yo
g
ee
s
h Am
ba
la
g
er
e
Cha
nd
ra
s
hek
a
ra
ia
h
3
1
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
G
S
S
S
I
n
st
i
t
u
t
e
o
f
En
g
i
n
e
e
r
i
n
g
a
n
d
Te
c
h
n
o
l
o
g
y
f
o
r
W
o
me
n
,
M
y
s
o
r
e
,
I
n
d
i
a
2
D
e
p
a
r
t
me
n
t
o
f
A
r
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
a
n
d
D
a
t
a
S
c
i
e
n
c
e
,
G
S
S
S
I
n
st
i
t
u
t
e
o
f
E
n
g
i
n
e
e
r
i
n
g
a
n
d
Te
c
h
n
o
l
o
g
y
f
o
r
W
o
me
n
,
M
y
so
r
e
,
V
i
sv
e
sv
a
r
a
y
a
T
e
c
h
n
o
l
o
g
i
c
a
l
U
n
i
v
e
r
s
i
t
y
,
B
e
l
a
g
a
v
i
,
I
n
d
i
a
3
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
G
o
v
e
r
n
me
n
t
E
n
g
i
n
e
e
r
i
n
g
C
o
l
l
e
g
e
,
C
h
a
mara
j
a
n
a
g
a
r
,
V
i
sv
e
sv
a
r
a
y
a
T
e
c
h
n
o
l
o
g
i
c
a
l
U
n
i
v
e
r
s
i
t
y
,
B
e
l
a
g
a
v
i
,
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
Ma
r
2
4
,
2
0
2
5
R
ev
is
ed
Dec
2
2
,
2
0
2
5
Acc
ep
ted
J
an
1
6
,
2
0
2
6
Cu
rre
n
t
tec
h
n
iq
u
e
s
fo
r
d
e
tec
ti
n
g
d
e
p
re
ss
io
n
a
re
lab
o
r
-
i
n
ten
siv
e
a
n
d
su
b
jec
ti
v
e
,
d
e
p
e
n
d
i
n
g
o
n
c
li
n
ic
a
l
in
terv
iew
s
o
r
se
lf
-
re
p
o
rts.
T
h
e
re
is
a
g
ro
wi
n
g
a
d
o
p
t
io
n
o
f
m
a
c
h
in
e
lea
r
n
in
g
(M
L)
a
n
d
n
a
t
u
ra
l
lan
g
u
a
g
e
p
ro
c
e
ss
in
g
(NLP
)
t
o
a
u
to
m
a
ti
c
a
ll
y
id
e
n
ti
fy
d
e
p
re
ss
io
n
in
tex
t
u
a
l
d
a
ta.
T
h
e
lac
k
o
f
in
terp
re
tab
il
it
y
,
wh
ich
is
e
ss
e
n
ti
a
l
fo
r
h
e
a
lt
h
c
a
re
a
p
p
li
c
a
ti
o
n
s
,
is
stil
l
a
m
a
jo
r
o
b
sta
c
le,
th
o
u
g
h
.
B
y
c
o
m
b
i
n
in
g
c
o
n
v
o
lu
t
io
n
n
e
u
ra
l
n
e
tw
o
rk
(
CNN
)
fo
r
fe
a
tu
re
e
x
trac
ti
o
n
,
b
i
d
irec
ti
o
n
a
l
lo
n
g
s
h
o
rt
-
term
m
e
m
o
ry
(B
iL
S
TM
)
f
o
r
c
a
p
t
u
rin
g
se
q
u
e
n
ti
a
l
d
e
p
e
n
d
e
n
c
ies
,
a
n
d
tra
n
sfo
rm
e
r
-
b
a
se
d
p
re
-
train
e
d
lan
g
u
a
g
e
m
o
d
e
l
(P
TL
M
)
f
o
r
c
o
n
tex
tu
a
l
u
n
d
e
rsta
n
d
i
n
g
,
t
h
is
st
u
d
y
o
ffe
rs
a
n
in
terp
re
tab
le
fra
m
e
wo
rk
fo
r
e
a
rly
d
e
p
re
ss
io
n
id
e
n
ti
fica
ti
o
n
.
Ad
d
it
i
o
n
a
ll
y
,
th
e
sy
ste
m
u
se
s
a
n
o
v
e
l
in
terp
re
t
a
b
il
it
y
m
e
th
o
d
t
o
g
u
a
ra
n
tee
tran
sp
a
re
n
t
d
e
c
isio
n
-
m
a
k
i
n
g
.
T
h
e
o
u
tc
o
m
e
o
f
th
e
p
r
o
p
o
se
d
sy
ste
m
is
fo
u
n
d
t
o
a
c
h
ie
v
e
9
6
.
2
%
a
c
c
u
ra
c
y
,
9
4
.
5
%
p
re
c
isio
n
,
9
5
.
1
%
re
c
a
ll
,
a
n
d
9
4
.
8
%
F
1
-
sc
o
re
,
wh
ich
is
a
sig
n
ifi
c
a
n
t
im
p
ro
v
e
m
e
n
t
o
v
e
r
c
u
rre
n
t
m
e
th
o
d
.
T
h
is
fra
m
e
wo
rk
a
c
ts
a
s
a
u
se
fu
l
t
o
o
l
fo
r
e
a
rly
m
e
n
tal
h
e
a
lt
h
in
terv
e
n
ti
o
n
.
K
ey
w
o
r
d
s
:
C
o
n
tex
tu
al
u
n
d
e
r
s
tan
d
in
g
Dec
is
io
n
m
ak
in
g
Dee
p
lear
n
in
g
Dep
r
ess
io
n
Ma
ch
in
e
lear
n
in
g
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
:
C
h
aith
r
a
I
n
d
av
a
r
a
Ven
k
atesh
a
g
o
wd
a
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
an
d
E
n
g
in
ee
r
in
g
,
GSSS
I
n
s
titu
te
o
f
E
n
g
in
ee
r
i
n
g
an
d
T
ec
h
n
o
lo
g
y
f
o
r
W
o
m
en
,
Vis
v
esv
ar
ay
a
T
ec
h
n
o
lo
g
ical
Un
iv
er
s
ity
KR
S
R
d
,
Me
tag
alli,
My
s
u
r
u
,
Kar
n
atak
a
–
5
7
0
0
1
6
,
I
n
d
ia
E
m
ail:
ch
aith
r
a.
iv
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Dep
r
ess
io
n
is
a
c
o
m
m
o
n
b
u
t
d
an
g
er
o
u
s
m
e
n
tal
h
ea
lth
illn
e
s
s
ch
ar
ac
ter
ized
b
y
p
er
s
is
ten
t
f
ee
lin
g
s
o
f
s
ad
n
ess
,
h
o
p
eless
n
ess
,
an
d
a
l
o
s
s
o
f
in
ter
est
o
r
p
leasu
r
e
i
n
f
o
r
m
er
ly
en
jo
y
a
b
le
ac
tiv
ities
[
1
]
.
I
t
ca
n
in
f
lu
en
ce
h
o
w
a
p
er
s
o
n
th
i
n
k
s
,
f
ee
ls
,
an
d
b
e
h
av
es,
f
r
eq
u
en
tly
r
esu
ltin
g
in
p
h
y
s
ical
an
d
em
o
tio
n
al
i
s
s
u
es
th
at
in
ter
f
er
e
with
d
aily
f
u
n
ctio
n
in
g
[
2
]
.
Sin
ce
p
r
o
m
p
t
id
en
tific
atio
n
ca
n
g
r
ea
tly
im
p
r
o
v
e
o
u
tco
m
es
an
d
s
to
p
th
e
d
is
o
r
d
er
f
r
o
m
g
ettin
g
wo
r
s
e,
ac
cu
r
ate
an
d
r
eliab
le
d
ep
r
ess
io
n
d
etec
tio
n
is
ess
en
tial
f
o
r
ea
r
ly
m
an
ag
em
en
t.
Ho
wev
er
,
s
u
b
jectiv
ity
,
b
iases
,
an
d
th
e
ti
m
e
-
co
n
s
u
m
in
g
n
at
u
r
e
o
f
e
v
alu
atio
n
s
ar
e
s
o
m
e
o
f
th
e
p
r
o
b
l
em
s
th
at
p
lag
u
e
th
e
ex
is
tin
g
ap
p
r
o
ac
h
es
f
o
r
d
iag
n
o
s
in
g
d
ep
r
ess
io
n
,
s
u
ch
as
s
elf
-
r
ep
o
r
ts
an
d
p
r
o
f
ess
io
n
al
i
n
ter
v
iews.
Du
e
to
s
tig
m
a,
m
an
y
p
eo
p
le
m
ig
h
t
n
o
t
s
ee
k
ca
r
e
o
r
ad
eq
u
ately
c
o
m
m
u
n
icate
th
eir
p
r
o
b
lem
s
,
wh
ich
co
u
ld
r
esu
lt
in
u
n
d
er
d
iag
n
o
s
is
[
3
]
,
[
4
]
.
Fu
r
t
h
er
m
o
r
e
,
d
iag
n
o
s
in
g
d
e
p
r
ess
io
n
is
ch
allen
g
in
g
d
u
e
to
its
co
m
p
lex
ity
,
w
h
ich
in
clu
d
es
a
wid
e
s
p
ec
tr
u
m
o
f
s
y
m
p
to
m
s
an
d
i
n
d
iv
id
u
al
v
ar
ia
tio
n
s
in
p
r
esen
tatio
n
.
B
y
au
to
m
atin
g
th
e
d
etec
tio
n
p
r
o
ce
s
s
th
r
o
u
g
h
th
e
u
s
e
o
f
n
atu
r
al
lan
g
u
a
g
e
p
r
o
ce
s
s
in
g
(
NL
P)
an
d
m
ac
h
in
e
lear
n
in
g
(
ML
)
tech
n
iq
u
es,
ar
tific
ial
in
tellig
en
ce
(
AI
)
ca
n
s
ig
n
if
ican
tly
co
n
tr
ib
u
te
to
th
e
r
eso
lu
tio
n
o
f
th
ese
is
s
u
es.
AI
ca
n
p
r
o
v
id
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
2
,
Ap
r
il
20
2
6
:
8
9
5
-
904
896
q
u
ick
er
,
m
o
r
e
u
n
b
iased
,
an
d
s
ca
lab
le
s
o
lu
tio
n
s
b
y
an
aly
zin
g
v
ast
v
o
lu
m
es
o
f
d
ata
f
r
o
m
d
ig
ital
p
latf
o
r
m
s
,
in
clu
d
in
g
wr
itten
tex
t
an
d
s
o
c
ial
m
ed
ia
p
o
s
ts
,
to
f
i
n
d
p
atter
n
s
s
u
g
g
esti
v
e
o
f
s
ad
n
ess
[
5
]
,
[
6
]
.
Fu
r
th
er
m
o
r
e,
AI
m
o
d
els
ca
n
b
e
m
ad
e
t
r
an
s
p
ar
en
t
an
d
in
ter
p
r
etab
le,
m
ak
i
n
g
th
em
r
eliab
le
an
d
u
s
ef
u
l
in
s
tr
u
m
en
ts
in
m
ed
ical
s
itu
atio
n
s
[
7
]
,
[
8
]
.
T
h
e
r
elate
d
wo
r
k
ca
r
r
ied
o
u
t
in
th
is
d
ir
ec
tio
n
is
d
is
cu
s
s
ed
as
f
o
llo
ws:
Sev
er
al
ac
ad
e
m
ics
h
av
e
wo
r
k
ed
o
v
er
th
e
y
ea
r
s
to
im
p
r
o
v
e
th
e
d
etec
tio
n
an
d
d
iag
n
o
s
is
o
f
d
ep
r
ess
io
n
u
s
in
g
m
ac
h
in
e
lear
n
in
g
(
ML
)
an
d
d
ee
p
lear
n
in
g
(
DL
)
a
p
p
r
o
ac
h
es
[
9
]
–
[
1
2
]
.
Ullah
et
a
l
.
[
1
3
]
in
v
esti
g
ated
th
e
u
s
e
o
f
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
es
(
SVM)
an
d
r
an
d
o
m
f
o
r
ests
f
o
r
class
if
y
in
g
d
ep
r
ess
io
n
b
ased
o
n
lin
g
u
is
tic
v
ar
iab
les
d
er
iv
ed
f
r
o
m
tex
t,
an
d
f
o
u
n
d
en
c
o
u
r
a
g
in
g
ac
cu
r
ac
y
r
esu
lts
.
Similar
ly
,
v
ar
io
u
s
ex
i
s
tin
g
s
tu
d
ies
s
h
o
wed
t
h
at
d
ee
p
n
e
u
r
al
n
etwo
r
k
s
,
n
am
ely
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NN)
,
m
ay
p
r
ed
ict
d
ep
r
ess
io
n
f
r
o
m
s
o
cial
m
ed
ia
p
o
s
ts
,
s
h
o
win
g
th
eir
ab
ilit
y
to
ca
p
tu
r
e
co
m
p
le
x
p
at
ter
n
s
in
b
ig
tex
t
d
atasets
[
1
4
]
–
[
1
7
]
.
B
id
ir
ec
t
io
n
al
en
co
d
er
r
ep
r
es
en
tat
io
n
s
f
r
o
m
tr
a
n
s
f
o
r
m
e
r
s
(
B
E
R
T
)
h
a
v
e
r
e
c
eiv
e
d
at
te
n
ti
o
n
f
o
r
its
co
n
t
e
x
t
u
aliz
e
d
c
o
m
p
r
e
h
e
n
s
i
o
n
o
f
l
a
n
g
u
ag
e,
d
e
m
o
n
s
tr
ati
n
g
its
s
u
p
e
r
i
o
r
it
y
in
a
v
a
r
ie
ty
o
f
NL
P
t
ask
s
,
i
n
cl
u
d
i
n
g
s
en
ti
m
en
t
a
n
al
y
s
is
f
o
r
d
e
p
r
ess
i
o
n
i
d
en
t
if
ica
ti
o
n
.
So
m
e
s
tu
d
ies
h
av
e
co
u
p
led
b
id
ir
ec
ti
o
n
al
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
B
iLST
M)
n
etwo
r
k
s
with
C
N
Ns
to
ca
p
tu
r
e
b
o
th
s
eq
u
en
tial
an
d
l
o
ca
l
s
ig
n
als
in
tex
tu
al
d
ata,
r
esu
ltin
g
in
b
ett
er
d
iag
n
o
s
is
o
f
d
e
p
r
ess
io
n
s
y
m
p
to
m
s
[
1
8
]
.
T
h
ese
d
ev
elo
p
m
e
n
ts
u
n
d
er
s
co
r
e
th
e
ex
p
an
d
i
n
g
p
o
ten
tial
o
f
m
ac
h
in
e
lear
n
in
g
an
d
d
ee
p
l
ea
r
n
in
g
m
o
d
els
in
au
to
m
atin
g
d
ep
r
ess
io
n
d
etec
t
io
n
,
as
well
as
th
e
im
p
o
r
tan
ce
o
f
in
ter
p
r
etab
ilit
y
an
d
tr
a
n
s
p
ar
en
cy
i
n
s
u
ch
s
en
s
itiv
e
ap
p
licatio
n
s
.
