I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l
I
nte
llig
ence
(
I
J
-
AI
)
Vo
l.
13
,
No
.
4
,
Dec
em
b
er
2
0
2
4
,
p
p
.
4
6
7
5
~
4
6
8
7
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
13
.i
4
.
p
p
4
6
7
5
-
4
6
8
7
4675
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Sentimen
t
-
el
ectr
o
encepha
lo
g
ra
m
f
usio
n f
o
r efficien
t
product
revi
ew pre
diction
using
co
rrelatio
n
-
ba
sed deep
learn
ing
neural
network
Ra
hu
l K
um
a
r
Sh
a
rm
a
,
Arv
i
nd
Da
g
ur
S
c
h
o
o
l
o
f
C
o
m
p
u
t
i
n
g
S
c
i
e
n
c
e
a
n
d
En
g
i
n
e
e
r
i
n
g
,
G
a
l
g
o
t
i
a
s
U
n
i
v
e
r
si
t
y
,
G
r
e
a
t
e
r
N
o
i
d
a
,
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
Dec
8
,
2
0
2
3
R
ev
is
ed
Mar
29
,
2
0
2
4
Acc
ep
ted
Ap
r
17
,
2
0
2
4
Va
rio
u
s
tec
h
n
iq
u
e
s
h
a
v
e
b
e
e
n
p
r
o
p
o
se
d
a
n
d
im
p
lem
e
n
ted
i
n
p
re
v
io
u
s
w
o
rk
fo
r
se
n
ti
m
e
n
t
a
n
a
l
y
sis
p
re
d
icti
o
n
.
Ho
we
v
e
r,
a
c
h
iev
i
n
g
sa
ti
sfa
c
to
ry
q
u
a
li
t
y
o
f
d
e
sc
rip
ti
o
n
a
n
d
fa
u
l
t
p
re
d
ictio
n
re
m
a
in
s
a
c
h
a
ll
e
n
g
i
n
g
tas
k
.
T
o
o
v
e
rc
o
m
e
th
e
se
li
m
it
a
ti
o
n
s,
t
h
is
st
u
d
y
p
ro
p
o
se
s
a
n
e
fficie
n
t
p
re
d
ictio
n
tec
h
n
iq
u
e
t
h
a
t
u
ti
li
z
e
s
se
n
ti
m
e
n
t
a
n
a
l
y
sis
o
f
p
ro
d
u
c
t
re
v
iew
s
a
n
d
e
lec
tro
e
n
c
e
p
h
a
lo
g
ra
m
(
EEG
)
sig
n
a
ls
u
sin
g
c
o
rre
latio
n
-
b
a
se
d
d
e
e
p
lea
rn
in
g
n
e
u
ra
l
n
e
two
rk
(
CDN
N
)
.
Th
e
st
u
d
y
e
m
p
l
o
y
s
two
ty
p
e
s
o
f
d
a
tas
e
ts:
EE
G
sig
n
a
ls
a
n
d
Am
a
z
o
n
p
r
o
d
u
c
t
re
v
iew
s.
Du
rin
g
t
h
e
p
re
-
p
ro
c
e
ss
in
g
p
h
a
se
,
EE
G
sig
n
a
ls
u
n
d
e
rg
o
n
o
rm
a
li
z
a
ti
o
n
,
wh
il
e
Am
a
z
o
n
p
r
o
d
u
c
t
re
v
iew
s
u
n
d
e
rg
o
to
k
e
n
iza
t
io
n
,
st
o
p
wo
rd
re
m
o
v
a
l,
a
n
d
we
ig
h
ti
n
g
fa
c
to
r
a
p
p
li
c
a
ti
o
n
to
c
o
n
v
e
rt
u
n
str
u
c
tu
re
d
d
a
t
a
in
to
a
stru
c
tu
re
d
f
o
rm
a
t.
S
u
b
se
q
u
e
n
tl
y
,
th
e
p
re
-
p
r
o
c
e
ss
e
d
EE
G
si
g
n
a
ls
a
n
d
re
v
iew
s
a
re
a
n
a
ly
z
e
d
to
e
x
trac
t
fe
a
tu
re
s
li
k
e
e
m
o
ti
o
n
,
d
e
m
o
g
ra
p
h
ic
in
fo
rm
a
ti
o
n
,
p
e
rso
n
a
li
t
y
traits,
a
n
d
se
n
ti
m
e
n
t.
T
h
e
se
fe
a
tu
re
s
a
re
th
e
n
e
m
p
lo
y
e
d
i
n
se
n
t
ime
n
t
a
n
a
ly
sis
v
ia
a
n
e
n
tro
p
y
-
b
a
se
d
d
e
e
p
-
lea
rn
i
n
g
n
e
u
ra
l
n
e
two
rk
.
Th
e
p
ro
p
o
se
d
CDN
N
u
ti
li
z
e
s
th
e
g
ra
ss
h
o
p
p
e
r
o
p
ti
m
iza
ti
o
n
a
lg
o
rit
h
m
(EG
OA
)
to
o
p
ti
m
ize
h
y
p
e
rp
a
ra
m
e
ters
fo
r
e
a
c
h
lay
e
r.
C
o
m
p
a
ra
ti
v
e
p
e
rfo
rm
a
n
c
e
a
ss
e
ss
m
e
n
t
a
g
a
in
st
e
sta
b
li
sh
e
d
m
e
th
o
d
s
li
k
e
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
(
CNN
)
,
lo
n
g
sh
o
r
t
-
term
m
e
m
o
ry
(
L
S
TM
)
,
m
u
l
ti
c
las
s
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
(
M
-
S
VM
)
,
a
n
d
b
id
irec
ti
o
n
a
l
e
n
c
o
d
e
r
re
p
re
se
n
tat
io
n
s
f
ro
m
tran
sfo
rm
e
rs
(
BERT
)
is
c
o
n
d
u
c
ted
,
a
n
d
t
h
e
re
su
lt
s
a
re
e
v
a
lu
a
ted
.
Ex
p
e
rime
n
tal
re
su
lt
re
v
e
a
l
t
h
a
t
t
h
e
p
ro
p
o
se
d
sy
ste
m
o
u
tp
e
rf
o
rm
s
trad
it
io
n
a
l
a
p
p
ro
a
c
h
e
s.
K
ey
w
o
r
d
s
:
C
o
r
r
elatio
n
-
b
ased
d
ee
p
lear
n
i
n
g
n
eu
r
al
n
etwo
r
k
E
n
tr
o
p
y
v
alu
e
Gr
ass
h
o
p
p
er
o
p
tim
is
atio
n
alg
o
r
ith
m
Sen
tim
en
t a
n
aly
s
is
W
eig
h
tin
g
f
ac
to
r
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
:
R
ah
u
l K
u
m
ar
Sh
ar
m
a
Sch
o
o
l o
f
C
o
m
p
u
tin
g
Scien
ce
an
d
E
n
g
in
ee
r
in
g
,
Galg
o
tias
Un
iv
er
s
ity
Gr
ea
ter
No
id
a,
I
n
d
ia
E
m
ail:
r
ah
u
lp
cc
s
1
9
8
8
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
Peo
p
le
n
o
w
u
s
e
a
v
ar
iety
o
f
e
-
co
m
m
er
ce
web
s
ites
to
ex
ch
an
g
e
item
s
b
ec
au
s
e
th
e
co
m
m
er
cial
s
ec
to
r
h
as
m
ain
ly
m
o
v
ed
to
o
n
lin
e
p
l
atf
o
r
m
s
.
B
ef
o
r
e
m
ak
in
g
a
p
u
r
ch
ase,
an
aly
zin
g
p
r
o
d
u
cts
h
as
b
ec
o
m
e
a
co
m
m
o
n
p
r
ac
tice.
Ad
d
itio
n
ally
,
c
u
s
to
m
er
s
ar
e
in
c
r
ea
s
in
g
ly
r
ely
in
g
o
n
r
ev
iews
to
m
ak
e
th
eir
p
u
r
c
h
a
s
in
g
d
ec
is
io
n
s
[
1
]
.
B
u
s
in
ess
lead
er
s
ca
n
lear
n
m
o
r
e
a
b
o
u
t
h
o
w
co
n
s
u
m
er
s
f
ee
l
ab
o
u
t
th
eir
p
r
o
d
u
cts
b
y
u
s
in
g
th
e
r
elativ
ely
n
ew
ac
ad
em
ic
ar
ea
o
f
s
en
tim
en
t
an
aly
s
is
,
wh
ich
an
aly
s
es
tex
t
f
r
o
m
r
ev
iews,
b
lo
g
s
,
s
u
r
v
ey
r
ep
lie
s
,
an
d
o
th
er
s
o
u
r
ce
s
.
I
n
m
a
r
k
etin
g
an
d
cu
s
to
m
e
r
s
er
v
ice
ap
p
licatio
n
s
,
th
is
k
in
d
o
f
an
aly
s
is
is
f
r
eq
u
e
n
tly
u
tili
ze
d
to
co
m
p
r
eh
e
n
d
t
h
e
v
o
ice
o
f
t
h
e
co
n
s
u
m
er
[
2
]
,
[
3
]
.
Sen
tim
en
t
an
aly
s
is
also
k
n
o
wn
as
em
o
tio
n
an
aly
s
is
o
r
o
p
in
io
n
m
in
in
g
,
in
v
o
lv
es
id
en
tify
in
g
k
ey
attr
ib
u
tes
with
in
u
n
s
tr
u
ct
u
r
ed
,
r
aw
tex
t
c
o
m
m
o
n
ly
f
o
u
n
d
o
n
s
o
cial
m
ed
ia
p
latf
o
r
m
s
a
n
d
n
etwo
r
k
s
[
4
]
,
[
5
]
.
I
n
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
13
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
4
6
7
5
-
4
6
8
7
4676
th
e
class
if
icatio
n
p
r
o
ce
s
s
,
s
en
tim
en
t
an
aly
s
is
p
lay
s
a
cr
u
cial
r
o
le
in
o
p
i
n
io
n
m
in
in
g
,
as
it
i
n
v
o
lv
es
e
v
alu
atin
g
cu
s
to
m
er
o
p
i
n
io
n
s
ab
o
u
t
a
p
r
o
d
u
ct
an
d
ca
teg
o
r
izin
g
th
em
as
p
o
s
itiv
e
o
r
n
eg
ativ
e
f
ac
to
r
s
f
o
r
an
aly
s
is
[
6
]
,
[
7
]
.
Star
r
atin
g
s
ar
e
a
p
o
p
u
la
r
wa
y
t
o
q
u
ick
ly
c
o
n
v
e
y
th
e
em
o
t
io
n
al
to
n
e
o
f
a
r
ev
iew
in
c
o
n
s
u
m
er
ass
ess
m
en
ts
.
Nev
er
th
eless
,
s
tar
r
atin
g
s
p
o
s
s
ess
ce
r
tain
co
n
s
tr
ain
ts
.
Occ
a
s
io
n
ally
,
th
ey
m
ay
e
x
h
ib
it
b
ia
s
,
lead
in
g
to
r
atin
g
s
th
at
d
o
n
o
t
c
o
n
s
is
ten
tly
r
ef
lect
th
e
g
en
u
in
e
e
m
o
tio
n
co
n
v
ey
e
d
in
th
e
r
e
v
iew
[
8
]
.
I
n
s
o
m
e
c
ases
,
th
e
s
tar
r
atin
g
m
ay
b
e
h
ig
h
b
u
t
th
e
r
ev
iew
m
ay
b
e
n
e
g
ativ
e
o
r
th
e
s
tar
r
atin
g
m
ay
b
e
lo
w
b
u
t
th
e
r
e
v
iew
m
ay
b
e
p
o
s
itiv
e
[
9
]
,
[
1
0
]
.
As
a
r
esu
lt,
r
ev
iews
ar
e
o
f
ten
ac
co
m
p
a
n
ied
b
y
te
x
tu
a
l
in
f
o
r
m
atio
n
to
p
r
o
v
id
e
a
m
o
r
e
co
m
p
r
eh
e
n
s
iv
e
ev
alu
atio
n
.
E
m
o
tio
n
al
an
aly
s
is
is
ess
en
ti
al
to
th
e
s
u
cc
ess
o
f
a
co
m
p
a
n
y
as
it
g
u
id
es
its
ef
f
o
r
ts
to
i
m
p
r
o
v
e
its
o
p
er
atio
n
s
an
d
th
e
p
r
o
d
u
cts
it
o
f
f
er
s
[
1
1
]
,
[
1
2
]
.
T
o
e
x
tr
ac
t
e
m
o
tio
n
s
f
r
o
m
cu
s
to
m
er
r
ev
iew
s
,
two
d
ee
p
lear
n
in
g
m
o
d
els
wer
e
u
s
ed
.
E
x
tr
ac
tin
g
em
o
tio
n
s
f
r
o
m
s
o
cial
m
ed
ia
in
m
u
ltip
le
lan
g
u
ag
es
is
a
cr
itical
is
s
u
e
in
th
e
f
ield
o
f
s
en
tim
en
t
an
aly
s
is
an
d
p
er
c
ep
tio
n
an
aly
s
is
[
1
3
]
,
[
1
4
]
.
W
ith
th
e
in
cr
ea
s
in
g
o
n
lin
e
p
r
esen
ce
o
f
m
ajo
r
g
lo
b
al
ac
tiv
ities
,
s
o
cial
m
ed
ia
u
s
er
s
u
s
e
th
ese
p
latf
o
r
m
s
to
ex
p
r
ess
th
eir
v
iews a
n
d
o
p
in
io
n
s
ab
o
u
t
th
em
.
B
r
ain
s
ig
n
als ca
n
n
o
w
b
e
u
s
ed
to
e
v
alu
ate
p
r
o
d
u
cts th
r
o
u
g
h
a
tech
n
iq
u
e
k
n
o
wn
as
“
n
e
u
r
o
m
ar
k
etin
g
,
”
wh
ich
in
v
o
lv
es
m
o
n
ito
r
in
g
b
r
a
in
wav
es
to
u
n
d
er
s
tan
d
c
o
n
s
u
m
er
p
s
y
ch
o
lo
g
y
[
1
5
]
.
