I
AE
S In
t
er
na
t
io
na
l J
o
urna
l o
f
Art
if
icia
l In
t
ellig
ence
(
I
J
-
AI
)
Vo
l.
1
5
,
No
.
4
,
A
u
g
u
s
t 2
0
2
6
,
p
p
.
3
9
3
4
~
3
9
4
3
I
SS
N:
2
2
5
2
-
8
9
3
8
,
DOI
: 1
0
.
1
1
5
9
1
/ijai.v
15
.i
4
.
p
p
3
9
3
4
-
3
9
4
3
3934
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
i
.
ia
esco
r
e.
co
m
Em
o
tion recog
nit
io
n of elec
troence
pha
lo
g
ra
m using
hybrid
co
nv
o
lutiona
l neural networks
and
v
isio
n t
ra
nsfo
rme
r
E
s
m
er
a
lda
Co
nte
s
s
a
Dj
a
m
a
l
1
,
Rev
a
n Vio
E
nd
ria
ns
y
a
h
1
,
Da
s
wa
ra
Dj
a
j
a
s
a
s
m
it
a
2
1
D
e
p
a
r
t
me
n
t
o
f
I
n
f
o
r
mat
i
c
s
,
F
a
c
u
l
t
y
o
f
S
c
i
e
n
c
e
a
n
d
I
n
f
o
r
ma
t
i
c
s,
U
n
i
v
e
r
si
t
a
s
Jen
d
e
r
a
l
A
c
h
ma
d
Y
a
n
i
,
C
i
m
a
h
i
,
I
n
d
o
n
e
s
i
a
2
M
e
d
i
c
a
l
E
d
u
c
a
t
i
o
n
P
r
o
g
r
a
m
,
F
a
c
u
l
t
y
o
f
M
e
d
i
c
i
n
e
,
U
n
i
v
e
r
si
t
a
s J
e
n
d
e
r
a
l
A
c
h
ma
d
Y
a
n
i
,
C
i
m
a
h
i
,
I
n
d
o
n
e
si
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Dec
1
6
,
2
0
2
5
R
ev
is
ed
J
u
n
1
7
,
2
0
2
6
Acc
ep
ted
J
u
l 9
,
2
0
2
6
El
e
c
tro
e
n
c
e
p
h
a
lo
g
ra
m
(EE
G
)
sig
n
a
l
-
b
a
se
d
e
m
o
ti
o
n
c
las
sifica
ti
o
n
fa
c
e
s
c
h
a
ll
e
n
g
e
s
in
sp
a
ti
o
-
tem
p
o
ra
l
c
o
m
p
lex
it
y
a
n
d
in
ter
-
c
h
a
n
n
e
l
re
d
u
n
d
a
n
c
y
.
Th
is
p
a
p
e
r
p
r
o
p
o
se
s
a
2
D
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
ra
l
n
e
two
rk
(CNN
)
a
n
d
v
isio
n
tran
sfo
rm
e
r
(ViT)
a
p
p
ro
a
c
h
.
Re
lev
a
n
t
e
m
o
ti
o
n
s
re
q
u
ire
si
g
n
a
l
e
x
trac
ti
o
n
i
n
th
e
4
-
40
Hz
fre
q
u
e
n
c
y
b
a
n
d
u
si
n
g
t
h
e
d
isc
re
te
wa
v
e
let
tran
sfo
r
m
(DWT
).
Th
is
stu
d
y
u
se
s
th
e
S
EE
D
d
a
t
a
se
t
fro
m
1
5
su
b
jec
ts,
with
6
2
c
h
a
n
n
e
ls
re
d
u
c
e
d
t
o
1
2
.
E
v
e
ry
5
-
se
c
o
n
d
se
g
m
e
n
t
is
d
e
c
o
m
p
o
se
d
u
sin
g
DW
T
in
to
fo
u
r
fre
q
u
e
n
c
y
b
a
n
d
s,
wh
ich
a
re
th
e
n
m
a
p
p
e
d
to
a
2D
sp
a
ti
a
l
re
p
re
se
n
tatio
n
fo
r
CNN
s
a
n
d
ViTs.
E
x
p
e
rime
n
t
r
e
su
lt
s
sh
o
w
t
h
a
t
th
e
DW
T
-
2D
CNN
-
ViT
a
c
h
iev
e
s
th
e
b
e
st
a
c
c
u
ra
c
y
o
f
8
7
%
wit
h
a
fin
a
l
lo
ss
o
f
0
.
0
8
7
.
I
n
th
e
m
e
a
n
ti
m
e
,
th
e
CNN
-
o
n
l
y
m
o
d
e
l
(8
5
%
),
th
e
ViT
-
o
n
ly
m
o
d
e
l
(
7
7
%
),
a
n
d
th
e
DWT
-
o
n
l
y
m
o
d
e
l
(4
7
%
).
Tes
ti
n
g
wit
h
se
v
e
ra
l
o
p
ti
m
ize
rs
a
lso
s
h
o
ws
t
h
a
t
Ad
a
m
W
p
ro
v
id
e
s
t
h
e
h
i
g
h
e
st
p
e
rfo
rm
a
n
c
e
with
th
e
fa
ste
st
train
i
n
g
ti
m
e
o
f
1
2
.
3
4
m
in
u
tes
.
T
h
e
se
re
su
lt
s
d
e
m
o
n
stra
te
th
a
t
t
h
is
i
n
teg
ra
ti
o
n
c
a
n
p
ro
d
u
c
e
m
o
re
e
fficie
n
t
a
n
d
sta
b
le
lea
rn
in
g
.
Th
e
se
fin
d
in
g
s
in
d
ica
te
th
a
t
th
e
c
o
m
b
in
a
ti
o
n
o
f
DWT
a
n
d
th
e
C
NN
-
ViT
h
y
b
ri
d
a
rc
h
it
e
c
tu
re
is
e
ffe
c
ti
v
e
fo
r
a
c
c
u
ra
te,
sta
b
le
EE
G
-
b
a
se
d
e
m
o
ti
o
n
re
c
o
g
n
it
i
o
n
a
n
d
is
p
o
ten
ti
a
ll
y
a
p
p
li
c
a
b
le
to
re
a
l
-
ti
m
e
e
m
o
ti
o
n
-
m
o
n
i
to
ri
n
g
sy
ste
m
s
.
K
ey
w
o
r
d
s
:
C
o
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
E
lectr
o
en
ce
p
h
al
o
g
r
am
E
m
o
tio
n
Vis
io
n
tr
an
s
f
o
r
m
er
W
av
elet
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
:
E
s
m
er
ald
a
C
o
n
tess
a
Djam
al
Dep
ar
tm
en
t o
f
I
n
f
o
r
m
atics,
Facu
lty
o
f
Scien
ce
a
n
d
I
n
f
o
r
m
atics,
Un
iv
er
s
itas
J
en
d
er
al
Ach
m
ad
Yan
i
C
im
ah
i,
I
n
d
o
n
esia
E
m
ail:
esm
er
ald
a.
co
n
tess
a@
le
ctu
r
e.
u
n
jan
i.a
c
.
id
1.
I
NT
RO
D
UCT
I
O
N
Au
to
m
atic
em
o
tio
n
r
ec
o
g
n
itio
n
is
b
ec
o
m
in
g
n
ec
ess
ar
y
f
o
r
e
ar
ly
d
etec
tio
n
o
f
em
o
tio
n
al
d
is
o
r
d
er
s
an
d
th
er
ap
y
m
o
n
ito
r
in
g
[
1
]
,
a
n
d
as
a
d
r
iv
e
r
o
f
b
r
ain
-
c
o
m
p
u
te
r
in
ter
f
ac
es
(
B
C
I
s
)
[
2
]
.
O
n
e
s
o
u
r
ce
o
f
em
o
tio
n
al
in
f
o
r
m
atio
n
r
ef
lectin
g
a
p
er
s
o
n
'
s
tr
u
e
an
d
o
b
jectiv
e
c
o
n
d
iti
o
n
is
elec
tr
ical
s
ig
n
als
in
th
e
b
r
ain
[
3
]
,
r
ec
o
r
d
e
d
v
ia
an
elec
tr
o
en
ce
p
h
alo
g
r
am
(
E
E
G)
.
Acc
o
r
d
in
g
ly
,
c
u
r
r
e
n
t r
e
s
ea
r
ch
o
n
em
o
tio
n
r
ec
o
g
n
itio
n
is
o
f
in
ter
est
[
4
]
.
T
h
e
ch
allen
g
e
in
e
m
o
tio
n
r
e
co
g
n
itio
n
is
th
at
E
E
G
s
ig
n
al
s
d
o
n
o
t
h
a
v
e
a
s
tan
d
ar
d
s
h
a
p
e
an
d
ar
e
non
-
s
tatio
n
ar
y
[
5
]
,
an
d
ac
r
o
s
s
m
u
ltip
le
ch
an
n
els,
d
ata
r
e
d
u
n
d
a
n
cy
in
cr
ea
s
es
th
e
co
m
p
lex
ity
o
f
em
o
tio
n
id
en
tific
atio
n
f
r
o
m
E
E
G
s
ig
n
als
[
6
]
.
Featu
r
e
ex
tr
ac
tio
n
is
a
cr
u
cial
s
tep
in
em
o
tio
n
r
ec
o
g
n
itio
n
.
Usu
ally
,
it
u
s
es
th
e
f
r
eq
u
en
cy
o
r
tim
e
-
f
r
e
q
u
en
cy
d
o
m
ai
n
,
co
n
s
id
er
in
g
t
h
e
f
r
eq
u
en
cy
b
an
d
s
r
elev
an
t to
em
o
tio
n
class
[
7
]
.
T
h
is
r
esear
ch
u
s
ed
d
ata
f
r
o
m
th
e
SJ
T
U
em
o
tio
n
E
E
G
d
ataset
(
SEE
D)
,
p
r
o
v
id
ed
b
y
th
e
B
r
ain
-
lik
e
C
o
m
p
u
tin
g
an
d
Ma
ch
in
e
I
n
tellig
en
ce
(
B
C
MI
)
l
ab
o
r
ato
r
y
,
with
th
r
ee
class
es:
p
o
s
itiv
e,
n
eu
tr
al,
an
d
n
eg
ativ
e
[
8
]
,
[
9
]
.
T
h
ese
th
r
e
e
class
es
ex
h
ib
it
d
is
tin
ct
f
r
eq
u
en
cy
b
an
d
s
:
th
eta,
alp
h
a,
b
eta,
an
d
g
am
m
a,
4
-
4
8
Hz
[
8
]
,
[
1
0
]
.
Po
s
itiv
e
em
o
tio
n
s
ar
e
g
en
e
r
ally
ass
o
ciate
d
with
in
cr
ea
s
es
in
b
eta
an
d
g
am
m
a
co
m
p
o
n
en
ts
,
r
esu
ltin
g
in
co
g
n
itiv
e
e
n
h
an
ce
m
en
t.
Po
s
itiv
e
em
o
tio
n
s
a
r
e
g
en
e
r
ally
ass
o
ciate
d
with
i
n
cr
ea
s
ed
b
eta
an
d
g
am
m
a
wav
es,
in
d
icatin
g
co
g
n
itiv
e
en
h
an
ce
m
en
t.
Neu
t
r
al
e
m
o
tio
n
s
ar
e
b
alan
ce
d
ac
r
o
s
s
all
wav
es.
Neg
ativ
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
E
mo
tio
n
r
ec
o
g
n
itio
n
o
f e
lectro
en
ce
p
h
a
lo
g
r
a
m
u
s
in
g
h
yb
r
id
co
n
vo
lu
tio
n
a
l
…
(
E
s
mera
ld
a
C
o
n
tess
a
Dja
ma
l
)
3935
em
o
tio
n
s
,
o
n
th
e
o
th
er
h
a
n
d
,
a
r
e
ch
ar
ac
ter
ized
b
y
in
cr
ea
s
ed
t
h
eta
an
d
alp
h
a
wav
es.
T
h
e
r
ef
o
r
e,
f
r
eq
u
en
cy
b
a
n
d
ex
tr
ac
tio
n
p
r
o
v
id
es
d
is
tin
g
u
is
h
in
g
f
ea
tu
r
es
f
o
r
em
o
tio
n
cla
s
s
es
[
1
1
]
.
W
av
elet
tr
an
s
f
o
r
m
s
ar
e
ef
f
ec
tiv
e
f
o
r
an
aly
zin
g
th
e
E
E
G
s
ig
n
als,
en
ab
lin
g
tim
e
-
f
r
e
q
u
en
c
y
d
ec
o
m
p
o
s
itio
n
.
T
h
is
is
ac
h
iev
ed
b
y
d
ec
o
m
p
o
s
in
g
s
p
ec
if
ic
f
r
eq
u
e
n
cy
b
an
d
s
to
p
r
eser
v
e
th
e
tem
p
o
r
al
d
y
n
am
ics
ess
en
tial
to
em
o
tio
n
.
It
p
r
o
v
id
es
r
o
b
u
s
tn
ess
ag
ain
s
t n
o
is
e
[
1
0
]
.
P
r
ev
io
u
s
s
tu
d
ies
o
n
em
o
tio
n
class
if
icatio
n
h
av
e
u
s
ed
co
n
v
o
l
u
tio
n
al
n
e
u
r
al
n
etwo
r
k
s
(
C
NN)
,
r
ec
u
r
r
en
t
n
e
u
r
al
n
etwo
r
k
s
[
1
2
]
,
an
d
g
r
a
p
h
n
e
u
r
al
n
etwo
r
k
s
(
GNN)
.
C
NNs
ex
p
lo
it
lo
ca
l
s
p
atial
an
d
s
p
ec
tr
al
co
r
r
elatio
n
s
am
o
n
g
E
E
G
c
h
a
n
n
els
an
d
f
r
e
q
u
en
c
y
b
an
d
s
.
C
NNs
ar
e
ef
f
ec
tiv
e
in
ca
p
tu
r
in
g
d
is
cr
im
in
ativ
e
em
o
tio
n
al
p
atter
n
s
in
E
E
G
s
ig
n
als
[
8
]
,
f
o
llo
wed
b
y
2
D
C
NNs
f
o
r
m
u
lti
-
c
h
an
n
el
a
n
aly
s
is
[
1
3
]
,
[
1
4
]
,
asy
m
m
etr
ic
v
ar
iab
les
ac
r
o
s
s
ch
an
n
el
p
air
s
[
1
5
]
.
E
E
G
s
ig
n
al
s
f
r
o
m
m
u
lti
-
ch
a
n
n
el
ar
r
ay
s
o
f
ten
r
eq
u
i
r
e
s
p
atial
f
ea
tu
r
e
ex
tr
ac
tio
n
,
wh
ich
ca
n
b
e
ac
co
m
p
lis
h
ed
with
a
C
NN.
T
h
er
ef
o
r
e,
h
y
b
r
id
C
NN
-
r
ec
u
r
r
en
t
n
eu
r
al
n
etwo
r
k
(
R
NN
)
is
o
f
ten
u
s
ed
as
s
eq
u
en
tial
[
1
6
]
an
d
p
ar
allel
[
1
1
]
ap
p
r
o
ac
h
es.