Var
io
u
s
ex
is
tin
g
r
esear
ch
wo
r
k
s
h
av
e
in
tr
o
d
u
ce
d
a
h
y
b
r
id
d
ee
p
le
ar
n
in
g
m
o
d
el
th
at
co
m
b
in
es
C
NN
an
d
l
o
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
L
STM
)
n
etw
o
r
k
s
to
d
etec
t
d
ep
r
ess
io
n
in
t
ex
t
d
ata
[
1
9
]
–
[
2
2
]
.
T
h
eir
ap
p
r
o
ac
h
e
n
h
an
ce
d
th
e
ac
cu
r
ac
y
o
f
d
iag
n
o
s
in
g
d
ep
r
e
s
s
io
n
s
y
m
p
to
m
s
b
y
u
tili
zin
g
b
o
th
lo
ca
l
an
d
lo
n
g
-
r
an
g
e
d
ata
elem
en
ts
.
Var
io
u
s
r
esear
ch
er
s
h
av
e
cr
ea
ted
a
h
y
b
r
id
m
o
d
el
th
at
co
m
b
in
es
B
E
R
T
with
a
B
iLST
M
n
etwo
r
k
to
d
etec
t
d
ep
r
ess
io
n
f
r
o
m
s
o
cial
m
ed
ia
p
o
s
ts
,
w
ith
a
f
o
cu
s
o
n
ca
p
tu
r
in
g
co
m
p
licated
lin
g
u
is
tic
p
atter
n
s
[
2
3
]
–
[
2
5
]
.
T
h
e
r
esear
ch
p
r
o
b
lem
s
ar
e
as
f
o
llo
ws:
i)
an
y
e
x
is
tin
g
d
ep
r
ess
io
n
d
ete
ctio
n
m
eth
o
d
s
r
ely
lar
g
ely
o
n
d
o
m
ain
-
s
p
ec
if
ic
d
a
tasets
(
s
u
ch
as
s
o
cial
m
ed
ia
an
d
clin
ical
n
o
tes)
th
at
f
ails
to
g
en
er
alize
ac
r
o
s
s
m
u
ltip
le
p
latf
o
r
m
s
o
r
d
o
m
ain
s
wh
er
e
lin
g
u
is
tic
tr
aits
an
d
ex
p
r
ess
io
n
p
atter
n
s
v
a
r
y
,
ii)
lack
o
f
in
te
r
p
r
etab
ilit
y
in
th
ese
m
o
d
els
m
ak
es
it
ch
allen
g
in
g
f
o
r
h
ea
lth
ca
r
e
wo
r
k
er
s
to
t
r
u
s
t
an
d
ef
f
ec
tiv
ely
d
ep
lo
y
AI
-
p
o
wer
e
d
m
en
tal
h
ea
lth
d
iag
n
o
s
tic
s
y
s
tem
s
,
iii)
d
ep
r
ess
io
n
-
r
elate
d
d
atasets
ar
e
f
r
eq
u
en
tly
u
n
b
ala
n
ce
d
,
r
esu
ltin
g
to
class
if
ier
b
ias
,
an
d
iv
)
t
r
ad
it
io
n
al
p
r
o
ce
d
u
r
es,
s
u
ch
as
s
e
lf
-
r
ep
o
r
ts
an
d
clin
ical
test
s
,
f
r
eq
u
e
n
tly
r
ely
o
n
s
u
b
jectiv
e
ju
d
g
m
e
n
t,
wh
ich
ca
n
r
esu
lt in
in
co
n
s
is
ten
cies a
n
d
b
iases
in
d
iag
n
o
s
in
g
d
ep
r
ess
io
n
.
T
h
is
r
esear
ch
aim
s
to
p
r
o
v
id
e
an
in
ter
p
r
etab
le
f
r
am
ewo
r
k
f
o
r
th
e
ea
r
ly
id
en
tific
atio
n
o
f
d
ep
r
ess
io
n
u
s
in
g
ad
v
a
n
ce
d
d
ee
p
lear
n
in
g
tech
n
iq
u
es
an
d
p
r
o
v
id
e
tr
an
s
p
ar
en
t
in
s
ig
h
ts
in
to
th
e
m
o
d
el
'
s
d
ec
is
io
n
-
m
ak
in
g
p
r
o
ce
s
s
.
T
h
e
v
alu
e
-
ad
d
e
d
co
n
tr
ib
u
tio
n
o
f
th
is
s
tu
d
y
ar
e
as
f
o
llo
ws:
i)
t
h
e
wo
r
k
co
m
b
in
es
tr
an
s
f
o
r
m
er
-
b
ased
p
r
e
-
tr
ain
ed
lan
g
u
ag
e
m
o
d
el
(
PTL
M)
f
o
r
s
tr
o
n
g
co
n
tex
tu
al
u
n
d
er
s
tan
d
i
n
g
,
B
iLST
M
f
o
r
c
ap
tu
r
in
g
s
eq
u
en
tia
l
r
elatio
n
s
h
ip
s
in
tex
t,
a
n
d
C
NN
f
o
r
f
ast
f
ea
t
u
r
e
ex
t
r
ac
tio
n
,
i
i)
t
h
e
n
ew
u
s
e
o
f
a
tten
tio
n
we
ig
h
t
v
is
u
aliza
tio
n
b
y
to
k
en
im
p
o
r
tan
ce
(
AW
VT
I
)
an
d
n
o
r
m
alize
d
atten
tio
n
s
co
r
es
(
NAS)
m
eth
o
d
o
lo
g
ies
ad
d
s
an
ex
tr
a
lay
e
r
o
f
in
ter
p
r
etab
ilit
y
allo
win
g
t
h
e
f
r
am
ewo
r
k
to
h
ig
h
lig
h
t
t
h
e
m
o
s
t
im
p
o
r
ta
n
t
elem
e
n
ts
o
f
th
e
in
p
u
t
tex
t
th
at
co
n
tr
ib
u
te
to
t
h
e
m
o
d
el'
s
ju
d
g
m
en
t,
g
iv
in
g
p
h
y
s
ician
s
a
clea
r
g
r
asp
o
f
th
e
v
a
r
iab
les
d
r
iv
in
g
d
ep
r
ess
io
n
f
o
r
ec
asts
,
iii)
th
is
in
teg
r
ated
m
o
d
el
in
cr
ea
s
es
th
e
d
etec
tio
n
o
f
d
ep
r
ess
io
n
-
r
elate
d
p
atter
n
s
in
tex
tu
al
d
ata
b
y
ac
co
u
n
tin
g
f
o
r
b
o
th
co
n
tex
tu
a
l
n
u
an
ce
s
an
d
s
eq
u
e
n
tial
lin
k
s
,
an
d
iv
)
th
is
wo
r
k
b
r
id
g
es
th
e
g
ap
b
etwe
en
h
ig
h
-
p
er
f
o
r
m
an
ce
d
ep
r
ess
io
n
d
etec
tio
n
m
o
d
els an
d
th
e
n
ee
d
f
o
r
tr
an
s
p
ar
en
cy
,
m
a
k
in
g
it a
n
ex
cit
in
g
to
o
l f
o
r
m
e
n
tal
h
ea
lth
p
r
o
f
ess
io
n
als lo
o
k
in
g
f
o
r
au
to
m
ate
d
,
in
ter
p
r
etab
le,
a
n
d
r
eliab
le
d
ep
r
ess
io
n
d
iag
n
o
s
is
o
p
tio
n
s
.
T
h
e
p
r
o
p
o
s
ed
s
tu
d
y
in
tr
o
d
u
ce
s
a
h
y
b
r
id
ar
ch
itectu
r
e
ca
p
ab
le
o
f
s
y
n
er
g
izin
g
th
e
lo
c
al
p
atter
n
d
etec
tio
n
o
f
C
NN,
s
eq
u
en
tial
lear
n
in
g
o
f
B
iLST
M,
an
d
co
n
tex
tu
al
p
o
ten
tials
o
f
PTL
M.
T
h
e
co
llab
o
r
atio
n
o
f
th
e
two
p
r
esen
ted
in
ter
p
r
etab
l
e
m
eth
o
d
s
AW
VT
I
an
d
NAS
is
d
ir
ec
tly
co
n
n
ec
ted
to
th
e
in
t
er
n
al
r
ep
r
esen
tatio
n
o
f
th
e
m
o
d
el.
T
h
e
s
tu
d
y
also
ex
h
ib
ited
th
e
p
er
f
o
r
m
an
ce
e
n
h
an
ce
m
e
n
t
in
c
o
n
tr
ast
to
b
a
s
elin
e
m
o
d
els
with
p
o
ten
tial
in
ter
p
r
etab
ilit
y
in
p
er
s
p
ec
tiv
e
to
t
h
e
d
iag
n
o
s
is
o
f
th
e
m
en
tal
h
ea
lth
ar
ea
,
wh
er
e
it
is
h
ig
h
ly
u
n
s
u
itab
le
to
d
e
p
lo
y
b
lack
b
o
x
m
o
d
els.
Dif
f
er
e
n
t
f
r
o
m
co
n
v
en
tio
n
al
atten
tio
n
-
b
ased
ex
p
l
an
atio
n
o
r
u
s
ag
e
o
f
Sh
ap
ley
ad
d
itiv
e
ex
p
la
n
atio
n
(
SHAP)
o
r
lo
ca
l
in
ter
p
r
etab
le
m
o
d
el
ag
n
o
s
tic
ex
p
lan
atio
n
(
L
I
ME
)
th
at
o
f
f
er
s
p
o
s
t
-
h
o
c
in
ter
p
r
etab
ilit
y
,
th
e
d
is
tr
ib
u
tio
n
o
f
an
in
ter
n
al
atten
tio
n
is
lev
er
ag
ed
b
y
AW
VT
I
ac
r
o
s
s
all
lay
er
s
o
f
th
e
tr
an
s
f
o
r
m
e
r
f
o
r
d
ete
r
m
in
i
n
g
th
e
im
p
o
r
tan
ce
o
f
to
k
en
wh
ile
atten
tio
n
is
s
im
p
lifie
d
ad
o
p
tin
g
NAS
b
y
o
b
tain
in
g
m
ea
n
o
v
er
all
lay
er
s
an
d
h
ea
d
s
.
T
h
is
in
n
o
v
ativ
e
m
eth
o
d
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
f
ac
ilit
ates
g
lo
b
al
in
ter
p
r
etab
ilit
y
with
f
in
er
g
r
an
u
lar
ity
,
m
ea
n
t
f
o
r
aid
in
g
p
r
ac
t
itio
n
er
s
to
u
n
d
e
r
s
tan
d
th
e
p
o
te
n
tially
in
f
lu
e
n
cin
g
in
p
u
t te
x
t.
2.
M
E
T
H
O
D
I
n
th
is
p
a
r
t,
we
o
u
tli
n
e
t
h
e
m
e
th
o
d
s
u
s
ed
in
o
u
r
s
u
g
g
est
e
d
f
r
am
e
wo
r
k
f
o
r
d
e
p
r
ess
i
o
n
e
ar
l
y
d
et
ec
ti
o
n
,
wh
i
ch
c
o
m
b
i
n
es
s
o
p
h
is
tic
ate
d
in
t
er
p
r
et
a
b
ili
ty
te
c
h
n
iq
u
es
wi
t
h
i
n
t
er
p
r
et
a
b
le
d
ee
p
l
ea
r
n
i
n
g
m
o
d
els.
T
o
i
m
p
r
o
v
e
m
o
d
el
p
er
f
o
r
m
a
n
ce
,
th
e
f
r
a
m
ewo
r
k
e
x
h
ib
ite
d
i
n
Fi
g
u
r
e
1
m
a
k
es
u
s
e
o
f
P
T
L
M
-
b
as
ed
m
o
d
els
i
n
a
d
d
iti
o
n
t
o
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
n
in
terp
r
eta
b
le
d
ee
p
lea
r
n
in
g
fr
a
mewo
r
k
fo
r
ea
r
ly
d
etec
tio
n
o
f
…
(
C
h
a
ith
r
a
I
n
d
a
va
r
a
V
e
n
ka
tesh
a
g
o
w
d
a
)
897
ex
t
r
a
l
a
y
e
r
s
l
ik
e
C
N
N
an
d
B
i
L
STM
.
W
e
u
s
e
NAS
a
n
d
att
en
t
io
n
we
ig
h
t
v
is
u
al
iz
ati
o
n
b
y
to
k
e
n
i
m
p
o
r
ta
n
ce
(
AW
VT
I
)
as
i
n
te
r
p
r
eta
b
i
lit
y
m
o
d
u
l
es
t
o
g
u
a
r
a
n
te
e
i
n
t
er
p
r
eta
b
il
it
y
.
T
h
e
p
r
o
p
o
s
e
d
m
o
d
e
l
s
u
r
p
ass
es
e
x
is
t
in
g
s
y
s
te
m
s
b
y
in
te
g
r
ati
n
g
th
e
s
t
r
e
n
g
t
h
s
o
f
PT
L
M
,
B
i
L
ST
M,
a
n
d
C
NN,
r
es
u
lti
n
g
in
s
u
p
er
io
r
f
e
atu
r
e
e
x
t
r
a
cti
o
n
a
n
d
class
i
f
i
ca
t
io
n
ac
c
u
r
a
cy
.
U
n
li
k
e
s
t
a
n
d
ar
d
m
o
d
els,
wh
ic
h
s
tr
u
g
g
l
e
t
o
c
ap
tu
r
e
c
o
n
te
x
t
u
al
r
ela
ti
o
n
s
h
i
p
s
,
t
h
e
p
r
o
p
o
s
e
d
m
et
h
o
d
a
cc
u
r
at
el
y
p
r
e
d
ic
ts
b
o
th
s
h
o
r
t
-
te
r
m
a
n
d
lo
n
g
-
t
e
r
m
te
x
t
p
atte
r
n
s
,
i
n
c
r
e
asin
g
p
r
e
cisi
o
n
a
n
d
r
e
ca
ll
.
F
u
r
t
h
er
m
o
r
e,
t
h
e
u
s
e
o
f
in
te
r
p
r
eta
b
il
it
y
te
ch
n
i
q
u
es
s
u
c
h
as
AW
VT
I
a
n
d
NAS
im
p
r
o
v
es
t
h
e
m
o
d
el'
s
tr
a
n
s
p
a
r
e
n
cy
,
m
a
k
i
n
g
it
m
o
r
e
d
e
p
e
n
d
a
b
l
e
f
o
r
r
ea
l
-
wo
r
l
d
a
p
p
l
icat
io
n
s
t
h
a
n
p
r
e
v
i
o
u
s
b
l
ac
k
-
b
o
x
ap
p
r
o
ac
h
es
.
Fig
u
r
e
1
.
I
n
ter
p
r
etab
le
f
r
am
ew
o
r
k
f
o
r
ea
r
l
y
d
etec
tio
n
o
f
d
e
p
r
ess
io
n
2
.
1
.
Da
t
a
s
et
a
nd
p
re
pro
ce
s
s
i
ng
T
h
e
p
r
o
p
o
s
ed
s
tu
d
y
u
s
es
a
s
tan
d
ar
d
p
u
b
licly
av
ailab
le
d
a
taset
[
2
6
]
t
h
at
co
n
s
is
ts
o
f
s
o
cial
m
ed
ia
p
o
s
ts
lab
elled
f
o
r
i
n
d
icatio
n
s
o
f
d
e
p
r
ess
io
n
.