Ob
tai
n
in
g
ac
cu
r
ate
r
esp
o
n
s
es
th
r
o
u
g
h
th
is
m
eth
o
d
ca
n
h
elp
co
m
p
an
ies
s
av
e
m
o
n
e
y
o
n
p
r
o
d
u
ct
ad
v
er
tis
in
g
[
1
6
]
.
Ma
n
y
p
eo
p
le
d
o
n
o
t
wan
t
to
f
u
lly
co
m
m
u
n
icate
th
eir
s
en
tim
en
ts
an
d
p
r
e
f
er
en
ce
s
r
e
g
ar
d
in
g
item
s
wh
en
ex
p
licitly
q
u
esti
o
n
ed
.
Ho
wev
er
,
n
eu
r
o
im
a
g
in
g
in
s
tr
u
m
en
ts
lik
e
elec
tr
o
en
ce
p
h
alo
g
r
am
(
E
E
G
)
ca
n
g
ath
er
in
f
o
r
m
atio
n
f
r
o
m
a
cu
s
to
m
e
r
'
s
b
r
ain
ev
en
d
u
r
in
g
th
e
g
en
er
atio
n
o
f
d
esire
o
r
o
b
s
er
v
atio
n
o
f
a
p
r
o
d
u
ct.
T
h
e
r
ef
o
r
e,
th
ese
b
r
ain
im
ag
in
g
to
o
ls
aid
m
ar
k
etin
g
r
esear
ch
er
s
in
m
a
k
in
g
d
ec
is
io
n
s
ab
o
u
t f
u
t
u
r
e
p
r
o
d
u
ct
ad
v
er
tis
em
en
ts
[
1
7
]
,
[
1
8
]
.
2.
RE
L
AT
E
D
WO
RK
Gau
b
a
et
a
l.
[
1
8
]
watc
h
a
v
id
eo
ad
v
er
tis
in
g
wh
ile
s
im
u
ltan
eo
u
s
ly
r
ec
o
r
d
in
g
th
eir
E
E
G
s
ig
n
als.
Fo
r
ev
er
y
m
o
v
ie,
v
alen
ce
s
co
r
es
wer
e
ca
lcu
lated
b
ased
o
n
s
elf
-
r
ep
o
r
t.
A
h
i
g
h
er
v
alen
ce
is
in
d
icativ
e
o
f
th
e
u
s
er
'
s
in
h
er
en
t
b
ea
u
ty
.
Ad
d
itio
n
ally
,
th
e
n
atu
r
al
lan
g
u
ag
e
p
r
o
ce
s
s
in
g
(
NL
P)
tech
n
iq
u
e
was
u
s
ed
to
ex
tr
ac
t
an
d
ev
alu
ate
th
e
m
u
ltime
d
ia
d
ata
,
wh
ich
in
clu
d
ed
th
e
c
o
m
m
e
n
ts
lef
t
b
y
v
iewe
r
s
wo
r
ld
wid
e,
to
d
o
s
en
tim
en
t
an
aly
s
is
.
T
h
e
s
en
tim
en
t
o
f
t
h
e
v
id
eo
was
ev
alu
ated
t
h
r
o
u
g
h
t
h
e
an
aly
s
is
o
f
tex
tu
al
f
ee
d
b
ac
k
i
n
r
ev
iew
co
m
m
en
ts
to
ass
ig
n
a
s
co
r
e.
E
E
G
d
ata
was
u
tili
ze
d
,
em
p
lo
y
in
g
a
r
an
d
o
m
f
o
r
est
r
eg
r
ess
io
n
tech
n
iq
u
e
to
f
o
r
ec
ast
th
e
r
atin
g
o
f
an
ad
v
er
tis
em
en
t.
T
o
b
o
ls
ter
th
e
p
r
ed
ictiv
e
ac
cu
r
ac
y
,
th
e
E
E
G
-
d
er
iv
ed
r
atin
g
an
d
th
e
s
en
tim
en
t
s
co
r
e
f
r
o
m
NL
P we
r
e
u
ltima
tely
co
m
b
in
e
d
.
Fan
et
a
l
.
[
1
9
]
s
u
g
g
ested
an
in
n
o
v
ativ
e
ap
p
r
o
ac
h
to
p
r
o
d
u
ct
p
r
ed
ictiv
e
an
aly
s
is
,
in
teg
r
atin
g
th
e
b
ass
/n
o
r
to
n
m
o
d
el
with
s
en
ti
m
en
t
an
aly
s
is
u
s
in
g
h
is
to
r
ical
s
ales
d
ata
an
d
o
n
lin
e
r
ev
ie
w
d
ata.
T
o
en
h
a
n
ce
p
r
ed
ictio
n
ac
cu
r
ac
y
,
th
ey
au
g
m
en
ted
t
h
e
im
itatio
n
co
ef
f
icien
t
o
f
th
e
b
ass
/n
o
r
to
n
m
o
d
el
with
th
e
n
aiv
e
B
ay
es
tech
n
iq
u
e
an
d
m
o
o
d
in
d
e
x
.
T
h
eir
s
tu
d
y
s
p
an
n
e
d
b
o
th
p
h
y
s
ical
an
d
o
n
lin
e
au
t
o
m
o
tiv
e
in
d
u
s
tr
ies.
Ho
wev
er
,
th
e
p
ag
e
lack
s
s
tatis
tics
r
eg
ar
d
in
g
u
s
er
ag
r
ee
m
en
t
o
r
d
is
ag
r
ee
m
en
t
with
th
e
r
e
v
iews,
as
well
as
th
e
f
r
eq
u
en
cy
o
f
r
ev
iew
r
ea
d
er
s
h
ip
.
Sh
r
esth
a
an
d
Naso
z
[
2
0
]
e
m
p
l
o
y
ed
s
en
tim
e
n
t
an
aly
s
is
to
ass
ess
th
e
alig
n
m
en
t b
etwe
en
A
m
az
o
n
.
co
m
r
ev
iews
an
d
th
eir
ass
o
ciate
d
r
atin
g
s
.
T
h
eir
s
tu
d
y
aim
ed
t
o
id
en
tify
an
d
ca
teg
o
r
ize
th
e
p
o
s
itiv
e
an
d
n
eg
ativ
e
em
o
tio
n
s
co
n
v
ey
e
d
in
tex
tu
al
d
ata
th
r
o
u
g
h
s
en
tim
en
t a
n
aly
s
is
.
Desp
ite
cu
s
to
m
er
s
p
r
o
v
id
in
g
p
o
la
r
ity
r
atin
g
s
i
n
th
eir
r
ev
iews
o
n
e
-
co
m
m
er
ce
p
latf
o
r
m
s
lik
e
Am
az
o
n
.
co
m
,
d
is
cr
ep
an
cies
b
etwe
en
th
e
r
ev
iew
co
n
ten
t
an
d
th
e
r
atin
g
m
ay
ar
is
e.
B
y
lev
e
r
ag
i
n
g
d
ee
p
lear
n
i
n
g
s
en
ti
m
en
t
a
n
aly
s
is
o
n
Am
az
o
n
.
co
m
p
r
o
d
u
ct
r
ev
iew
d
ata,
th
e
r
esear
ch
er
s
id
en
tifie
d
r
e
v
iews
s
p
an
n
in
g
d
iv
er
s
e
r
atin
g
s
.
T
h
e
y
tr
an
s
f
o
r
m
ed
p
r
o
d
u
ct
r
ev
iew
s
in
to
v
ec
t
o
r
s
u
s
in
g
a
tex
t v
ec
to
r
izatio
n
p
r
o
ce
s
s
an
d
s
u
b
s
eq
u
en
tly
t
r
ain
ed
a
g
ated
r
ec
u
r
r
e
n
t u
n
it o
f
a
r
ec
u
r
r
en
t
n
e
u
r
al
n
etwo
r
k
.
Asg
h
ar
et
a
l
.
[
2
1
]
in
v
esti
g
at
ed
s
en
tim
en
t
an
aly
s
is
o
f
p
u
b
lic
p
r
o
d
u
ct
f
ee
d
b
ac
k
,
an
aly
zin
g
wo
r
d
s
em
an
tic
o
r
ien
tatio
n
to
class
if
y
em
o
tico
n
s
,
m
o
d
if
ier
s
,
g
e
n
er
al
-
p
u
r
p
o
s
e
ter
m
s
,
an
d
d
o
m
ain
-
s
p
ec
if
ic
wo
r
d
s
in
t
o
p
o
s
itiv
e
an
d
n
eg
ativ
e
ca
teg
o
r
ies.
T
h
ey
also
in
tr
o
d
u
ce
d
an
en
h
an
ce
d
s
en
tim
en
t
an
al
y
s
is
m
eth
o
d
u
tili
zin
g
a
r
u
le
-
b
ased
class
if
icatio
n
f
r
am
e
wo
r
k
g
r
o
u
n
d
e
d
in
lex
ico
n
s
to
i
m
p
r
o
v
e
ef
f
ec
tiv
en
ess
with
in
o
n
lin
e
co
m
m
u
n
ities
.
Ho
wev
er
,
a
n
o
tab
le
lim
itatio
n
o
f
th
eir
s
tu
d
y
is
th
e
n
ec
ess
it
y
f
o
r
au
to
m
atic
wo
r
d
ca
teg
o
r
i
za
tio
n
an
d
s
co
r
in
g
,
p
o
ten
tially
im
p
ac
tin
g
th
e
class
if
icatio
n
ac
cu
r
ac
y
o
f
d
o
m
ain
-
s
p
ec
if
ic
ter
m
s
.
Kau
r
et
a
l
.
[
2
2
]
in
v
esti
g
ated
th
e
r
esp
o
n
s
e
o
f
E
E
G
to
p
o
s
it
iv
e
an
d
n
eg
ati
v
e
em
o
tio
n
s
,
s
p
ec
if
ically
tr
an
q
u
ilit
y
,
an
g
e
r
,
an
d
h
a
p
p
in
ess
.
T
h
ey
r
ec
o
r
d
ed
E
E
G
s
ig
n
als
f
r
o
m
ten
v
o
lu
n
teer
s
in
r
ea
l
tim
e
wh
ile
th
ey
watc
h
ed
em
o
tio
n
al
v
id
eo
clip
s
f
o
r
two
m
i
n
u
tes.
T
o
ca
teg
o
r
iz
e
th
e
s
ig
n
als
g
ath
er
ed
to
id
en
ti
f
y
em
o
tio
n
al
s
tates,
f
r
ac
tal
d
im
en
s
io
n
f
ea
tu
r
es
we
r
e
d
er
iv
e
d
f
r
o
m
th
e
r
aw
E
E
G
d
ata.
A
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM
)
u
s
in
g
a
r
ad
ial
b
asis
f
u
n
ctio
n
(
R
B
F)
k
e
r
n
el
was
em
p
lo
y
ed
to
o
b
tain
g
r
ea
ter
ac
cu
r
ac
y
,
with
an
av
e
r
a
g
e
ac
c
u
r
ac
y
o
f
6
0
%.
Ho
wev
er
,
it
s
h
o
u
ld
b
e
h
ig
h
lig
h
ted
th
at
f
in
d
in
g
a
m
o
d
el
th
at
is
o
p
tim
al
f
o
r
id
en
tify
in
g
em
o
tio
n
s
g
ets
h
ar
d
er
as
th
e
n
u
m
b
er
o
f
in
d
iv
i
d
u
als r
is
es.
W
an
g
et
al
.
[
2
3
]
u
tili
ze
d
ey
e
tr
ac
k
in
g
an
d
E
E
G
r
esp
o
n
s
e
d
ata
as
in
p
u
t
f
o
r
a
m
u
ltim
o
d
al
f
u
s
io
n
tech
n
iq
u
e
aim
e
d
at
r
ec
o
g
n
iz
in
g
d
esig
n
d
ec
is
io
n
-
m
ak
in
g
.
R
esu
lts
f
r
o
m
th
e
ex
p
er
im
e
n
t
s
u
g
g
est
th
at
th
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
S
en
timen
t
-
elec
tr
o
en
ce
p
h
a
lo
g
r
a
m
fu
s
io
n
fo
r
efficien
t p
r
o
d
u
c
t revie
w
p
r
ed
ictio
n
…
(
R
a
h
u
l
K
u
ma
r
S
h
a
r
ma
)
4677
p
er
f
o
r
m
an
ce
o
f
th
e
f
u
s
io
n
s
tr
ateg
y
alig
n
s
well
with
ex
p
er
t
d
e
cisi
o
n
-
m
ak
in
g
o
u
tco
m
es
wh
e
n
in
co
r
p
o
r
atin
g
E
E
G
s
ig
n
als
an
d
ey
e
m
o
v
em
en
t
c
h
ar
ac
ter
is
tics
.
B
y
in
teg
r
atin
g
ey
e
tr
ac
k
in
g
d
ata
an
d
E
E
G,
th
e
m
u
ltimo
d
al
f
u
s
io
n
ap
p
r
o
ac
h
h
o
ld
s
s
ig
n
if
ican
t
p
o
ten
tial
to
m
it
ig
ate
th
e
ef
f
e
cts
o
f
s
u
b
jectiv
ity
,
b
ias,
an
d
s
u
p
er
f
iciality
in
d
ec
is
io
n
-
m
ak
in
g
,
o
f
f
er
in
g
a
n
o
v
el
ap
p
r
o
ac
h
t
o
g
u
i
d
in
g
d
esig
n
p
r
ac
tice
with
o
b
jectiv
e
ev
i
d
en
ce
an
d
s
u
p
p
o
r
t.