H
o
wev
er
,
t
h
ese
ap
p
r
o
ac
h
es
a
r
e
s
u
s
ce
p
tib
le
to
f
lu
ctu
atio
n
s
an
d
h
av
e
h
i
g
h
co
m
p
u
tatio
n
al
tim
e
.
C
NNs
ar
e
lim
ited
to
l
o
ca
l
r
ec
o
g
n
itio
n
,
wh
ile
R
NNs
p
r
o
ce
s
s
s
eq
u
en
tial
d
ata
a
n
d
lo
s
e
p
ar
allel
in
f
o
r
m
atio
n
.
T
r
a
n
s
f
o
r
m
er
s
al
lo
w
ea
ch
elem
en
t
to
b
e
s
i
m
u
ltan
eo
u
s
ly
a
n
d
g
l
o
b
ally
s
p
atially
co
n
n
ec
ted
,
s
im
ilar
to
th
e
co
n
n
ec
tio
n
s
o
f
n
eu
r
o
n
s
in
th
e
b
r
ain
[
1
7
]
.
T
r
an
s
f
o
r
m
er
s
m
o
r
e
ef
f
ec
tiv
ely
ca
p
tu
r
e
g
lo
b
al
r
elatio
n
s
th
r
o
u
g
h
s
elf
-
atten
ti
o
n
m
ec
h
an
is
m
s
,
m
a
k
in
g
th
e
m
well
-
s
u
ited
f
o
r
e
m
o
tio
n
class
if
icatio
n
f
r
o
m
E
E
G
s
ig
n
als.
Fu
r
th
er
m
o
r
e,
tr
an
s
f
o
r
m
er
co
m
p
u
tatio
n
is
f
aster
f
o
r
tr
ain
in
g
an
d
ca
p
tu
r
in
g
s
p
atio
-
tem
p
o
r
al
p
atter
n
s
.
R
esear
ch
s
h
o
ws
th
at
tr
an
s
f
o
r
m
er
-
b
ased
m
o
d
els
ac
h
iev
e
s
u
p
er
io
r
p
er
f
o
r
m
a
n
ce
an
d
r
o
b
u
s
tn
ess
in
E
E
G
s
ig
n
al
class
if
icatio
n
,
allo
win
g
f
o
r
d
y
n
am
ic
g
e
n
er
ali
za
tio
n
d
u
r
in
g
lear
n
in
g
b
y
r
ed
u
cin
g
ir
r
ele
v
an
t
co
m
p
o
n
en
ts
o
r
n
o
is
e
[
1
0
]
,
[
1
7
]
.
T
h
e
v
is
io
n
tr
a
n
s
f
o
r
m
er
(
ViT
)
h
as
attr
ac
ted
atten
tio
n
i
n
E
E
G
-
b
ased
em
o
tio
n
r
ec
o
g
n
itio
n
d
u
e
to
its
r
o
b
u
s
t
s
elf
-
atten
tio
n
m
ec
h
an
i
s
m
,
wh
ich
en
a
b
les
ef
f
ec
tiv
e
m
o
d
elin
g
o
f
g
lo
b
al
d
ep
e
n
d
en
cies
ac
r
o
s
s
b
r
ain
r
eg
io
n
s
an
d
tim
e
s
eq
u
e
n
ce
s
.
W
h
en
th
e
tim
e
-
f
r
e
q
u
en
c
y
f
ea
t
u
r
e
ex
tr
ac
tio
n
r
esu
lts
ar
e
p
ar
titi
o
n
ed
in
to
p
atch
es,
ViT
ca
n
ca
p
tu
r
e
b
r
o
a
d
co
n
te
x
tu
al
r
elatio
n
s
h
ip
s
th
at
ar
e
d
if
f
icu
lt
to
m
o
d
el
u
s
in
g
C
NNs
o
r
R
NNs
[
1
7
]
,
[
1
8
]
.
ViT
r
elies
s
o
lely
o
n
d
ata
-
d
r
i
v
en
lear
n
in
g
,
m
ak
in
g
it
less
ef
f
ec
tiv
e
wh
en
tr
ain
ed
o
n
m
ed
iu
m
-
s
ized
E
E
G
d
atasets
s
u
ch
as
SEE
D
a
n
d
p
r
o
n
e
t
o
o
v
er
f
itti
n
g
.
T
h
is
is
esp
ec
ially
tr
u
e
wh
en
in
ter
-
s
u
b
ject
v
ar
iab
ilit
y
is
h
ig
h
,
a
ch
allen
g
e
th
at
b
ec
o
m
es
cr
it
ical
as
ViT
's
p
er
f
o
r
m
an
ce
d
e
clin
es
s
ig
n
if
ican
tly
[
1
7
]
.
T
h
e
s
im
p
licity
o
f
ViT
co
m
p
u
tatio
n
allo
ws
f
o
r
f
ast
er
lear
n
in
g
.
T
h
is
is
d
u
e
to
its
f
lex
ib
ilit
y
in
h
an
d
lin
g
v
ar
io
u
s
E
E
G
s
ig
n
al
r
ep
r
esen
tatio
n
s
b
y
co
n
v
er
tin
g
th
em
in
to
im
ag
e
f
o
r
m
ats,
m
a
k
in
g
it
co
m
p
atib
le
with
ViT
p
atch
to
k
en
izatio
n
.
T
h
e
p
atch
s
ize
is
ad
ju
s
tab
le,
allo
win
g
ViT
to
ef
f
ec
tiv
ely
le
ar
n
h
ier
a
r
ch
ical
r
e
p
r
esen
tatio
n
s
wh
ile
p
r
eser
v
in
g
g
lo
b
al
s
tr
u
ctu
r
al
r
elatio
n
s
h
ip
s
in
m
u
lti
-
ch
an
n
el
E
E
G
s
ig
n
als an
d
f
r
eq
u
en
cy
-
b
an
d
v
a
r
iab
ilit
y
.
ViT
co
m
p
u
tatio
n
ca
n
b
e
d
y
n
am
ically
s
ca
led
an
d
allo
ws
f
o
r
p
ar
allel
co
m
p
u
tati
o
n
b
y
p
r
o
ce
s
s
in
g
all
in
p
u
t
to
k
en
s
s
im
u
ltan
eo
u
s
ly
.
T
h
is
m
eth
o
d
f
ac
ilit
ates
lo
n
g
-
ter
m
tem
p
o
r
al
m
o
d
elin
g
,
ad
d
r
ess
in
g
R
N
N
is
s
u
es.
ViT
allo
ws
m
ap
p
in
g
ac
r
o
s
s
E
E
G
s
ig
n
al
ch
an
n
els an
d
f
r
e
q
u
en
cy
b
an
d
s
.
I
t c
o
n
tr
ib
u
tes to
class
if
icatio
n
in
b
r
ain
m
ap
p
in
g
.
Pre
v
io
u
s
r
esear
ch
u
s
ed
a
s
i
n
g
le
-
ch
an
n
el
ViT
[
1
9
]
.
Ho
w
ev
er
,
E
E
G
s
ig
n
als
ar
e
r
ec
o
r
d
ed
f
r
o
m
m
u
ltip
le
ch
an
n
els
an
d
r
eq
u
i
r
e
ex
p
lo
r
atio
n
o
f
th
e
in
f
o
r
m
atio
n
co
n
tain
e
d
in
ea
c
h
f
r
eq
u
en
cy
b
an
d
[
1
8
]
.
T
h
er
ef
o
r
e,
w
av
elet
an
d
C
NN
n
ee
d
to
b
e
co
m
b
in
ed
with
Vi
T
.
T
h
e
h
y
b
r
id
m
o
d
el
h
elp
s
i
m
p
r
o
v
e
p
atch
-
b
ased
ViT
lear
n
in
g
.
Hen
ce
,
th
e
in
te
r
p
r
etatio
n
a
n
d
p
h
y
s
io
lo
g
ical
r
elev
an
ce
o
f
t
h
e
lear
n
e
d
atten
t
io
n
m
ap
s
ar
e
m
o
r
e
m
ea
n
in
g
f
u
l.
C
NN
-
ViT
was u
s
ed
f
o
r
em
o
tio
n
class
if
icatio
n
o
f
a
s
in
g
le
ch
an
n
el
[
2
0
]
,
[
2
1
]
.
C
NN
is
s
u
itab
le
f
o
r
f
ea
tu
r
e
e
x
tr
ac
tio
n
,
b
u
t
lear
n
i
n
g
g
ets
s
tu
ck
o
n
n
e
u
r
o
n
lo
ca
lizatio
n
.
T
h
er
ef
o
r
e,
class
if
icatio
n
is
p
er
f
o
r
m
ed
u
s
in
g
th
e
ViT
to
ca
p
tu
r
e
th
e
g
lo
b
al
co
n
n
ec
tiv
ity
o
f
m
u
lti
-
ch
a
n
n
el
E
E
G
s
ig
n
als
an
d
v
ar
io
u
s
f
r
eq
u
e
n
cy
b
an
d
s
with
b
r
ain
c
o
n
n
ec
tio
n
s
d
u
r
in
g
s
p
ec
i
f
ic
em
o
tio
n
s
.
T
h
e
in
teg
r
atio
n
o
f
DW
T
,
2
D
C
NN,
an
d
ViT
in
th
is
p
ap
er
r
ep
r
esen
ts
a
n
ew
s
tep
to
war
d
s
im
p
r
o
v
in
g
ac
cu
r
ac
y
,
tr
ain
in
g
s
tab
ilit
y
,
an
d
ef
f
icien
cy
,
as
well
as u
n
d
er
s
tan
d
in
g
t
h
e
r
ele
v
an
ce
o
f
ea
ch
f
r
eq
u
e
n
cy
b
a
n
d
to
em
o
tio
n
s
.
T
h
e
n
o
v
elty
o
f
th
is
r
esear
ch
is
th
e
in
teg
r
atio
n
o
f
a
2
D
-
C
NN
b
ased
o
n
w
av
elets
wit
h
ViT
f
o
r
EEG
-
b
ased
em
o
tio
n
class
if
icatio
n
.
W
av
elets
en
ab
le
th
e
e
x
tr
ac
tio
n
o
f
f
r
e
q
u
en
c
y
b
a
n
d
s
th
at
co
r
r
esp
o
n
d
t
o
d
is
tin
ct
em
o
tio
n
class
es.
T
h
e
2
D
C
NN
-
ViT
ar
ch
itectu
r
e
allo
ws
f
o
r
m
o
r
e
s
tab
le
an
d
r
o
b
u
s
t
g
en
er
aliza
tio
n
o
f
th
e
E
E
G
s
ig
n
al
with
ce
r
tain
e
m
o
tio
n
s
.
A
h
y
b
r
id
C
NN
-
tr
an
s
f
o
r
m
er
ar
ch
itectu
r
e
h
as
b
ee
n
a
p
p
lied
to
th
e
DE
AP
d
ataset
f
r
o
m
p
r
ev
io
u
s
r
esear
c
h
[
2
1
]
.
2.
M
E
T
H
O
D
2
.
1
.
Da
t
a
s
et
des
cr
iptio
n
T
h
is
s
tu
d
y
e
m
p
lo
y
s
SEE
D,
a
wid
ely
u
s
ed
b
en
c
h
m
ar
k
f
o
r
E
E
G
-
b
ased
em
o
tio
n
r
ec
o
g
n
itio
n
[
8
]
,
[
9
]
.
T
h
e
SEE
D
d
ataset
c
o
n
tain
s
E
E
G
r
ec
o
r
d
in
g
s
f
r
o
m
1
5
s
u
b
jec
ts
,
ea
ch
p
a
r
ticip
atin
g
in
3
ex
p
er
im
en
tal
s
ess
io
n
s
.
Su
b
jects
wer
e
ex
p
o
s
ed
to
s
tim
u
li
elicitin
g
th
r
e
e
em
o
tio
n
a
l
s
tates
(
p
o
s
itiv
e,
n
eu
tr
al,
an
d
n
eg
ati
v
e)
th
r
o
u
g
h
em
o
tio
n
al
f
ilm
v
id
e
o
s
.
E
E
G
s
ig
n
als
wer
e
r
ec
o
r
d
ed
u
s
in
g
6
2
ch
an
n
els
f
o
llo
win
g
th
e
in
ter
n
atio
n
al
10
–
2
0
s
y
s
tem
,
with
a
s
am
p
lin
g
r
ate
o
f
2
0
0
Hz.
T
h
e
r
e
d
u
ctio
n
f
r
o
m
6
2
c
h
an
n
els to
1
2
ch
a
n
n
els wa
s
p
er
f
o
r
m
ed
u
s
in
g
a
r
elev
a
n
t
r
eg
io
n
-
b
ased
ch
an
n
el
s
elec
tio
n
a
p
p
r
o
ac
h
[
1
5
]
.
E
m
o
tio
n
s
ar
e
cl
o
s
ely
r
el
ated
to
th
e
f
r
o
n
tal,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
5
,
No
.
4
,
Au
g
u
s
t 2
0
2
6
:
3
9
3
4
-
3
9
4
3
3936
tem
p
o
r
al,
a
n
d
p
ar
ietal
ar
ea
s
.
T
h
ey
ar
e
FT7
,
FT8
,
T
7
,
T
8
,
C
5
,
C
6
,
T
P7
,
T
P8
,
C
P5
,
C
P6
,
P7
,
an
d
P8
[
2
2
]
.
T
h
is
ap
p
r
o
ac
h
m
ain
tain
s
n
eu
r
o
p
h
y
s
io
lo
g
ical
r
elev
an
ce
wh
ile
r
ed
u
cin
g
r
ed
u
n
d
a
n
cy
an
d
co
m
p
u
t
atio
n
al
co
m
p
lex
ity
,
a
p
ar
ticu
lar
ly
b
en
ef
icial
p
r
o
p
e
r
ty
f
o
r
tr
an
s
f
o
r
m
er
-
b
ased
ar
ch
itectu
r
es.
L
et
(
)
∈
,
d
en
o
te
t
h
e
r
aw
E
E
G
s
ig
n
al
am
p
litu
d
e
r
ec
o
r
d
ed
f
r
o
m
th
e
ℎ
E
E
G
ch
an
n
el
at
a
d
is
cr
ete
tim
e
in
d
ex
t,
wh
er
e
c
∈
{
1
,
2
,
…
,
1
2
}.
An
E
E
G
tr
ial
co
n
s
is
tin
g
o
f
1
2
ch
an
n
els an
d
T
=
1
0
0
0
p
o
i
n
ts
o
r
5
s
ec
o
n
d
s
is
r
ep
r
esen
ted
i
n
(
1
)
.
[
1
(
1
)
⋯
1
(
1000
)
⋮
⋱
⋮
12
(
1
)
⋯
12
(
1000
)
]
(
1
)
2
.
2
.
M
o
del
d
esig
n
EEG
-
b
ased
em
o
tio
n
r
ec
o
g
n
it
io
n
in
teg
r
ates
tim
e
-
f
r
eq
u
e
n
cy
an
aly
s
is
,
lo
ca
l
s
p
atial
-
s
p
ec
tr
al
f
ea
tu
r
e
lear
n
in
g
,
an
d
g
lo
b
al
co
n
tex
t
u
al
m
o
d
elin
g
.