T
h
e
d
ataset
c
o
n
s
is
ts
o
f
1
3
,
0
0
0
tex
t
u
al
in
f
o
r
m
atio
n
f
r
o
m
u
s
er
p
o
s
ts
th
at
ar
e
f
u
r
th
er
class
if
ied
in
to
6
,
5
0
0
p
o
s
ts
o
f
co
n
tr
o
ll
ed
u
s
er
s
an
d
6
,
5
0
0
p
o
s
ts
o
f
u
s
er
s
with
a
p
o
s
itiv
e
ca
s
e
o
f
d
ep
r
ess
io
n
.
Pre
p
r
o
ce
s
s
in
g
o
p
er
atio
n
is
in
itiated
b
y
elim
in
atin
g
all
non
-
E
n
g
lis
h
c
o
n
ten
ts
,
f
o
llo
wed
b
y
f
ilter
in
g
all
ir
r
elev
an
t
co
n
ten
t
s
,
lo
wer
ca
s
in
g
,
tr
u
n
ca
tin
g
lin
k
s
an
d
s
p
ec
ial
ch
ar
ac
ter
s
.
T
o
k
en
izer
co
n
n
ec
ted
with
th
e
o
p
te
d
PTL
M
,
B
E
R
T
is
u
s
ed
f
o
r
to
k
e
n
izatio
n
w
h
ile
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
)
is
u
s
ed
f
o
r
ad
d
r
ess
in
g
class
im
b
alan
ce
.
Fu
r
th
er
,
p
o
ten
tial
b
iases
ar
e
id
en
tifie
d
an
d
ad
d
r
ess
ed
b
y
in
clu
d
in
g
d
iv
er
s
e
s
am
p
les
wh
ile
s
tr
atif
ied
cr
o
s
s
-
v
alid
atio
n
is
ap
p
lied
,
co
n
s
id
er
in
g
8
0
%
o
f
th
e
tr
ain
an
d
2
0
%
o
f
th
e
test
d
ata.
T
h
e
p
r
o
p
o
s
ed
h
y
b
r
id
f
r
am
ewo
r
k
is
tr
ai
n
ed
to
e
n
ca
p
s
u
late
lin
g
u
is
tic
p
atter
n
s
th
at
ar
e
in
d
ep
en
d
en
t o
f
p
latf
o
r
m
s
(
e.
g
.
,
em
o
tio
n
al
c
u
es a
n
d
s
en
tim
en
t m
ar
k
er
s
)
f
o
r
en
h
an
cin
g
g
en
e
r
aliza
tio
n
.
2
.
2
.
P
r
o
po
s
ed
hy
brid m
o
del
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
tar
g
ets
to
em
p
lo
y
a
p
r
e
-
tr
ain
e
d
lan
g
u
ag
e
m
o
d
el
(
PTL
M)
to
r
ec
o
g
n
ize
lin
g
u
is
tic
co
n
tex
t
a
n
d
g
en
er
ate
m
ea
n
i
n
g
f
u
l
tex
t
r
ep
r
esen
tatio
n
s
.
E
ac
h
in
p
u
t
to
k
en
x
i
is
r
e
p
r
esen
ted
b
y
a
to
k
e
n
em
b
ed
d
in
g
e
i
in
PTL
M,
wh
ich
em
p
lo
y
s
th
e
T
r
an
s
f
o
r
m
er
a
r
ch
itectu
r
e.
N
lev
els
o
f
m
u
lti
-
h
ea
d
s
elf
-
atten
tio
n
ar
e
ap
p
lied
to
th
e
in
p
u
t to
k
e
n
s
eq
u
en
ce
:
(
,
,
)
=
(
√
)
.
(
1
)
I
n
(
1
)
,
Q
s
tan
d
s
f
o
r
t
h
e
q
u
e
r
y
,
K
f
o
r
th
e
k
e
y
,
an
d
V
f
o
r
th
e
v
alu
e.
A
weig
h
ted
s
u
m
o
f
v
alu
es
V
is
ca
lcu
lated
b
y
th
e
atten
tio
n
m
e
ch
an
is
m
,
wh
er
e
th
e
weig
h
ts
ar
e
b
ased
o
n
h
o
w
s
im
ilar
th
e
q
u
er
ies
an
d
k
ey
s
ar
e.
T
h
e
o
u
tp
u
ts
o
f
th
e
m
o
d
el
u
n
d
er
g
o
la
y
er
n
o
r
m
aliza
tio
n
an
d
m
an
y
f
ee
d
f
o
r
war
d
n
eu
r
al
n
etwo
r
k
lay
e
r
s
.
Fo
llo
win
g
r
ef
in
em
e
n
t,
we
em
p
lo
y
th
e
class
if
icatio
n
to
k
e
n
'
s
(
C
L
S)
o
u
tp
u
t a
s
(
2
)
:
ℎ
=
PT
L
M
(
1
,
2
,
…
)
(
2
)
I
n
(
2
)
,
th
e
in
p
u
t
s
eq
u
en
ce
i
n
th
e
f
ea
tu
r
e
s
p
ac
e
is
r
ep
r
esen
te
d
b
y
th
e
f
in
al
h
id
d
en
s
tate
(
h
C
LS
)
,
wh
ich
co
r
r
esp
o
n
d
s
to
th
e
C
L
S
to
k
e
n
.
PTL
M'
s
awa
r
en
ess
o
f
lan
g
u
ag
e
n
u
an
ce
s
en
a
b
les
th
e
m
o
d
el
to
ex
t
r
ac
t
d
ee
p
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
2
,
Ap
r
il
20
2
6
:
8
9
5
-
904
898
co
n
tex
tu
al
c
h
ar
ac
ter
is
tics
f
r
o
m
tex
t,
h
e
n
ce
im
p
r
o
v
in
g
th
e
m
ac
h
in
e'
s
ca
p
ac
ity
to
d
etec
t
d
ep
r
ess
iv
e
cu
es.
T
h
is
b
u
ild
s
a
s
o
lid
f
o
u
n
d
atio
n
f
o
r
d
ep
r
ess
io
n
id
e
n
tific
atio
n
b
y
en
co
d
in
g
tex
t
in
to
c
o
m
p
lex
,
co
n
tex
t
-
awa
r
e
r
ep
r
esen
tatio
n
s
,
wh
ich
im
p
r
o
v
es
task
p
er
f
o
r
m
an
ce
.
T
h
e
s
u
g
g
ested
s
y
s
tem
'
s
B
iLST
M
lay
er
is
m
ad
e
to
r
ec
o
r
d
co
n
tex
tu
al
in
f
o
r
m
atio
n
in
tex
t
s
eq
u
en
ce
s
f
r
o
m
th
e
p
ast
an
d
th
e
f
u
tu
r
e.
B
iLST
Ms
p
r
o
ce
s
s
th
e
in
p
u
t
s
eq
u
en
ce
b
o
th
f
o
r
war
d
a
n
d
b
ac
k
war
d
s
,
en
ab
lin
g
t
h
e
m
o
d
el
t
o
u
s
e
b
o
th
p
ast
a
n
d
f
u
tu
r
e
d
ep
en
d
en
cies
to
e
n
h
an
c
e
p
r
ed
ictio
n
s
,
i
n
c
o
n
tr
ast
to
r
eg
u
lar
L
STM
s
,
wh
ich
o
n
ly
ta
k
e
in
to
ac
co
u
n
t
p
ast
c
o
n
tex
t.
A
B
iLST
M
ca
lcu
lates
h
id
d
en
s
tates in
b
o
t
h
f
o
r
war
d
(
ℎ
⃗
⃗
⃗
)
an
d
b
ac
k
war
d
s
(
ℎ
⃖
⃗
⃗
⃗
)
d
ir
ec
tio
n
s
g
iv
en
an
in
p
u
t seq
u
en
ce
{
x
1
,
x
2
,
…,
x
n
}:
ℎ
⃗
⃗
⃗
=
(
,
ℎ
−
1
⃗
⃗
⃗
⃗
⃗
⃗
⃗
⃗
)
(
3
)
ℎ
⃖
⃗
⃗
⃗
=
=
(
,
ℎ
+
1
⃖
⃗
⃗
⃗
⃗
⃗
⃗
⃗
⃗
)
(
4
)
I
n
(
3
)
a
n
d
(
4
)
,
th
e
c
o
n
ca
ten
at
io
n
o
f
th
e
f
o
r
war
d
a
n
d
b
ac
k
w
ar
d
s
h
id
d
e
n
s
tates
is
th
e
f
in
al
o
u
tp
u
t
at
ea
ch
tim
e
s
tep
t
.
T
h
e
n
o
tio
n
is
to
in
clu
d
e
a
B
iLST
M
lay
er
th
at
ca
n
d
etec
t
b
o
th
f
o
r
w
ar
d
an
d
b
ac
k
war
d
s
d
ep
en
d
e
n
cies
in
th
e
in
p
u
t
te
x
t
s
eq
u
en
ce
.
B
iLST
M
en
ab
le
s
th
e
m
o
d
el
to
lear
n
f
r
o
m
b
o
th
p
ast
a
n
d
f
u
tu
r
e
co
n
tex
t in
a
s
en
ten
ce
,
wh
ich
i
m
p
r
o
v
es its
g
r
asp
o
f
s
eq
u
e
n
ce
-
b
ased
v
ar
iab
les s
u
ch
as e
m
o
tio
n
al
to
n
e.
ℎ
=
[
ℎ
,
⃗
⃗
⃗
⃗
ℎ
⃖
⃗
⃗
⃗
]
(
5
)
I
n
(
5
)
,
u
n
d
er
s
tan
d
in
g
th
e
s
u
b
tl
e
lan
g
u
ag
e
o
f
d
ep
r
ess
ed
co
n
te
n
t
is
m
ad
e
ea
s
ier
b
y
th
e
m
o
d
e
l's
ab
ilit
y
to
g
r
asp
in
tr
icate
d
ep
e
n
d
en
ci
es
in
th
e
in
p
u
t
s
eq
u
en
ce
d
u
e
to
th
ese
b
id
ir
ec
tio
n
al
r
ep
r
esen
tatio
n
s
.
T
h
is
lay
er
h
elp
s
to
d
etec
t
d
ep
r
ess
io
n
m
o
r
e
ac
cu
r
ately
b
y
ta
k
in
g
i
n
to
ac
co
u
n
t
th
e
co
m
p
lete
co
n
tex
t
o
f
th
e
tex
t
r
ath
e
r
th
an
ju
s
t
th
e
p
r
ev
io
u
s
o
r
n
ex
t
lin
es.
Fu
r
th
er
,
t
h
e
s
tu
d
y
m
o
d
el
im
p
lem
en
ts
a
C
NN
lay
er
to
d
etec
t
lo
ca
l
p
atter
n
s
an
d
ess
en
tial
ch
ar
ac
ter
is
tics
in
tex
t
in
p
u
t.
L
o
ca
l
p
atter
n
s
in
th
e
in
p
u
t
tex
t
a
r
e
ca
p
tu
r
ed
b
y
C
NN
lay
er
,
wh
ich
is
cr
u
cial
f
o
r
r
ec
o
g
n
izin
g
p
ar
tic
u
lar
wo
r
d
s
o
r
p
h
r
ases
th
at
ar
e
s
u
g
g
esti
v
e
o
f
d
ep
r
ess
io
n
.
T
o
id
en
tify
d
if
f
er
e
n
t
n
-
g
r
am
f
ea
tu
r
es
in
th
e
tex
t,
t
h
e
C
NN
lay
er
em
p
lo
y
s
a
n
u
m
b
er
o
f
f
ilter
s
o
f
v
ar
ied
wid
th
s
.
A
C
NN
lay
er
u
s
es
a
f
ilter
W
k
o
f
s
ize
k
to
p
er
f
o
r
m
a
co
n
v
o
lu
tio
n
o
p
er
atio
n
o
n
th
e
in
p
u
t
s
eq
u
en
ce
.
At
ev
er
y
lo
ca
tio
n
t
,
th
e
co
n
v
o
l
u
tio
n
o
p
er
atio
n
is
p
r
o
v
i
d
ed
b
y
(
6
)
,
=
σ(
⋅
:
+
−
1
+
)
(
6
)
I
n
(
6
)
,
:
+
−
1
is
th
e
s
u
b
s
eq
u
en
ce
o
f
l
en
g
th
f
r
o
m
th
e
in
p
u
t te
x
t,
is
t
h
e
co
n
v
o
l
u
tio
n
f
ilter
,
an
d
σ
is
th
e
ac
tiv
atio
n
f
u
n
ctio
n
s
u
c
h
as
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
e
L
U)
.
A
c
o
llectio
n
o
f
f
ea
tu
r
e
m
ap
s
th
at
i
d
en
tify
r
eg
io
n
al
tr
en
d
s
is
th
e
en
d
p
r
o
d
u
ct
o
f
th
e
co
n
v
o
lu
tio
n
.
T
h
e
m
o
s
t
im
p
o
r
tan
t
f
ea
tu
r
es
ar
e
th
en
ex
tr
ac
ted
u
s
in
g
m
ax
-
p
o
o
lin
g
.
C
o
n
ca
ten
atin
g
t
h
e
f
ea
tu
r
e
m
ap
s
f
r
o
m
s
ev
er
al
co
n
v
o
l
u
tio
n
f
ilter
s
y
ield
s
th
e
f
in
al
r
ep
r
esen
tatio
n
:
=
ma
x
−
pool
(
{
1
,
2
,
…
}
)
(
7
)
I
n
(
7
)
,
th
r
o
u
g
h
t
h
is
ap
p
r
o
ac
h
,
t
h
e
m
o
d
el
ca
n
r
ec
o
g
n
ize
im
p
o
r
tan
t,
s
h
o
r
t
-
r
an
g
e
p
atter
n
s
in
th
e
tex
t th
at
m
ay
in
d
icate
lan
g
u
ag
e
c
o
n
n
e
cted
to
d
ep
r
ess
io
n
.
T
h
e
C
NN
lay
er
d
etec
ts
cr
u
cial
lin
g
u
is
tic
p
atter
n
s
,
s
u
c
h
as
s
en
tim
en
t
-
ca
r
r
y
in
g
p
h
r
ases
,
w
h
ich
aid
in
b
etter
d
is
cr
im
in
atin
g
b
etwe
en
d
e
p
r
ess
iv
e
ex
p
r
es
s
io
n
s
.
B
y
f
o
cu
s
in
g
o
n
lo
ca
l
te
x
t
f
ea
tu
r
es,
th
is
l
ay
er
in
cr
ea
s
es
th
e
m
o
d
el'
s
s
en
s
itiv
ity
to
in
d
iv
i
d
u
al
p
h
r
as
es
an
d
k
e
y
wo
r
d
s
,
allo
win
g
f
o
r
m
o
r
e
ac
cu
r
ate
d
e
p
r
ess
io
n
id
en
tific
atio
n
.
T
h
e
n
ex
t
p
ar
t
o
f
th
e
im
p
lem
en
tatio
n
is
ass
o
ciate
d
with
th
e
in
teg
r
atio
n
o
f
lay
er
s
an
d
t
h
e
f
in
al
p
r
ed
ictio
n
.
T
h
e
g
o
al
o
f
th
is
p
ar
t
o
f
th
e
im
p
lem
en
ta
tio
n
is
to
m
er
g
e
th
e
o
u
tp
u
ts
o
f
th
e
PTL
M,
B
iLST
M,
an
d
C
NN
lay
er
s
in
to
a
s
in
g
le
u
n
if
ied
r
ep
r
esen
tatio
n
an
d
m
ak
e
th
e
f
i
n
al
p
r
ed
ictio
n
.