Sh
ar
m
a
an
d
Dag
u
r
[2
4
]
u
tili
ze
d
a
v
ast
d
ataset
co
m
p
r
is
in
g
o
v
er
5
0
0
,
0
0
0
o
n
lin
e
in
s
tan
t
v
id
e
o
r
ev
iews
to
d
ev
elo
p
a
s
en
tim
en
t
p
o
la
r
ity
ca
teg
o
r
izatio
n
tech
n
iq
u
e.
T
h
e
r
ev
iews
wer
e
ca
teg
o
r
ized
in
t
o
f
iv
e
g
r
o
u
p
s
,
with
co
n
s
id
er
atio
n
g
iv
en
t
o
th
r
ee
p
o
lar
ity
f
ea
tu
r
es
(
v
er
b
s
,
a
d
v
er
b
s
,
an
d
ad
jectiv
es)
f
o
r
r
ev
iew
-
l
ev
el
class
if
icatio
n
.
T
h
e
f
in
d
in
g
s
d
em
o
n
s
tr
ated
a
p
r
o
m
is
in
g
ac
cu
r
ac
y
r
ate
o
f
8
1
%,
s
u
r
p
ass
in
g
m
an
y
p
r
i
o
r
s
y
s
tem
s
b
y
3
%.
W
h
ile
au
to
m
ated
s
en
tim
en
t
an
aly
s
is
aid
s
in
p
r
o
ce
s
s
in
g
lar
g
e
v
o
lu
m
es
o
f
tex
t
d
ata,
it
s
till
en
co
u
n
ter
s
lim
itatio
n
s
;
f
o
r
ex
am
p
le,
th
e
s
o
f
twar
e
em
p
lo
y
ed
in
th
is
s
tu
d
y
s
tr
u
g
g
les
with
in
ter
p
r
etin
g
d
iv
er
s
e
s
ty
les
lik
e
s
ar
ca
s
m
.
R
ef
er
to
T
ab
le
1
f
o
r
a
co
m
p
ar
is
o
n
a
n
d
ev
alu
atio
n
o
f
p
r
e
v
io
u
s
wo
r
k
.
T
ab
le
1
.
An
aly
zi
n
g
an
d
co
m
p
ar
in
g
p
r
ev
io
u
s
wo
r
k
s
R
e
f
e
r
e
n
c
e
s
Te
c
h
n
i
q
u
e
Ef
f
i
c
i
e
n
c
y
S
e
c
u
r
i
t
y
l
e
v
e
l
C
o
m
p
u
t
a
t
i
o
n
a
l
o
v
e
r
h
e
a
d
G
a
u
b
a
e
t
a
l
.
[
1
8
]
N
LP
M
e
d
i
u
m
M
e
d
i
u
m
M
e
d
i
u
m
F
a
n
e
t
a
l
.
[
19]
B
N
m
o
d
e
l
M
e
d
i
u
m
H
i
g
h
H
i
g
h
A
sg
h
a
r
e
t
a
l
.
[
2
1
]
R
B
C
sc
h
e
me
H
i
g
h
M
e
d
i
u
m
H
i
g
h
K
a
u
r
e
t
a
l
.
[
2
2
]
R
B
F
&
B
C
I
A
v
e
r
a
g
e
M
e
d
i
u
m
H
i
g
h
W
a
n
g
e
t
a
l
.
[
2
3
]
M
u
l
t
i
m
o
d
a
l
f
u
si
o
n
s
t
r
a
t
e
g
y
H
i
g
h
M
e
d
i
u
m
H
i
g
h
K
a
u
sar
e
t
a
l
.
[2
5
]
R
B
a
n
d
R
B
S
A
v
e
r
a
g
e
H
i
g
h
H
i
g
h
3.
P
RO
P
O
SE
D
M
E
T
H
O
DO
L
O
G
Y
Ma
ch
in
e
lear
n
in
g
-
b
ased
s
en
ti
m
en
t a
n
al
y
s
is
th
at
in
co
r
p
o
r
ate
s
cu
s
to
m
er
f
ee
d
b
ac
k
a
n
d
E
E
G
s
ig
n
al
d
ata
h
as
g
r
o
wn
i
n
im
p
o
r
tan
ce
in
r
e
ce
n
t
y
ea
r
s
.
T
h
e
s
u
g
g
ested
s
tu
d
y
o
f
f
er
s
to
ad
d
r
ess
th
e
cu
r
r
en
t
is
s
u
es
b
y
u
tili
zin
g
a
co
r
r
elatio
n
-
b
ased
d
ee
p
lear
n
in
g
n
e
u
r
al
n
etwo
r
k
(
C
DNN)
.
T
h
e
m
ac
h
i
n
e
lear
n
in
g
an
d
p
atter
n
r
ec
o
g
n
itio
n
co
m
m
u
n
ities
ar
e
cu
r
r
en
tly
ac
tiv
ely
r
esear
ch
in
g
d
ee
p
lear
n
in
g
.
NL
P
,
s
p
ee
ch
r
ec
o
g
n
itio
n
,
a
n
d
co
m
p
u
ter
v
is
io
n
r
ep
r
esen
t o
n
ly
a
h
a
n
d
f
u
l o
f
d
o
m
ain
s
in
wh
ich
it h
as d
em
o
n
s
tr
ated
ex
ce
p
tio
n
al
p
e
r
f
o
r
m
an
ce
.
W
ith
th
e
g
r
o
win
g
s
ize
o
f
d
ata,
d
ee
p
lear
n
in
g
is
b
ec
o
m
in
g
i
n
cr
ea
s
in
g
ly
im
p
o
r
tan
t in
p
r
o
v
i
d
in
g
p
r
e
d
ictiv
e
an
al
y
tics
s
o
lu
tio
n
s
.
T
h
e
d
ataset
u
s
ed
in
th
e
p
r
o
p
o
s
ed
s
tu
d
y
is
th
e
in
p
u
t
an
d
h
as
b
ee
n
p
r
e
-
p
r
o
ce
s
s
ed
u
s
in
g
s
to
p
wo
r
d
r
em
o
v
al,
to
k
en
izatio
n
,
an
d
a
weig
h
tin
g
f
ac
to
r
f
o
r
p
r
o
d
u
ct
r
ev
iews.
No
r
m
aliza
tio
n
is
u
s
ed
to
co
n
v
er
t
u
n
s
tr
u
ctu
r
ed
d
ata
in
to
a
s
tr
u
ctu
r
ed
m
an
n
er
wh
e
n
an
E
E
G
s
ig
n
al
is
em
p
lo
y
ed
.
Var
io
u
s
attr
ib
u
tes
s
u
ch
as
em
o
tio
n
,
d
em
o
g
r
ap
h
ic
in
d
icato
r
s
,
p
er
s
o
n
ality
tr
aits
,
an
d
s
en
tim
en
t
f
ea
tu
r
es
ar
e
d
er
iv
ed
f
r
o
m
t
h
e
p
r
e
-
p
r
o
ce
s
s
ed
d
ata.
T
h
e
s
tu
d
y
in
tr
o
d
u
ce
s
a
h
y
b
r
id
C
DNN,
wh
ich
is
f
in
e
-
tu
n
ed
an
d
o
p
ti
m
ized
f
o
r
h
y
p
er
p
ar
am
eter
s
u
s
in
g
th
e
g
r
ass
h
o
p
p
er
o
p
tim
izatio
n
alg
o
r
ith
m
b
ased
o
n
en
tr
o
p
y
(
E
GOA)
f
o
r
ea
c
h
lay
er
.
Fig
u
r
e
1
illu
s
tr
ates
t
h
e
s
u
g
g
ested
b
lo
ck
d
iag
r
am
.
T
h
is
m
eth
o
d
’
s
o
b
jectiv
e
is
to
ad
d
r
ess
th
e
p
r
o
b
lem
s
with
s
en
tim
en
t
an
aly
s
is
an
d
p
r
o
d
u
ce
m
o
r
e
ac
cu
r
ate
o
u
tco
m
es.
Fig
u
r
e
1
.
Pro
p
o
s
ed
b
lo
ck
d
iag
r
am
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
13
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
4
6
7
5
-
4
6
8
7
4678
As
s
h
o
wn
in
Fig
u
r
e
1
;
p
r
o
d
u
ct
r
ev
iew,
as
well
as
E
E
G
s
ig
n
al
d
atasets
,
ar
e
u
s
ed
.
Del
ib
er
ate
an
d
is
th
e
Am
az
o
n
p
r
o
d
u
ct
r
e
v
iew
d
ata
s
et
∈
{
1
1
,
2
2
,
3
3
,
.
.
.
}
an
d
is
th
e
E
E
G
s
ig
n
al
d
ata
s
et
∈
{
1
1
,
2
2
,
3
3
,
.
.
.
}
wh
er
e
n
is
th
e
to
tal
am
o
u
n
t
o
f
em
o
tio
n
s
ex
p
er
ien
c
ed
b
y
h
u
m
an
s
b
ased
o
n
E
E
G
s
ig
n
als
an
d
th
e
r
ev
iew
o
f
p
r
o
d
u
cts.
T
h
e
v
ec
to
r
f
u
n
ctio
n
o
f
ea
ch
s
ig
n
al
an
d
th
e
p
r
o
d
u
ct
r
ev
iews
o
f
a
d
atab
ase
is
g
iv
en
as
i=1
to
n
.
E
E
G
s
ig
n
als
is
f
i
r
s
t
p
r
e
-
p
r
o
ce
s
s
ed
,
f
o
llo
wed
b
y
p
r
e
-
p
r
o
ce
s
s
in
g
o
f
th
e
f
o
llo
win
g
s
et
o
f
d
ata,
s
u
c
h
as p
r
o
d
u
ct
r
ev
iews.
3
.
1
.
E
lect
ro
ence
ph
a
lo
g
ra
m
s
ig
na
l:
pre
-
pro
ce
s
s
ing
3
.
1
.
1
.
Sig
na
l t
ra
ns
f
o
rma
t
io
n
Sig
n
al
co
n
v
er
s
io
n
is
an
im
p
o
r
tan
t
s
tep
in
p
r
ep
ar
in
g
th
e
E
E
G
s
ig
n
al
d
ata
f
o
r
f
u
r
th
er
p
r
o
c
ess
in
g
an
d
an
aly
s
is
.
T
h
is
en
tails
co
n
v
e
r
tin
g
th
e
s
ig
n
al
d
ata
f
r
o
m
it
s
in
itial
f
o
r
m
at,
w
h
ich
m
a
y
b
e
ca
teg
o
r
ized
o
r
non
-
n
u
m
er
ic,
in
t
o
a
n
u
m
e
r
ical
r
ep
r
esen
tatio
n
th
at
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
ca
n
an
aly
ze
.
On
ce
th
e
s
ig
n
al
d
ata
h
as
b
ee
n
c
o
n
v
er
te
d
,
it
c
an
b
e
r
ep
r
esen
te
d
as
a
f
ea
tu
r
e
v
ec
to
r
th
at
c
o
n
tain
s
n
u
m
er
i
ca
l
v
alu
es
f
o
r
ea
c
h
co
m
p
o
n
en
t
o
f
t
h
e
s
ig
n
al.
T
h
e
s
e
f
ea
tu
r
e
v
ec
to
r
s
ca
n
t
h
en
b
e
u
s
ed
to
tr
ain
a
class
if
ier
to
r
ec
o
g
n
ize
d
if
f
er
e
n
t
p
atter
n
s
in
th
e
E
E
G
s
ig
n
als,
s
u
ch
as th
e
p
r
esen
ce
o
r
ab
s
en
ce
o
f
s
p
ec
if
ic
em
o
tio
n
s
o
r
co
g
n
iti
v
e
s
tates.
3
.
1
.
2
.
No
rma
liza
t
io
n
No
r
m
aliza
tio
n
o
f
a
n
E
E
G
s
ig
n
al
r
ef
er
s
to
th
e
p
r
o
ce
s
s
o
f
c
o
n
v
er
tin
g
an
u
n
s
tr
u
ctu
r
e
d
s
ig
n
al
in
to
a
s
tr
u
ctu
r
ed
f
o
r
m
at.
T
h
e
E
E
G
s
i
g
n
al
is
lin
ea
r
ly
tr
a
n
s
f
o
r
m
e
d
to
f
it
in
to
a
s
p
ec
if
ic
r
an
g
e
d
u
r
i
n
g
n
o
r
m
aliza
tio
n
.
Fo
r
th
e
s
tan
d
ar
d
izatio
n
o
f
d
ata,
m
in
-
m
a
x
n
o
r
m
aliza
tio
n
is
u
s
ed
,
wh
ich
lin
ea
r
ly
c
h
an
g
e
s
th
e
s
ig
n
als.
T
h
e
ac
co
m
p
an
y
in
g
c
o
n
d
itio
n
is
u
s
ed
to
p
er
f
o
r
m
m
in
-
m
a
x
n
o
r
m
a
lizatio
n
r
eg
u
lar
ly
.
=
−
(
1
)
W
h
er
e,
an
d
ar
e
th
e
m
in
im
u
m
an
d
m
ax
im
u
m
v
alu
es in
,
an
d
∈
is
th
e
s
et
o
f
v
alu
es
in
th
e
in
p
u
t E
E
G
s
ig
n
al
d
atase
t.
Af
ter
co
m
p
letin
g
t
h
e
p
r
e
-
p
r
o
ce
s
s
in
g
s
tag
e
o
f
E
E
G
s
ig
n
als,
th
e
p
r
o
d
u
ct
r
e
v
iew
o
f
d
atasets
is
p
r
e
-
p
r
o
ce
s
s
ed
w
h
ich
is
r
ep
r
esen
ted
as f
o
llo
ws.
3.
2
.
A
m
a
zo
n
pro
du
ct
re
v
iew
pre
-
pro
ce
s
s
ing
T
h
e
p
r
o
d
u
ct
r
ev
iew
d
ataset
s
e
r
v
es a
s
th
e
in
p
u
t so
u
r
ce
.