R
aw
s
ig
n
als
ar
e
ex
tr
ac
ted
i
n
th
e
th
eta,
alp
h
a,
b
eta,
an
d
g
am
m
a
f
r
eq
u
e
n
cy
b
an
d
s
u
s
in
g
d
is
cr
ete
wav
elet
tr
an
s
f
o
r
m
(
DW
T
)
.
T
h
e
p
r
esen
ce
o
f
ea
ch
o
f
th
ese
b
an
d
s
is
s
tr
o
n
g
ly
co
r
r
elate
d
with
e
m
o
tio
n
al
s
tates,
f
ac
ilit
atin
g
th
e
id
e
n
tific
atio
n
s
tag
e
[
2
3
]
.
T
h
e
p
r
o
p
o
s
ed
m
o
d
el,
b
ased
o
n
a
5
-
s
ec
o
n
d
d
ataset
o
r
1
,
0
0
0
p
o
in
ts
,
is
d
ec
o
m
p
o
s
ed
u
s
in
g
D
W
T
.
T
h
en
,
th
e
1
2
ch
an
n
els
ar
e
ex
tr
ac
ted
u
s
in
g
a
2
D
C
NN
an
d
m
ap
p
e
d
u
s
in
g
V
iT,
as sh
o
wn
in
Fig
u
r
e
1
.
D
e
c
om
pos
i
t
i
on
T
h
e
t
a
A
l
p
h
a
Be
t
a
G
a
m
m
a
D
i
s
cret
e
W
av
el
et
T
ran
s
fo
rm
E
E
G
s
i
g
n
a
l
s
e
v
e
ry
5
s
e
c
o
n
d
s
o
f
1
2
c
h
a
n
n
e
l
s
T
ra
i
n
i
n
g
d
a
t
a
S
p
a
t
i
a
l
o
f
D
W
T
o
f
3
2
x
3
2
2
D
T
o
p
o
m
a
p
F
e
a
t
ure
e
m
be
ddi
ng
V
i
s
i
o
n
T
ran
s
fo
rme
r
M
L
P
H
e
a
d
S
o
ft
m
a
x
N
e
g
a
t
i
v
e
N
e
u
t
r
a
l
P
o
s
i
t
i
v
e
E
m
o
t
i
o
n
c
l
a
s
s
E
E
G
s
i
g
n
a
l
s
e
v
e
ry
5
s
e
c
o
n
d
s
o
f
1
2
c
h
a
n
n
e
l
s
D
W
T
2
D
M
a
t
ri
x
Re
p
re
s
e
n
t
a
t
i
o
n
2
D
C
N
N
F
e
a
t
u
r
e
E
m
b
e
d
i
n
g
P
ool
i
ng
2
D
CN
N
Co
nv+
Re
L
u
Fig
u
r
e
1
.
Hy
b
r
id
2
D
C
NN
-
ViT
m
o
d
el
Fig
u
r
e
1
d
escr
ib
es
th
e
co
m
p
lete
ar
ch
itectu
r
e
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el,
in
clu
d
i
n
g
th
e
tr
ain
in
g
an
d
id
en
tific
atio
n
s
tag
es.
Du
r
in
g
th
e
tr
ain
in
g
s
tag
e,
E
E
G
d
ata
f
r
o
m
1
5
s
u
b
jects
wer
e
co
llected
wh
ile
th
ey
v
iewe
d
s
tim
u
lu
s
v
id
eo
s
f
r
o
m
th
r
ee
e
m
o
tio
n
class
es
(
p
o
s
itiv
e,
n
eu
t
r
al,
an
d
n
e
g
ativ
e)
ac
r
o
s
s
1
2
m
ain
ch
an
n
els.
T
h
e
E
E
G
s
ig
n
als
wer
e
s
eg
m
en
te
d
in
to
f
iv
e
-
s
ec
o
n
d
s
eg
m
e
n
ts
,
a
tim
e
win
d
o
w
d
u
r
atio
n
p
r
o
v
en
t
o
p
r
o
v
id
e
a
n
o
p
tim
al
b
alan
ce
b
etwe
en
te
m
p
o
r
al
r
eso
lu
tio
n
an
d
s
tab
ili
ty
o
f
em
o
tio
n
al
f
ea
tu
r
es
[
2
4
]
.
T
h
en
,
th
e
y
wer
e
p
r
o
ce
s
s
ed
u
s
in
g
th
e
DW
T
to
ex
tr
ac
t
f
iv
e
f
r
eq
u
e
n
cy
b
a
n
d
s
,
in
clu
d
in
g
th
eta,
alp
h
a
,
b
eta,
a
n
d
g
am
m
a
wav
es,
wh
ich
r
ep
r
esen
t
co
g
n
itiv
e
an
d
em
o
tio
n
al
ac
tiv
ity
[
2
5
]
.
T
h
e
DW
T
r
esu
lt
s
ar
e
ex
p
r
ess
ed
as
a
2
D
to
p
o
m
ap
m
ea
s
u
r
in
g
3
2
×
3
2
p
ix
els
th
at
d
is
p
lay
s
th
e
d
is
tr
ib
u
tio
n
o
f
f
r
eq
u
en
cy
en
e
r
g
y
ag
ai
n
s
t
E
E
G
ch
an
n
el
p
o
s
itio
n
s
,
an
ef
f
ec
tiv
e
tech
n
iq
u
e
f
o
r
p
r
es
er
v
in
g
s
p
atial
in
f
o
r
m
atio
n
[
2
6
]
.
T
h
is
to
p
o
m
ap
is
u
s
ed
as
in
p
u
t
to
a
2
D
C
NN
to
ex
tr
ac
t
lo
ca
l
s
p
atial
p
atter
n
s
ac
r
o
s
s
ch
an
n
els
an
d
f
r
eq
u
en
cy
b
an
d
s
.
T
h
e
r
esu
lt
is
an
8
×
8
×
2
5
6
f
ea
tu
r
e
m
ap
,
wh
ich
is
th
en
f
latten
ed
in
to
a
f
ea
tu
r
e
em
b
ed
d
in
g
(
6
4
p
atc
h
e
s
×
2
5
6
d
im
en
s
io
n
s
)
an
d
s
en
t
to
ViT
.
ViT
lear
n
s
g
lo
b
al
r
elatio
n
s
h
i
p
s
b
etwe
en
p
atch
es
u
s
in
g
a
m
u
lti
-
h
ea
d
s
elf
-
atten
tio
n
m
ec
h
an
is
m
t
h
at
is
s
u
p
er
io
r
in
ca
p
tu
r
in
g
lo
n
g
-
ter
m
d
ep
e
n
d
en
cies
co
m
p
ar
ed
to
r
ec
u
r
r
en
t
m
eth
o
d
s
[
2
7
]
.
T
h
e
o
u
tp
u
t
o
f
Vi
T
is
p
ass
ed
to
t
h
e
m
u
lti
-
lay
er
p
er
ce
p
tr
o
n
(
ML
P)
h
ea
d
an
d
s
o
f
tm
ax
class
if
ier
to
d
eter
m
in
e
t
h
r
ee
em
o
tio
n
cla
s
s
es.
T
h
e
p
r
o
p
o
s
ed
ar
ch
itectu
r
e
s
u
p
p
o
r
ts
r
ea
l
-
tim
e
em
o
tio
n
p
r
ed
ictio
n
,
as
s
h
o
w
n
b
y
th
e
b
o
tto
m
p
ath
in
Fig
u
r
e
1
.
New
E
E
G
d
ata
r
ec
o
r
d
e
d
in
r
ea
l
tim
e
is
th
en
p
r
o
ce
s
s
ed
with
DW
T
→
2
D
C
NN
→
ViT
,
an
d
ev
er
y
5
s
e
co
n
d
s
,
it
g
en
e
r
ates
a
lab
el
f
o
r
o
n
e
o
f
th
e
th
r
ee
em
o
t
io
n
class
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
E
mo
tio
n
r
ec
o
g
n
itio
n
o
f e
lectro
en
ce
p
h
a
lo
g
r
a
m
u
s
in
g
h
yb
r
id
co
n
vo
lu
tio
n
a
l
…
(
E
s
mera
ld
a
C
o
n
tess
a
Dja
ma
l
)
3937
2
.
3
.
Dis
cr
et
e
wa
v
ele
t
t
ra
ns
f
o
rm
s
Af
ter
b
an
d
-
p
ass
f
ilter
in
g
in
th
e
r
an
g
e
o
f
4
–
4
8
Hz,
th
e
E
E
G
s
ig
n
al
is
ex
p
r
ess
ed
as (
2
)
.
̃
(
)
=
ℱ
4
−
48
{
(
)
}
(
2
)
T
h
e
p
r
ep
r
o
ce
s
s
in
g
s
tag
e
u
s
es
th
e
DW
T
.
T
h
is
m
eth
o
d
was
ch
o
s
en
b
ec
au
s
e
it
ca
n
an
al
y
ze
n
o
n
-
s
tatio
n
ar
y
s
ig
n
als
s
u
ch
as
E
E
G
an
d
p
r
o
v
id
es
d
etailed
in
f
o
r
m
atio
n
o
n
f
r
eq
u
en
cy
ch
a
n
g
es
o
v
er
t
im
e.
W
av
elet
-
b
ased
ap
p
r
o
ac
h
es
ar
e
s
u
p
er
io
r
at
ca
p
tu
r
in
g
a
b
r
u
p
t
tr
an
s
itio
n
s
in
b
io
m
ed
ical
s
ig
n
als
co
m
p
ar
ed
to
s
tan
d
a
r
d
Fo
u
r
ier
-
b
ased
m
eth
o
d
s
[
2
8
]
.
E
ac
h
E
E
G
s
ig
n
al
s
eg
m
en
t
f
r
o
m
th
e
p
r
ev
io
u
s
s
tag
e
is
d
ec
o
m
p
o
s
ed
to
p
r
o
d
u
ce
f
iv
e
m
ain
f
r
eq
u
en
cy
b
an
d
s
:
th
eta
(
4
–
8
Hz)
,
alp
h
a
(
8
–
1
3
Hz)
,
b
eta
(
1
4
–
3
0
Hz)
,
an
d
g
am
m
a
(
3
1
–
4
8
Hz)
[
2
9
]
,
wh
ich
ar
e
r
elate
d
t
o
em
o
tio
n
a
l
s
tates.
Fo
r
ex
am
p
le,
th
e
alp
h
a
b
an
d
o
f
ten
a
p
p
ea
r
s
in
r
elax
e
d
s
tates,
wh
ile
th
e
b
eta
an
d
g
am
m
a
b
an
d
s
ar
e
m
o
r
e
d
o
m
in
an
t
d
u
r
i
n
g
tim
es
o
f
h
ig
h
m
en
tal
ac
tiv
ity
o
r
s
tr
o
n
g
em
o
tio
n
s
[
3
0
]
.
R
ec
en
t
co
m
p
u
tatio
n
al
n
e
u
r
o
p
h
y
s
io
lo
g
y
s
tu
d
ies
co
n
f
ir
m
th
a
t
p
o
wer
f
lu
ctu
atio
n
s
in
th
e
h
i
g
h
-
f
r
e
q
u
en
c
y
b
an
d
s
(
b
eta
an
d
g
am
m
a)
ar
e
th
e
m
o
s
t
co
n
s
is
ten
t
b
io
m
ar
k
er
s
f
o
r
d
is
tin
g
u
is
h
in
g
p
o
s
itiv
e
an
d
n
eg
ativ
e
em
o
tio
n
a
l
v
alen
ce
[
2
5
]
.
Ap
p
ly
in
g
th
e
D
W
T
to
th
e
f
ilter
ed
E
E
G
s
ig
n
al
y
ield
s
wav
elet
co
ef
f
icien
ts
as sh
o
wn
in
(
3
)
.
̃
(
)
=
∑
,
(
)
,
(
)
+
∑
∑
,
(
)
=
1
,
(
)
(
3
)
W
h
er
e
,
(
)
is
ch
an
n
el
ap
p
r
o
x
im
a
tio
n
co
ef
f
icien
t
c
,
,
(
)
is
d
etail
co
ef
f
icien
t
o
n
s
ca
le
j
,
an
d
(
)
,
(
)
ar
e
s
ca
le
f
u
n
ctio
n
an
d
wav
elet
.
W
av
elet
co
ef
f
icien
ts
ar
e
g
r
o
u
p
e
d
in
to
em
o
tio
n
-
r
elate
d
E
E
G
b
a
n
d
s
as (
4
)
.
W
=
[
W
,
W
,
W
,
W
]
(
4
)
W
h
er
e
θ
is
4
–
8
Hz,
α
is
8
–
1
3
Hz,
β
is
13
–
3
0
Hz,
a
n
d
γ
is
30
–
4
8
Hz
.
T
h
e
DW
T
d
ec
o
m
p
o
s
itio
n
r
es
u
lts
f
o
r
ea
ch
E
E
G
c
h
an
n
el
a
r
e
th
en
ar
r
an
g
e
d
in
t
o
a
2
D
m
atr
ix
th
at
r
ep
r
esen
ts
th
e
en
er
g
y
d
is
tr
ib
u
t
io
n
o
f
ea
ch
f
r
eq
u
e
n
cy
b
an
d
ac
r
o
s
s
th
e
ch
an
n
els.
T
h
is
2
D
r
ep
r
esen
tatio
n
is
u
s
ed
as
in
p
u
t
f
o
r
a
2
D
-
C
NN
to
r
ec
o
g
n
ize
s
p
atial
p
atter
n
s
b
etwe
en
ch
an
n
els
an
d
f
r
e
q
u
en
cy
c
h
ar
a
cter
is
tics
r
elate
d
to
em
o
tio
n
s
.
T
h
is
tr
an
s
f
o
r
m
atio
n
to
an
im
a
g
e
m
atr
ix
f
o
r
m
at
allo
ws
th
e
C
NN
m
o
d
el
to
ex
p
l
o
it
s
p
atial
co
r
r
elatio
n
s
b
etwe
en
elec
tr
o
d
es
th
at
ar
e
o
f
te
n
lo
s
t
wh
en
u
s
in
g
o
n
ly
o
n
e
-
d
im
e
n
s
io
n
al
v
ec
to
r
s
[
2
6
]
.
Fo
r
ea
ch
tr
ial,
th
e
wav
elet
f
ea
tu
r
es a
r
e
a
r
r
an
g
e
d
in
to
a
2
D
r
ep
r
esen
tatio
n
as (
5
)
.
∈
ℝ
×
(
5
)
W
h
er
e
is
1
2
,
is
4
(
n
u
m
b
e
r
o
f
f
r
eq
u
e
n
cy
b
a
n
d
s
)
,
an
d
is
ch
an
n
el
–
f
r
eq
u
en
cy
E
E
G
f
ea
tu
r
e
m
ap
.
2
.
4
.