I
n
teg
r
atin
g
th
ese
l
ay
er
s
en
ab
les
th
e
m
o
d
el
to
b
en
ef
it
f
r
o
m
lo
n
g
-
r
an
g
e
d
e
p
en
d
en
cies
B
iLST
M,
lo
ca
lized
p
atter
n
s
C
NN,
an
d
d
ee
p
co
n
tex
tu
al
awa
r
en
ess
PTL
M.
T
h
e
f
in
al
d
ep
r
ess
io
n
class
if
icatio
n
is
g
en
er
ated
b
y
co
m
b
in
in
g
th
e
o
u
tp
u
t
f
r
o
m
th
e
C
NN,
B
iLST
M,
an
d
PTL
M
lay
er
s
in
to
a
f
u
lly
lin
k
ed
lay
e
r
an
d
th
en
ap
p
l
y
in
g
a
s
o
f
tm
ax
f
u
n
cti
o
n
:
=
⋅
[
ℎ
C
L
S
;
ℎ
B
i
L
S
T
M
;
C
N
N
]
+
(
8
)
̂
=
s
oft
ma
x
(
z
)
(
9
)
I
n
(
8
)
a
n
d
(
9
)
,
̂
is
th
e
ex
p
ec
ted
p
r
o
b
ab
ilit
y
d
is
tr
ib
u
tio
n
o
v
er
th
e
d
ep
r
ess
io
n
a
n
d
n
o
n
-
d
e
p
r
ess
io
n
class
es.
T
h
e
v
ar
iab
les
W
fc
an
d
b
fc
r
ep
r
esen
ts
th
e
f
u
lly
co
n
n
e
cted
lay
er
'
s
weig
h
ts
an
d
b
iase
s
.
T
h
is
co
m
b
in
atio
n
in
cr
ea
s
es
o
v
er
all
m
o
d
el
ac
cu
r
ac
y
an
d
p
r
ed
ictio
n
r
eliab
ilit
y
wh
en
id
en
tify
in
g
d
ep
r
ess
io
n
-
r
elate
d
m
ater
ial
in
tex
t.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
n
in
terp
r
eta
b
le
d
ee
p
lea
r
n
in
g
fr
a
mewo
r
k
fo
r
ea
r
ly
d
etec
tio
n
o
f
…
(
C
h
a
ith
r
a
I
n
d
a
va
r
a
V
e
n
ka
tesh
a
g
o
w
d
a
)
899
2
.
3
.
I
nte
rpre
t
a
bil
it
y
m
o
dellin
g
T
h
is
p
ar
t
o
f
th
e
o
p
er
atio
n
t
ar
g
ets
to
d
is
p
lay
th
e
s
ig
n
if
i
ca
n
ce
o
f
in
d
iv
id
u
al
to
k
en
s
i
n
th
e
tex
t,
in
d
icatin
g
wh
ich
wo
r
d
s
o
r
p
h
r
ases
co
n
tr
ib
u
te
th
e
m
o
s
t
to
t
h
e
d
ep
r
ess
io
n
class
if
icatio
n
.
W
e
u
s
e
AW
VT
I
,
a
tech
n
iq
u
e
th
at
s
h
o
ws
t
h
e
atten
tio
n
r
atin
g
s
g
iv
e
n
to
ea
ch
to
k
en
in
th
e
i
n
p
u
t
s
eq
u
en
ce
,
to
i
m
p
r
o
v
e
th
e
m
o
d
el'
s
in
ter
p
r
etab
ilit
y
.
AW
VT
I
ca
lcu
lates
th
e
im
p
o
r
tan
ce
s
co
r
e
f
o
r
ea
ch
to
k
en
x
i
b
y
ad
d
in
g
u
p
th
e
atten
tio
n
v
alu
es,
tak
in
g
in
to
ac
c
o
u
n
t t
h
e
atten
tio
n
weig
h
ts
ac
r
o
s
s
all
lay
er
s
an
d
h
ea
d
s
o
f
th
e
PTL
M
m
o
d
el:
(
)
=
∑
∑
ℎ
(
)
ℎ
−
1
−
1
(
10
)
I
n
(
1
0
)
,
th
e
atten
tio
n
weig
h
t
o
f
to
k
en
x
i
in
lay
er
l
a
n
d
h
e
ad
h
is
r
ep
r
esen
ted
as
ℎ
(
x
i
)
,
wh
er
e
L
is
t
h
e
n
u
m
b
e
r
o
f
lay
e
r
s
an
d
H
is
th
e
n
u
m
b
e
r
o
f
atte
n
tio
n
h
ea
d
s
.
I
n
s
ig
h
ts
in
to
wh
i
ch
wo
r
d
s
o
r
p
h
r
ases
ar
e
ess
en
tial
f
o
r
d
ep
r
ess
io
n
d
e
tectio
n
ar
e
p
r
o
v
i
d
ed
b
y
t
h
e
s
ig
n
if
ican
ce
s
co
r
es,
wh
ic
h
r
a
n
k
t
h
e
to
k
e
n
s
th
at
h
a
v
e
th
e
g
r
ea
test
in
f
lu
en
ce
o
n
th
e
m
o
d
el'
s
ch
o
ice.
AW
VT
I
aid
s
in
th
e
m
o
d
el'
s
d
ec
is
io
n
-
m
ak
in
g
b
y
p
r
esen
tin
g
tr
an
s
p
ar
en
t,
h
u
m
an
-
r
e
ad
ab
le
v
is
u
als
o
f
wh
at
th
e
alg
o
r
ith
m
f
in
d
s
s
ig
n
if
ican
t.
T
h
i
s
im
p
r
o
v
es
m
o
d
el
in
ter
p
r
etab
ilit
y
,
all
o
win
g
u
s
er
s
to
tr
u
s
t
a
n
d
co
m
p
r
eh
en
d
th
e
s
y
s
tem
'
s
p
r
ed
ictio
n
s
,
p
ar
tic
u
lar
ly
in
s
en
s
itiv
e
m
en
tal
h
ea
lth
s
ettin
g
s
.
Fu
r
th
er
,
NAS
av
er
ag
es
atten
tio
n
weig
h
ts
ac
r
o
s
s
all
lay
er
s
an
d
h
ea
d
s
to
p
r
o
d
u
ce
a
co
n
d
en
s
ed
f
o
r
m
o
f
atten
tio
n
v
is
u
aliza
tio
n
.
W
ith
o
u
t
th
e
g
r
an
u
lar
ity
p
r
o
v
id
ed
b
y
AW
VT
I
,
th
is
m
eth
o
d
p
r
o
d
u
ce
s
a
h
ig
h
-
lev
el
u
n
d
er
s
t
an
d
in
g
o
f
wh
ich
to
k
e
n
s
ar
e
s
ig
n
if
ican
t.
T
o
k
e
n
x
i
'
s
atten
tio
n
s
co
r
e
in
NAS
i
s
d
eter
m
in
ed
b
y
(
1
1
)
:
(
)
=
1
.
.
∑
∑
ℎ
(
)
ℎ
−
1
−
1
(
1
1
)
I
n
(
1
1
)
,
it
is
n
o
te
d
th
at
NAS
o
f
f
er
s
a
h
elp
f
u
l
s
u
m
m
ar
y
o
f
wh
ich
to
k
en
s
ar
e
m
o
s
t
im
p
o
r
ta
n
t f
o
r
m
o
d
el
p
r
ed
ictio
n
s
,
ev
e
n
th
o
u
g
h
it
d
o
es
n
o
t
r
ef
lect
th
e
in
tr
icate
in
ter
ac
tio
n
s
b
etwe
en
atten
tio
n
we
ig
h
ts
ac
r
o
s
s
lay
er
s
an
d
h
ea
d
s
.
NAS
s
im
p
lifie
s
th
e
u
n
d
er
s
tan
d
i
n
g
o
f
to
k
e
n
im
p
o
r
tan
ce
b
y
av
er
a
g
in
g
atten
tio
n
ac
r
o
s
s
th
e
m
o
d
el'
s
lay
er
s
,
r
esu
ltin
g
in
a
clea
r
p
i
ctu
r
e
o
f
k
ey
t
o
k
en
s
.
T
h
is
m
e
th
o
d
p
r
o
v
id
es
a
s
im
p
ler
alter
n
ativ
e
to
AW
VT
I
,
m
ak
in
g
t
h
e
m
o
d
el'
s
d
ec
is
io
n
-
m
ak
in
g
p
r
o
ce
s
s
m
o
r
e
ac
c
ess
ib
le
an
d
u
n
d
e
r
s
tan
d
ab
le
t
o
p
r
ac
titi
o
n
e
r
s
an
d
co
n
s
u
m
er
s
.
2
.
4
.
L
o
s
s
f
un
ct
io
n a
nd
o
ptim
iza
t
io
n
T
h
is
o
p
er
atio
n
tar
g
ets
to
s
p
e
cif
y
th
e
lo
s
s
f
u
n
ctio
n
a
n
d
o
p
tim
izatio
n
m
eth
o
d
s
u
ch
as
A
d
am
W
f
o
r
tr
ain
in
g
t
h
e
m
o
d
el
to
r
ed
u
ce
p
r
ed
ictio
n
er
r
o
r
s
.
T
h
e
c
r
o
s
s
-
en
tr
o
p
y
lo
s
s
f
u
n
ctio
n
,
wh
ich
w
o
r
k
s
well
f
o
r
b
in
a
r
y
class
if
icatio
n
task
s
lik
e
d
ep
r
ess
io
n
d
etec
tio
n
,
is
u
s
ed
to
t
r
ain
th
e
f
in
al
m
o
d
el.
T
h
e
d
e
f
in
itio
n
o
f
th
e
lo
s
s
f
u
n
ctio
n
is
:
=
−
1
∑
[
l
og
(
̂
)
.
l
og
(
1
−
̂
)
]
=
1
(
1
2
)
I
n
(
1
2
)
,
it
is
n
o
ted
th
at
if
y
i
is
th
e
g
r
o
u
n
d
tr
u
th
lab
el,
̂
is
th
e
ex
p
ec
ted
p
r
o
b
a
b
ilit
y
f
o
r
th
e
p
o
s
itiv
e
class
(
d
ep
r
ess
io
n
)
,
an
d
N
is
t
h
e
n
u
m
b
er
o
f
s
am
p
les.
T
h
e
c
r
o
s
s
-
en
tr
o
p
y
lo
s
s
f
u
n
ctio
n
en
s
u
r
es
th
at
th
e
m
o
d
el
lear
n
s
to
ap
p
r
o
p
r
iately
class
i
f
y
d
ep
r
ess
io
n
-
r
elate
d
in
f
o
r
m
a
tio
n
b
y
co
m
p
ar
in
g
ex
p
ec
ted
o
u
tco
m
es
to
ac
tu
al
lab
els.
T
h
e
Ad
am
W
o
p
tim
ize
r
,
wh
ich
m
o
d
if
ies
th
e
lear
n
i
n
g
r
ate
in
r
esp
o
n
s
e
to
p
r
e
v
io
u
s
g
r
ad
ien
t
u
p
d
ates,
is
u
s
ed
to
u
p
d
ate
th
e
m
o
d
el
p
a
r
a
m
eter
s
:
+
1
=
−
.
√
+
(
1
3
)
I
n
(
1
3
)
,
t
h
e
m
o
d
e
l
p
a
r
a
m
e
t
e
r
s
a
r
e
d
e
n
o
t
e
d
b
y
θ
t
,
t
h
e
l
e
a
r
n
i
n
g
r
a
t
e
i
s
d
e
n
o
t
e
d
b
y
η
,
t
h
e
f
i
r
s
t
an
d
s
e
c
o
n
d
m
o
m
e
n
t
e
s
t
i
m
a
t
es
o
f
t
h
e
g
r
a
d
ie
n
t
s
b
y
m
t
a
n
d
v
t
,
a
n
d
ϵ
i
s
a
ti
n
y
c
o
n
s
t
a
n
t f
o
r
n
u
m
e
r
i
c
a
l
s
ta
b
i
l
it
y
.
T
h
e
o
p
t
i
m
iz
a
t
i
o
n
a
p
p
r
o
a
c
h
i
n
c
r
e
a
s
e
s
m
o
d
e
l
a
c
cu
r
a
c
y
a
n
d
g
u
a
r
a
n
t
e
e
s
t
h
a
t
t
h
e
n
e
t
w
o
r
k
l
e
a
r
n
s
e
f
f
e
c
t
i
v
el
y
,
r
e
s
u
l
t
i
n
g
i
n
i
m
p
r
o
v
e
d
d
e
p
r
e
s
s
i
o
n
d
e
te
c
t
i
o
n
p
e
r
f
o
r
m
an
c
e
.
T
h
e
n
e
x
t
s
e
ct
i
o
n
p
r
e
s
e
n
ts
a
d
i
s
c
u
s
s
i
o
n
o
f
a
c
c
o
m
p
li
s
h
e
d
o
u
t
c
o
m
e
s
.
3.
RE
SU
L
T
Fro
m
th
e
p
er
s
p
ec
tiv
e
o
f
s
o
f
t
war
e
u
s
ag
e,
T
en
s
o
r
Flo
w
an
d
Py
T
o
r
ch
,
two
Py
th
o
n
-
b
ased
f
r
am
ewo
r
k
s
,
ar
e
u
s
ed
b
y
t
h
e
en
v
ir
o
n
m
en
t
t
o
b
u
ild
an
d
tr
ain
th
e
m
o
d
el.
B
y
u
s
in
g
ef
f
ec
tiv
e
g
r
ap
h
ics
p
r
o
ce
s
s
in
g
u
n
it
(
GPU)
s
u
p
p
o
r
t
f
o
r
q
u
ic
k
er
t
r
ain
in
g
an
d
in
f
er
en
ce
,
th
ese
f
r
am
ew
o
r
k
s
en
a
b
le
th
e
d
ep
l
o
y
m
en
t
o
f
a
PTL
M
-
b
ased
ar
ch
itectu
r
e
in
co
n
j
u
n
ctio
n
with
ex
tr
a
la
y
er
s
lik
e
C
NN
an
d
B
iLST
M.
Fro
m
a
h
ar
d
war
e
u
s
ag
e
p
er
s
p
ec
tiv
e,
a
co
m
p
u
ter
r
u
n
n
in
g
Ub
u
n
tu
2
0
.
0
4
with
Py
t
h
o
n
3
.
8
,
an
I
n
tel
C
o
r
e
i9
-
1
0
9
0
0
K
p
r
o
ce
s
s
o
r
,
an
NVI
DI
A
R
T
X
3
0
9
0
24
GB
GPU,
an
d
6
4
GB
DDR4
r
an
d
o
m
ac
ce
s
s
m
em
o
r
y
(
R
AM
)
was u
s
ed
f
o
r
all
ex
p
er
im
en
ts
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
2
,
Ap
r
il
20
2
6
:
8
9
5
-
904
900
3
.
1
.
H
y
perpa
ra
m
e
t
er
s
a
nd
t
ra
ini
ng
s
et
up
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
u
s
es
th
e
f
o
llo
win
g
h
y
p
er
p
a
r
am
eter
s
w
ith
a
lear
n
i
n
g
r
ate
o
f
0
.