T
h
e
in
itial st
ep
in
h
an
d
lin
g
th
is
d
ataset,
wh
ich
co
n
tain
s
u
n
s
tr
u
ctu
r
e
d
s
ig
n
als,
in
v
o
lv
es
co
n
v
e
r
tin
g
th
ese
s
ig
n
als
in
to
s
tr
u
ctu
r
ed
o
n
es.
T
h
is
p
r
o
ce
s
s
en
tails
em
p
lo
y
in
g
to
k
en
izatio
n
,
s
to
p
wo
r
d
r
em
o
v
al,
a
n
d
weig
h
tin
g
v
ar
iab
les
to
ac
h
iev
e
s
tr
u
ctu
r
ed
d
ata.
T
h
e
f
ir
s
t
s
tep
,
s
o
lu
tio
n
cr
ea
tio
n
,
is
d
escr
ib
ed
in
d
etail
as
f
o
llo
ws
{
1
,
2
,
3
,
.
.
.
}
im
p
lies
th
e
d
o
cu
m
en
ts
o
n
th
e
Am
az
o
n
p
r
o
d
u
ct
r
ev
iew
d
ataset.
=
{
1
,
2
,
3
,
.
.
.
}
(
2
)
3
.
2
.
1
.
T
o
k
eniza
t
io
n
T
o
k
en
izatio
n
is
th
e
f
ir
s
t step
i
n
b
r
ea
k
in
g
d
o
wn
len
g
th
y
s
tr
in
g
s
o
f
p
r
o
d
u
ct
r
ev
iews in
to
s
m
a
ller
to
k
en
s
o
r
u
n
its
.
T
h
is
in
v
o
lv
es
s
p
litt
in
g
lar
g
e
r
s
ec
tio
n
s
o
f
a
p
r
o
d
u
ct
r
ev
iew
in
to
p
h
r
ases
an
d
b
r
ea
k
in
g
d
o
wn
s
en
ten
ce
s
in
to
in
d
iv
id
u
al
wo
r
d
s
.
T
o
k
e
n
izatio
n
is
als
o
r
ef
er
r
ed
to
as
p
r
o
d
u
ct
s
eg
m
en
tatio
n
an
d
lex
ical
an
aly
s
is
.
T
o
k
en
izatio
n
is
a
m
eth
o
d
u
s
ed
to
b
r
ea
k
d
o
wn
a
s
et
o
f
in
p
u
t
d
a
tasets
in
to
m
ea
n
in
g
f
u
l
s
eg
m
en
t
s
,
o
r
to
k
en
s
,
wh
ich
ca
n
in
clu
d
e
wo
r
d
s
,
p
h
r
ases
,
an
d
s
y
m
b
o
ls
.
T
h
e
r
esu
ltin
g
i
n
p
u
t
te
x
t
is
th
e
n
o
r
g
a
n
ized
i
n
to
a
lis
t
o
f
to
k
e
n
co
llectio
n
s
f
o
r
f
u
r
th
er
p
r
o
ce
s
s
in
g
.
Ho
wev
er
,
th
e
n
atu
r
e
o
f
n
atu
r
al
lan
g
u
a
g
e
p
r
esen
ts
ch
allen
g
es
d
u
r
in
g
to
k
en
izatio
n
; w
h
ile
m
a
n
y
lan
g
u
ag
es u
s
e
wh
ite
s
p
ac
es to
s
ep
ar
ate
wo
r
d
s
,
o
th
e
r
s
d
o
n
o
t.
C
h
allen
g
es
in
to
k
e
n
izatio
n
:
i
n
d
ee
d
,
th
e
s
y
s
tem
o
f
wr
itin
g
an
d
ty
p
o
g
r
ap
h
ic
lay
o
u
t
o
f
te
x
t,
s
u
ch
as
p
r
o
d
u
ct
r
ev
iews,
ca
n
s
ig
n
if
ican
tly
in
f
lu
en
ce
to
k
e
n
izatio
n
.
T
h
ese
f
ac
to
r
s
ca
n
in
tr
o
d
u
ce
ad
d
itio
n
al
ch
allen
g
es
th
at
n
ee
d
to
b
e
ad
d
r
ess
ed
wh
e
n
to
k
e
n
izin
g
tex
t.
Fu
r
th
er
m
o
r
e
,
lan
g
u
ag
es
ca
n
b
e
ca
teg
o
r
ized
in
to
th
r
ee
p
r
im
a
r
y
ty
p
es b
ased
o
n
t
h
eir
s
tr
u
ctu
r
e:
−
I
s
o
latin
g
:
in
is
o
latin
g
lan
g
u
a
g
es,
th
e
Am
az
o
n
p
r
o
d
u
ct
r
ev
iew
wo
r
d
s
ar
e
ty
p
ically
co
m
p
o
s
ed
o
f
s
in
g
l
e
m
o
r
p
h
e
m
es,
an
d
w
o
r
d
o
r
d
e
r
an
d
co
n
tex
t
ar
e
c
r
u
cial
f
o
r
co
n
v
ey
in
g
m
ea
n
in
g
.
T
h
er
e
is
litt
le
to
n
o
u
s
e
o
f
in
f
lectio
n
al
o
r
d
er
i
v
atio
n
al
m
o
r
p
h
o
lo
g
y
.
C
h
in
ese
an
d
Vietn
am
ese
ar
e
ex
am
p
les
o
f
is
o
latin
g
lan
g
u
a
g
es.
I
n
th
ese
lan
g
u
ag
es,
wo
r
d
s
ar
e
n
o
t
in
f
lecte
d
f
o
r
ten
s
e,
n
u
m
b
er
,
g
e
n
d
er
,
o
r
ca
s
e,
b
u
t
in
s
tead
,
co
n
t
ex
t
o
r
ad
d
itio
n
al
wo
r
d
s
ar
e
u
s
ed
to
co
n
v
ey
th
es
e
g
r
am
m
atica
l f
ea
tu
r
es.
−
Ag
g
lu
tin
ativ
e:
ag
g
lu
tin
ativ
e
lan
g
u
ag
es
u
s
e
af
f
i
x
es
ex
ten
s
iv
ely
to
d
en
o
te
g
r
am
m
atica
l
r
el
atio
n
s
h
ip
s
,
b
u
t
ea
ch
af
f
ix
g
en
er
ally
r
ep
r
esen
t
s
a
s
in
g
le
g
r
am
m
atica
l
m
ea
n
i
n
g
.
T
h
ese
lan
g
u
ag
es
ten
d
to
h
av
e
a
r
elativ
ely
f
ix
ed
w
o
r
d
o
r
d
er
,
an
d
wo
r
d
s
ca
n
h
a
v
e
m
a
n
y
af
f
ix
es
attac
h
ed
to
th
em
.
T
u
r
k
is
h
,
Fin
n
is
h
,
an
d
Swah
ili
a
r
e
ex
am
p
les o
f
a
g
g
lu
tin
ativ
e
lan
g
u
ag
es.
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
S
en
timen
t
-
elec
tr
o
en
ce
p
h
a
lo
g
r
a
m
fu
s
io
n
fo
r
efficien
t p
r
o
d
u
c
t revie
w
p
r
ed
ictio
n
…
(
R
a
h
u
l
K
u
ma
r
S
h
a
r
ma
)
4679
−
I
n
f
lectio
n
al
:
f
u
s
io
n
al
la
n
g
u
a
g
es
o
f
ten
ex
p
r
ess
m
u
ltip
le
g
r
am
m
atica
l
m
ea
n
in
g
s
with
in
a
s
in
g
le
af
f
ix
,
r
esu
ltin
g
in
co
m
p
lex
wo
r
d
f
o
r
m
s
.
T
h
ese
lan
g
u
ag
es
ty
p
ically
h
av
e
m
o
r
e
f
lex
i
b
le
wo
r
d
o
r
d
er
co
m
p
ar
e
d
to
is
o
latin
g
an
d
ag
g
lu
tin
ativ
e
lan
g
u
ag
es.
E
x
a
m
p
les o
f
f
u
s
io
n
al
lan
g
u
ag
es in
clu
d
e
L
atin
,
Gr
ee
k
,
an
d
Sp
a
n
is
h
.
3
.
2
.
2
.
Sto
p
wo
rd
re
mo
v
al
Sto
p
wo
r
d
s
ar
e
co
m
m
o
n
l
y
u
s
ed
wo
r
d
s
th
at
o
f
ten
d
o
n
o
t
s
ig
n
if
ican
tly
ch
an
g
e
th
e
m
ea
n
i
n
g
o
r
co
n
tex
t
o
f
a
p
h
r
ase.
Alth
o
u
g
h
th
ey
a
r
e
f
r
eq
u
e
n
tly
em
p
lo
y
ed
in
s
en
te
n
ce
s
to
co
n
n
ec
t
o
th
er
wo
r
d
s
,
th
ey
g
en
e
r
ally
d
o
n
o
t
ad
d
m
ea
n
in
g
f
u
l
c
o
n
ten
t
t
o
t
h
e
tex
t.
I
n
tex
t
m
in
in
g
,
th
e
p
r
ev
ale
n
ce
o
f
s
to
p
wo
r
d
s
ca
n
co
m
p
licate
th
e
u
n
d
er
s
tan
d
i
n
g
o
f
d
o
c
u
m
en
t
c
o
n
ten
t.
T
h
e
r
ef
o
r
e,
it
is
wid
ely
ac
ce
p
ted
th
at
elim
in
atin
g
s
to
p
wo
r
d
s
d
u
r
in
g
p
r
e
-
p
r
o
ce
s
s
in
g
en
h
an
c
es
th
e
p
r
ec
is
io
n
an
d
ef
f
ec
tiv
e
n
ess
o
f
tex
t
an
aly
s
is
.
E
x
am
p
les
o
f
s
to
p
wo
r
d
s
in
cl
u
d
e
‘
a
n
d
’,
‘
ar
e
’,
a
n
d
‘
t
h
is
’.
As
th
ey
d
o
n
o
t
co
n
t
r
ib
u
te
s
u
b
s
tan
tially
t
o
a
d
o
cu
m
en
t
’
s
o
v
er
all
m
ea
n
in
g
,
s
to
p
wo
r
d
s
ar
e
ty
p
ically
r
em
o
v
ed
f
r
o
m
th
e
te
x
t.
Ho
wev
er
,
c
o
m
p
ilin
g
a
c
o
m
p
r
eh
en
s
iv
e
lis
t
o
f
s
to
p
wo
r
d
s
ca
n
b
e
ch
allen
g
in
g
d
u
e
to
in
co
n
s
is
ten
cies a
cr
o
s
s
d
if
f
er
en
t te
x
t
s
o
u
r
ce
s
an
d
th
eir
f
r
eq
u
e
n
t o
cc
u
r
r
en
ce
in
tex
t.
T
h
ese
ch
allen
g
es c
an
im
p
ed
e
th
e
p
r
o
ce
s
s
in
g
o
f
tex
tu
al
d
ata
an
d
th
e
im
p
lem
en
tatio
n
o
f
ef
f
ec
tiv
e
s
to
p
-
wo
r
d
f
ilter
i
n
g
.
3
.
2
.
3
.
Weig
hting
f
a
ct
o
r
T
h
is
in
v
o
lv
es a
s
s
ig
n
in
g
a
wei
g
h
t to
ea
ch
ter
m
with
in
a
d
o
cu
m
en
t,
in
d
icatin
g
its
r
elativ
e
s
i
g
n
if
ican
ce
.
T
h
is
p
r
o
ce
s
s
o
f
ass
ig
n
in
g
weig
h
ts
to
ea
ch
p
h
r
ase
is
co
m
m
o
n
ly
r
ef
er
r
ed
to
as
th
e
“
weig
h
tin
g
m
eth
o
d
”.
T
h
e
f
r
eq
u
e
n
cy
o
f
th
e
weig
h
tin
g
f
ac
to
r
is
q
u
an
tifie
d
n
u
m
e
r
ically
a
s
:
(
)
=
[
]
(
3
)
[
]
=
(
4
)
L
et
th
e
o
b
tain
ed
[
]
b
e
s
to
ck
ed
u
p
o
n
th
e
eq
u
i
v
alen
t
G.
G
s
ig
n
if
ies
th
e
f
r
eq
u
en
cy
s
et
o
f
an
Am
az
o
n
p
r
o
d
u
ct
r
ev
iew.
T
h
e
p
r
e
-
p
r
o
c
ess
ed
p
r
o
d
u
ct
r
ev
iew
d
ata
th
e
n
m
o
v
es
o
n
to
th
e
s
tep
o
f
f
ea
tu
r
e
ex
tr
ac
tio
n
.
T
h
e
p
r
e
-
p
r
o
ce
s
s
ed
d
ata
is
u
tili
ze
d
f
o
r
ex
tr
ac
tin
g
s
en
tim
en
t
ch
a
r
a
cter
is
tics
,
p
er
s
o
n
ality
tr
aits
,
d
em
o
g
r
ap
h
ic
f
ac
to
r
s
,
an
d
em
o
tio
n
al
attr
ib
u
tes.
3.
3
.
F
e
a
t
ure
ex
t
r
a
ct
io
n
Fo
llo
win
g
p
r
e
-
p
r
o
ce
s
s
in
g
,
b
o
t
h
th
e
E
E
G
d
ata
an
d
Am
az
o
n
p
r
o
d
u
ct
r
ev
iews
p
r
o
ce
ed
to
t
h
e
f
ea
tu
r
e
ex
tr
ac
tio
n
p
h
ase.
Featu
r
e
ex
tr
a
ctio
n
is
p
iv
o
tal
in
th
e
class
if
icatio
n
p
r
o
ce
s
s
,
in
v
o
lv
in
g
t
h
e
ex
tr
ac
tio
n
o
f
p
e
r
tin
en
t
s
ig
n
als
an
d
r
ev
iew
attr
ib
u
tes
f
r
o
m
th
e
p
r
e
-
pr
o
ce
s
s
ed
d
ata
to
f
ac
ilit
ate
ca
teg
o
r
izatio
n
.