H
y
brid 2
D
co
nv
o
lutio
n
a
l neura
l net
wo
rk
–
v
is
io
n t
ra
ns
f
o
rm
er
A
h
y
b
r
id
m
o
d
el
co
m
b
i
n
in
g
C
NN
an
d
ViT
.
At
th
i
s
s
tag
e
,
a
co
n
v
o
lu
tio
n
o
p
e
r
atio
n
is
ap
p
lied
to
ex
tr
ac
t lo
ca
l p
atter
n
s
f
r
o
m
th
e
2D
DW
T
r
ep
r
esen
tatio
n
.
T
h
e
co
n
v
o
l
u
tio
n
o
p
er
atio
n
is
f
o
r
m
u
lated
in
(
6
)
.
(
,
)
=
(
×
)
(
,
)
=
∑
∑
(
−
,
−
)
.
(
,
)
(
6
)
W
h
er
e
is
th
e
in
p
u
t
m
atr
ix
r
esu
ltin
g
f
r
o
m
t
h
e
DW
T
tr
an
s
f
o
r
m
atio
n
,
is
th
e
co
n
v
o
lu
tio
n
k
e
r
n
el,
an
d
(
,
)
is
th
e
co
n
v
o
lu
tio
n
al
f
ea
tu
r
e
m
ap
.
C
NNs
e
x
t
r
a
ct
im
p
o
r
ta
n
t
e
m
o
tio
n
a
l
f
e
at
u
r
es
in
s
p
ec
if
ic
f
r
e
q
u
e
n
c
y
b
an
d
s
f
r
o
m
m
u
l
ti
-
c
h
a
n
n
el
E
E
G
s
ig
n
als
.
M
u
lt
ip
le
co
n
v
o
l
u
ti
o
n
a
l
l
a
y
e
r
s
,
ac
t
iv
ati
o
n
f
u
n
ct
io
n
s
,
a
n
d
m
a
x
p
o
o
li
n
g
o
p
e
r
a
ti
o
n
s
r
e
d
u
c
e
d
im
en
s
io
n
a
lit
y
wit
h
o
u
t
l
o
s
i
n
g
i
n
f
o
r
m
ati
o
n
,
w
h
ic
h
a
r
e
k
e
y
a
d
v
a
n
t
a
g
es
o
f
C
NNs.
M
ea
n
w
h
il
e,
V
iT
m
o
d
e
ls
g
l
o
b
al
r
el
ati
o
n
s
h
ip
s
am
o
n
g
f
ea
t
u
r
es
v
ia
a
s
el
f
-
att
e
n
ti
o
n
m
e
ch
a
n
is
m
[
3
1
]
.
T
h
e
o
u
t
p
u
t
o
f
t
h
e
C
NN
is
co
n
v
e
r
t
e
d
i
n
t
o
a
f
i
x
ed
-
s
i
ze
p
at
c
h
e
m
b
e
d
d
i
n
g
.
V
iT
u
t
ili
ze
s
two
at
te
n
ti
o
n
h
e
ad
s
a
n
d
o
n
e
e
n
c
o
d
e
r
l
a
y
e
r
,
ea
ch
e
q
u
i
p
p
ed
wit
h
m
u
lti
-
h
e
ad
s
elf
-
a
tte
n
t
io
n
a
n
d
a
f
e
e
d
-
f
o
r
w
a
r
d
n
et
wo
r
k
.
T
h
e
C
N
N
f
e
at
u
r
e
m
a
p
is
d
i
v
i
d
e
d
i
n
t
o
p
at
ch
es
as
(
7
)
.
0
=
[
0
1
,
0
2
,
…
,
0
]
(
7
)
0
=
+
(
8
)
W
h
er
e
is
p
atch
E
E
G
i
-
th
, E
is
m
atr
ix
em
b
e
d
d
in
g
,
an
d
is
p
o
s
itio
n
al
en
co
d
in
g
.
T
h
e
o
u
tp
u
t
o
f
ViT
is
p
ass
ed
t
o
a
f
u
lly
co
n
n
ec
ted
lay
er
an
d
f
in
ally
d
eter
m
in
es
th
e
em
o
tio
n
ca
teg
o
r
y
:
p
o
s
itiv
e,
n
eu
tr
al,
o
r
n
e
g
ativ
e
u
s
in
g
s
o
f
tm
ax
[
3
2
]
.
T
h
e
f
u
n
ctio
n
u
s
e
(
9
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
5
,
No
.
4
,
Au
g
u
s
t 2
0
2
6
:
3
9
3
4
-
3
9
4
3
3938
̂
=
(
+
)
(
9
)
W
h
er
e
̂
∈
ℝ
3
is
th
e
p
r
o
b
ab
ilit
y
o
f
cl
ass
:
p
o
s
itiv
e,
n
eu
tr
al,
an
d
n
eg
ativ
e.
T
h
e
d
ataset
is
s
p
lit
8
0
%
f
o
r
tr
ai
n
in
g
an
d
2
0
%
f
o
r
test
in
g
,
a
s
tan
d
a
r
d
s
p
lit
r
atio
u
s
ed
in
t
h
e
SEE
D
d
ataset
to
av
o
i
d
o
v
er
f
itti
n
g
an
d
v
alid
ate
m
o
d
el
g
en
er
aliza
tio
n
[
3
3
]
.
T
h
e
p
r
e
-
ex
p
er
im
e
n
tal
r
esu
lts
d
eter
m
in
ed
th
e
p
ar
am
eter
s
o
f
b
atch
s
ize
4
,
A
d
am
W
o
p
tim
izer
,
an
d
lear
n
in
g
r
ate
0
.
0
0
0
1
,
as in
T
ab
le
1
.
T
ab
le
1
.
Ar
c
h
itectu
r
e
o
f
h
y
b
r
i
d
DW
T
-
2
D
C
NN
-
ViT
La
y
e
r
C
o
n
f
i
g
u
r
a
t
i
o
n
O
u
t
p
u
t
s
h
a
p
e
N
u
mb
e
r
o
f
p
a
r
a
me
t
e
r
s
I
n
p
u
t
l
a
y
e
r
EEG
(
3
2
×
32
×
3)
32
×
32
×
3
0
C
o
n
v
2
D
_
1
3
2
f
i
l
t
e
r
,
k
e
r
n
e
l
3
×
3
,
R
e
LU
30
×
30
×
32
8
9
6
M
a
x
P
o
o
l
i
n
g
2
D
_
1
P
o
o
l
2
×
2
15
×
15
×
32
0
C
o
n
v
2
D
_
2
6
4
f
i
l
t
e
r
,
k
e
r
n
e
l
3
×
3
,
R
e
LU
13
×
13
×
64
1
8
,
4
9
6
F
l
a
t
t
e
n
-
1
0
8
1
6
0
P
a
t
c
h
e
mb
e
d
d
i
n
g
16
×
1
6
p
a
t
c
h
16
×
2
5
6
-
V
i
T
e
n
c
o
d
e
r
1
l
a
y
e
r
,
2
h
e
a
d
,
M
LP
d
i
m
2
5
6
16
×
2
5
6
-
D
e
n
se
(
s
o
f
t
ma
x
)
O
u
t
p
u
t
=
3
c
l
a
ss
e
s
3
7
7
1
N
u
mb
e
r
o
f
p
a
r
a
me
t
e
r
s
2
0
,
0
0
0
T
ab
le
1
s
h
o
ws
th
at
th
e
C
NN
co
m
p
o
n
en
t
e
x
tr
ac
ts
s
p
atial
f
e
atu
r
es
f
r
o
m
th
e
DW
T
r
esu
lts
,
wh
ile
ViT
lear
n
s
g
lo
b
al
r
elatio
n
s
h
ip
s
ac
r
o
s
s
ch
an
n
els
an
d
tim
e.
T
h
is
m
o
d
el
co
n
f
i
g
u
r
atio
n
im
p
r
o
v
es
th
e
m
o
d
el'
s
ab
ilit
y
to
r
ec
o
g
n
ize
em
o
tio
n
al
p
atte
r
n
s
,
as
m
ea
s
u
r
ed
b
y
ac
c
u
r
ac
y
,
s
tab
ilit
y
,
an
d
tim
e.
T
h
r
o
u
g
h
o
u
t
th
e
tr
ai
n
in
g
p
r
o
ce
s
s
,
ac
cu
r
ac
y
v
alu
es
in
cr
ea
s
ed
s
tead
ily
,
in
d
icatin
g
th
a
t
th
e
m
o
d
el
ef
f
ec
tiv
ely
an
d
ef
f
icien
tly
lear
n
ed
s
p
atial
p
atter
n
s
an
d
r
elatio
n
s
h
ip
s
am
o
n
g
E
E
G
ch
an
n
els.
T
h
e
Ad
am
W
o
p
tim
izer
was
s
e
lecte
d
b
ased
o
n
its
ab
ilit
y
to
s
ep
ar
ate
weig
h
t
d
e
ca
y
f
r
o
m
th
e
g
r
a
d
ien
t
ad
a
p
tatio
n
s
tep
,
a
c
o
m
m
o
n
wea
k
n
e
s
s
o
f
th
e
s
tan
d
ar
d
Ad
am
.
C
o
m
p
ar
ativ
e
s
tu
d
ies
h
av
e
s
h
o
wn
t
h
at
Ad
am
W
co
n
v
er
g
es
f
aster
an
d
g
en
e
r
alize
s
b
etter
th
an
class
ical
o
p
tim
izatio
n
m
eth
o
d
s
s
u
ch
as
R
MSPr
o
p
o
r
r
eg
u
lar
Ad
am
[
3
4
]
.
T
h
e
m
o
d
el
tr
ain
in
g
p
r
o
ce
s
s
u
s
es
a
cr
o
s
s
-
en
tr
o
p
y
lo
s
s
f
u
n
ctio
n
to
m
in
im
ize
p
r
ed
ictio
n
er
r
o
r
s
in
m
u
lti
-
class
class
if
icatio
n
.
T
h
is
f
u
n
ctio
n
wo
r
k
s
b
y
ca
lcu
l
atin
g
th
e
lo
g
a
r
ith
m
ic
d
if
f
er
en
ce
b
etwe
en
th
e
s
o
f
tm
ax
-
p
r
e
d
icted
p
r
o
b
ab
ilit
ies
an
d
th
e
ac
tu
al
class
lab
els,
th
er
eb
y
f
o
r
cin
g
th
e
m
o
d
el
to
in
cr
ea
s
e
th
e
co
n
f
id
en
ce
in
th
e
co
r
r
ec
t
class
[
3
5
]
.
All
e
x
p
er
im
e
n
ts
wer
e
im
p
lem
en
ted
u
s
in
g
th
e
Py
T
o
r
ch
f
r
a
m
ewo
r
k
o
n
a
g
r
ap
h
ics
p
r
o
ce
s
s
in
g
u
n
it
(
GPU
)
-
p
o
wer
ed
c
o
m
p
u
tin
g
en
v
ir
o
n
m
en
t
f
o
r
ac
ce
ler
ated
C
NN
an
d
s
elf
-
atten
tio
n
m
atr
ix
o
p
e
r
atio
n
s
.
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
3
.
1
.
Co
nfi
g
ura
t
io
n
des
ig
n e
x
perim
ent
T
o
o
b
tain
th
e
o
p
tim
al
m
o
d
el
co
n
f
ig
u
r
atio
n
,
e
x
p
er
im
e
n
ts
wer
e
co
n
d
u
cted
o
n
s
ev
er
al
k
ey
p
ar
am
eter
s
o
f
th
e
h
y
b
r
id
C
NN
–
ViT
ar
c
h
itectu
r
e,
in
clu
d
i
n
g
p
o
s
itio
n
al
em
b
ed
d
in
g
s
,
ViT
d
im
en
s
io
n
s
,
th
e
n
u
m
b
er
o
f
atten
tio
n
h
ea
d
s
,
th
e
n
u
m
b
e
r
o
f
en
co
d
er
lay
er
s
,
an
d
th
e
lear
n
in
g
r
ate,
wh
ich
ar
e
p
r
esen
ted
i
n
T
ab
le
2
.
T
ab
le
2
s
h
o
ws
th
at
in
cr
ea
s
in
g
m
o
d
el
ca
p
ac
ity
d
o
es
n
o
t
alwa
y
s
r
esu
lt
in
b
etter
p
er
f
o
r
m
a
n
ce
.
I
n
c
o
n
f
ig
u
r
atio
n
B
,
th
e
ViT
d
im
en
s
io
n
was
in
cr
ea
s
ed
to
2
5
6
with
eig
h
t
atten
tio
n
h
ea
d
s
,
b
u
t
ac
c
u
r
ac
y
ac
t
u
ally
d
ec
r
ea
s
ed
b
y
3
3
%.
C
o
n
f
ig
u
r
atio
n
s
A
an
d
B
ar
e
to
o
co
m
p
lex
,
lea
d
in
g
to
u
n
s
tab
le
lear
n
in
g
a
n
d
o
v
e
r
f
i
ttin
g
.
C
o
n
v
e
r
s
ely
,
co
n
f
ig
u
r
atio
n
C
is
m
o
r
e
lig
h
tweig
h
t
an
d
ac
h
iev
es
8
7
%
ac
cu
r
ac
y
.
T
h
is
s
im
p
ler
ar
c
h
itectu
r
e
ca
n
a
d
ju
s
t
th
e
lear
n
in
g
r
ate
an
d
im
p
r
o
v
e
tr
ain
i
n
g
e
f
f
icien
cy
wh
il
e
m
ain
tain
in
g
g
r
ad
ien
t
s
tab
ilit
y
.
T
h
e
o
p
tim
al
co
n
f
ig
u
r
atio
n
(
C
)
ac
h
iev
e
d
t
h
e
h
ig
h
est
p
er
f
o
r
m
an
ce
with
8
7
%
ac
cu
r
ac
y
,
s
ig
n
if
ican
tly
o
u
tp
e
r
f
o
r
m
in
g
t
h
e
in
itial
s
etu
p
(
A)
b
y
2
8
%.
Fu
r
t
h
er
m
o
r
e
,
th
e
m
in
im
al
lo
s
s
o
f
0
.
1
0
in
co
n
f
ig
u
r
atio
n
C
in
d
ic
ates
a
s
ig
n
if
ican
tly
m
o
r
e
p
r
ec
is
e
m
o
d
el
c
o
m
p
a
r
ed
to
th
e
o
th
er
s
.
T
h
ese
r
esu
lts
co
n
f
ir
m
t
h
at
th
e
a
p
p
lied
o
p
tim
i
za
tio
n
s
u
cc
ess
f
u
lly
s
tab
ilized
th
e
tr
ain
in
g
p
r
o
ce
s
s
an
d
m
ax
im
ize
d
lear
n
in
g
ab
ilit
y
.
T
ab
le
2
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
b
etwe
en
co
n
f
ig
u
r
atio
n
s
C
o
n
f
i
g
u
r
a
t
i
o
n
P
o
si
t
i
o
n
a
l
e
m
b
e
d
d
i
n
g
V
i
T
(
d
i
m
-
h
e
a
d
-
l
a
y
e
r
)
Le
a
r
n
i
n
g
r
a
t
e
A
c
c
u
r
a
c
y
(
%)
Lo
ss
C
o
n
v
e
r
g
e
n
c
e
r
e
mar
k
s
A
(
i
n
i
t
i
a
l
)
R
a
n
d
o
m
1
2
8
–
4
–
2
10
-
3
59
0
.