0
0
0
2
,
b
atch
s
ize
o
f
3
2
,
1
0
e
p
o
ch
s
,
0
.
3
d
r
o
p
o
u
t
r
ate,
an
d
1
2
8
as
th
e
m
ax
im
u
m
s
eq
u
e
n
ce
len
g
th
.
T
h
e
m
o
d
el
u
s
es
th
e
Ad
a
m
o
p
tim
izer
,
wh
ile
B
E
R
T
-
b
ased
u
n
ca
s
ed
is
u
s
ed
b
y
PTL
M
f
o
r
ad
ju
s
tm
en
t w
h
ile
p
er
f
o
r
m
in
g
t
r
ain
in
g
.
A
win
d
o
w
s
ize
o
f
(
2
,
3
,
4
)
is
u
s
ed
f
o
r
th
e
C
NN
lay
er
f
ilter
,
wh
er
e
th
e
r
e
ar
e
1
0
0
f
ilter
s
ea
ch
,
an
d
th
er
e
ar
e
2
5
6
h
id
d
en
u
n
its
p
r
esen
t in
th
e
B
iLST
M
lay
er
.
3
.
2
.
Arc
hite
ct
ure
d
et
a
ils
T
h
e
p
r
o
p
o
s
ed
s
y
s
tem
m
ain
ly
co
n
s
is
ts
o
f
f
o
u
r
ess
en
tial
co
m
p
o
n
e
n
ts
o
f
ar
ch
itectu
r
al
lay
er
s
,
wh
o
s
e
d
etails
ar
e
as
f
o
llo
ws:
I
n
th
e
PTL
M
lay
er
,
a
p
r
etr
ain
ed
B
E
R
T
is
u
s
ed
with
a
h
id
d
en
s
ize
o
f
lay
er
as
7
6
8
,
1
2
atten
tio
n
h
ea
d
s
,
an
d
1
2
tr
an
s
f
o
r
m
er
lay
er
s
.
I
n
th
e
B
iLST
M
lay
er
,
1
2
8
h
id
d
en
u
n
its
ca
p
tu
r
e
b
o
th
b
ac
k
war
d
s
an
d
f
o
r
war
d
d
ep
e
n
d
en
cies
f
r
o
m
th
e
i
n
p
u
t
s
eq
u
en
ce
s
.
Ov
e
r
f
itti
n
g
is
m
in
im
ized
b
y
ad
o
p
tin
g
a
d
r
o
p
o
u
t
r
ate
im
m
ed
iately
af
ter
th
e
B
iLST
M
lay
er
.
I
n
th
e
C
NN
lay
er
,
th
er
e
ar
e
o
n
e
-
d
im
en
s
io
n
al
co
n
v
o
lu
tio
n
lay
er
s
ar
r
an
g
e
d
in
p
a
r
allel
with
3
,
4
,
an
d
5
f
ilter
s
izes,
with
o
v
er
all
1
0
0
f
ilter
s
i
n
v
o
lv
e
d
i
n
it.
T
h
e
m
o
d
el
u
s
es
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
eL
U)
a
s
an
ac
tiv
atio
n
f
u
n
ctio
n
,
wh
ic
h
is
f
u
r
th
er
r
esu
m
ed
b
y
a
m
a
x
-
p
o
o
lin
g
o
p
er
atio
n
f
o
llo
wed
b
y
c
o
n
ca
ten
atin
g
th
e
p
o
o
le
d
o
u
tco
m
es.
I
n
t
h
e
f
u
l
ly
co
n
n
ec
ted
la
y
er
,
all
th
e
c
o
n
ca
ten
ated
f
ea
t
u
r
es
ar
e
f
o
r
war
d
ed
v
ia
a
f
u
lly
co
n
n
ec
ted
d
en
s
e
lay
er
co
n
s
is
tin
g
o
f
R
eL
U
an
d
2
5
6
u
n
its
th
at
is
f
in
ally
s
u
b
jecte
d
to
a
So
f
tMa
x
class
if
ier
to
f
ac
ilit
ate
f
in
al
b
in
ar
y
class
if
icatio
n
.
3
.
3
.
Acc
o
m
pli
s
hed
r
esu
lt
s
T
ab
le
1
s
h
o
wca
s
es
th
e
n
u
m
er
i
ca
l
o
u
tco
m
e
o
f
th
e
s
tu
d
y
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
tem
(
PTL
M
+
B
iLST
M
+
C
NN)
ad
o
p
tin
g
th
e
s
tan
d
ar
d
d
ataset.
E
x
is
tin
g
s
y
s
tem
s
1
(
E
S
1
)
r
ep
r
esen
ts
f
in
e
-
tu
n
ed
PTL
M,
e
x
is
tin
g
s
y
s
tem
s
2
(
E
S2
)
r
e
p
r
esen
ts
PTL
M
-
B
i
L
STM
,
an
d
ex
is
tin
g
s
y
s
tem
s
3
(
E
S3
)
r
e
p
r
esen
ts
PTL
M
-
C
NN.
All
th
ese
ar
e
b
aselin
e
m
o
d
els
an
d
u
s
ed
f
o
r
co
m
p
ar
is
o
n
s
in
ce
th
e
y
ar
e
we
ll
-
estab
lis
h
ed
ap
p
r
o
ac
h
es
f
o
r
t
ex
t
ca
teg
o
r
izatio
n
an
d
d
ep
r
ess
io
n
d
etec
tio
n
th
at
u
s
e
PTL
M
-
b
ased
m
o
d
els.
T
h
ese
s
y
s
tem
s
p
r
o
v
id
e
a
b
aselin
e
f
o
r
test
in
g
th
e
ef
f
icac
y
o
f
o
th
er
ar
c
h
itectu
r
e
s
,
s
u
ch
as
PTL
M
'
s
b
as
ic
f
in
e
-
tu
n
in
g
,
th
e
ad
d
itio
n
o
f
s
eq
u
en
tial
d
ep
en
d
e
n
cy
m
o
d
ellin
g
with
B
iLST
M,
an
d
f
ea
tu
r
e
ex
tr
ac
tio
n
a
u
g
m
e
n
tatio
n
with
C
NN.
W
h
en
co
m
p
ar
e
d
to
th
ese
ex
is
tin
g
s
y
s
tem
s
,
th
e
s
u
g
g
ested
m
o
d
el'
s
p
er
f
o
r
m
an
ce
ad
v
a
n
tag
es
in
a
cc
u
r
ac
y
,
r
ec
all,
p
r
ec
is
io
n
,
an
d
in
ter
p
r
etab
ilit
y
a
r
e
r
ea
d
ily
ev
id
en
t
.
Alth
o
u
g
h
v
a
r
io
u
s
s
tatis
tical
s
ig
n
if
ican
ce
test
s
(
e.
g
.
,
W
ilco
x
o
n
test
s
,
t
-
test
s
)
ar
e
ad
o
p
ted
f
r
eq
u
e
n
tly
f
o
r
ass
ess
in
g
im
p
r
o
v
em
en
t
in
s
y
s
tem
p
er
f
o
r
m
an
ce
,
th
e
p
r
o
p
o
s
ed
ass
ess
m
en
t
h
as
witn
ess
ed
s
u
b
s
tan
tial
an
d
co
n
s
is
ten
t
g
ain
s
in
p
er
f
o
r
m
a
n
ce
,
with
a
r
o
u
n
d
3
%
p
er
f
o
r
m
an
ce
g
ain
i
n
a
cc
u
r
ac
y
a
n
d
a
r
o
u
n
d
5
%
p
er
f
o
r
m
an
ce
g
ain
in
r
ec
al
l/p
r
ec
is
io
n
.
Fo
r
th
e
s
tated
m
e
asu
r
em
en
t
an
d
co
n
s
is
ten
cy
o
f
en
h
an
ce
m
e
n
ts
with
s
tan
d
ar
d
ized
m
etr
ics
o
f
ev
al
u
atio
n
,
s
u
c
h
a
s
tatis
tical
s
ig
n
if
ican
ce
test
is
n
o
t
ess
en
tial
to
ex
h
ib
it
m
o
d
el
s
u
p
er
io
r
ity
.
T
h
e
in
f
er
en
ce
o
f
ac
co
m
p
lis
h
ed
o
u
tco
m
es
s
h
o
wn
in
Fig
u
r
e
2
is
as
f
o
llo
ws:
with
an
ac
cu
r
ac
y
o
f
9
6
%,
th
e
s
u
g
g
ested
m
eth
o
d
o
u
t
p
er
f
o
r
m
s
all
cu
r
r
e
n
t
s
y
s
tem
s
(
Fi
g
u
r
e
2
(
a)
)
.
Giv
e
n
th
at
it
g
ai
n
s
f
r
o
m
co
n
tex
tu
al
awa
r
en
ess
as
well
as
th
e
ca
p
ac
ity
to
id
en
tify
b
o
th
lo
ca
l
an
d
d
is
tan
t
p
atter
n
s
,
th
is
im
p
lies
th
at
th
e
co
m
b
in
atio
n
o
f
PTL
M
with
B
iLST
M
an
d
C
NN
lay
er
s
is
q
u
ite
s
u
cc
es
s
f
u
l
in
ca
teg
o
r
izin
g
c
o
n
ten
t
p
e
r
ta
in
in
g
to
d
e
p
r
ess
io
n
.
T
h
e
s
u
g
g
ested
s
y
s
tem
p
e
r
f
o
r
m
s
b
etter
th
an
th
e
o
th
er
m
o
d
els
with
a
p
r
ec
is
io
n
s
co
r
e
o
f
0
.
9
8
a
n
d
s
h
o
wn
in
Fig
u
r
e
2
(
b
)
.
T
h
e
R
ec
all
an
d
F1
-
Sco
r
e
is
illu
s
tr
ated
in
Fig
u
r
e
s
2
(
c
)
an
d
2
(
d
)
r
esp
ec
tiv
ely
.
T
h
is
s
u
g
g
ests
th
at
i
t
m
in
im
izes
f
alse
p
o
s
itiv
es
an
d
is
q
u
ite
s
u
cc
ess
f
u
l
in
d
etec
tin
g
d
ep
r
ess
iv
e
ep
is
o
d
es
wh
en
it
p
r
ed
icts
th
em
.
I
n
ap
p
licatio
n
s
r
elate
d
to
m
en
ta
l
h
ea
lth
,
wh
er
e
f
alse
p
o
s
itiv
es
m
ay
r
esu
lt
in
n
ee
d
less
in
ter
v
en
tio
n
s
,
t
h
is
is
p
ar
ticu
lar
ly
cr
u
cial.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
h
as
also
b
ee
n
co
m
p
ar
ed
with
c
u
r
r
en
t
s
tate
-
of
-
th
e
-
ar
t
m
o
d
els,
as
ex
h
ib
ited
in
T
ab
le
2
.
T
h
e
r
e
ar
e
v
a
r
io
u
s
m
o
d
els
[
1
8
]
,
[
2
1
]
,
an
d
[
2
3
]
t
h
at
h
av
e
ac
co
m
p
lis
h
ed
in
c
r
ea
s
ed
a
cc
u
r
ac
y
m
o
r
e
th
an
9
4
%,
an
d
y
et
in
ter
p
r
etab
ilit
y
i
s
s
ig
n
if
ican
tly
lack
in
g
in
th
em
,
wh
ich
r
en
d
er
s
th
em
u
n
s
u
itab
le
f
o
r
th
e
d
ia
g
n
o
s
is
o
f
m
en
tal
illn
ess
,
wh
ich
is
co
n
s
id
er
ed
a
s
en
s
itiv
e
d
o
m
ain
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
o
f
f
er
s
h
ig
h
er
ac
cu
r
ac
y
m
o
r
e
th
an
9
6
%
an
d
also
f
ac
ilit
ates
in
ter
p
r
etab
le
m
eth
o
d
s
u
s
in
g
A
W
VT
I
an
d
NAS.
Hen
ce
,
it
f
a
cilitates
ex
p
lain
ab
le
an
d
tr
an
s
p
ar
e
n
t
p
r
e
d
ictio
n
.
I
r
r
esp
ec
tiv
e
o
f
th
e
h
y
b
r
i
d
ar
ch
it
ec
tu
r
e,
a
m
in
im
al
c
o
m
p
lex
ity
is
witn
ess
ed
in
th
e
p
r
o
p
o
s
ed
m
o
d
el,
wh
ich
is
m
ain
ly
d
u
e
to
th
e
ef
f
icien
t
co
llab
o
r
atio
n
o
f
all
co
m
p
o
n
en
ts
,
m
a
k
in
g
it
s
u
itab
le
f
o
r
p
r
ac
tical
wo
r
ld
s
ce
n
ar
io
s
.
T
ab
le
1
.
C
o
m
p
a
r
is
o
n
with
b
as
elin
e
m
o
d
els
M
o
d
e
l
/
M
e
t
r
i
c
P
r
o
p
o
se
d
s
y
st
e
m
ES1
ES2
ES3
A
c
c
u
r
a
c
y
0
.
9
6
0
.
9
3
0
.
9
4
0
.
9
5
P
r
e
c
i
s
i
o
n
0
.
9
8
0
.
9
4
0
.
9
6
0
.
9
7
R
e
c
a
l
l
0
.
9
7
0
.
9
1
0
.
9
3
0
.
9
5
F1
-
S
c
o
r
e
0
.
9
7
0
.
9
2
0
.
9
4
0
.
9
6
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
n
in
terp
r
eta
b
le
d
ee
p
lea
r
n
in
g
fr
a
mewo
r
k
fo
r
ea
r
ly
d
etec
tio
n
o
f
…
(
C
h
a
ith
r
a
I
n
d
a
va
r
a
V
e
n
ka
tesh
a
g
o
w
d
a
)
901
(
a)
(
b
)
(
c)
(
d
)
Fig
u
r
e
2
.
Acc
o
m
p
lis
h
ed
o
u
tco
m
e
o
f
s
tu
d
y
in
(
a
)
ac
cu
r
ac
y
,
(
b
)
p
r
ec
is
io
n
,
(
c)
r
ec
all
,
a
n
d
(
d
)
F1
-
Sco
r
e
T
ab
le
2
.
C
o
m
p
a
r
is
o
n
with
s
tate
-
of
-
th
e
-
ar
t m
o
d
els
R
e
f
.
N
o
.
M
o
d
e
l
t
y
p
e
A
c
c
u
r
a
c
y
I
n
t
e
r
p
r
e
t
a
b
i
l
i
t
y
M
o
d
e
l
c
o
m
p
l
e
x
i
t
y
[
1
3
]
S
V
M
+
Li
n
g
u
i
st
i
c
f
e
a
t
u
r
e
s
0
.
8
5
Y
e
s (f
e
a
t
u
r
e
-
b
a
s
e
d
)
Lo
w
[
1
4
]
C
N
N
0
.
9
0
No
M
e
d
i
u
m
[
1
5
]
LSTM
0
.
8
9
No
M
e
d
i
u
m
[
1
6
]
B
ER
T
F
i
n
e
-
t
u
n
e
d
0
.
9
3
No
H
i
g
h
[
1
7
]
C
N
N
+
W
o
r
d
2
V
e
c
0
.
9
1
No
M
e
d
i
u
m
[
1
8
]
B
i
LST
M
+
C
N
N
0
.
9
4
No
H
i
g
h
[
2
0
]
D
i
st
i
l
B
ER
T
+
G
R
U
0
.