E
x
tr
ac
tin
g
u
s
ef
u
l
f
ea
t
u
r
es
f
r
o
m
E
E
G
s
ig
n
als
an
d
p
r
o
d
u
ct
r
ev
iews
ca
n
b
e
ch
allen
g
in
g
,
an
d
th
er
e
ar
e
v
ar
io
u
s
ap
p
r
o
ac
h
es
to
f
ea
tu
r
e
ex
tr
ac
tio
n
.
I
n
th
is
s
tu
d
y
,
we
ex
tr
ac
t
em
o
tio
n
,
d
em
o
g
r
ap
h
ic,
p
er
s
o
n
ality
,
an
d
s
en
tim
en
t
f
ea
tu
r
es
f
r
o
m
t
h
e
p
r
e
-
p
r
o
ce
s
s
ed
E
E
G
s
ig
n
als an
d
Am
az
o
n
p
r
o
d
u
ct
r
e
v
iews.
3
.
3
.
1
.
F
ea
t
ure
e
x
t
ra
ct
i
o
n ba
s
ed
o
n e
m
o
t
io
n f
e
a
t
ures
T
h
e
m
eth
o
d
s
a
n
d
r
es
o
u
r
ce
s
i
n
th
is
s
ec
tio
n
c
o
n
ce
n
tr
ate
o
n
c
o
llectin
g
em
o
ti
o
n
o
r
m
o
o
d
s
ta
tes
f
r
o
m
a
tex
t
ex
ce
r
p
t
u
s
in
g
e
m
o
tio
n
al
q
u
alities
.
A
m
ess
ag
e
n
ee
d
s
to
b
e
p
u
t
in
t
o
a
p
r
o
ce
s
s
th
at
p
r
o
d
u
ce
s
s
p
ec
if
ic
em
o
tio
n
al
ca
teg
o
r
ies,
s
u
ch
as
s
u
r
p
r
is
e,
jo
y
,
o
r
g
r
ief
,
am
o
n
g
m
an
y
o
th
e
r
s
,
an
d
is
em
o
tio
n
-
o
r
ien
te
d
.
A
lis
t
o
f
wo
r
d
s
o
r
ex
p
r
ess
io
n
s
ca
teg
o
r
ized
b
y
v
ar
i
o
u
s
em
o
tio
n
s
tates
s
h
o
u
ld
b
e
p
r
o
v
id
e
d
b
y
em
o
ti
o
n
-
o
r
ie
n
ted
lex
ical
r
eso
u
r
ce
s
in
T
ab
le
2
.
T
ab
le
2
.
E
m
o
tio
n
f
ea
t
u
r
es a
r
e
ca
teg
o
r
ized
in
to
th
e
s
u
b
s
eq
u
e
n
t c
lass
es
S
c
o
p
e
F
e
a
t
u
r
e
S
o
u
r
c
e
D
e
scri
p
t
i
o
n
R
a
n
g
e
Emo
t
i
o
n
N
JO
N
R
C
n
u
m
b
e
r
o
f
p
r
o
d
u
c
t
r
e
v
i
e
w
s
a
n
d
E
EG
s
i
g
n
a
l
s
t
h
a
t
mat
c
h
t
h
e
j
o
y
w
o
r
d
a
n
d
mo
me
n
t
l
i
st
{0
,
1
,
2
.
.
.
,
n
}
N
TR
.
.
.
m
a
t
c
h
e
s
t
h
e
t
r
u
s
t
w
o
r
d
a
n
d
m
o
m
e
n
t
l
i
s
t
{0
,
1
,
2
.
.
.
,
n
}
N
S
A
.
.
.
m
a
t
c
h
e
s
t
h
e
s
a
d
n
e
ss
w
o
r
d
a
n
d
m
o
men
t
l
i
s
t
{0
,
1
,
2
.
.
.
,
n
}
NANG
.
.
.
m
a
t
c
h
e
s
t
h
e
a
n
g
e
r
w
o
r
d
a
n
d
m
o
me
n
t
l
i
s
t
{0
,
1
,
2
.
.
.
,
n
}
N
S
U
.
.
.
m
a
t
c
h
e
s
t
h
e
s
u
r
p
r
i
se
w
o
r
d
a
n
d
m
o
men
t
l
i
s
t
{0
,
1
,
2
.
.
.
,
n
}
N
F
E
.
.
.
m
a
t
c
h
e
s
t
h
e
f
e
a
r
w
o
r
d
a
n
d
m
o
m
e
n
t
l
i
st
{0
,
1
,
2
.
.
.
,
n
}
NANT
.
.
.
m
a
t
c
h
e
s
t
h
e
a
n
t
i
c
i
p
a
t
i
o
n
w
o
r
d
a
n
d
mo
me
n
t
l
i
st
{0
,
1
,
2
.
.
.
,
n
}
NDIS
.
.
.
m
a
t
c
h
e
s
t
h
e
d
i
s
g
u
st
w
o
r
d
a
n
d
m
o
m
e
n
t
l
i
st
{0
,
1
,
2
.
.
.
,
n
}
3
.
3
.
3
.
F
ea
t
ure
e
x
t
ra
ct
i
o
n ba
s
ed
o
n per
s
o
na
lity
t
ra
it
s
Per
s
o
n
ality
tr
ait
s
co
r
es
a
r
e
c
o
n
tin
u
o
u
s
v
alu
es
th
at
ca
n
b
e
u
s
ed
as
in
p
u
t
f
o
r
E
E
G
s
ig
n
als
an
d
Am
az
o
n
p
r
o
d
u
ct
r
atin
g
s
r
ig
h
t
awa
y
;
h
o
wev
er
,
to
d
o
class
if
icatio
n
,
th
ese
s
co
r
es
m
u
s
t
b
e
d
is
cr
etize
d
.
I
n
th
is
s
tu
d
y
,
f
iv
e
d
is
tin
ct
ap
p
r
o
ac
h
es a
r
e
test
ed
,
an
d
th
e
r
es
u
lts
ar
e
e
v
alu
ated
b
ased
o
n
th
e
c
h
ar
ac
ter
tr
aits
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
13
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
4
6
7
5
-
4
6
8
7
4680
−
C
o
n
tin
u
o
u
s
:
co
n
tin
u
o
u
s
in
th
e
co
n
tex
t
m
ea
n
s
th
at
p
er
s
o
n
ality
tr
ait
s
co
r
es
ar
e
u
s
ed
in
t
h
eir
o
r
i
g
in
al,
u
n
alter
ed
f
o
r
m
,
with
o
u
t
ca
teg
o
r
izatio
n
o
r
tr
an
s
f
o
r
m
atio
n
.
I
n
r
eg
r
ess
io
n
an
aly
s
is
,
th
ese
s
co
r
es
ar
e
d
ir
e
ctly
u
tili
ze
d
as
n
u
m
er
ical
v
a
r
iab
les
to
u
n
d
er
s
tan
d
h
o
w
th
e
y
r
elate
to
o
t
h
er
v
ar
ia
b
les,
with
o
u
t
c
o
n
v
e
r
tin
g
th
em
in
to
ca
teg
o
r
ical
o
r
d
is
cr
ete
g
r
o
u
p
s
.
−
L
AH:
wh
ich
s
tan
d
s
f
o
r
“
lo
w
av
er
ag
e
h
ig
h
,
”
is
a
m
eth
o
d
o
f
c
ateg
o
r
izin
g
in
d
iv
id
u
als
in
to
th
r
ee
g
r
o
u
p
s
b
ased
o
n
th
eir
s
co
r
es
o
r
p
er
f
o
r
m
an
c
e.
T
h
ese
g
r
o
u
p
s
ar
e
ty
p
ically
lab
eled
as
lo
w,
av
er
a
g
e,
a
n
d
h
i
g
h
.
T
h
e
ess
en
ce
o
f
L
AH
is
to
d
iv
id
e
p
ar
ticip
an
ts
in
to
th
ese
th
r
ee
ca
teg
o
r
ies
b
y
estab
lis
h
in
g
p
ar
titi
o
n
b
o
u
n
d
ar
ies
wh
ich
ar
e
d
ef
in
ed
m
at
h
em
atica
lly
as
(
5
)
:
(
)
=
{
ℎ
ℎ
,
>
+
2
;
,
<
+
2
;
,
ℎ
.
(
5
)
−
L
H:
L
H
is
an
ab
b
r
ev
iatio
n
f
o
r
lo
w
h
ig
h
,
wh
ich
r
ef
er
s
to
th
e
d
iv
is
io
n
o
f
p
ar
ticip
a
n
ts
in
to
t
wo
g
r
o
u
p
s
:
th
o
s
e
with
h
ig
h
e
r
s
co
r
es a
n
d
th
o
s
e
with
lo
wer
s
co
r
es.
Ho
wev
er
,
p
ar
ticip
an
ts
wh
o
ar
e
cl
o
s
er
to
th
e
b
o
u
n
d
ar
y
m
ay
r
ec
eiv
e
s
im
ilar
s
co
r
es.
T
h
is
ap
p
r
o
ac
h
is
s
cien
tific
ally
ch
ar
ac
t
er
ized
as a
b
in
ar
y
class
if
icatio
n
m
eth
o
d
.
(
)
=
{
ℎ
ℎ
,
>
;
,
<
;
(
6
)
−
L
HNA:
“
s
tan
d
s
f
o
r
“lo
w
h
ig
h
,
n
o
a
v
er
ag
e”
.
It
i
s
a
ter
m
u
s
ed
to
d
if
f
e
r
en
tiate
b
etwe
en
m
ax
im
u
m
a
n
d
m
in
im
u
m
s
co
r
in
g
b
y
d
is
r
eg
ar
d
in
g
an
y
av
er
a
g
e
v
alu
es.
T
h
is
ap
p
r
o
ac
h
f
o
c
u
s
es
s
o
lely
o
n
ex
tr
em
es,
wh
eth
er
h
ig
h
o
r
lo
w,
with
o
u
t
co
n
s
id
er
i
n
g
th
e
m
id
d
le
g
r
o
u
n
d
r
ep
r
esen
ted
b
y
av
er
ag
e
s
co
r
es.
Ma
th
e
m
atica
lly
,
L
HNA
is
r
ep
r
esen
ted
as
(
7
)
:
(
)
=
{
ℎ
ℎ
,
>
+
2
;
,
<
−
2
;
,
ℎ
.
(
7
)
−
L
HNASD:
lo
w
h
ig
h
,
n
o
a
v
er
a
g
e,
an
d
ev
en
w
h
o
le
s
tan
d
ar
d
d
ev
iatio
n
,
is
a
s
tatis
tical
m
eth
o
d
u
s
ed
to
an
aly
ze
d
ata
b
y
f
o
cu
s
in
g
o
n
ex
tr
em
e
v
alu
es
wh
ile
d
is
r
eg
ar
d
in
g
av
er
ag
es.
I
n
L
HNASD,
th
e
h
ig
h
est
an
d
lo
west
v
alu
es a
r
e
em
p
h
asized
b
y
s
ettin
g
a
th
r
esh
o
ld
at
±
1
s
tan
d
a
r
d
d
ev
iatio
n
f
r
o
m
t
h
e
m
ea
n
.
T
h
is
m
ea
n
s
th
at
d
ata
p
o
in
ts
b
ey
o
n
d
th
is
th
r
esh
o
ld
a
r
e
co
n
s
id
er
ed
s
ig
n
i
f
ican
t
o
u
tli
er
s
.
B
y
d
o
in
g
s
o
,
L
HNASD
aim
s
to
h
ig
h
lig
h
t
ex
tr
em
e
v
alu
es
an
d
d
etec
t
an
y
d
ev
iatio
n
s
f
r
o
m
th
e
ty
p
ical
p
atter
n
in
th
e
d
ata.
L
HNASD
wh
ich
i
s
m
ath
em
atica
lly
r
ep
r
esen
ted
as
(
8
)
:
(
)
=
{
ℎ
ℎ
,
>
+
;
,
<
−
;
,
ℎ
.
(
8
)
3
.
3
.
4
.
F
ea
t
ure
e
x
t
ra
ct
i
o
n ba
s
ed
o
n sentim
ent
f
ea
t
ures
Sen
tim
en
t
r
ef
er
s
to
a
p
er
s
o
n
'
s
f
ee
lin
g
s
o
r
ju
d
g
m
en
ts
.
T
r
ad
itio
n
ally
,
wh
e
n
ev
er
a
p
er
s
o
n
is
l
o
o
k
in
g
to
p
u
r
ch
ase
a
co
m
p
u
ter
,
th
er
e
ar
e
s
ev
er
al
f
ac
t
o
r
s
to
co
n
s
id
er
,
He
co
n
s
u
lts
h
is
f
r
ien
d
s
a
n
d
f
am
ily
m
em
b
e
r
s
f
o
r
ad
v
ice.
Peo
p
le
n
o
wad
a
y
s
,
o
n
t
h
e
o
th
er
h
an
d
,
r
ea
d
r
ev
iews
o
n
th
e
in
te
r
n
et
b
e
f
o
r
e
d
ec
id
in
g
wh
ich
co
m
p
u
ter
t
o
b
u
y
.
Op
i
n
io
n
m
i
n
in
g
/s
en
tim
en
t
an
aly
s
is
ca
p
tu
r
es
a
p
er
s
o
n
'
s
im
p
r
ess
io
n
s
an
d
o
th
er
w
is
e
f
ee
lin
g
s
ab
o
u
t
s
o
m
eth
in
g
lik
e
a
p
r
o
d
u
ct.
Op
i
n
io
n
m
in
i
n
g
an
d
s
en
tim
en
t a
n
aly
s
is
ar
e
two
in
ter
ch
an
g
ea
b
le
ter
m
s
.
Sco
p
e:
en
tr
ep
r
en
eu
r
s
ca
n
u
s
e
s
en
tim
en
t
an
aly
s
is
to
ev
alu
ate
wh
eth
er
a
n
ew
p
r
o
d
u
ct
will
s
u
cc
ee
d
o
r
f
ail.