8
0
H
i
g
h
l
y
f
l
u
c
t
u
a
t
i
n
g
B
R
a
n
d
o
m
2
5
6
–
8
–
2
10
-
3
33
1
.
1
0
M
o
d
e
l
n
o
t
l
e
a
r
n
i
n
g
C (
o
p
t
i
ma
l
)
N
u
l
l
1
2
8
–
2
–
1
10
-
4
87
0
.
1
0
S
t
a
b
i
l
i
t
y
a
n
d
f
a
st
Fig
u
r
e
2
s
h
o
ws
th
e
tr
an
s
ien
t
-
s
tate
p
er
f
o
r
m
a
n
ce
f
r
o
m
T
a
b
le
3
o
f
th
r
ee
m
o
d
el
d
esig
n
s
.
Fi
g
u
r
e
2
(
a)
s
h
o
ws
th
e
ac
cu
r
ac
y
p
er
f
o
r
m
a
n
ce
.
I
n
itially
,
th
e
y
h
av
e
an
ac
cu
r
ac
y
o
f
5
9
%
with
f
l
u
ctu
atio
n
s
.
C
o
n
f
ig
u
r
atio
n
A
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
E
mo
tio
n
r
ec
o
g
n
itio
n
o
f e
lectro
en
ce
p
h
a
lo
g
r
a
m
u
s
in
g
h
yb
r
id
co
n
vo
lu
tio
n
a
l
…
(
E
s
mera
ld
a
C
o
n
tess
a
Dja
ma
l
)
3939
ap
p
ea
r
s
u
n
s
tab
le
d
u
e
to
th
e
r
a
n
d
o
m
e
m
b
ed
d
in
g
.
Ho
we
v
er
,
t
h
e
o
p
tim
al
co
n
f
ig
u
r
atio
n
B
co
n
tin
u
es
to
im
p
r
o
v
e
an
d
r
ea
ch
es
8
7
%.
C
o
n
f
ig
u
r
ati
o
n
B
r
em
ain
s
lo
w,
in
d
icatin
g
u
n
d
er
f
itti
n
g
d
u
e
to
ex
ce
s
s
iv
e
co
m
p
lex
ity
an
d
a
p
o
o
r
lear
n
in
g
r
ate.
Fo
r
th
e
l
o
s
s
v
alu
e
s
h
o
wn
in
Fig
u
r
e
2
(
b
)
,
th
e
o
r
an
g
e
cu
r
v
e
d
ec
r
e
ases
r
ap
id
ly
f
r
o
m
0
.
8
0
9
t
o
0
.
0
8
7
,
in
d
icatin
g
s
o
li
d
co
n
v
e
r
g
en
ce
is
o
cc
u
r
r
in
g
.
T
h
e
in
itializatio
n
r
esu
lts
r
em
ain
h
ig
h
an
d
u
n
s
tab
le,
wh
ile
co
n
f
ig
u
r
atio
n
B
f
ails
to
s
ig
n
if
ican
tly
r
ed
u
ce
th
e
lo
s
s
,
in
d
icatin
g
u
n
d
er
f
itti
n
g
.
(
a)
(
b
)
Fig
u
r
e
2
.
Per
f
o
r
m
an
c
e
o
f
h
y
b
r
id
C
NN
-
ViT
of
(
a
)
ac
cu
r
ac
y
a
n
d
(
b
)
lo
s
s
T
ab
le
3
p
r
esen
ts
a
d
etailed
co
m
p
ar
is
o
n
o
f
p
r
ec
is
io
n
,
r
ec
all,
an
d
F1
-
s
co
r
es
f
o
r
ea
ch
class
.
A
s
o
b
s
er
v
ed
,
th
e
p
r
o
p
o
s
ed
im
p
r
o
v
em
en
ts
r
esu
lted
i
n
s
ig
n
if
ica
n
t
p
er
f
o
r
m
an
ce
g
ain
s
ac
r
o
s
s
all
ca
teg
o
r
ies.
I
n
t
h
e
n
eg
ativ
e
class
,
p
r
ec
is
io
n
in
cr
e
ased
s
h
ar
p
ly
f
r
o
m
3
6
%
to
8
4
%,
wh
ile
r
ec
all
r
o
s
e
f
r
o
m
1
2
%
to
8
0
%,
in
d
icatin
g
th
at
th
e
o
p
tim
ized
m
o
d
el
s
u
cc
ess
f
u
lly
id
en
tifie
s
p
r
ev
io
u
s
ly
m
is
s
ed
n
eg
ativ
e
em
o
tio
n
s
.
Similar
en
h
an
ce
m
en
ts
ar
e
ev
id
en
t
in
th
e
n
eu
tr
al
class
,
wh
er
e
p
r
ec
is
io
n
im
p
r
o
v
e
d
f
r
o
m
4
7
%
to
8
2
%,
with
r
ec
all
r
em
ain
in
g
s
tab
le
at
a
h
ig
h
le
v
el
(
8
5
%
to
8
6
%).
Me
an
wh
ile,
th
e
p
o
s
itiv
e
class
ac
h
iev
ed
t
h
e
m
o
s
t
co
n
s
is
ten
t
r
es
u
lts
,
with
p
r
ec
is
io
n
r
is
in
g
f
r
o
m
8
0
%
to
9
4
%
an
d
r
ec
all
f
r
o
m
6
6
%
to
9
3
%,
cu
lm
i
n
atin
g
in
a
h
i
g
h
F1
-
s
co
r
e
o
f
9
3
%.
T
h
ese
m
etr
ics
co
n
f
ir
m
t
h
at
th
e
o
p
tim
al
co
n
f
i
g
u
r
atio
n
o
f
f
e
r
s
s
u
p
er
io
r
s
en
s
itiv
ity
an
d
p
r
ec
is
io
n
co
m
p
ar
ed
to
th
e
in
itial m
o
d
el.
T
ab
le
3
.
Per
f
o
r
m
an
ce
co
m
p
a
r
is
o
n
b
etwe
en
th
e
in
itial a
n
d
th
e
im
p
r
o
v
is
atio
n
C
l
a
s
s
P
r
e
c
i
s
i
o
n
(
%)
R
e
c
a
l
l
(
%)
F1
-
sc
o
r
e
(
%)
I
n
i
t
i
a
l
O
p
t
i
mal
I
n
i
t
i
a
l
O
p
t
i
mal
I
n
i
t
i
a
l
O
p
t
i
mal
N
e
g
a
t
i
v
e
36
84
12
80
18
82
N
e
u
t
r
a
l
47
82
85
86
60
84
P
o
si
t
i
v
e
80
94
66
93
72
93
3
.
2
.
E
f
f
ec
t
o
f
hy
brid co
m
po
nent:
a
bla
t
io
n t
esting
T
h
e
m
o
d
el
co
n
f
ig
u
r
atio
n
was
ab
lated
b
y
r
em
o
v
i
n
g
o
n
e
o
f
th
e
co
m
p
o
n
en
ts
o
f
DW
T
,
2
D
C
NN,
an
d
ViT
,
o
n
its
p
er
f
o
r
m
an
ce
,
as
s
h
o
wn
in
T
ab
le
4
.
I
t
is
s
h
o
wn
th
at
DW
T
im
p
r
o
v
es
ac
cu
r
a
cy
b
y
3
8
.
7
6
%
an
d
r
ed
u
ce
s
tim
e
b
y
2
1
.
6
8
m
in
u
t
es.
Me
an
wh
ile,
2
D
C
NN
s
p
atial
ex
tr
ac
tio
n
im
p
r
o
v
es
ac
cu
r
ac
y
b
y
9
.
5
3
%
an
d
r
ed
u
ce
s
tim
e
b
y
1
7
.
3
m
in
u
tes.
ViT
ac
h
iev
es
an
ac
cu
r
ac
y
i
n
cr
ea
s
e
o
f
1
.
5
8
%
b
u
t
r
ed
u
ce
s
co
m
p
u
tatio
n
tim
e
s
ig
n
if
ican
tly
,
n
am
ely
4
0
%,
ch
ar
ac
ter
is
tic
o
f
tr
ain
in
g
.
T
ab
le
4
.
Ab
latio
n
t
esti
n
g
of
D
W
T
,
2
D
C
NN,
ViT
M
o
d
e
l
A
c
c
u
r
a
c
y
(
%)
Lo
ss
Le
a
r
n
i
n
g
t
i
me
(
mi
n
u
t
e
s)
2
D
C
N
N
+
V
i
T
(
w
i
t
h
o
u
t
D
W
T)
4
7
.
9
3
(
↓
)
0
.
4
2
3
4
.
0
2
(
↓
)
D
W
T+
V
i
T
7
7
.
1
6
(
↑
)
0
.
4
8
2
9
.
7
0
(
↓
)
D
W
T+
2
D
C
N
N
8
5
.
1
1
(
↑
)
0
.
1
8
2
9
.
6
5
(
↓
)
D
W
T+
2
D
C
N
N
+
V
i
T
8
6
.
6
9
(
↑
)
0
.
0
8
1
2
.
3
4
(
↓
)
Fig
u
r
e
3
s
h
o
ws
th
at
th
e
m
o
d
el
with
o
u
t
DW
T
h
as
an
ac
cu
r
ac
y
b
elo
w
5
0
%
an
d
a
lo
s
s
ab
o
v
e
1
.
0
d
u
r
in
g
tr
ain
in
g
.
I
n
co
n
tr
ast,
t
h
e
DW
T
+
2
D
C
NN
+
ViT
m
o
d
el
s
h
o
ws
f
ast
an
d
s
tab
le
co
n
v
er
g
en
ce
s
tar
tin
g
f
r
o
m
th
e
1
0
t
h
ep
o
ch
.
T
h
e
co
m
b
in
atio
n
o
f
DW
T
,
C
NN,
an
d
ViT
is
ef
f
icien
t
an
d
r
o
b
u
s
t.
T
h
e
p
er
f
o
r
m
a
n
ce
in
ea
ch
ca
s
e
in
T
a
b
le
4
ca
n
b
e
i
llu
s
tr
ated
in
Fig
u
r
e
3
(
a)
f
o
r
a
cc
u
r
ac
y
,
w
h
ich
p
r
esen
ts
a
m
a
r
k
ed
im
p
r
o
v
em
en
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
5
,
No
.
4
,
Au
g
u
s
t 2
0
2
6
:
3
9
3
4
-
3
9
4
3
3940
o
v
er
co
n
f
ig
u
r
atio
n
s
with
o
u
t
DW
T
.
T
h
en
,
th
e
f
ea
tu
r
e
r
e
d
u
ce
s
d
im
en
s
io
n
ality
wh
ile
s
till
p
r
eser
v
in
g
lo
ca
l
s
p
atial
f
ea
tu
r
es
f
o
r
u
s
e
in
C
NNs.
Simu
ltan
eo
u
s
ly
,
ViT
is
also
ad
d
ed
,
ac
h
iev
in
g
t
h
e
h
i
g
h
est
ac
cu
r
ac
y
,
with
im
p
r
o
v
e
d
tr
ain
in
g
s
tab
ilit
y
an
d
r
ed
u
ce
d
co
m
p
u
tatio
n
tim
e.
Fig
u
r
e
3
(
b
)
p
r
esen
ts
th
e
co
r
r
esp
o
n
d
in
g
lo
s
s
co
m
p
ar
is
o
n
.
T
h
e
DW
T
–
C
NN
–
ViT
m
o
d
el
p
r
o
v
id
es
th
e
lo
west
lo
s
s
v
alu
e,
co
n
v
er
g
es,
a
n
d
is
s
tab
le.
T
h
ese
r
esu
lts
co
n
f
ir
m
th
at
DW
T
p
l
ay
s
a
cr
u
cial
r
o
le
in
e
x
tr
ac
ti
n
g
f
ea
t
u
r
es
in
r
elev
an
t
f
r
e
q
u
en
cy
b
an
d
s
.
C
NN
ef
f
ec
tiv
ely
m
o
d
els
lo
ca
l
s
p
atial
p
atter
n
s
,
an
d
ViT
ca
p
tu
r
es
s
im
u
ltan
eo
u
s
s
p
atio
-
tem
p
o
r
al
r
elatio
n
s
h
ip
s
th
r
o
u
g
h
s
elf
-
atten
tio
n
.
(
a)
(
b
)
Fig
u
r
e
3
.
Per
f
o
r
m
an
c
e
o
f
ea
c
h
co
m
p
o
n
en
t
of
(
a)
ac
c
u
r
ac
y
an
d
(
b
)
l
o
s
s
3
.
3
.
E
f
f
ec
t
o
f
le
a
rning
weig
ht
co
rr
ec
t
io
n t
ec
hn
iqu
es
Af
ter
d
eter
m
in
in
g
th
e
co
n
f
ig
u
r
atio
n
a
n
d
m
ain
co
m
p
o
n
en
t
s
,
s
ev
er
al
o
p
tim
izatio
n
alg
o
r
i
th
m
s
wer
e
ass
es
s
ed
f
o
r
th
eir
im
p
ac
t
o
n
t
r
ain
in
g
.
T
h
e
o
p
tim
izer
s
co
m
p
a
r
ed
wer
e
Ad
am
,
Ad
am
W
,
Ad
aM
ax
,
NAd
am
,
an
d
R
MSPr
o
p
,
all
with
lear
n
in
g
r
a
te
o
f
1
×
10
-
4
.
T
h
e
p
er
f
o
r
m
a
n
c
e
co
m
p
ar
is
o
n
r
esu
lts
ar
e
p
r
ese
n
ted
in
T
ab
le
5
.
T
ab
le
5
.
C
o
m
p
a
r
is
o
n
o
f
s
o
m
e
o
p
tim
izer
s
O
p
t
i
mi
z
e
r
A
c
c
u
r
a
c
y
(
%)
Lo
ss
Le
a
r
n
i
n
g
t
i
me
(
mi
n
u
t
e
s)
A
d
a
m
86
0
.
0
9
1
1
2
.
3
5
A
d
a
mW
87
0
.
0
8
7
1
2
.
3
4
A
d
a
M
a
x
83
0
.
1
2
8
1
3
.
5
0
N
A
d
a
m
86
0
.
0
9
3
1
8
.
0
4
R
M
S
P
r
o
p
86
0
.
1
0
5
1
5
.
3
8
B
ased
o
n
T
a
b
le
6
,
all
o
p
tim
i
ze
r
s
d
em
o
n
s
tr
ated
r
elativ
ely
s
tab
le
p
er
f
o
r
m
a
n
ce
with
ac
cu
r
ac
y
ab
o
v
e
8
2
%.