9
2
No
H
i
g
h
[
2
1
]
B
ER
T
+
C
N
N
0
.
9
5
No
H
i
g
h
[
2
3
]
B
ER
T
+
B
i
LST
M
0
.
9
4
No
H
i
g
h
[
2
5
]
X
LN
e
t
+
LST
M
0
.
9
3
No
H
i
g
h
P
r
o
p
o
se
d
P
TLM
+
B
i
LSTM
+
C
N
N
+
A
W
V
TI
/
N
A
S
0
.
9
6
y
e
s (
v
i
a
A
W
V
TI
a
n
d
N
A
S
)
Lo
w
T
h
e
co
r
e
r
esu
lt
s
h
o
wca
s
e
th
at
a
h
ig
h
r
ec
all
s
co
r
e
o
f
0
.
9
7
is
a
ls
o
attain
ed
b
y
th
e
s
u
g
g
ested
tech
n
iq
u
e,
s
u
g
g
esti
n
g
th
at
it
ac
cu
r
ately
d
etec
ts
a
s
ig
n
if
ican
t
p
er
ce
n
ta
g
e
o
f
d
e
p
r
ess
iv
e
ca
s
es.
A
s
tr
o
n
g
r
ec
all
g
u
ar
a
n
tees
th
at
th
e
s
y
s
tem
will
n
o
t
o
v
er
lo
o
k
m
an
y
ca
s
es
o
f
d
ep
r
ess
io
n
,
wh
ich
is
ess
en
tial
f
o
r
ea
r
ly
m
en
tal
h
ea
lth
s
u
p
p
o
r
t
d
etec
tio
n
.
T
h
e
r
ec
o
m
m
e
n
d
ed
ap
p
r
o
ac
h
h
as
a
g
o
o
d
r
ec
all
an
d
p
r
ec
is
io
n
with
a
n
F1
-
s
co
r
e
o
f
0
.
9
7
.
I
n
ev
e
r
y
d
im
en
s
io
n
,
it
p
er
f
o
r
m
s
b
etter
th
an
ex
is
tin
g
m
o
d
els,
p
r
o
v
i
n
g
th
e
r
o
b
u
s
tn
ess
o
f
th
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
in
d
etec
tin
g
s
o
r
r
o
w
in
tex
tu
al
d
ata.
T
h
e
s
y
s
tem
's
co
m
p
r
eh
en
s
io
n
o
f
in
tr
icate
lan
g
u
ag
e
p
att
er
n
s
in
in
f
o
r
m
atio
n
r
elev
an
t
to
d
ep
r
ess
io
n
is
en
h
an
ce
d
b
y
th
e
i
n
co
r
p
o
r
atio
n
o
f
PTL
M
as
th
e
b
asic
m
o
d
el,
wh
ich
o
f
f
er
s
r
ich
co
n
tex
tu
alize
d
r
e
p
r
esen
tatio
n
s
o
f
tex
t.
W
h
ile
C
NN
lay
er
s
f
i
n
d
lo
ca
l
p
atter
n
s
an
d
p
h
r
ases
lik
e
“
h
o
p
eless
”
o
r
“
s
u
icid
al,
”
B
iLST
M
ca
p
tu
r
es lo
n
g
-
r
a
n
g
e
d
ep
en
d
en
cies.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
2
,
Ap
r
il
20
2
6
:
8
9
5
-
904
902
3
.
4
.
Dis
cus
s
io
n
W
e
tak
e
a
s
u
itab
le
ca
s
e
s
tu
d
y
to
u
n
d
er
s
tan
d
i
n
ter
p
r
etab
ilit
y
in
th
e
o
u
tco
m
es.
C
o
n
s
id
er
a
n
ex
am
p
le
tex
t
as
“I
f
ee
l
lo
s
t
an
d
h
o
p
el
ess
.
No
th
in
g
ex
cites
m
e
an
y
m
o
r
e.
I
ju
s
t
wan
t
t
o
d
is
ap
p
e
a
r
.
”
T
h
e
to
k
e
n
s
lik
e
“d
is
ap
p
ea
r
,
”
“h
o
p
eless
,
”
an
d
“lo
s
t”
ar
e
h
ig
h
ly
ess
en
tial
an
d
ar
e
d
etec
ted
b
y
AW
VT
I
with
m
o
r
e
th
an
0
.
7
5
atten
tio
n
s
co
r
e.
NAS
ass
ig
n
s
a
g
lo
b
al
im
p
o
r
tan
ce
s
co
r
e
to
“h
o
p
eless
”
an
d
“n
o
th
in
g
”
.
T
h
e
ac
cu
r
ac
y
o
f
th
e
m
o
d
el
is
im
p
r
o
v
e
d
b
y
t
h
is
co
m
b
in
atio
n
o
f
m
eth
o
d
s
,
s
u
r
p
ass
in
g
ea
r
lier
s
y
s
tem
s
with
less
r
esil
ien
t d
esig
n
s
.
T
h
e
m
o
d
el'
s
ex
ce
llen
t
ac
cu
r
ac
y
s
t
em
s
f
r
o
m
its
ca
p
ac
ity
to
i
d
en
tify
im
p
o
r
tan
t
c
h
ar
ac
ter
is
tics
an
d
tr
e
n
d
s
th
at
ar
e
clo
s
ely
ass
o
ciate
d
with
d
e
p
r
e
s
s
io
n
.
W
h
ile
PTL
M'
s
atten
tio
n
m
ec
h
a
n
is
m
s
co
n
ce
n
t
r
ate
o
n
th
e
m
o
s
t
p
er
tin
e
n
t
p
o
r
tio
n
s
o
f
th
e
te
x
t,
C
NN
lay
er
s
f
in
d
ce
r
tain
wo
r
d
s
o
r
p
h
r
a
s
es
ass
o
ciate
d
with
d
ep
r
ess
ed
co
n
ten
t,
l
o
wer
in
g
f
alse
p
o
s
itiv
es.
T
h
is
g
u
ar
an
tees
p
r
ec
is
e
f
o
r
ec
asts
,
r
en
d
er
in
g
th
e
s
y
s
tem
ex
tr
em
ely
ef
f
icien
t
f
o
r
p
r
ac
tical
u
s
es
wh
er
e
r
ed
u
cin
g
f
alse
p
o
s
itiv
es
is
e
s
s
en
tial.
T
h
e
h
ig
h
r
ec
all
s
co
r
e
r
ef
lects
th
e
B
iL
STM
lay
er
'
s
ab
ili
ty
to
ca
p
tu
r
e
l
o
n
g
-
r
an
g
e
d
e
p
en
d
e
n
c
ies
an
d
o
v
er
all
s
en
tim
en
t
in
t
h
e
tex
t.
T
h
is
allo
ws
t
h
e
m
o
d
e
l
to
u
n
d
e
r
s
tan
d
th
e
em
o
tio
n
al
co
n
te
x
t,
wh
ich
is
cr
u
cial
f
o
r
ea
r
ly
d
ep
r
ess
io
n
d
ete
ctio
n
,
esp
ec
ially
wh
en
s
y
m
p
to
m
s
ar
e
s
u
b
tle.
T
h
e
s
tr
o
n
g
r
ec
all
en
s
u
r
es
th
at
d
ep
r
ess
iv
e
in
s
tan
ce
s
ar
e
n
o
t
m
is
s
e
d
,
h
elp
in
g
p
r
e
v
en
t
u
n
d
iag
n
o
s
ed
d
ep
r
ess
io
n
f
r
o
m
g
o
in
g
u
n
n
o
ticed
.
T
h
e
h
ig
h
F
1
-
s
co
r
e
d
em
o
n
s
tr
ates
th
e
s
y
s
tem
'
s
ef
f
ec
tiv
en
ess
in
d
ep
r
ess
io
n
d
etec
tio
n
,
as
it
b
alan
ce
s
f
alse
p
o
s
itiv
es
an
d
f
a
ls
e
n
eg
ativ
es.
Un
lik
e
ac
c
u
r
ac
y
,
th
e
F1
-
s
co
r
e
en
s
u
r
es
a
m
o
r
e
r
eliab
le
ev
alu
atio
n
o
f
th
e
m
o
d
el'
s
p
er
f
o
r
m
an
ce
,
e
s
p
ec
ially
with
im
b
alan
ce
d
d
atasets
.
T
h
e
co
m
b
in
atio
n
o
f
PTL
M,
B
iLST
M,
an
d
C
NN
lay
er
s
h
elp
s
th
e
m
o
d
el
ex
ce
l
in
b
o
th
p
r
ec
is
io
n
an
d
r
e
ca
ll
b
y
e
x
tr
ac
tin
g
lo
ca
l
an
d
g
l
o
b
al
f
ea
tu
r
es
f
r
o
m
th
e
tex
t.
T
r
ad
itio
n
al
m
o
d
els
lik
e
Fin
e
-
tu
n
ed
PTL
M
a
n
d
P
T
L
M
-
B
iLST
M
m
ay
ac
h
iev
e
h
ig
h
class
if
icatio
n
p
er
f
o
r
m
an
ce
b
u
t
lack
tr
an
s
p
a
r
en
cy
in
t
h
eir
d
ec
is
io
n
-
m
ak
i
n
g
.
T
h
is
lack
o
f
in
ter
p
r
etab
il
ity
ca
n
lim
it
th
eir
ad
o
p
tio
n
in
s
en
s
itiv
e
f
ield
s
l
ik
e
m
en
tal
h
ea
lth
ca
r
e,
wh
er
e
u
n
d
er
s
tan
d
in
g
a
m
o
d
el'
s
r
atio
n
ale
is
cr
u
cial.
E
n
s
u
r
in
g
th
at
i
n
ter
v
en
tio
n
s
ar
e
b
o
th
a
p
p
r
o
p
r
iate
an
d
eth
ical
r
eq
u
ir
es m
o
d
els th
at
ca
n
b
e
ea
s
ily
in
ter
p
r
eted
.
T
o
ad
d
r
ess
th
e
id
en
tifie
d
g
ap
s
,
th
e
p
r
o
p
o
s
ed
s
tu
d
y
in
tr
o
d
u
ce
s
a
h
y
b
r
id
m
o
d
el
b
y
co
llab
o
r
at
in
g
C
NN,
B
iLST
M,
an
d
PTL
M
with
a
tar
g
et
to
war
d
s
en
h
an
cin
g
g
en
er
aliza
tio
n
o
v
er
v
a
r
io
u
s
s
o
u
r
c
es
o
f
tex
t.
T
h
is
is
d
o
n
e
b
y
en
ca
p
s
u
latin
g
b
o
th
g
l
o
b
al
an
d
lo
ca
l
p
atter
n
s
o
f
lan
g
u
ag
e.
T
h
e
in
ter
p
r
etab
ilit
y
ch
al
len
g
e
is
o
v
er
co
m
e
b
y
in
tr
o
d
u
cin
g
NAS
an
d
AW
VT
I
to
o
f
f
er
a
s
u
m
m
ar
ized
to
k
en
-
lev
el
ex
p
la
n
atio
n
with
in
cr
ea
s
ed
g
r
an
u
lar
ity
ass
o
ciate
d
with
o
f
m
o
d
el.
T
h
e
is
s
u
es
ab
o
u
t
d
ata
im
b
alan
ce
ar
e
a
d
d
r
ess
ed
b
y
ad
o
p
tin
g
a
b
alan
ce
d
d
ataset
as
well,
an
d
th
e
p
r
esen
ted
ap
p
r
o
ac
h
is
also
ca
p
ab
le
o
f
b
alan
ci
n
g
an
y
d
ataset
th
at
is
n
ativ
ely
f
o
u
n
d
im
b
alan
ce
d
,
f
o
llo
wed
b
y
s
tr
atif
ied
s
am
p
l
in
g
p
e
r
f
o
r
m
ed
.
Fin
ally
,
o
b
je
ctiv
e
lin
g
u
is
tic
cu
es
ar
e
u
s
ed
f
o
r
a
u
to
n
o
m
o
u
s
d
etec
tio
n
o
f
d
e
p
r
ess
io
n
th
at
m
in
im
izes
b
o
th
in
co
n
s
is
ten
cies
an
d
s
u
b
jectiv
ity
f
o
u
n
d
in
co
n
v
en
tio
n
al
clin
ical
ass
es
s
m
en
t a
n
d
s
elf
-
r
ep
o
r
tin
g
s
y
s
tem
s
.
4.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
o
f
f
e
r
s
an
in
ter
p
r
et
ab
le
f
r
a
m
ewo
r
k
f
o
r
ea
r
ly
d
ep
r
ess
io
n
id
en
tific
atio
n
u
s
in
g
c
u
ttin
g
-
ed
g
e
d
ee
p
lear
n
in
g
m
eth
o
d
s
,
s
p
ec
if
ically
C
NN,
PTL
M,
an
d
B
iL
S
T
M
m
o
d
els.
B
y
in
teg
r
atin
g
th
ese
tech
n
iq
u
es,
th
e
s
u
g
g
ested
s
y
s
tem
o
u
tp
er
f
o
r
m
s
cu
r
r
en
t
m
o
d
els
i
n
a
n
u
m
b
er
o
f
im
p
o
r
tan
t
p
ar
am
eter
s
,
i
n
clu
d
in
g
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
ec
all,
a
n
d
F1
-
s
co
r
e.
Ad
d
itio
n
ally
,
th
e
s
y
s
tem
in
co
r
p
o
r
ates
s
o
p
h
is
ticated
in
ter
p
r
etab
ilit
y
m
eth
o
d
s
s
u
ch
as
AW
VT
I
an
d
NAS,
wh
ich
im
p
r
o
v
e
th
e
d
ec
is
io
n
-
m
ak
in
g
p
r
o
ce
s
s
o
f
th
e
m
o
d
el
'
s
tr
an
s
p
ar
en
cy
an
d
in
ter
p
r
etab
ilit
y
.
T
h
is
is
p
ar
tic
u
lar
ly
cr
u
cial
in
th
e
d
elica
te
f
ield
o
f
m
en
tal
h
ea
lth
,
wh
er
e
p
r
ac
tical
an
d
eth
ical
ap
p
licatio
n
s
d
ep
en
d
o
n
k
n
o
win
g
wh
y
a
m
o
d
el
p
r
ed
icts
a
p
a
r
ticu
lar
o
u
tco
m
e.
T
h
e
lim
itatio
n
o
f
th
e
p
r
o
p
o
s
e
d
s
y
s
tem
is
th
at
it
d
o
es
n
o
t
f
ac
ilit
ate
a
co
m
p
r
eh
en
s
iv
e
ev
a
lu
atio
n
f
o
r
f
ac
ilit
atin
g
th
e
p
e
r
f
o
r
m
an
ce
o
f
class
if
icatio
n
th
at
is
n
ec
ess
ar
y
to
war
d
s
a
h
i
g
h
er
d
eg
r
ee
o
f
co
m
p
le
x
d
iag
n
o
s
is
o
f
d
ep
r
ess
io
n
.
Fu
r
th
e
r
,
th
e
i
n
teg
r
atio
n
o
f
m
u
ltimo
d
al
d
at
a
s
o
u
r
ce
s
,
lik
e
au
d
io
an
d
v
i
s
u
al
cu
es,
m
ay
b
e
in
v
esti
g
ated
in
f
u
tu
r
e
r
esear
ch
to
d
ev
elo
p
a
m
o
r
e
th
o
r
o
u
g
h
a
n
d
in
te
g
r
ated
m
o
d
el
f
o
r
d
ep
r
e
s
s
io
n
id
en
tific
atio
n
.