I
t
ca
n
ass
is
t
th
em
in
d
eter
m
in
in
g
wh
ich
f
ea
tu
r
es
clien
ts
l
ik
e.
Fo
r
e
x
am
p
le,
a
web
s
ite
ev
alu
atio
n
claim
s
th
at
th
e
lap
to
p
is
ef
f
icien
t
b
u
t
th
e
b
atter
y
is
in
ad
eq
u
ate.
T
h
is
ty
p
e
o
f
d
ata
p
r
o
v
id
es
a
clea
r
er
p
i
ctu
r
e
o
f
e
n
d
u
s
er
s
’
o
p
in
io
n
s
,
wh
ic
h
m
ay
b
e
u
s
ed
t
o
im
p
r
o
v
e
f
ea
tu
r
es in
f
u
tu
r
e
r
e
leases.
C
h
allen
g
es
:
th
e
m
o
s
t
d
if
f
icu
l
t
asp
ec
t
o
f
s
en
tim
e
n
t
an
aly
s
i
s
is
am
b
ig
u
o
u
s
w
o
r
d
m
ea
n
in
g
.
I
n
o
n
e
s
ce
n
ar
io
,
a
wo
r
d
m
ay
b
e
p
o
s
it
iv
e,
wh
ile
in
an
o
th
er
,
it
m
ay
b
e
n
eg
ativ
e.
T
ak
e
th
e
ter
m
“
s
h
o
r
t,
”
f
o
r
e
x
am
p
le.
Po
s
itiv
e
f
ee
d
b
ac
k
wo
u
ld
b
e
i
f
a
cu
s
to
m
er
r
ep
o
r
ts
th
at
th
e
co
m
p
u
te
r
’
s
s
tar
tu
p
tim
e
is
f
ast,
wh
ile
n
eg
ativ
e
f
ee
d
b
ac
k
wo
u
ld
b
e
if
a
cu
s
to
m
er
co
m
p
lain
s
ab
o
u
t
th
e
co
m
p
u
ter
’
s
s
h
o
r
t
b
atter
y
life
.
I
t
is
cr
u
cial
to
r
em
em
b
er
th
at
with
o
u
t
a
p
p
r
o
p
r
iate
tr
ain
in
g
an
d
cu
s
to
m
izatio
n
,
a
s
y
s
tem
th
at
h
as
b
ee
n
tr
ain
ed
o
n
o
n
e
s
o
r
t
o
f
p
r
o
d
u
ct
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
S
en
timen
t
-
elec
tr
o
en
ce
p
h
a
lo
g
r
a
m
fu
s
io
n
fo
r
efficien
t p
r
o
d
u
c
t revie
w
p
r
ed
ictio
n
…
(
R
a
h
u
l
K
u
ma
r
S
h
a
r
ma
)
4681
f
ee
d
b
ac
k
ca
n
n
o
t
b
e
u
s
ed
f
o
r
an
o
th
er
p
r
o
d
u
ct.
Su
r
v
ey
s
,
n
e
ws,
wo
r
d
-
of
-
m
o
u
t
h
,
an
d
in
ter
n
al
s
tatis
tic
s
s
u
ch
as
clien
t o
p
in
io
n
s
in
ce
co
n
tact
ce
n
ter
s
,
em
ail,
an
d
o
th
er
s
o
u
r
ce
s
ar
e
u
s
ed
to
o
b
tain
o
p
in
io
n
s
.
3.
4
.
Cla
s
s
if
ica
t
io
n sta
g
e
us
in
g
co
rr
ela
t
io
n
-
ba
s
ed
deep
lea
rning
neura
l net
wo
rk
A
co
r
r
elatio
n
an
al
y
s
is
en
tails
a
s
tatis
t
ical
ev
alu
atio
n
o
f
th
e
s
tr
en
g
th
an
d
d
ir
ec
tio
n
o
f
th
e
r
e
latio
n
s
h
ip
b
etwe
en
two
v
ar
iab
les.
T
h
e
co
r
r
elatio
n
v
alu
es
ty
p
ically
r
a
n
g
e
b
etwe
en
-
1
.
0
a
n
d
1
.
0
,
with
a
n
y
v
alu
e
b
ey
o
n
d
th
is
r
an
g
e
i
n
d
icatin
g
an
er
r
o
r
in
th
e
co
r
r
elat
io
n
ca
lcu
latio
n
.
C
o
r
r
elatio
n
is
a
m
ea
s
u
r
e
o
f
th
e
r
el
atio
n
s
h
ip
b
etwe
e
n
two
s
ets
o
f
d
ata,
an
d
th
e
m
o
s
t
c
o
m
m
o
n
m
ath
em
atica
l
m
ea
s
u
r
e
o
f
co
r
r
elatio
n
is
d
e
p
icted
b
y
(
9
)
,
wh
ic
h
m
ea
s
u
r
es
th
e
lin
ea
r
r
elatio
n
s
h
ip
b
etwe
e
n
th
e
in
p
u
t E
E
G
s
ig
n
als an
d
t
h
e
p
r
o
d
u
ct
r
ev
iews
.
=
∑
(
1
−
2
−
)
(
1
−
2
−
)
√
∑
(
1
−
2
−
)
2
∑
(
1
−
2
−
)
2
(
9
)
W
h
er
e
z
→
co
r
r
elatio
n
co
ef
f
icien
t
o
f
E
E
G
s
ig
n
al
an
d
Am
az
o
n
p
r
o
d
u
ct
r
ev
iew,
1
→
v
alu
es
o
f
th
e
‘
a’
v
ar
iab
l
e
in
a
s
am
p
le,
−
2
→
m
ea
n
o
f
th
e
v
alu
e
o
f
th
e
‘
a’
v
a
r
iab
le,
1
→
v
alu
es
o
f
th
e
‘
b
’
v
a
r
iab
le
in
a
s
am
p
le,
−
2
→
m
ea
n
o
f
th
e
v
alu
e
o
f
th
e
'
b
'
v
ar
iab
le.
Af
ter
f
in
d
in
g
th
e
co
r
r
elatio
n
co
ef
f
icien
ts
o
f
th
e
E
E
G
s
ig
n
al
as
well
as
Am
az
o
n
p
r
o
d
u
ct
r
ev
iews
th
e
o
u
tco
m
es
ar
e
m
o
v
ed
to
th
e
n
e
x
t
s
tag
e
o
f
class
if
icatio
n
wh
ich
is
ex
p
lain
ed
in
d
etail
as
:
f
o
llo
win
g
f
ea
tu
r
e
s
elec
tio
n
u
s
in
g
co
r
r
elatio
n
co
ef
f
icien
ts
,
class
if
icat
io
n
is
p
er
f
o
r
m
ed
u
s
in
g
a
d
ee
p
lear
n
in
g
n
eu
r
al
n
etwo
r
k
(
DNN)
.
Mu
lti
-
l
ay
er
ed
ar
tific
ial
n
eu
r
al
n
etwo
r
k
s
with
h
id
d
en
in
p
u
ts
an
d
o
u
tp
u
ts
m
ak
e
u
p
DNNs.
T
o
m
ax
im
ize
p
a
r
am
eter
lear
n
in
g
,
th
e
lear
n
in
g
p
r
o
c
ess
in
clu
d
es
n
u
m
er
o
u
s
p
r
e
-
tr
ai
n
in
g
a
n
d
f
in
e
-
tu
n
i
n
g
s
tag
es.
T
h
e
p
u
r
p
o
s
e
o
f
th
is
s
tu
d
y
is
t
o
d
is
co
v
e
r
th
e
b
est
weig
h
ts
th
at
ca
n
b
e
u
s
ed
to
p
r
ec
is
ely
cla
s
s
if
y
th
e
o
u
t
p
u
t
b
y
tr
ain
in
g
th
e
d
ata
f
ea
tu
r
es,
s
u
ch
as
E
E
G
s
ig
n
als
a
n
d
Am
az
o
n
p
r
o
d
u
ct
r
ev
iew
d
ata
s
ets.
T
h
e
DNN
ca
teg
o
r
izes
th
e
d
ata
f
o
r
s
en
tim
en
t
an
aly
s
is
b
ased
o
n
th
e
ex
tr
ac
ted
f
ea
tu
r
es.
Fig
u
r
e
2
s
h
o
ws
th
e
DNN
s
tr
u
ctu
r
e
in
d
iag
r
am
m
atic
f
o
r
m
.
Fig
u
r
e
2
.
DNN
s
tr
u
ctu
r
e
3
.
4
.
1
.
Usi
ng
a
deep
belief
net
wo
rk
a
s
a
pre
-
t
ra
ini
ng
s
t
a
g
e
W
e
em
p
lo
y
t
h
e
d
ee
p
a
r
ch
itect
u
r
e
f
ee
d
-
f
o
r
war
d
n
eu
r
al
n
etwo
r
k
(
DB
N)
,
wh
ic
h
c
o
n
tin
u
es
t
o
b
e
a
d
ee
p
ar
ch
itectu
r
e
n
etwo
r
k
with
s
e
v
er
al
d
e
g
r
ee
s
o
f
c
o
n
ce
alm
en
t,
d
u
r
i
n
g
th
e
tr
ai
n
in
g
p
h
ase.
T
h
e
DB
N
in
p
u
t
is
cu
r
r
en
tly
o
n
l
y
o
f
f
er
e
d
with
a
lim
ited
s
et
o
f
f
ea
tu
r
es.
T
h
e
D
B
N
is
g
en
er
ativ
ely
p
r
e
-
t
r
ain
e
d
u
s
in
g
th
e
lear
n
ed
weig
h
ts
af
ter
wh
ich
it
is
u
s
ed
as
its
in
itial
weig
h
ts
.
T
h
e
DB
N
m
o
d
el
ass
er
ts
th
at
it
em
p
lo
y
s
a
n
etwo
r
k
b
ased
o
n
th
e
s
tates
o
f
its
h
id
d
en
u
n
its
an
d
p
er
m
its
th
e
ac
tiv
atio
n
s
to
b
e
m
ad
e
v
is
ib
le.
T
h
is
lay
er
is
m
ad
e
u
p
o
f
an
in
p
u
t
lay
er
th
at
en
clo
s
es
th
e
in
p
u
t
u
n
its
,
s
ev
er
al
h
id
d
en
u
n
its
,
an
d
an
o
u
tp
u
t
lay
er
th
at
m
ea
s
u
r
e
s
o
n
e
u
n
it
f
o
r
ea
ch
class
.
Am
o
n
g
s
t
th
e
lay
er
in
g
u
n
its
−
1
an
d
,
th
e
b
iases
(
)
→
lay
er
,
weig
h
ts
ar
e
th
e
p
ar
am
eter
s
o
f
a
DB
N
.
Op
tim
izin
g
th
e
p
ar
am
eter
s
is
o
n
e
o
f
th
e
m
o
s
t
d
if
f
icu
lt
p
ar
ts
o
f
DNN
tr
ain
in
g
.
T
o
d
ay
,
a
h
y
p
er
p
a
r
am
eter
o
p
tim
izatio
n
tech
n
iq
u
e
is
p
r
o
d
u
ce
d
v
ia
r
an
d
o
m
in
itializatio
n
.
W
e
em
p
lo
y
a
r
estricte
d
b
o
ltzm
an
n
m
ac
h
in
e
t
o
ad
d
r
ess
th
e
is
s
u
e.
DB
N
ca
n
le
ar
n
u
s
in
g
a
s
tack
ed
s
et
o
f
ea
s
y
-
to
-
u
n
d
er
s
tan
d
lear
n
i
n
g
m
o
d
u
l
es.
T
h
e
n
am
e
o
f
th
is
s
tr
aig
h
tf
o
r
war
d
lea
r
n
in
g
m
o
d
u
le
is
r
esu
lts
-
b
ased
m
an
ag
em
en
t
(
RBM
)
.
3
.
4
.
2
.
F
ine
-
t
un
ing
s
t
a
g
e
T
h
e
weig
h
ts
ar
e
f
in
e
-
t
u
n
ed
v
i
a
b
ac
k
p
r
o
p
a
g
atio
n
t
o
en
h
an
ce
th
e
s
ch
em
e
’
s
ac
cu
r
ac
y
o
n
ce
t
h
e
n
etwo
r
k
h
as
b
ee
n
p
r
e
-
tr
ain
ed
as
R
B
M
-
s
tack
m
o
d
el.
T
o
c
h
ar
ac
ter
ize
t
h
e
s
y
s
tem
’
s
p
er
f
o
r
m
a
n
ce
,
an
o
u
tp
u
t
lay
er
is
p
lace
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
13
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
4
6
7
5
-
4
6
8
7
4682
o
n
to
p
o
f
th
e
DNN.
Mo
r
eo
v
er
,
th
e
tr
ain
in
g
d
ataset
is
r
ef
in
ed
u
n
til
th
e
o
p
tim
al
weig
h
t
o
r
h
ig
h
p
er
f
o
r
m
a
n
ce
is
ac
h
iev
ed
.
T
h
e
DNN
class
if
ier
o
f
f
er
s
s
ev
er
al
ad
v
an
tag
es,
in
clu
d
in
g
th
e
ab
ilit
y
to
h
an
d
le
m
is
s
in
g
d
ata
d
u
r
in
g
class
if
icatio
n
,
wh
er
e
it
au
to
m
atica
lly
in
co
r
p
o
r
ates
av
ailab
le
i
n
f
o
r
m
atio
n
f
o
r
f
u
r
th
e
r
p
r
o
ce
s
s
in
g
.
L
ev
er
a
g
in
g
th
e
ac
q
u
ir
ed
f
ea
tu
r
es,
a
d
ee
p
-
lear
n
in
g
n
eu
r
al
n
etwo
r
k
ca
teg
o
r
iz
es
th
e
d
ata
f
o
r
s
en
tim
en
t
a
n
aly
s
is
.