T
h
e
d
i
f
f
er
en
ce
s
in
r
esu
lts
ac
r
o
s
s
o
p
tim
izer
s
we
r
e
n
o
t
s
i
g
n
if
ican
t,
in
d
icatin
g
th
at
th
e
m
o
d
el
h
ad
ac
h
iev
e
d
ar
ch
itectu
r
al
s
tab
ilit
y
;
th
u
s
,
ch
an
g
es
to
th
e
o
p
tim
izatio
n
alg
o
r
ith
m
d
id
n
o
t
s
ig
n
i
f
ican
t
ly
af
f
ec
t
t
h
e
f
i
n
al
p
er
f
o
r
m
an
ce
.
T
h
e
m
o
r
e
in
f
l
u
en
tial
f
ac
to
r
s
af
f
ec
tin
g
ac
c
u
r
ac
y
wer
e
m
o
d
el
co
n
f
ig
u
r
atio
n
,
ar
ch
itectu
r
al
co
m
p
o
n
en
ts
(
DW
T
,
2
D
C
NN,
an
d
ViT
)
,
an
d
lear
n
in
g
r
ate
a
d
ju
s
tm
en
ts
.
Nev
er
th
eless
,
Ad
a
m
W
was
ch
o
s
en
as
th
e
p
r
im
ar
y
o
p
tim
izer
b
ec
au
s
e
it
p
r
o
v
i
d
ed
th
e
b
est
b
alan
ce
b
etwe
en
g
r
ad
ien
t
s
tab
ilit
y
,
8
7
%
ac
cu
r
ac
y
,
a
n
d
an
ef
f
icien
t le
ar
n
in
g
tim
e
o
f
1
2
m
in
u
tes.
3
.
4
.
Co
m
pa
riso
n wit
h
prev
io
us
re
s
ea
rc
h
T
o
ass
ess
th
e
ef
f
ec
tiv
en
ess
o
f
th
e
p
r
o
p
o
s
ed
m
o
d
el,
a
c
o
m
p
ar
is
o
n
was
co
n
d
u
cted
w
ith
s
ev
er
al
p
r
ev
io
u
s
s
tu
d
ies
o
n
th
e
SEE
D
d
ataset.
T
h
e
g
o
al
was
to
ev
alu
ate
th
e
p
er
f
o
r
m
a
n
ce
i
m
p
r
o
v
e
m
en
t
o
f
t
h
e
DW
T
+
h
y
b
r
id
2
D
C
NN
-
ViT
m
eth
o
d
c
o
m
p
a
r
ed
to
C
N
N
-
an
d
R
NN
-
b
ased
ap
p
r
o
ac
h
es,
as
p
r
esen
ted
in
T
ab
le
6
.
T
h
e
p
er
f
o
r
m
an
ce
o
f
t
h
e
p
r
o
p
o
s
ed
m
eth
o
d
was
test
ed
o
n
th
e
s
am
e
d
ataset,
SEE
D.
T
ab
le
6
s
h
o
ws
th
at
th
e
DW
T
+
h
y
b
r
id
2
D
C
NN
-
ViT
ap
p
r
o
ac
h
s
u
cc
ess
f
u
lly
s
u
r
p
ass
ed
p
r
ev
i
o
u
s
r
esear
ch
.
T
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
ac
h
iev
ed
an
ac
c
u
r
ac
y
o
f
8
6
.
6
9
%,
h
ig
h
er
th
a
n
th
e
C
NN
-
o
n
l
y
an
d
C
NN
-
R
NN
ap
p
r
o
ac
h
es.
T
h
e
in
teg
r
atio
n
o
f
DW
T
,
C
N
N,
an
d
ViT
im
p
r
o
v
es
th
e
m
o
d
el'
s
ab
ilit
y
to
r
ec
o
g
n
ize
em
o
tio
n
al
p
atter
n
s
m
o
r
e
ac
cu
r
ately
,
ef
f
icien
tly
,
an
d
s
tab
ly
an
d
h
as
th
e
p
o
ten
tial to
b
e
ap
p
lied
to
e
m
o
tio
n
m
o
n
ito
r
in
g
.
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
E
mo
tio
n
r
ec
o
g
n
itio
n
o
f e
lectro
en
ce
p
h
a
lo
g
r
a
m
u
s
in
g
h
yb
r
id
co
n
vo
lu
tio
n
a
l
…
(
E
s
mera
ld
a
C
o
n
tess
a
Dja
ma
l
)
3941
T
ab
le
6
.
C
o
m
p
a
r
is
o
n
with
p
r
e
v
io
u
s
r
esear
ch
M
e
t
h
o
d
A
c
c
u
r
a
c
y
(
%)
D
i
f
f
e
r
e
n
t
i
a
l
e
n
t
r
o
p
y
(
D
E)
+
S
V
M
a
n
d
g
r
a
p
h
r
e
g
u
l
a
r
i
z
e
d
e
x
t
r
e
m
e
l
e
a
r
n
i
n
g
m
a
c
h
i
n
e
(
G
E
LM
)
[
3
6
]
79
W
a
v
e
l
e
t
+
2
D
C
N
N
[
1
3
]
83
W
a
v
e
l
e
t
+
m
u
l
t
i
p
l
e
C
N
N
[
1
4
]
81
H
y
b
r
i
d
p
a
r
a
l
l
e
l
2
D
C
N
N
-
G
R
U
[
1
1
]
84
D
W
T
+
h
y
b
r
i
d
2
D
C
N
N
-
V
i
T
(
p
r
o
p
o
s
e
d
met
h
o
d
s)
87
4.
CO
NCLU
SI
O
N
T
h
e
r
esear
ch
s
h
o
ws
a
way
to
em
o
tio
n
r
ec
o
g
n
itio
n
o
f
E
E
G
s
ig
n
als
u
s
in
g
a
h
y
b
r
i
d
wav
elet
tr
an
s
f
o
r
m
,
th
e
2
D
C
NN,
an
d
th
e
ViT
.
T
h
is
s
ee
m
s
to
im
p
r
o
v
e
t
h
e
r
ec
o
g
n
itio
n
o
f
em
o
tio
n
al
p
atter
n
s
an
d
th
e
r
o
b
u
s
tn
ess
o
v
er
all.
W
av
elet
tr
an
s
f
o
r
m
with
DW
T
g
ets
th
e
r
elev
an
t
f
r
eq
u
e
n
cy
b
a
n
d
s
an
d
h
an
d
le
s
th
e
n
o
n
-
s
tatio
n
ar
y
asp
ec
ts
o
f
th
e
E
E
G
s
ig
n
als.
T
h
e
2
D
C
NN
ca
p
tu
r
es
lo
ca
l
s
p
atial
f
ea
tu
r
es,
wh
ile
th
e
ViT
lear
n
s
g
lo
b
al
r
elatio
n
s
h
ip
s
b
etwe
en
f
ea
tu
r
e
s
th
r
o
u
g
h
s
elf
-
atten
tio
n
,
en
ab
lin
g
it
to
ca
p
tu
r
e
m
o
r
e
co
m
p
lex
p
atter
n
s
th
an
a
s
in
g
le
C
NN
m
eth
o
d
.
T
h
e
co
m
b
in
atio
n
o
f
t
h
ese
th
r
ee
m
eth
o
d
s
p
r
o
v
id
ed
g
o
o
d
class
if
icatio
n
p
er
f
o
r
m
an
ce
i
n
d
is
tin
g
u
is
h
in
g
em
o
tio
n
al
s
tat
es.
T
h
e
e
v
alu
atio
n
m
etr
ics
in
clu
d
e
ac
cu
r
ac
y
,
p
r
ec
is
io
n
,
r
e
ca
ll,
an
d
F1
-
s
co
r
e,
wh
ich
g
av
e
g
o
o
d
p
er
f
o
r
m
an
ce
.
T
h
is
ap
p
r
o
ac
h
d
em
o
n
s
tr
ates
h
ig
h
p
o
ten
tial
f
o
r
ap
p
licatio
n
i
n
B
C
I
s
y
s
tem
s
an
d
r
ea
l
-
tim
e
em
o
tio
n
al
m
o
n
ito
r
in
g
.
Ar
ch
itectu
r
al
co
n
f
ig
u
r
atio
n
s
s
u
ch
as
th
e
n
u
m
b
er
o
f
atten
t
io
n
h
ea
d
s
,
en
co
d
er
lay
er
s
,
p
ar
am
ete
r
s
elec
tio
n
,
an
d
weig
h
t
co
r
r
ec
tio
n
tec
h
n
iq
u
es
o
r
o
p
tim
izer
s
s
ig
n
if
ica
n
tly
in
f
lu
e
n
ce
th
e
ac
cu
r
ac
y
,
s
tab
ilit
y
,
an
d
r
o
b
u
s
tn
ess
o
f
em
o
tio
n
r
ec
o
g
n
itio
n
.
T
h
e
Ad
a
m
W
o
p
tim
izer
p
r
o
v
id
es
ac
c
u
r
ac
y
im
p
r
o
v
em
e
n
ts
,
alth
o
u
g
h
n
o
t
s
ig
n
if
ican
tly
m
o
r
e
th
an
th
e
o
th
er
ar
c
h
itectu
r
al
co
n
f
ig
u
r
atio
n
s
.
Ov
er
all,
th
e
r
ec
o
g
n
itio
n
p
er
f
o
r
m
an
ce
o
f
th
e
th
r
ee
em
o
tio
n
class
es
is
r
elativ
ely
s
im
ilar
.
Ho
wev
er
,
th
e
p
o
s
itiv
e
em
o
tio
n
class
p
r
o
v
id
es
s
u
p
er
i
o
r
ac
c
u
r
a
cy
co
m
p
ar
ed
to
t
h
e
o
t
h
er
clas
s
es.
T
h
is
is
n
o
t
o
n
ly
d
u
e
t
o
th
e
ch
ar
ac
ter
is
tics
o
f
tr
ain
in
g
d
ata,
b
u
t
also
b
ec
au
s
e
th
e
p
o
s
itiv
e
em
o
tio
n
class
is
r
elativ
ely
ea
s
y
to
r
ec
o
g
n
ize
f
r
o
m
its
m
o
r
e
d
is
tin
ctiv
e
f
r
eq
u
e
n
cy
b
an
d
,
w
h
ich
is
en
h
an
ce
d
b
y
th
e
u
s
e
o
f
wav
elet
tr
an
s
f
o
r
m
s
.
Fu
tu
r
e
wo
r
k
will
f
o
cu
s
o
n
ex
ten
d
in
g
th
e
p
r
o
p
o
s
ed
m
o
d
els
f
o
r
cr
o
s
s
-
d
ataset
ev
alu
ati
o
n
,
ex
p
lo
r
in
g
ad
a
p
tiv
e
ch
a
n
n
el
s
elec
tio
n
,
an
d
in
v
esti
g
atin
g
lig
h
tweig
h
t tr
a
n
s
f
o
r
m
er
v
ar
ian
ts
to
im
p
r
o
v
e
g
e
n
er
aliza
tio
n
an
d
r
ea
l
-
tim
e
p
e
r
f
o
r
m
an
ce
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
r
esear
ch
is
f
u
n
d
e
d
b
y
t
h
e
p
en
elitia
n
fu
n
d
a
men
ta
l
r
eg
u
la
r
2
0
2
5
f
r
o
m
th
e
I
n
d
o
n
esian
Dir
ec
to
r
ate
Gen
er
al
o
f
R
esear
ch
an
d
C
o
m
m
u
n
ity
Ser
v
ice,
Min
is
tr
y
o
f
Hig
h
er
E
d
u
c
atio
n
,
Scien
ce
,
an
d
T
ec
h
n
o
lo
g
y
u
n
d
er
co
n
tr
ac
t 1
2
5
/C
3
/DT
.
0
5
.
0
0
/PL/2
0
2
5
.
AUTHO
R
CO
NT
RI
B
UT
I
O
NS ST
A
T
E
M
E
N
T
T
h
is
jo
u
r
n
al
u
s
es
th
e
C
o
n
tr
ib
u
to
r
R
o
les
T
ax
o
n
o
m
y
(
C
R
ed
iT)
to
r
ec
o
g
n
ize
in
d
iv
id
u
al
au
th
o
r
co
n
tr
ib
u
tio
n
s
,
r
ed
u
ce
au
th
o
r
s
h
ip
d
is
p
u
tes,
an
d
f
ac
ilit
ate
co
llab
o
r
atio
n
.
Na
m
e
o
f
Aut
ho
r
C
M
So
Va
Fo
I
R
D
O
E
Vi
Su
P
Fu
E
s
m
er
ald
a
C
o
n
tess
a
Djam
al
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
R
ev
an
Vio
E
n
d
r
ia
n
s
y
ah
✓
✓
✓
✓
✓
✓
✓
✓
Daswar
a
Djajasa
s
m
ita
✓
✓
✓
✓
✓
✓
✓
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
s
u
p
p
o
r
tin
g
th
e
SEE
D
d
ataset
co
n
s
is
ts
o
f
E
E
G
d
a
tasets
p
r
o
v
id
ed
b
y
th
e
B
C
MI
lab
o
r
ato
r
y
,
av
ailab
le
at
h
ttp
s
://b
cm
i.sjt
u
.
ed
u
.
cn
/h
o
m
e/seed
/,
with
5
-
s
e
co
n
d
s
eg
m
e
n
tatio
n
s
.
C
o
d
in
g
o
f
th
e
ex
p
er
im
en
t
av
ailab
le
in
GitHu
b
at
h
ttp
s
://g
ith
u
b
.
c
o
m
/R
ev
an
VE
/eeg
-
em
o
tio
n
-
class
if
icatio
n
-
with
-
VI
T
.
g
it
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
9
3
8
I
n
t J Ar
tif
I
n
tell
,
Vo
l.
1
5
,
No
.
4
,
Au
g
u
s
t 2
0
2
6
:
3
9
3
4
-
3
9
4
3
3942
RE
F
E
R
E
NC
E
S
[
1
]
P
.
Le
szc
z
e
ł
o
w
s
k
a
a
n
d
N
.
D
a
w
i
d
o
w
sk
a
,
“
S
y
s
t
e
m
a
t
i
c
l
i
t
e
r
a
t
u
r
e
r
e
v
i
e
w
f
o
r
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
f
r
o
m
EEG
si
g
n
a
l
s
,
”
N
e
w
T
r
e
n
d
s
i
n
D
a
t
a
b
a
se
a
n
d
I
n
f
o
rm
a
t
i
o
n
S
y
st
e
m
s
(
AD
BI
S
2
0
2
2
)
.
C
h
a
m
,
S
w
i
t
z
e
r
l
a
n
d
:
S
p
r
i
n
g
e
r
,
p
p
.
4
6
7
–
4
7
5
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
3
-
0
3
1
-
1
5
7
4
3
-
1
_
4
3
.
[
2
]
Y
.
W
a
n
g
e
t
a
l
.
,
“
A
s
y
s
t
e
ma
t
i
c
r
e
v
i
e
w
o
n
a
f
f
e
c
t
i
v
e
c
o
m
p
u
t
i
n
g
:
e
m
o
t
i
o
n
mo
d
e
l
s
,
d
a
t
a
b
a
ses
,
a
n
d
r
e
c
e
n
t
a
d
v
a
n
c
e
s,
”
I
n
f
o
rm
a
t
i
o
n
Fu
si
o
n
,
v
o
l
.