T
h
is
wo
u
ld
e
n
ab
le
t
h
e
m
ac
h
in
e
to
in
te
r
p
r
et
a
n
d
co
m
p
r
eh
en
d
d
e
p
r
ess
io
n
m
o
r
e
ac
c
u
r
at
ely
ac
r
o
s
s
v
ar
i
o
u
s
ch
an
n
els
o
f
c
o
m
m
u
n
icatio
n
.
T
h
e
s
y
s
tem
m
ig
h
t
also
b
e
e
x
ten
d
ed
t
o
ac
co
m
m
o
d
ate
v
a
r
io
u
s
lan
g
u
ag
es
an
d
cu
ltu
r
al
s
itu
atio
n
s
to
in
cr
ea
s
e
its
g
en
er
aliza
b
ilit
y
an
d
s
u
itab
ilit
y
f
o
r
u
s
e
in
in
ter
n
atio
n
al
m
en
tal
h
ea
lth
ca
r
e
s
ettin
g
s
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
e
au
th
o
r
s
r
ec
eiv
ed
n
o
f
in
a
n
cial
s
u
p
p
o
r
t
f
o
r
th
e
r
esear
ch
,
au
th
o
r
s
h
ip
,
an
d
/o
r
p
u
b
licatio
n
o
f
th
is
ar
ticle.
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
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2088
-
8
7
0
8
A
n
in
terp
r
eta
b
le
d
ee
p
lea
r
n
in
g
fr
a
mewo
r
k
fo
r
ea
r
ly
d
etec
tio
n
o
f
…
(
C
h
a
ith
r
a
I
n
d
a
va
r
a
V
e
n
ka
tesh
a
g
o
w
d
a
)
903
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
C
h
aith
r
a
I
n
d
av
a
r
a
Ven
k
atesh
ag
o
wd
a
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
R
o
o
p
ash
r
ee
Hejjajji
R
an
g
an
ath
ash
ar
m
a
✓
✓
✓
✓
✓
✓
✓
Yo
g
ee
s
h
Am
b
alag
er
e
C
h
an
d
r
ash
ek
ar
aiah
✓
✓
✓
✓
✓
✓
✓
✓
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
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.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
in
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
av
aila
b
le
f
r
o
m
th
e
c
o
r
r
esp
o
n
d
in
g
au
th
o
r
u
p
o
n
r
ea
s
o
n
ab
le
r
eq
u
est.
RE
F
E
R
E
NC
E
S
[
1
]
G
.
R
u
f
f
i
n
i
,
F
.
C
a
s
t
a
l
d
o
,
E.
L
o
p
e
z
-
S
o
l
a
,
R
.
S
a
n
c
h
e
z
-
To
d
o
,
a
n
d
J
.
V
o
h
r
y
z
e
k
,
“
T
h
e
a
l
g
o
r
i
t
h
mi
c
a
g
e
n
t
p
e
r
sp
e
c
t
i
v
e
a
n
d
c
o
m
p
u
t
a
t
i
o
n
a
l
n
e
u
r
o
p
s
y
c
h
i
a
t
r
y
:
F
r
o
m e
t
i
o
l
o
g
y
t
o
a
d
v
a
n
c
e
d
t
h
e
r
a
p
y
i
n
maj
o
r
d
e
p
r
e
ss
i
v
e
d
i
so
r
d
e
r
,
”
En
t
r
o
p
y
,
v
o
l
.
2
6
,
n
o
.
1
1
,
p
p
.
1
–
7
1
,
N
o
v
.
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
e
2
6
1
1
0
9
5
3
.
[
2
]
A
.
A
l
b
e
r
t
i
e
t
a
l
.
,
“
F
a
c
t
o
r
s
a
ss
o
c
i
a
t
e
d
w
i
t
h
t
h
e
d
e
v
e
l
o
p
m
e
n
t
o
f
d
e
p
r
e
ss
i
o
n
a
n
d
t
h
e
i
n
f
l
u
e
n
c
e
o
f
o
b
e
s
i
t
y
o
n
d
e
p
r
e
s
si
v
e
d
i
s
o
r
d
e
r
s
:
a
n
a
r
r
a
t
i
v
e
r
e
v
i
e
w
,
”
B
i
o
m
e
d
i
c
i
n
e
s
,
v
o
l
.
1
2
,
n
o
.
9
,
p
p
.
1
–
2
7
,
S
e
p
.
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
b
i
o
m
e
d
i
c
i
n
e
s
1
2
0
9
1
9
9
4
.
[
3
]
M
.
T.
A
z
i
z
e
t
a
l
.
,
“
Te
x
t
u
a
l
a
n
d
n
u
meric
a
l
d
a
t
a
f
u
si
o
n
f
o
r
d
e
p
r
e
ssi
o
n
d
e
t
e
c
t
i
o
n
:
a
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
f
r
a
mew
o
r
k
,
”
I
n
d
o
n
e
s
i
a
n
J
o
u
rn
a
l
o
f
E
l
e
c
t
ri
c
a
l
E
n
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
.
3
8
,
n
o
.
2
,
p
p
.
1
2
3
1
–
1
2
4
4
,
M
a
y
2
0
2
5
,
d
o
i
:
1
0
.
1
1
5
9
1
/
i
j
e
e
c
s.
v
3
8
.
i
2
.
p
p
1
2
3
1
-
1
2
4
4
.
[
4
]
N
.
A
.
A
.
R
e
z
a
l
,
N
.
Y
a
h
y
a
,
F
.
D
.
A
z
man
,
M
.
A
.
A
.
H
a
n
a
p
i
,
A
.
A
.
A
z
i
z
,
a
n
d
D
.
M
.
K
h
a
n
,
“
M
a
j
o
r
d
e
p
r
e
ss
i
v
e
d
i
s
o
r
d
e
r
d
e
t
e
c
t
i
o
n
u
si
n
g
e
f
f
e
c
t
i
v
e
c
o
n
n
e
c
t
i
v
i
t
y
o
f
EE
G
si
g
n
a
l
s
a
n
d
d
e
e
p
l
e
a
r
n
i
n
g
t
r
a
n
sf
o
r
mer
mo
d
e
l
,
”
2
0
2
4
I
E
EE
S
y
m
p
o
si
u
m
o
n
I
n
d
u
st
r
i
a
l
El
e
c
t
r
o
n
i
c
s
& A
p
p
l
i
c
a
t
i
o
n
s (I
S
I
EA),
K
u
a
l
a
L
u
m
p
u
r,
M
a
l
a
y
si
a
,
p
p
.
1
–
6
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
I
S
I
EA
6
1
9
2
0
.
2
0
2
4
.
1
0
6
0
7
2
2
4
.
[
5
]
A
n
a
n
d
,
Y
.
S
h
a
r
ma
,
V
.
J
a
i
n
,
a
n
d
S
.
T
a
r
w
a
n
i
,
“
E
n
se
mb
l
e
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
mo
d
e
l
f
o
r
p
r
e
d
i
c
t
i
n
g
p
o
st
p
a
r
t
u
m
d
e
p
r
e
ssi
o
n
d
i
s
o
r
d
e
r
,
”
2
0
2
4
I
EE
E
R
e
g
i
o
n
1
0
S
y
m
p
o
s
i
u
m
(
T
E
N
S
Y
MP
)
,
N
e
w
D
e
l
h
i
,
I
n
d
i
a
,
p
p
.
1
–
6
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
TEN
S
Y
M
P
6
1
1
3
2
.
2
0
2
4
.
1
0
7
5
2
3
0
5
.
[
6
]
S
.
J.
P
i
n
t
o
a
n
d
M
.
P
a
r
e
n
t
e
,
“
C
o
mp
r
e
h
e
n
si
v
e
r
e
v
i
e
w
o
f
d
e
p
r
e
ss
i
o
n
d
e
t
e
c
t
i
o
n
t
e
c
h
n
i
q
u
e
s
b
a
se
d
o
n
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
,
”
Ap
p
l
i
c
a
t
i
o
n
o
f
so
f
t
c
o
m
p
u
t
i
n
g
,
v
o
l
.
2
8
,
p
p
.
1
0
7
0
1
–
1
0
7
2
5
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s
0
0
5
0
0
-
0
2
4
-
0
9
8
6
2
-
1.
[
7
]
Q
.
D
e
n
g
,
S
.
Lu
z
,
a
n
d
S
.
d
e
l
a
F
u
e
n
t
e
G
a
r
c
i
a
,
“
A
f
r
a
me
-
b
a
se
d
a
t
t
e
n
t
i
o
n
i
n
t
e
r
p
r
e
t
a
t
i
o
n
m
e
t
h
o
d
f
o
r
r
e
l
e
v
a
n
t
a
c
o
u
st
i
c
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
i
n
l
o
n
g
s
p
e
e
c
h
d
e
p
r
e
ss
i
o
n
d
e
t
e
c
t
i
o
n
,
”
a
rX
i
v
e
-
p
r
i
n
t
s
.
p
p
.
1
–
5
,
2
0
2
4
,
d
o
i
:
1
0
.
4
8
5
5
0
/
a
r
X
i
v
.
2
4
0
6
.
0
3
1
3
8
[
8
]
U
.
A
h
me
d
,
G
.
S
r
i
v
a
s
t
a
v
a
,
U
.
Y
u
n
,
a
n
d
J.
C
.
-
W
.
L
i
n
,
“
EA
N
D
C
:
a
n
e
x
p
l
a
i
n
a
b
l
e
a
t
t
e
n
t
i
o
n
n
e
t
w
o
r
k
b
a
se
d
d
e
e
p
a
d
a
p
t
i
v
e
c
l
u
st
e
r
i
n
g
mo
d
e
l
f
o
r
me
n
t
a
l
h
e
a
l
t
h
t
r
e
a
t
me
n
t
,
”
F
u
t
u
r
e
G
e
n
e
r
a
t
i
o
n
C
o
m
p
u
t
e
r
S
y
s
t
e
m
s
,
v
o
l
.
1
3
0
,
p
p
.
1
0
6
–
1
1
3
,
M
a
y
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
f
u
t
u
r
e
.
2
0
2
1
.
1
2
.
0
0
8
.
[
9
]
Y
.
Li
u
,
C
.
P
u
,
S
.
X
i
a
,
D
.
D
e
n
g
,
X
.
W
a
n
g
,
a
n
d
M
.
Li
,
“
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
e
s
f
o
r
d
i
a
g
n
o
s
i
n
g
d
e
p
r
e
ss
i
o
n
u
s
i
n
g
E
EG
:
a
r
e
v
i
e
w
,
”
T
ra
n
s
l
a
t
i
o
n
a
l
N
e
u
ro
s
c
i
e
n
c
e
,
v
o
l
.
1
3
,
n
o
.
1
,
p
p
.
2
2
4
–
2
3
5
,
A
u
g
.
2
0
2
2
,
d
o
i
:
1
0
.
1
5
1
5
/
t
n
sc
i
-
2
0
2
2
-
0
2
3
4
.
[
1
0
]
L.
B
e
n
d
e
b
a
n
e
,
Z
.
La
b
o
u
d
i
,
A
.
S
a
i
g
h
i
,
H
.
A
l
-
T
a
r
a
w
n
e
h
,
A
.
O
u
a
n
n
a
s,
a
n
d
G
.
G
r
a
ssi
,
“
A
m
u
l
t
i
-
c
l
a
ss
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
f
o
r
e
a
r
l
y
d
e
t
e
c
t
i
o
n
o
f
d
e
p
r
e
ss
i
v
e
a
n
d
a
n
x
i
e
t
y
d
i
so
r
d
e
r
s
u
si
n
g
Tw
i
t
t
e
r
d
a
t
a
,
”
Al
g
o
ri
t
h
m
s
,
v
o
l
.
1
6
,
n
o
.
1
2
,
p
p
.
1
–
2
4
,
N
o
v
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
a
1
6
1
2
0
5
4
3
.
[
1
1
]
S
.
A
l
e
e
m,
N
.
u
l
H
u
d
a
,
R
.
A
m
i
n
,
S
.
K
h
a
l
i
d
,
S
.
S
.
A
l
s
h
a
mr
a
n
i
,
a
n
d
A
.
A
l
sh
e
h
r
i
,
“
M
a
c
h
i
n
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
ms
f
o
r
d
e
p
r
e
ss
i
o
n
:
d
i
a
g
n
o
si
s,
i
n
si
g
h
t
s
,
a
n
d
r
e
s
e
a
r
c
h
d
i
r
e
c
t
i
o
n
s
,
”
El
e
c
t
ro
n
i
c
s
,
v
o
l
.
1
1
,
n
o
.
7
,
p
p
.
1
–
2
0
,
M
a
r
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
e
l
e
c
t
r
o
n
i
c
s1
1
0
7
1
1
1
1
.
[
1
2
]
K
.
E
l
n
a
g
g
a
r
,
M
.
E
l
-
G
a
y
a
r
,
a
n
d
M
.
E
l
mo
g
y
,
“
D
e
p
r
e
ssi
o
n
d
e
t
e
c
t
i
o
n
a
n
d
d
i
a
g
n
o
s
i
s
b
a
s
e
d
o
n
e
l
e
c
t
r
o
e
n
c
e
p
h
a
l
o
g
r
a
m
(
EEG
)
a
n
a
l
y
s
i
s:
a
sy
st
e
mat
i
c
r
e
v
i
e
w
,
”
D
i
a
g
n
o
s
t
i
c
s
,
v
o
l
.
1
5
,
n
o
.
2
,
p
p
.
1
–
2
8
,
Ja
n
.
2
0
2
5
,
d
o
i
:
1
0
.
3
3
9
0
/
d
i
a
g
n
o
s
t
i
c
s1
5
0
2
0
2
1
0
.
[
1
3
]
W
.
U
l
l
a
h
,
P
.
O
l
i
v
e
i
r
a
-
S
i
l
v
a
,
M
.
N
a
w
a
z
,
R
.
M
.
Zu
l
q
a
r
n
a
i
n
,
I
.
S
i
d
d
i
q
u
e
,
a
n
d
M
.
S
a
l
l
a
h
,
“
I
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
d
e
p
r
e
s
si
n
g
t
w
e
e
t
s
u
si
n
g
n
a
t
u
r
a
l
l
a
n
g
u
a
g
e
p
r
o
c
e
ssi
n
g
a
n
d
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
:
a
p
p
l
i
c
a
t
i
o
n
o
f
g
r
e
y
r
e
l
a
t
i
o
n
a
l
g
r
a
d
e
s,”
J
o
u
r
n
a
l
o
f
R
a
d
i
a
t
i
o
n
Re
s
e
a
r
c
h
a
n
d
Ap
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
1
8
,
n
o
.
1
,
p
p
.
1
–
1
3
,
M
a
r
.
2
0
2
5
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
r
r
a
s
.
2
0
2
5
.
1
0
1
2
9
9
.
[
1
4
]
C
.
H
.
Esp
i
n
o
-
S
a
l
i
n
a
s
e
t
a
l
.
,
“
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
f
o
r
d
e
p
r
e
ss
i
o
n
a
n
d
sc
h
i
z
o
p
h
r
e
n
i
a
d
e
t
e
c
t
i
o
n
,
”
D
i
a
g
n
o
s
t
i
c
s
,
v
o
l
.