Su
b
s
eq
u
en
tly
,
th
e
weig
h
t p
ar
am
ete
r
f
o
r
ea
ch
lay
er
is
ad
ju
s
ted
u
s
in
g
th
e
E
GOA
.
3
.
5
.
E
ntr
o
py
-
ba
s
ed
g
ra
s
s
ho
pp
er
o
ptim
iza
t
io
n a
lg
o
rit
h
m
I
n
th
is
ap
p
r
o
ac
h
,
an
E
GOA
is
em
p
lo
y
ed
to
o
p
tim
ize
h
y
p
e
r
p
ar
am
eter
s
.
Utilizin
g
th
e
en
t
r
o
p
y
-
b
ased
g
r
ass
h
o
p
p
er
alg
o
r
ith
m
,
th
e
f
o
l
lo
win
g
p
r
o
ce
s
s
in
g
s
tep
s
ar
e
ex
ec
u
ted
to
y
ield
o
p
tim
al
r
esu
lts
.
B
y
em
p
lo
y
in
g
a
h
y
b
r
id
DNN
th
at
in
teg
r
ates
m
a
ch
in
e
lear
n
in
g
with
E
E
G
s
ig
n
a
ls
an
d
Am
az
o
n
p
r
o
d
u
ct
r
e
v
iew
d
ata,
th
e
o
b
jectiv
e
is
to
co
n
d
u
ct
s
en
tim
en
t a
n
aly
s
is
ef
f
ec
tiv
ely
.
3
.
5
.
1
.
E
ntr
o
py
ca
lcula
t
io
n
C
alcu
latin
g
en
tr
o
p
y
is
a
k
ey
tak
ea
way
:
en
tr
o
p
y
(
E
NT
R
)
is
a
p
r
o
b
ab
ilis
tic
m
ea
s
u
r
e
o
f
d
is
o
r
d
er
o
r
r
an
d
o
m
n
ess
in
a
m
ac
r
o
s
co
p
i
c
s
y
s
tem
,
r
ef
lectin
g
th
e
n
u
m
b
er
o
f
p
o
s
s
ib
le
ar
r
an
g
em
e
n
ts
o
f
its
co
n
s
titu
en
t
m
o
lecu
les.
I
t
q
u
an
tifie
s
th
e
s
y
s
tem
’
s
ten
d
en
cy
to
m
o
v
e
to
war
d
s
a
m
o
r
e
d
is
o
r
d
e
r
ed
s
tate.
Ma
th
em
atica
lly
,
en
tr
o
p
y
ca
n
b
e
ca
lcu
lated
as
th
e
n
atu
r
al
lo
g
ar
ith
m
o
f
th
e
n
u
m
b
e
r
o
f
p
o
s
s
ib
le
co
n
f
ig
u
r
atio
n
s
d
iv
id
ed
b
y
B
o
ltzm
an
n
’
s
co
n
s
tan
t a
s
s
u
m
in
g
ea
ch
co
n
f
ig
u
r
atio
n
is
eq
u
all
y
p
r
o
b
ab
le:
=
=
(
1
0
)
E
NT
R
→
en
tr
o
p
y
,
→
B
o
ltzm
an
n
c
o
n
s
tan
t,
NL
→
n
atu
r
al
lo
g
ar
ith
m
,
→
n
u
m
b
e
r
o
f
m
icr
o
s
co
p
ic
co
n
f
ig
u
r
atio
n
s
.
T
h
e
b
est s
o
lu
tio
n
is
cr
ea
ted
u
s
in
g
th
e
p
r
o
ce
d
u
r
es lis
ted
b
elo
w
af
ter
th
e
en
tr
o
p
y
v
alu
e
h
as b
ee
n
ca
lcu
lated
.
T
h
e
f
i
r
s
t step
,
s
o
lu
tio
n
cr
ea
tio
n
,
is
d
escr
ib
ed
in
d
e
tail a
s
f
o
llo
ws
:
Step
1
:
I
n
itializatio
n
:
T
o
o
p
ti
m
ize
th
e
h
y
p
e
r
p
ar
am
ete
r
s
,
th
e
E
GOA
alg
o
r
ith
m
b
eg
i
n
s
b
y
p
o
p
u
latin
g
th
e
s
o
lu
tio
n
with
an
y
n
u
m
b
er
o
f
p
eo
p
le.
An
o
p
tim
izatio
n
al
g
o
r
ith
m
’
s
s
o
lu
tio
n
d
ev
elo
p
m
e
n
t
is
a
cr
u
cial
s
tag
e
th
at
aid
s
in
th
e
r
ap
id
id
en
tific
atio
n
o
f
th
e
b
est
s
o
lu
tio
n
.
W
e
s
elec
t
o
p
tim
al
weig
h
t
s
ettin
g
s
f
r
o
m
am
o
n
g
th
e
n
r
an
g
e
o
f
f
ea
t
u
r
es
an
d
weig
h
ts
in
ea
ch
d
ata
s
et.
So
lu
tio
n
g
e
n
er
atio
n
is
b
ased
o
n
th
e
c
h
o
s
en
f
ea
tu
r
es
a
n
d
t
h
eir
weig
h
ts
.
Ass
u
m
e
th
at
ea
ch
f
lo
wer
h
as
N
p
o
llen
a
n
d
th
at
th
e
r
e
ar
e
M
f
lo
wer
s
o
r
s
o
lu
tio
n
s
in
t
h
e
to
tal
p
o
p
u
latio
n
.
T
h
e
p
o
p
u
latio
n
o
f
f
l
o
wer
s
as
a
wh
o
le
is
d
ep
icted
as
=
(
1
,
2
,
.
.
.
,
)
,
an
y
wh
er
e,
is
th
e
ℎ
f
lo
wer
,
an
d
∈
[
1
,
]
is
th
e
f
l
o
wer
’
s
p
o
p
u
latio
n
’
s
o
r
d
er
n
u
m
b
er
.
E
ac
h
f
lo
wer
h
as
a
d
if
f
er
e
n
t
r
e
p
r
es
en
tatio
n
=
(
1
,
2
,
.
.
.
.
,
)
,
wh
er
e,
is
th
e
ℎ
p
o
llen
(
weig
h
t
p
ar
am
eter
)
in
t
h
e
ℎ
f
lo
wer
,
an
d
∈
[
1
,
]
is
th
e
p
o
llen
in
d
e
x
.
T
h
e
o
b
tain
e
d
an
s
wer
will b
e
u
s
ed
in
th
e
n
e
x
t stag
e,
wh
ich
is
th
e
f
itn
ess
ass
es
s
m
en
t.
Step
2
:
f
itn
ess
ca
lcu
latio
n
:
th
e
f
itn
ess
f
u
n
ctio
n
e
v
alu
ates
th
e
s
o
lu
tio
n
af
ter
it
is
g
e
n
er
ated
,
an
d
th
e
n
s
elec
ts
th
e
b
est o
p
tio
n
.
T
o
f
i
n
d
th
e
o
p
tim
u
m
s
o
lu
tio
n
,
t
h
e
o
p
t
im
izatio
n
alg
o
r
ith
m
p
r
im
ar
ily
r
elies o
n
its
f
itn
ess
f
u
n
ctio
n
.
A
m
ajo
r
f
ea
tu
r
e
o
f
E
GOA
is
th
e
ch
o
ice
o
f
f
itn
ess
.
=
1
∗
(
1
1
)
W
h
er
e,
W
F→
weig
h
t
f
ac
to
r
,
E
NT
R
V→
en
tr
o
p
y
v
alu
e
,
ea
c
h
p
er
s
o
n
’
s
f
itn
ess
lev
el
is
ass
e
s
s
ed
.
an
d
s
av
ed
f
o
r
f
u
tu
r
e
r
e
f
er
en
ce
o
n
ce
t
h
e
f
ir
s
t
s
o
lu
tio
n
,
as
well
as
th
e
o
p
p
o
s
in
g
s
o
lu
tio
n
,
h
as
b
ee
n
d
ev
el
o
p
ed
.
T
h
e
e
q
u
atio
n
is
u
s
ed
to
cr
ea
te
t
h
e
f
itn
ess
f
u
n
c
tio
n
(
1
1
)
.
T
h
e
s
o
lu
tio
n
o
b
tain
ed
af
ter
th
e
f
itn
ess
f
u
n
ctio
n
c
o
m
p
u
tatio
n
is
th
e
n
g
iv
en
f
o
r
th
e
e
v
alu
atio
n
o
f
th
e
u
p
d
ate
p
h
ase,
wh
ich
c
o
m
es n
ex
t.
Step
3
:
OGWF
S
-
b
ased
u
p
d
ati
n
g
s
o
lu
tio
n
:
in
t
h
e
g
r
ass
h
o
p
p
er
o
p
tim
izatio
n
alg
o
r
ith
m
(
GOA
)
,
u
p
d
atin
g
th
e
s
o
lu
tio
n
b
ased
o
n
en
tr
o
p
y
ty
p
ically
in
v
o
lv
es
ad
j
u
s
tin
g
th
e
p
o
s
itio
n
s
o
f
th
e
g
r
ass
h
o
p
p
er
s
in
th
e
p
o
p
u
latio
n
to
en
co
u
r
a
g
e
ex
p
lo
r
atio
n
o
f
t
h
e
s
o
lu
tio
n
s
p
ac
e.
In
(
1
2
)
is
lik
ely
a
m
ath
em
atica
l
r
ep
r
esen
tatio
n
u
s
ed
f
o
r
t
h
is
p
u
r
p
o
s
e.
L
et
’
s
d
en
o
te
t
h
e
cu
r
r
en
t p
o
s
itio
n
o
f
a
g
r
ass
h
o
p
p
er
i
.
=
+
+
,
i
=
1,2,...,N
(
1
2
)
W
h
er
e,
in
d
icate
s
th
e
p
o
s
itio
n
o
f
th
e
ℎ
g
r
ass
h
o
p
p
er
,
r
ep
r
esen
ts
th
e
s
o
cial
in
te
r
ac
tio
n
b
etwe
en
g
r
ass
h
o
p
p
er
s
,
d
en
o
tes
th
e
g
r
av
itatio
n
al
f
o
r
ce
o
n
th
e
ℎ
g
r
ass
h
o
p
p
er
,
an
d
is
th
e
win
d
ad
v
ec
tio
n
.
In
(
12
)
ca
n
b
e
r
ewr
itten
t
o
in
tr
o
d
u
ce
a
r
an
d
o
m
b
eh
a
v
io
r
o
f
g
r
ass
h
o
p
p
er
s
as
(
1
3
)
:
=
∑
(
)
^
,
d
b
ij
=
|
−
|
=
1
,
≠
(
1
3
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
Ar
tif
I
n
tell
I
SS
N:
2252
-
8
9
3
8
S
en
timen
t
-
elec
tr
o
en
ce
p
h
a
lo
g
r
a
m
fu
s
io
n
fo
r
efficien
t p
r
o
d
u
c
t revie
w
p
r
ed
ictio
n
…
(
R
a
h
u
l
K
u
ma
r
S
h
a
r
ma
)
4683
A
r
an
g
e
b
etwe
en
ℎ
as
well
as
ℎ
g
r
ass
h
o
p
p
er
s
is
in
d
icate
d
b
y
in
a
n
y
s
itu
atio
n
,
d
b
ij
=
|
−
|
d
en
o
tes
th
e
E
u
clid
ea
n
d
is
tan
ce
b
etwe
en
ℎ
an
d
ℎ
g
r
ass
h
o
p
p
er
,
d
en
o
tes th
e
n
u
m
b
er
o
f
g
r
ass
h
o
p
p
e
r
s
,
s
,
o
n
th
e
o
th
er
h
an
d
,
d
en
o
tes th
e
in
te
n
s
ity
o
f
t
h
e
s
o
cial
f
o
r
ce
s
f
u
n
ctio
n
,
t
h
at
ca
n
b
e
ex
p
r
ess
ed
m
ath
em
atica
lly
as
(
1
4
)
:
(
)
=
−
−
−
(
1
4
)
An
y
wh
er
e,
⟶
in
ten
s
ity
attr
ac
ti
o
n
,
an
d
⟶
s
ca
le
with
an
ap
p
ea
lin
g
len
g
th
,
r
esp
ec
tiv
ely
.
T
h
e
s
o
cial
in
ter
ac
tio
n
am
o
n
g
g
r
ass
h
o
p
p
e
r
s
ca
n
b
e
ch
ar
ac
ter
ized
b
y
attr
ac
tio
n
an
d
r
ep
u
ls
io
n
.
an
d
d
e
s
cr
ib
es
b
o
th
g
r
av
ity
f
o
r
ce
as
well
as
win
d
ad
v
ec
tio
n
ag
ai
n
f
o
r
th
i
g
r
ass
h
o
p
p
e
r
s
,
wh
ich
co
u
ld
b
e
q
u
a
n
titativ
ely
d
escr
ib
ed
as
m
o
n
ito
r
s
.
=
−
^
,
=
^
(
1
5
)
=
(
∑
−
2
(
|
−
|
)
−
=
1
,
≠
)
+
^
(
1
6
)
W
h
er
ev
er
,
⟶
co
n
s
tan
t
o
f
g
r
av
itatio
n
an
d
⟶
co
n
s
tan
t
d
r
if
t,
alth
o
u
g
h
^
an
d
^
r
ep
r
esen
t
th
e
win
d
d
ir
ec
tio
n
an
d
also
th
e
u
n
ity
v
ec
to
r
to
th
e
ea
r
t
h
'
s
ce
n
ter
,
⟶
Up
p
er
b
o
u
n
d
o
f
th
e
weig
h
t,
⟶
lo
wer
b
o
u
n
d
o
f
th
e
weig
h
t,
d
T
^
⟶
th
e
b
est
s
o
lu
tio
n
in
t
h
e
ℎ
d
im
en
s
io
n
s
p
ac
e,
is
th
e
s
o
cial
in
ter
ac
tio
n
b
et
wee
n
g
r
ass
h
o
p
p
er
s
r
esp
ec
tiv
ely
.