8
3
–
8
4
,
p
p
.
1
9
–
5
2
,
J
u
l
.
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
i
n
f
f
u
s.
2
0
2
2
.
0
3
.
0
0
9
.
[
3
]
J.
N
.
S
c
h
n
e
i
d
e
r
,
M
.
M
a
t
y
j
e
k
,
A
.
W
e
i
g
a
n
d
,
I
.
D
z
i
o
b
e
k
,
a
n
d
T.
R
.
B
r
i
c
k
,
“
S
u
b
j
e
c
t
i
v
e
a
n
d
o
b
j
e
c
t
i
v
e
d
i
f
f
i
c
u
l
t
y
o
f
e
m
o
t
i
o
n
a
l
f
a
c
i
a
l
e
x
p
r
e
ss
i
o
n
p
e
r
c
e
p
t
i
o
n
f
r
o
m
d
y
n
a
m
i
c
s
t
i
m
u
l
i
,
”
PLO
S
O
N
E
,
v
o
l
.
1
7
,
n
o
.
6
,
J
u
n
.
2
0
2
2
,
d
o
i
:
1
0
.
1
3
7
1
/
j
o
u
r
n
a
l
.
p
o
n
e
.
0
2
6
9
1
5
6
.
[
4
]
W
.
M
a
,
Y
.
Z
h
e
n
g
,
T.
L
i
,
Z
.
Li
,
Y
.
Li
,
a
n
d
L.
W
a
n
g
,
“
A
c
o
mp
r
e
h
e
n
s
i
v
e
r
e
v
i
e
w
o
f
d
e
e
p
l
e
a
r
n
i
n
g
i
n
EE
G
-
b
a
se
d
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
:
c
l
a
ssi
f
i
c
a
t
i
o
n
s,
t
r
e
n
d
s
,
a
n
d
p
r
a
c
t
i
c
a
l
i
m
p
l
i
c
a
t
i
o
n
s
,
”
P
e
e
rJ
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
1
0
,
M
a
y
2
0
2
4
,
d
o
i
:
1
0
.
7
7
1
7
/
p
e
e
r
j
-
c
s
.
2
0
6
5
.
[
5
]
A
.
C
h
a
d
d
a
d
,
Y
.
W
u
,
R
.
K
a
t
e
b
,
a
n
d
A
.
B
o
u
r
i
d
a
n
e
,
“
El
e
c
t
r
o
e
n
c
e
p
h
a
l
o
g
r
a
p
h
y
si
g
n
a
l
p
r
o
c
e
ssi
n
g
:
a
c
o
mp
r
e
h
e
n
si
v
e
r
e
v
i
e
w
a
n
d
a
n
a
l
y
si
s
o
f
m
e
t
h
o
d
s
a
n
d
t
e
c
h
n
i
q
u
e
s,”
S
e
n
so
rs
,
v
o
l
.
2
3
,
n
o
.
1
4
,
Ju
l
.
2
0
2
3
,
d
o
i
:
1
0
.
3
3
9
0
/
s2
3
1
4
6
4
3
4
.
[
6
]
Z.
Zh
a
n
g
,
J
.
M
.
F
o
r
t
,
a
n
d
L.
G
.
M
a
t
e
u
,
“
M
i
n
i
r
e
v
i
e
w
:
c
h
a
l
l
e
n
g
e
s
i
n
EEG
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
,
”
Fr
o
n
t
i
e
rs
i
n
Psy
c
h
o
l
o
g
y
,
v
o
l
.
1
4
,
Jan
.
2
0
2
4
,
d
o
i
:
1
0
.
3
3
8
9
/
f
p
s
y
g
.
2
0
2
3
.
1
2
8
9
8
1
6
.
[
7
]
J.
W
a
n
g
a
n
d
M
.
W
a
n
g
,
“
R
e
v
i
e
w
o
f
t
h
e
e
m
o
t
i
o
n
a
l
f
e
a
t
u
r
e
e
x
t
r
a
c
t
i
o
n
a
n
d
c
l
a
ssi
f
i
c
a
t
i
o
n
u
si
n
g
EEG
s
i
g
n
a
l
s,
”
C
o
g
n
i
t
i
v
e
R
o
b
o
t
i
c
s
,
v
o
l
.
1
,
p
p
.
2
9
–
4
0
,
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
o
g
r
.
2
0
2
1
.
0
4
.
0
0
1
.
[
8
]
W.
-
L.
Z
h
e
n
g
a
n
d
B
.
-
L.
Lu
,
“
I
n
v
e
st
i
g
a
t
i
n
g
c
r
i
t
i
c
a
l
f
r
e
q
u
e
n
c
y
b
a
n
d
s
a
n
d
c
h
a
n
n
e
l
s
f
o
r
EEG
-
b
a
se
d
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
w
i
t
h
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
s,
”
I
EEE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
Au
t
o
n
o
m
o
u
s
M
e
n
t
a
l
D
e
v
e
l
o
p
m
e
n
t
,
v
o
l
.
7
,
n
o
.
3
,
p
p
.
1
6
2
–
1
7
5
,
S
e
p
.
2
0
1
5
,
d
o
i
:
1
0
.
1
1
0
9
/
TA
M
D
.
2
0
1
5
.
2
4
3
1
4
9
7
.
[
9
]
R.
-
N
.
D
u
a
n
,
J
.
-
Y
.
Zh
u
,
a
n
d
B
.
-
L
.
L
u
,
“
D
i
f
f
e
r
e
n
t
i
a
l
e
n
t
r
o
p
y
f
e
a
t
u
r
e
f
o
r
E
EG
-
b
a
s
e
d
e
mo
t
i
o
n
c
l
a
ssi
f
i
c
a
t
i
o
n
,
”
i
n
2
0
1
3
6
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
I
EEE
/
EM
BS
C
o
n
f
e
r
e
n
c
e
o
n
N
e
u
r
a
l
En
g
i
n
e
e
r
i
n
g
,
N
o
v
.
2
0
1
3
,
p
p
.
8
1
–
84
,
d
o
i
:
1
0
.
1
1
0
9
/
N
ER
.
2
0
1
3
.
6
6
9
5
8
7
6
.
[
10
]
K
.
K
a
mb
l
e
a
n
d
J.
S
e
n
g
u
p
t
a
,
“
A
c
o
m
p
r
e
h
e
n
si
v
e
s
u
r
v
e
y
o
n
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
b
a
se
d
o
n
e
l
e
c
t
r
o
e
n
c
e
p
h
a
l
o
g
r
a
p
h
(
EEG
)
si
g
n
a
l
s,
”
Mu
l
t
i
m
e
d
i
a
T
o
o
l
s
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
8
2
,
n
o
.
1
8
,
p
p
.
2
7
2
6
9
–
2
7
3
0
4
,
Ju
l
.
2
0
2
3
,
d
o
i
:
1
0
.
1
0
0
7
/
s1
1
0
4
2
-
0
2
3
-
1
4
4
8
9
-
9.
[1
1
]
N
.
A
.
P
u
t
r
i
,
E
.
C
.
D
j
a
m
a
l
,
F
.
N
u
g
r
a
h
a
,
a
n
d
F
.
K
a
s
y
i
d
i
,
“
E
EG
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
u
s
i
n
g
p
a
r
a
l
l
e
l
h
y
b
r
i
d
c
o
n
v
o
l
u
t
i
o
n
a
l
-
r
e
c
u
r
r
e
n
t
n
e
u
r
a
l
n
e
t
w
o
r
k
s,
”
i
n
2
0
2
2
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
D
a
t
a
S
c
i
e
n
c
e
a
n
d
I
t
s
Ap
p
l
i
c
a
t
i
o
n
s
,
Ju
l
.
2
0
2
2
,
p
p
.
2
4
–
29
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
o
D
S
A
5
5
8
7
4
.
2
0
2
2
.
9
8
6
2
8
5
3
.
[1
2
]
H
.
M
.
M
.
H
a
sa
n
a
n
d
M
.
A
.
I
sl
a
m
,
“
Emo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
f
r
o
m
B
e
n
g
a
l
i
sp
e
e
c
h
u
s
i
n
g
R
N
N
mo
d
u
l
a
t
i
o
n
-
b
a
se
d
c
a
t
e
g
o
r
i
z
a
t
i
o
n
,
”
i
n
2
0
2
0
T
h
i
r
d
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
S
m
a
r
t
S
y
st
e
m
s
a
n
d
I
n
v
e
n
t
i
v
e
T
e
c
h
n
o
l
o
g
y
,
A
u
g
.
2
0
2
0
,
p
p
.
1
1
3
1
–
1
1
3
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
S
S
I
T4
8
9
1
7
.
2
0
2
0
.
9
2
1
4
1
9
6
.
[
1
3
]
R
.
H
.
D
.
H
a
q
q
u
e
,
E.
C
.
D
j
a
ma
l
,
a
n
d
A
.
W
u
l
a
n
d
a
r
i
,
“
Emo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
o
f
EEG
si
g
n
a
l
s
u
si
n
g
w
a
v
e
l
e
t
f
i
l
t
e
r
a
n
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
,
”
i
n
2
0
2
1
8
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
re
n
c
e
o
n
Ad
v
a
n
c
e
d
I
n
f
o
r
m
a
t
i
c
s:
C
o
n
c
e
p
t
s,
T
h
e
o
r
y
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
,
S
e
p
.
2
0
2
1
,
p
p
.
1
–
6
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
A
I
C
TA
5
3
2
1
1
.
2
0
2
1
.
9
6
4
0
2
7
9
.
[1
4
]
H
.
D
.
K
h
a
i
r
u
n
i
s
sa,
E.
C
.
D
j
a
ma
l
,
a
n
d
A
.
W
u
l
a
n
d
a
r
i
,
“
M
u
l
t
i
p
l
e
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
i
n
EEG
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
,
”
i
n
2
0
2
1
4
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
f
C
o
m
p
u
t
e
r
a
n
d
I
n
f
o
r
m
a
t
i
c
s
En
g
i
n
e
e
r
i
n
g
,
S
e
p
.
2
0
2
1
,
p
p
.
3
0
–
35
,
d
o
i
:
1
0
.
1
1
0
9
/
I
C
2
I
E5
3
2
1
9
.
2
0
2
1
.
9
6
4
9
2
1
5
.
[1
5
]
Y
.
Li
,
W
.
Z
h
e
n
g
,
Z
.
C
u
i
,
T.
Zh
a
n
g
,
a
n
d
Y
.
Zo
n
g
,
“
A
n
o
v
e
l
n
e
u
r
a
l
n
e
t
w
o
r
k
m
o
d
e
l
b
a
se
d
o
n
c
e
r
e
b
r
a
l
h
e
mi
s
p
h
e
r
i
c
a
sy
mm
e
t
r
y
f
o
r
EEG
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
,
”
i
n
Pr
o
c
e
e
d
i
n
g
s
o
f
t
h
e
T
w
e
n
t
y
-
S
e
v
e
n
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
i
n
t
C
o
n
f
e
r
e
n
c
e
o
n
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
,
Ju
l
.
2
0
1
8
,
p
p
.
1
5
6
1
–
1
5
6
7
,
d
o
i
:
1
0
.
2
4
9
6
3
/
i
j
c
a
i
.
2
0
1
8
/
2
1
6
.
[1
6
]
Y
.
M
a
e
t
a
l
.
,
“
Em
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
m
o
d
e
l
o
f
E
EG
s
i
g
n
a
l
s
b
a
se
d
o
n
d
o
u
b
l
e
a
t
t
e
n
t
i
o
n
mec
h
a
n
i
sm,
”
Bra
i
n
S
c
i
e
n
c
e
s
,
v
o
l
.
1
4
,
n
o
.
1
2
,
D
e
c
.
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
b
r
a
i
n
s
c
i
1
4
1
2
1
2
8
9
.
[1
7
]
W
.
Ta
o
e
t
a
l
.
,
“
EEG
-
b
a
se
d
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
v
i
a
c
h
a
n
n
e
l
-
w
i
se
a
t
t
e
n
t
i
o
n
a
n
d
se
l
f
a
t
t
e
n
t
i
o
n
,
”
I
E
EE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
Af
f
e
c
t
i
v
e
C
o
m
p
u
t
i
n
g
,
v
o
l
.
1
4
,
n
o
.
1
,
p
p
.
3
8
2
–
3
9
3
,
J
a
n
.
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
TA
F
F
C
.
2
0
2
0
.
3
0
2
5
7
7
7
.
[
18
]
A
.
D
o
s
o
v
i
t
s
k
i
y
e
t
a
l
.
,
“
A
n
i
m
a
g
e
i
s
w
o
r
t
h
1
6
x
1
6
w
o
r
d
s
:
t
r
a
n
sf
o
r
mers
f
o
r
i
m
a
g
e
r
e
c
o
g
n
i
t
i
o
n
a
t
sca
l
e
,
”
i
n
I
C
L
R
2
0
2
1
-
9
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
L
e
a
r
n
i
n
g
Re
p
res
e
n
t
a
t
i
o
n
s
,
2
0
2
1
,
p
p
.
1
–
2
1
.
[
19
]
A
.
A
r
j
u
n
,
A
.
S
.
R
a
j
p
o
o
t
,
a
n
d
M
.
R
.
P
a
n
i
c
k
e
r
,
“
I
n
t
r
o
d
u
c
i
n
g
a
t
t
e
n
t
i
o
n
m
e
c
h
a
n
i
sm
f
o
r
EEG
si
g
n
a
l
s:
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
w
i
t
h
v
i
si
o
n
t
r
a
n
sf
o
r
mers
,
”
i
n
2
0
2
1
4
3
r
d
A
n
n
u
a
l
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
f
t
h
e
I
EEE
En
g
i
n
e
e
ri
n
g
i
n
M
e
d
i
c
i
n
e
&
Bi
o
l
o
g
y
S
o
c
i
e
t
y
,
N
o
v
.
2
0
2
1
,
p
p
.
5
7
2
3
–
5
7
2
6
,
d
o
i
:
1
0
.
1
1
0
9
/
E
M
B
C
4
6
1
6
4
.
2
0
2
1
.
9
6
2
9
8
3
7
.
[2
0
]
M
.
M
a
n
s
o
u
r
,
L.
N
.
M
a
h
m
o
u
d
,
I
.
S
a
d
e
k
,
a
n
d
N
.
M
.
S
a
l
e
m,
“
E
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
f
r
o
m
EEG
si
g
n
a
l
s
u
s
i
n
g
v
i
s
i
o
n
t
r
a
n
sf
o
r
m
e
r
a
n
d
C
N
N
,
”
i
n
2
0
2
5
I
n
t
e
r
n
a
t
i
o
n
a
l
T
e
l
e
c
o
m
m
u
n
i
c
a
t
i
o
n
s
C
o
n
f
e
re
n
c
e
(
I
T
C
-
E
g
y
p
t
)
,
Ju
l
.
2
0
2
5
,
p
p
.
2
5
9
–
2
6
4
,
d
o
i
:
1
0
.