1
5
,
n
o
.
3
,
p
p
.
1
–
2
2
,
Ja
n
.
2
0
2
5
,
d
o
i
:
1
0
.
3
3
9
0
/
d
i
a
g
n
o
s
t
i
c
s1
5
0
3
0
3
1
9
.
[
1
5
]
C
.
H
.
Es
p
i
n
o
-
S
a
l
i
n
a
s
e
t
a
l
.
,
“
Tw
o
-
d
i
men
s
i
o
n
a
l
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
f
o
r
d
e
p
r
e
ssi
o
n
e
p
i
so
d
e
d
e
t
e
c
t
i
o
n
i
n
r
e
a
l
t
i
me
u
s
i
n
g
mo
t
o
r
a
c
t
i
v
i
t
y
t
i
me
seri
e
s
o
f
t
h
e
d
e
p
r
e
sj
o
n
d
a
t
a
se
t
,
”
B
i
o
e
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
9
,
n
o
.
9
,
p
p
.
1
–
1
7
,
S
e
p
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
b
i
o
e
n
g
i
n
e
e
r
i
n
g
9
0
9
0
4
5
8
.
[
1
6
]
C
.
L
i
n
e
t
a
l
.
,
“
A
u
t
o
m
a
t
i
c
d
i
a
g
n
o
s
i
s
o
f
l
a
t
e
-
l
i
f
e
d
e
p
r
e
ss
i
o
n
b
y
3
D
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
a
n
d
c
r
o
ss
-
s
a
m
p
l
e
E
n
t
r
o
p
y
a
n
a
l
y
s
i
s
f
r
o
m
r
e
s
t
i
n
g
-
s
t
a
t
e
f
M
R
I
,
”
Br
a
i
n
I
m
a
g
i
n
g
a
n
d
Be
h
a
v
i
o
r
,
v
o
l
.
1
7
,
n
o
.
1
,
p
p
.
1
2
5
–
1
3
5
,
F
e
b
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
1
6
8
2
-
022
-
00748
-
0.
[
1
7
]
M
.
N
a
r
i
g
i
n
a
,
A
.
R
o
ma
n
o
v
s,
a
n
d
Y
.
M
e
r
k
u
r
y
e
v
,
“
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
-
b
a
s
e
d
d
i
g
i
t
a
l
d
i
a
g
n
o
st
i
c
t
o
o
l
f
o
r
t
h
e
i
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
p
sy
c
h
o
s
o
ma
t
i
c
i
l
l
n
e
ss
e
s,
”
A
l
g
o
r
i
t
h
m
s
,
v
o
l
.
1
7
,
n
o
.
8
,
p
p
.
1
–
1
6
,
Ju
l
.
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
a
1
7
0
8
0
3
2
9
.
[
1
8
]
A
.
B
.
K
．
R
e
t
a
l
.
,
“
A
m
u
l
t
i
-
d
i
m
e
n
s
i
o
n
a
l
h
y
b
r
i
d
C
N
N
-
B
i
L
S
T
M
f
r
a
m
e
w
o
r
k
f
o
r
e
p
i
l
e
p
t
i
c
s
e
i
z
u
r
e
d
e
t
e
c
t
i
o
n
u
s
i
n
g
e
l
e
c
t
r
o
e
n
c
e
p
h
a
l
o
g
r
a
m
s
i
g
n
a
l
s
c
r
u
t
i
n
y
,
”
S
y
s
t
e
m
s
a
n
d
S
o
f
t
C
o
m
p
u
t
i
n
g
,
v
o
l
.
5
,
p
p
.
1
–
1
4
,
D
e
c
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
s
a
s
c
.
2
0
2
3
.
2
0
0
0
6
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
7
0
8
I
n
t J E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
1
6
,
No
.
2
,
Ap
r
il
20
2
6
:
8
9
5
-
904
904
[
1
9
]
H
.
Tu
f
a
i
l
,
S
.
M
.
C
h
e
e
ma,
M
.
A
l
i
,
I
.
M
.
P
i
r
e
s,
a
n
d
N
.
M
.
G
a
r
c
i
a
,
“
D
e
p
r
e
ss
i
o
n
d
e
t
e
c
t
i
o
n
w
i
t
h
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:
a
s
t
e
p
t
o
w
a
r
d
s
i
m
p
r
o
v
e
d
m
e
n
t
a
l
h
e
a
l
t
h
c
a
r
e
,
”
Pr
o
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
2
2
4
,
p
p
.
5
4
4
–
5
4
9
,
D
e
c
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s
.
2
0
2
3
.
0
9
.
0
7
9
.
[
2
0
]
W
.
X
i
e
e
t
a
l
.
,
“
M
u
l
t
i
mo
d
a
l
f
u
s
i
o
n
d
i
a
g
n
o
si
s
o
f
d
e
p
r
e
ss
i
o
n
a
n
d
a
n
x
i
e
t
y
b
a
se
d
o
n
C
N
N
-
LSTM
mo
d
e
l
,
”
C
o
m
p
u
t
e
ri
z
e
d
Me
d
i
c
a
l
I
m
a
g
i
n
g
a
n
d
G
r
a
p
h
i
c
s
,
v
o
l
.
1
0
2
,
p
p
.
1
–
7
,
D
e
c
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
m
p
m
e
d
i
ma
g
.
2
0
2
2
.
1
0
2
1
2
8
.
[
2
1
]
V
a
n
d
a
n
a
,
N
.
M
a
r
r
i
w
a
l
a
,
a
n
d
D
.
C
h
a
u
d
h
a
r
y
,
“
A
h
y
b
r
i
d
mo
d
e
l
f
o
r
d
e
p
r
e
ssi
o
n
d
e
t
e
c
t
i
o
n
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
,
”
Me
a
s
u
reme
n
t
:
S
e
n
so
rs
,
v
o
l
.
2
5
,
p
p
.
1
–
1
0
,
F
e
b
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
me
a
se
n
.
2
0
2
2
.
1
0
0
5
8
7
.
[
2
2
]
D
.
P
a
k
k
a
t
t
i
l
a
n
d
R
.
S
r
i
D
e
v
i
,
“
E
mp
o
w
e
r
i
n
g
m
e
n
t
a
l
h
e
a
l
t
h
:
C
N
N
a
n
d
LST
M
f
u
s
i
o
n
f
o
r
t
i
me
l
y
d
e
p
r
e
ssi
o
n
d
e
t
e
c
t
i
o
n
i
n
w
o
m
e
n
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
j
o
u
r
n
a
l
o
f
e
l
e
c
t
r
i
c
a
l
a
n
d
c
o
m
p
u
t
e
r
e
n
g
i
n
e
e
ri
n
g
sys
t
e
m
s
,
v
o
l
.
1
5
,
n
o
.
8
,
p
p
.
6
3
1
–
6
4
0
,
S
e
p
.
2
0
2
4
,
d
o
i
:
1
0
.
3
2
9
8
5
/
i
j
e
c
e
s.
1
5
.
8
.
1
.
[
2
3
]
S
.
M
.
P
a
d
ma
j
a
e
t
a
l
.
,
“
D
e
p
r
e
ssi
o
n
d
e
t
e
c
t
i
o
n
i
n
s
o
c
i
a
l
me
d
i
a
u
si
n
g
N
LP
a
n
d
h
y
b
r
i
d
d
e
e
p
l
e
a
r
n
i
n
g
m
o
d
e
l
s,”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
A
d
v
a
n
c
e
d
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
1
6
,
n
o
.
2
,
p
p
.
1
0
7
1
–
1
0
8
0
,
D
e
c
.
2
0
2
5
,
d
o
i
:
1
0
.
1
4
5
6
9
/
I
JA
C
S
A
.
2
0
2
5
.
0
1
6
0
2
1
0
6
.
[
2
4
]
F
.
I
.
K
u
r
n
i
a
d
i
,
N
.
L
.
P
.
S
.
P
.
P
a
r
a
mi
t
a
,
E
.
F
.
A
.
S
i
h
o
t
a
n
g
,
M
.
S
.
A
n
g
g
r
e
a
i
n
y
,
a
n
d
R
.
Zh
a
n
g
,
“
B
ER
T
a
n
d
R
o
B
E
R
Ta
mo
d
e
l
s
f
o
r
e
n
h
a
n
c
e
d
d
e
t
e
c
t
i
o
n
o
f
d
e
p
r
e
ssi
o
n
i
n
so
c
i
a
l
me
d
i
a
t
e
x
t
,
”
Pro
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
2
4
5
,
p
p
.
2
0
2
–
2
0
9
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
p
r
o
c
s
.
2
0
2
4
.
1
0
.
2
4
4
.
[
2
5
]
J.
P
h
i
l
i
p
T
h
e
k
k
e
k
a
r
a
,
S
.
Y
o
n
g
c
h
a
r
e
o
n
,
a
n
d
V
.
L
i
e
s
a
p
u
t
r
a
,
“
A
n
a
t
t
e
n
t
i
o
n
-
b
a
se
d
C
N
N
-
B
i
LST
M
m
o
d
e
l
f
o
r
d
e
p
r
e
s
si
o
n
d
e
t
e
c
t
i
o
n
o
n
so
c
i
a
l
me
d
i
a
t
e
x
t
,
”
Ex
p
e
r
t
S
y
s
t
e
m
s
w
i
t
h
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
2
4
9
,
p
p
.
1
–
9
,
S
e
p
.
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
e
sw
a
.
2
0
2
4
.
1
2
3
8
3
4
.
[
2
6
]
A
.
N
a
d
e
e
m
,
M
.
N
a
v
e
e
d
,
M
.
I
sl
a
m
S
a
t
t
i
,
H
.
A
f
z
a
l
,
T.
A
h
m
a
d
,
a
n
d
K
.
-
I
.
K
i
m
,
“
D
e
p
r
e
ss
i
o
n
d
e
t
e
c
t
i
o
n
b
a
se
d
o
n
h
y
b
r
i
d
d
e
e
p
l
e
a
r
n
i
n
g
S
S
C
L
f
r
a
mew
o
r
k
u
si
n
g
s
e
l
f
-
a
t
t
e
n
t
i
o
n
mec
h
a
n
i
sm
:
A
n
a
p
p
l
i
c
a
t
i
o
n
t
o
so
c
i
a
l
n
e
t
w
o
r
k
i
n
g
d
a
t
a
,
”
S
e
n
s
o
rs
,
v
o
l
.
2
2
,
n
o
.
2
4
,
p
p
.
1
–
2
8
,
D
e
c
.
2
0
2
2
,
d
o
i
:
1
0
.
3
3
9
0
/
s
2
2
2
4
9
7
7
5
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Cha
ith
r
a
I
n
d
a
v
a
r
a
Ve
n
k
a
te
sh
a
g
o
w
d
a
re
c
e
iv
e
d
th
e
m
a
ste
r’s
d
e
g
re
e
in
c
o
m
p
u
ter
sc
ien
c
e
a
n
d
e
n
g
in
e
e
ri
n
g
fr
o
m
Visv
e
sv
a
ra
y
a
Tec
h
n
o
lo
g
ica
l
Un
i
v
e
rsity
,
Be
lag
a
v
i,
Ka
rn
a
tak
a
,
In
d
ia
i
n
2
0
1
3
a
n
d
is
c
u
rre
n
tl
y
w
o
rk
i
n
g
t
o
wa
rd
s
P
h
.
D.
d
e
g
re
e
.
S
h
e
jo
in
e
d
th
e
c
o
m
p
u
ter
sc
ien
c
e
a
n
d
e
n
g
i
n
e
e
rin
g
d
e
p
a
rtme
n
t,
Ad
ich
u
n
c
h
a
n
a
g
ir
i
In
stit
u
te
o
f
Tec
h
n
o
lo
g
y
,
Ch
ik
k
a
m
a
g
a
lu
ru
a
s
a
n
a
ss
istan
t
p
ro
fe
ss
o
r,
in
2
0
1
5
.
He
r
re
se
a
rc
h
in
tere
sts
i
n
c
lu
d
e
m
a
c
h
in
e
lea
rn
in
g
,
d
e
e
p
lea
rn
i
n
g
,
a
rti
f
icia
l
in
telli
g
e
n
c
e
,
se
n
ti
m
e
n
t
a
n
a
l
y
sis
a
n
d
larg
e
lan
g
u
a
g
e
m
o
d
e
ls.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
c
h
a
it
h
ra
.
iv
@
g
m
a
il
.
c
o
m
.
Ro
o
p
a
shre
e
H
e
jja
jji
Ra
n
g
a
n
a
t
h
a
sha
r
m
a
c
o
m
p
lete
d
B.
E
.
(E
&
C)
a
n
d
M
.
Tec
h
.
(CS
&
E)
fro
m
VTU,
Be
lag
a
v
i,
Ka
rn
a
tak
a
,
In
d
ia
a
n
d
P
h
.
D.
fro
m
CHRIST
(De
e
m
e
d
to
b
e
Un
iv
e
rsity
)
Be
n
g
a
lu
r
u
,
Ka
rn
a
tak
a
,
In
d
ia.
S
h
e
h
a
s a
ro
u
n
d
1
3
y
e
a
rs o
f
in
d
u
strial
e
x
p
e
rien
c
e
a
n
d
5
y
e
a
rs
o
f
tea
c
h
i
n
g
e
x
p
e
rien
c
e
.
S
h
e
is
p
re
se
n
tl
y
wo
r
k
in
g
a
s
a
p
ro
fe
ss
o
r
a
n
d
h
e
a
d
o
f
,
th
e
d
e
p
a
rtme
n
t
o
f
a
rti
ficia
l
in
telli
g
e
n
c
e
a
n
d
d
a
ta
sc
ien
c
e
a
t
G
S
S
S
IET
W,
M
y
s
u
ru
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
ro
o
p
a
sh
re
e
h
r
@g
ss
s.e
d
u
.
i
n
.
Y
o
g
e
e
s
h
A
m
b
a
l
a
g
e
r
e
C
h
a
n
d
r
a
sh
e
k
a
r
a
i
a
h
h
a
s
c
o
m
p
l
e
te
d
B
.
E
,
M
.
T
e
c
h
,
a
n
d
P
h
.
D
.
f
r
o
m
V
i
s
v
e
s
v
a
r
a
y
a
Te
c
h
n
o
l
o
g
i
c
a
l
U
n
i
v
e
r
s
i
t
y
B
e
l
a
g
a
v
i
,
Ka
r
n
a
t
a
k
a
,
I
n
d
i
a
.
C
u
r
re
n
t
l
y
w
o
r
k
i
n
g
a
s
a
n
a
s
s
i
s
t
a
n
t
p
r
o
f
e
ss
o
r
i
n
C
S
&
E
,
G
o
v
e
r
n
m
e
n
t
E
n
g
i
n
e
e
r
i
n
g
C
o
l
l
e
g
e
,
C
h
a
m
a
r
a
j
a
n
a
g
a
r
,
K
a
r
n
a
ta
k
a
,
I
n
d
i
a
.
H
i
s
a
r
e
a
o
f
i
n
t
e
r
e
s
t
i
s
w
i
r
e
le
s
s
se
n
s
o
r
n
e
t
w
o
r
k
,
I
o
T
a
n
d
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
.
H
e
c
a
n
b
e
c
o
n
t
a
c
te
d
a
t
e
m
a
i
l
:
y
o
g
e
e
s
h
1
3
@
g
m
a
i
l
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.