Step
4
:
ter
m
in
atio
n
cr
iter
ia
Af
ter
ex
ec
u
tin
g
th
e
o
p
tim
izatio
n
alg
o
r
ith
m
,
th
e
d
escr
ib
e
d
p
r
o
ce
s
s
es,
s
u
ch
as
f
itn
ess
ev
alu
atio
n
,
s
o
lu
tio
n
u
p
d
ates,
an
d
ter
m
in
atio
n
cr
iter
ia,
ar
e
r
ep
ea
ted
iter
ativ
ely
u
n
til
th
e
d
esire
d
o
u
tco
m
e
is
ac
h
ie
v
ed
.
T
h
is
o
u
t
co
m
e
co
u
l
d
b
e
th
e
id
e
n
tific
atio
n
o
f
an
o
p
tim
al
r
esp
o
n
s
e
o
r
a
s
et
o
f
ch
ar
ac
ter
is
tics
th
at
m
ee
t
th
e
s
p
ec
if
ie
d
cr
iter
ia
o
r
o
b
jectiv
es
o
f
th
e
o
p
tim
izat
io
n
p
r
o
b
lem
.
On
ce
th
is
o
p
tim
i
ze
d
s
o
lu
tio
n
o
r
s
et
o
f
f
ea
tu
r
es
is
o
b
tain
e
d
,
it
ca
n
b
e
f
u
r
th
e
r
an
aly
ze
d
,
im
p
lem
e
n
ted
,
o
r
u
tili
ze
d
ac
co
r
d
in
g
to
t
h
e
s
p
ec
if
ic
g
o
als
o
f
th
e
o
p
tim
izatio
n
task
.
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
Ass
es
s
th
e
r
esu
lts
o
f
ef
f
ec
tiv
e
s
en
tim
en
t
an
aly
s
is
u
tili
zin
g
o
n
e
o
f
th
e
C
DNN
d
etailed
in
t
h
is
ch
ap
ter
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
o
l
o
g
y
i
s
im
p
lem
en
ted
u
s
in
g
Py
t
h
o
n
,
an
d
a
s
er
ies
o
f
test
s
wer
e
c
o
n
d
u
cte
d
o
n
a
PC
o
p
er
atin
g
o
n
W
in
d
o
ws
7
,
e
q
u
i
p
p
ed
with
a
2
GHz
d
u
al
-
co
r
e
p
r
o
ce
s
s
o
r
,
4
GB
R
AM
,
r
u
n
n
in
g
a
6
4
-
b
it
v
e
r
s
io
n
o
f
W
in
d
o
ws 2
0
0
7
.
4
.
1
.
Da
t
a
s
et
des
cr
iptio
n
4
.
1
.
1
.
Am
a
z
o
n pro
du
ct
re
v
iew
T
h
e
1
4
2
.
8
m
illi
o
n
r
ev
iews
co
llected
b
etwe
en
Ma
y
1
9
9
6
a
n
d
J
u
ly
2
0
1
4
f
o
r
th
e
Am
az
o
n
p
r
o
d
u
ct
r
ev
iew
d
ataset
en
co
m
p
ass
p
r
o
d
u
ct
r
e
v
iews
an
d
Am
az
o
n
-
r
elate
d
in
f
o
r
m
atio
n
.
T
h
is
d
ataset
p
r
o
v
id
es
d
iv
er
s
e
d
etails
ab
o
u
t
p
r
o
d
u
cts,
in
clu
d
in
g
t
h
eir
d
escr
ip
tio
n
s
,
ca
teg
o
r
ie
s
,
p
r
i
ce
,
b
r
an
d
,
an
d
im
a
g
e
p
r
o
p
er
ti
es.
Fu
r
th
er
m
o
r
e,
it
in
clu
d
es lin
k
s
to
r
elate
d
p
r
o
d
u
cts th
at
cu
s
to
m
er
s
h
av
e
v
iewe
d
o
r
p
u
r
ch
ased
.
4
.
1
.
2
.
E
lect
ro
ence
ph
a
lo
g
ra
m
s
ig
na
l
T
h
e
o
u
tco
m
es
o
f
o
u
r
an
aly
s
is
ar
e
p
r
esen
ted
in
th
is
s
ec
tio
n
u
s
in
g
th
e
C
DNN
-
b
ased
ef
f
icien
t
p
r
ed
ictio
n
tech
n
iq
u
e
f
o
r
an
aly
zi
n
g
th
e
s
en
tim
en
t
o
f
an
E
E
G
s
ig
n
al.
Fo
r
th
is
s
tu
d
y
,
we
u
tili
ze
d
E
E
G
s
i
g
n
al
d
atab
ases
.
T
h
e
d
ataset
co
n
tain
ed
1
1
,
5
0
0
s
am
p
les,
with
ea
ch
s
am
p
le
h
av
in
g
1
7
8
p
r
o
p
er
ties
th
at
f
o
llo
wed
a
n
o
r
m
al
d
is
tr
ib
u
tio
n
.
4
.
2
.
E
v
a
lua
t
i
o
n
re
s
ult
T
ab
les
3
an
d
4
p
r
esen
t
th
e
ac
cu
r
ac
y
an
d
lo
g
lo
s
s
v
alu
es
o
f
th
e
E
E
G
s
ig
n
al
an
d
Am
az
o
n
p
r
o
d
u
ct
r
ev
iew
d
atasets
ac
r
o
s
s
tr
ain
in
g
,
v
alid
atio
n
,
a
n
d
test
in
g
p
h
ases
.
Ou
r
s
u
g
g
ested
E
E
G
s
ig
n
al
d
ataset
ac
h
iev
ed
a
tr
ain
i
n
g
d
ata
ac
cu
r
ac
y
o
f
9
8
.
3
5
%
with
a
lo
g
l
o
s
s
o
f
0
.
0
4
5
,
wh
ile
v
alid
atio
n
d
ata
ac
c
u
r
ac
y
an
d
lo
g
lo
s
s
wer
e
9
8
.
2
0
%
an
d
0
.
0
4
6
,
r
esp
ec
tiv
ely
.
Fo
r
th
e
te
s
t
d
ata,
ac
cu
r
ac
y
an
d
lo
g
lo
s
s
v
alu
es
wer
e
9
8
.
2
0
%
an
d
0
.
0
5
7
,
r
esp
ec
tiv
ely
.
On
th
e
o
th
e
r
h
an
d
,
th
e
Am
az
o
n
p
r
o
d
u
ct
r
ev
iew
d
ataset
ex
h
i
b
ited
tr
ain
in
g
d
ata
ac
c
u
r
ac
y
an
d
l
o
g
lo
s
s
o
f
9
8
.
5
5
%
an
d
0
.
0
5
4
,
r
esp
ec
tiv
ely
.
Valid
atio
n
d
ata
ac
cu
r
ac
y
a
n
d
l
o
g
lo
s
s
wer
e
9
8
.
3
5
%
an
d
0
.
0
5
8
,
w
h
ile
test
d
ata
ac
cu
r
ac
y
an
d
lo
g
lo
s
s
s
to
o
d
at
9
8
.
4
0
%
an
d
0
.
0
6
1
,
r
esp
ec
tiv
ely
.
O
u
r
p
r
o
p
o
s
ed
s
y
s
tem
o
u
tp
e
r
f
o
r
m
s
e
x
is
tin
g
ap
p
r
o
ac
h
es
b
ased
o
n
th
ese
r
esu
lts
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
13
,
No
.
4
,
Dec
em
b
er
2
0
2
4
:
4
6
7
5
-
4
6
8
7
4684
T
ab
le
3
.
T
h
e
E
E
G
tr
ain
in
g
,
v
a
lid
atio
n
,
an
d
test
in
g
d
ataset’
s
ac
cu
r
ac
y
an
d
lo
g
lo
s
s
v
alu
es
T
ab
le
4
.
T
h
e
p
r
o
d
u
ct
r
ev
iew’
s
tr
ain
in
g
,
v
alid
atio
n
,
an
d
test
in
g
o
f
d
atasets
'
ac
cu
r
a
cy
an
d
l
o
g
lo
s
s
v
alu
es
D
a
t
a
s
e
t
A
c
c
u
r
a
c
y
(
%)
Lo
g
l
o
ss
Tr
a
i
n
i
n
g
d
a
t
a
9
8
.
3
5
0
.
0
4
5
V
a
l
i
d
a
t
i
o
n
d
a
t
a
9
8
.
2
0
0
.
0
4
6
Te
st
d
a
t
a
9
8
.
2
0
0
.
0
5
7
D
a
t
a
s
e
t
A
c
c
u
r
a
c
y
(
%)
Lo
g
l
o
ss
Tr
a
i
n
i
n
g
d
a
t
a
9
8
.
5
5
0
.
0
5
4
V
a
l
i
d
a
t
i
o
n
d
a
t
a
9
8
.
3
5
0
.
0
5
8
Te
st
d
a
t
a
9
8
.
4
0
0
.
0
6
1
4
.
3
.
E
x
perim
ent
a
l
re
s
ul
t
T
h
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
s
u
g
g
e
s
ted
s
tr
ateg
y
wh
en
u
tili
zin
g
th
e
f
o
llo
win
g
co
n
f
ig
u
r
atio
n
is
s
h
o
wn
in
t
h
e
ac
co
m
p
an
y
in
g
F
ig
u
r
es
3
to
9
.
Fig
u
r
e
3
d
ep
icts
th
e
p
er
f
o
r
m
a
n
ce
an
d
co
m
p
ar
ativ
e
a
n
aly
s
is
o
f
ac
cu
r
ac
y
v
alu
es.
Acc
u
r
ac
y
s
er
v
es
as
a
m
ea
s
u
r
e
o
f
tr
u
e
o
u
tco
m
es
with
in
a
g
en
er
al
p
o
p
u
latio
n
,
en
c
o
m
p
ass
in
g
tr
u
e
p
o
s
itiv
es
o
r
tr
u
e
n
eg
ativ
es,
an
d
it
ev
alu
ate
s
th
e
p
r
ec
is
io
n
o
f
d
ata
class
if
i
ca
tio
n
.
T
h
e
an
aly
s
is
in
d
icate
s
th
at
o
u
r
p
r
o
p
o
s
ed
m
eth
o
d
o
l
o
g
y
s
u
r
p
ass
es
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
(
C
NN
)
,
lo
n
g
s
h
o
r
t
-
te
r
m
m
e
m
o
r
y
(
L
STM
)
,
m
u
lticlas
s
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
M
-
SVM
)
,
an
d
b
id
ir
ec
tio
n
al
en
c
o
d
er
r
ep
r
esen
tatio
n
s
f
r
o
m
tr
a
n
s
f
o
r
m
er
s
(
B
E
R
T
)
m
eth
o
d
s
in
ter
m
s
o
f
r
esu
lts
.
Fig
u
r
e
3
. P
er
f
o
r
m
an
c
e
an
d
co
m
p
ar
ativ
e
an
aly
s
is
o
f
ac
c
u
r
ac
y
Fig
u
r
es
4
to
6
illu
s
tr
ate
th
e
p
r
ec
is
io
n
,
r
ec
all,
an
d
F
-
s
co
r
e
p
lo
ts
.
T
h
e
an
aly
s
is
r
ev
ea
ls
th
at
o
u
r
p
r
o
p
o
s
ed
m
eth
o
d
ex
h
ib
its
co
n
s
is
ten
t im
p
r
o
v
em
e
n
t a
cr
o
s
s
p
r
ec
is
io
n
,
r
e
ca
ll,
tim
e,
an
d
F
-
s
co
r
e
m
etr
ics
wh
en
co
m
p
ar
ed
to
o
th
er
m
eth
o
d
o
lo
g
ies.
T
h
ese
r
e
s
u
lts
d
em
o
n
s
tr
ate
th
e
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
o
f
o
u
r
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
o
v
er
C
NN,
L
STM
,
M
-
SVM,
an
d
B
E
R
T
m
eth
o
d
s
in
ter
m
s
o
f
p
r
ec
is
io
n
,
r
ec
all,
tim
e,
an
d
F1
-
s
co
r
e
r
ate
s
.
Fig
u
r
e
7
s
cr
u
tin
izes
th
e
p
er
f
o
r
m
an
ce
an
d
c
o
m
p
ar
is
o
n
o
f
lo
g
lo
s
s
v
alu
es.
L
o
g
lo
s
s
s
er
v
es
as
a
m
etr
ic
ass
es
s
in
g
th
e
p
r
o
x
im
ity
o
f
th
e
p
r
ed
icted
p
r
o
b
a
b
ilit
y
to
th
e
tr
u
e
n
u
m
b
e
r
,
ty
p
ically
0
o
r
1
in
b
i
n
ar
y
class
if
icatio
n
.
A
h
ig
h
er
lo
g
lo
s
s
v
alu
e
s
ig
n
if
i
es
a
g
r
ea
ter
d
is
p
ar
ity
b
etwe
en
th
e
ex
p
ec
ted
a
n
d
ac
tu
al
p
r
o
b
a
b
ilit
y
.
T
h
e
an
aly
s
is
r
ev
ea
ls
th
at
o
u
r
p
r
o
p
o
s
ed
m
et
h
o
d
o
lo
g
y
o
u
t
p
er
f
o
r
m
s
C
NN,
L
STM
,
M
-
SVM,
an
d
B
E
R
T
te
ch
n
iq
u
es i
n
ter
m
s
o
f
p
er
f
o
r
m
an
ce
.
Fig
u
r
e
4
.
Per
f
o
r
m
an
c
e
an
d
co
m
p
ar
ativ
e
an
aly
s
is
o
f
p
r
ec
is
io
n
Fig
u
r
e
5
.
Per
f
o
r
m
an
c
e
an
d
co
m
p
ar
ativ
e
an
aly
s
is
o
f
r
ec
all
Evaluation Warning : The document was created with Spire.PDF for Python.