1
1
0
9
/
I
TC
-
Eg
y
p
t
6
6
0
9
5
.
2
0
2
5
.
1
1
1
8
6
5
6
2
.
[2
1
]
K
.
D
e
v
a
r
a
j
a
n
,
S
.
P
o
n
n
a
n
,
a
n
d
S
.
P
e
r
u
mal
,
“
H
y
b
r
i
d
C
N
N
-
t
r
a
n
sf
o
r
m
e
r
a
r
c
h
i
t
e
c
t
u
r
e
f
o
r
e
n
h
a
n
c
e
d
EEG
-
b
a
s
e
d
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
:
c
a
p
t
u
r
i
n
g
l
o
c
a
l
a
n
d
g
l
o
b
a
l
d
e
p
e
n
d
e
n
c
i
e
s
w
i
t
h
s
e
l
f
-
a
t
t
e
n
t
i
o
n
me
c
h
a
n
i
sm
s,”
D
i
s
c
o
v
e
r
C
o
m
p
u
t
i
n
g
,
v
o
l
.
2
8
,
n
o
.
1
,
M
a
y
2
0
2
5
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
7
9
1
-
025
-
0
9
5
9
6
-
0.
[
2
2
]
Z.
-
M
.
W
a
n
g
,
S
.
-
Y
.
H
u
,
a
n
d
H
.
S
o
n
g
,
“
C
h
a
n
n
e
l
se
l
e
c
t
i
o
n
me
t
h
o
d
f
o
r
EEG
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
n
o
r
mal
i
z
e
d
mu
t
u
a
l
i
n
f
o
r
mat
i
o
n
,
”
I
E
EE
Ac
c
e
ss
,
v
o
l
.
7
,
p
p
.
1
4
3
3
0
3
–
1
4
3
3
1
1
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
1
9
.
2
9
4
4
2
7
3
.
[2
3
]
A
.
S
u
b
a
s
i
,
T
.
T
u
n
c
e
r
,
S
.
D
o
g
a
n
,
D
.
T
a
n
k
o
,
a
n
d
U
.
S
a
k
o
g
l
u
,
“
EEG
-
b
a
s
e
d
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
u
si
n
g
t
u
n
a
b
l
e
Q
w
a
v
e
l
e
t
t
r
a
n
sf
o
r
m
a
n
d
r
o
t
a
t
i
o
n
f
o
r
e
s
t
e
n
sem
b
l
e
c
l
a
ss
i
f
i
e
r
,
”
B
i
o
m
e
d
i
c
a
l
S
i
g
n
a
l
Pro
c
e
ssi
n
g
a
n
d
C
o
n
t
r
o
l
,
v
o
l
.
6
8
,
J
u
l
.
2
0
2
1
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
b
s
p
c
.
2
0
2
1
.
1
0
2
6
4
8
.
[2
4
]
P
.
K
e
e
l
a
w
a
t
,
N
.
T
h
a
mm
a
s
a
n
,
M
.
N
u
mao
,
a
n
d
B
.
K
i
j
s
i
r
i
k
u
l
,
“
A
c
o
m
p
a
r
a
t
i
v
e
s
t
u
d
y
o
f
w
i
n
d
o
w
si
z
e
a
n
d
c
h
a
n
n
e
l
a
r
r
a
n
g
e
me
n
t
o
n
EEG
-
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
u
s
i
n
g
d
e
e
p
C
N
N
,
”
S
e
n
s
o
rs
,
v
o
l
.
2
1
,
n
o
.
5
,
M
a
r
.
2
0
2
1
,
d
o
i
:
1
0
.
3
3
9
0
/
s
2
1
0
5
1
6
7
8
.
[2
5
]
S
.
K
.
K
h
a
r
e
a
n
d
V
.
B
a
j
a
j
,
“
T
i
me
–
f
r
e
q
u
e
n
c
y
r
e
p
r
e
se
n
t
a
t
i
o
n
a
n
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
-
b
a
s
e
d
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
,
”
I
E
EE
T
ra
n
s
a
c
t
i
o
n
s
o
n
N
e
u
r
a
l
N
e
t
w
o
rks
a
n
d
L
e
a
rn
i
n
g
S
y
s
t
e
m
s
,
v
o
l
.
3
2
,
n
o
.
7
,
p
p
.
2
9
0
1
–
2
9
0
9
,
J
u
l
.
2
0
2
1
,
d
o
i
:
1
0
.
1
1
0
9
/
TN
N
LS.
2
0
2
0
.
3
0
0
8
9
3
8
.
[2
6
]
S
.
K
.
G
a
t
f
a
n
,
“
A
r
e
v
i
e
w
o
n
d
e
e
p
l
e
a
r
n
i
n
g
f
o
r
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
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
J
o
u
r
n
a
l
o
f
Al
-
Q
a
d
i
si
y
a
h
f
o
r
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
Ma
t
h
e
m
a
t
i
c
s
,
v
o
l
.
1
6
,
n
o
.
1
,
p
p
.
1
3
7
–
1
5
1
,
M
a
r
.
2
0
2
4
,
d
o
i
:
1
0
.
2
9
3
0
4
/
j
q
c
sm.
2
0
2
4
.
1
6
.
1
1
4
5
3
.
[2
7
]
E.
V
a
f
a
e
i
a
n
d
M
.
H
o
ss
e
i
n
i
,
“
Tr
a
n
sf
o
r
mers
i
n
E
EG
a
n
a
l
y
si
s
:
a
r
e
v
i
e
w
o
f
a
r
c
h
i
t
e
c
t
u
r
e
s
a
n
d
a
p
p
l
i
c
a
t
i
o
n
s
i
n
m
o
t
o
r
i
ma
g
e
r
y
,
se
i
z
u
r
e
,
a
n
d
e
mo
t
i
o
n
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
S
e
n
s
o
rs
,
v
o
l
.
2
5
,
n
o
.
5
,
F
e
b
.
2
0
2
5
,
d
o
i
:
1
0
.
3
3
9
0
/
s2
5
0
5
1
2
9
3
.
[
28
]
S
.
Li
,
X
.
L
y
u
,
L
.
Z
h
a
o
,
Z
.
C
h
e
n
,
A
.
G
o
n
g
,
a
n
d
Y
.
F
u
,
“
I
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
e
mo
t
i
o
n
u
si
n
g
e
l
e
c
t
r
o
e
n
c
e
p
h
a
l
o
g
r
a
m
b
y
t
u
n
a
b
l
e
Q
-
f
a
c
t
o
r
w
a
v
e
l
e
t
t
r
a
n
sf
o
r
m
a
n
d
b
i
n
a
r
y
g
r
a
y
w
o
l
f
o
p
t
i
mi
z
a
t
i
o
n
,
”
Fr
o
n
t
i
e
rs
i
n
C
o
m
p
u
t
a
t
i
o
n
a
l
N
e
u
r
o
sc
i
e
n
c
e
,
v
o
l
.
1
5
,
S
e
p
.
2
0
2
1
,
d
o
i
:
1
0
.
3
3
8
9
/
f
n
c
o
m.
2
0
2
1
.
7
3
2
7
6
3
.
[
29
]
O
.
S
a
h
a
,
M
.
S
.
M
a
h
m
u
d
,
S
.
A
.
F
a
t
t
a
h
,
a
n
d
M
.
S
a
q
u
i
b
,
“
A
u
t
o
m
a
t
i
c
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
f
r
o
m
m
u
l
t
i
-
b
a
n
d
E
E
G
d
a
t
a
b
a
s
e
d
o
n
a
d
e
e
p
l
e
a
r
n
i
n
g
s
c
h
e
m
e
w
i
t
h
e
f
f
e
c
t
i
v
e
c
h
a
n
n
e
l
a
t
t
e
n
t
i
o
n
,
”
I
E
E
E
A
c
c
e
s
s
,
v
o
l
.
1
1
,
p
p
.
2
3
4
2
–
2
3
5
0
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
E
S
S
.
2
0
2
2
.
3
2
2
4
7
2
5
.
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
E
mo
tio
n
r
ec
o
g
n
itio
n
o
f e
lectro
en
ce
p
h
a
lo
g
r
a
m
u
s
in
g
h
yb
r
id
co
n
vo
lu
tio
n
a
l
…
(
E
s
mera
ld
a
C
o
n
tess
a
Dja
ma
l
)
3943
[
30
]
M
.
A
.
M
a
r
i
a
,
M
.
A
.
H
.
A
k
h
a
n
d
,
A
.
B
.
M
.
A
.
H
o
ss
a
i
n
,
M
.
A
.
S
.
K
a
mal
,
a
n
d
K
.
Y
a
ma
d
a
,
“
A
c
o
m
p
a
r
a
t
i
v
e
s
t
u
d
y
o
n
p
r
o
mi
n
e
n
t
c
o
n
n
e
c
t
i
v
i
t
y
f
e
a
t
u
r
e
s
f
o
r
e
m
o
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
f
r
o
m
E
EG
,
”
I
EE
E
Ac
c
e
ss
,
v
o
l
.
1
1
,
p
p
.
3
7
8
0
9
–
3
7
8
3
1
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
3
.
3
2
6
4
8
4
5
.
[
31
]
M
.
A
.
O
z
d
e
m
i
r
,
M
.
D
e
g
i
r
m
e
n
c
i
,
E
.
I
z
c
i
,
a
n
d
A
.
A
k
a
n
,
“
EEG
-
b
a
s
e
d
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
w
i
t
h
d
e
e
p
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
u
r
a
l
n
e
t
w
o
r
k
s,”
Bi
o
m
e
d
i
c
a
l
En
g
i
n
e
e
ri
n
g
/
Bi
o
m
e
d
i
zi
n
i
sc
h
e
T
e
c
h
n
i
k
,
v
o
l
.
6
6
,
n
o
.
1
,
p
p
.
4
3
–
5
7
,
F
e
b
.
2
0
2
1
,
d
o
i
:
1
0
.
1
5
1
5
/
b
mt
-
2
0
1
9
-
0
3
0
6
.
[
32
]
T.
D
h
a
r
a
a
n
d
P
.
K
.
S
i
n
g
h
,
“
Emo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
f
r
o
m
EEG
d
a
t
a
u
si
n
g
h
y
b
r
i
d
d
e
e
p
l
e
a
r
n
i
n
g
a
p
p
r
o
a
c
h
,
”
i
n
Fr
o
n
t
i
e
rs
o
f
I
C
T
i
n
H
e
a
l
t
h
c
a
re
,
S
i
n
g
a
p
o
r
e
:
S
p
r
i
n
g
e
r
,
2
0
2
3
,
p
p
.
1
7
9
–
1
8
9
,
d
o
i
:
1
0
.
1
0
0
7
/
9
7
8
-
981
-
19
-
5
1
9
1
-
6
_
1
5
.
[3
3
]
A
.
P
r
a
k
a
sh
a
n
d
A
.
P
o
u
l
o
s
e
,
“
El
e
c
t
r
o
e
n
c
e
p
h
a
l
o
g
r
a
m
-
b
a
se
d
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
:
a
c
o
m
p
a
r
a
t
i
v
e
a
n
a
l
y
s
i
s
o
f
su
p
e
r
v
i
se
d
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
a
l
g
o
r
i
t
h
ms,”
D
a
t
a
S
c
i
e
n
c
e
a
n
d
M
a
n
a
g
e
m
e
n
t
,
v
o
l
.
8
,
n
o
.
3
,
p
p
.
3
4
2
–
3
6
0
,
S
e
p
.
2
0
2
5
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
d
sm.
2
0
2
4
.
1
2
.
0
0
4
.
[3
4
]
C
.
D
e
sa
i
,
“
C
o
mp
a
r
a
t
i
v
e
a
n
a
l
y
s
i
s
o
f
o
p
t
i
m
i
z
e
r
s
i
n
d
e
e
p
n
e
u
r
a
l
n
e
t
w
o
r
k
s,”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
I
n
n
o
v
a
t
i
v
e
S
c
i
e
n
c
e
a
n
d
Re
se
a
rc
h
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
5
,
n
o
.
1
0
,
p
p
.
9
5
9
–
9
6
2
,
2
0
2
0
.
[3
5
]
Z.
F
a
n
,
F
.
C
h
e
n
,
X
.
X
i
a
,
a
n
d
Y
.
Li
u
,
“
EEG
e
mo
t
i
o
n
c
l
a
ssi
f
i
c
a
t
i
o
n
b
a
se
d
o
n
g
r
a
p
h
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
,
”
Ap
p
l
i
e
d
S
c
i
e
n
c
e
s
,
v
o
l
.
1
4
,
n
o
.
2
,
Ja
n
.
2
0
2
4
,
d
o
i
:
1
0
.
3
3
9
0
/
a
p
p
1
4
0
2
0
7
2
6
.
[
36
]
W.
-
L.
Z
h
e
n
g
,
J.
-
Y
.
Zh
u
,
a
n
d
B
.
-
L.
Lu
,
“
I
d
e
n
t
i
f
y
i
n
g
st
a
b
l
e
p
a
t
t
e
r
n
s
o
v
e
r
t
i
m
e
f
o
r
e
mo
t
i
o
n
r
e
c
o
g
n
i
t
i
o
n
f
r
o
m
EEG
,
”
I
E
E
E
T
ra
n
s
a
c
t
i
o
n
s
o
n
A
f
f
e
c
t
i
v
e
C
o
m
p
u
t
i
n
g
,
v
o
l
.
1
0
,
n
o
.
3
,
p
p
.
4
1
7
–
4
2
9
,
J
u
l
.
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
TA
F
F
C
.
2
0
1
7
.
2
7
1
2
1
4
3
.
B
I
O
G
RAP
H
I
E
S O
F
AUTH
O
RS
Esm
e
r
a
ld
a
Co
n
te
ss
a
Dja
m
a
l
re
c
e
iv
e
d
a
b
a
c
h
e
lo
r'
s
d
e
g
r
e
e
in
E
n
g
i
n
e
e
rin
g
P
h
y
sic
s
fro
m
I
n
stit
u
t
Tek
n
o
l
o
g
i
Ba
n
d
u
n
g
i
n
1
9
9
4
,
a
m
a
ste
r'
s
d
e
g
re
e
in
In
stru
m
e
n
t
a
n
d
C
o
n
tr
o
l
fro
m
th
e
sa
m
e
u
n
i
v
e
rsity
i
n
1
9
9
8
.
F
ro
m
h
e
r
d
isse
rtatio
n
u
n
ti
l
n
o
w,
h
e
r
re
se
a
rc
h
in
EE
G
sig
n
a
l
p
ro
c
e
ss
in
g
h
a
s
fin
is
h
e
d
h
e
r
d
o
c
t
o
ra
l
p
r
o
g
ra
m
a
t
In
stit
u
t
Tek
n
o
l
o
g
i
Ba
n
d
u
n
g
in
2
0
0
5
.
S
h
e
is
a
n
a
ss
o
c
iate
p
ro
fe
ss
o
r
in
th
e
De
p
a
rt
m
e
n
t
o
f
In
fo
rm
a
ti
c
s,
Un
i
v
e
rsitas
Je
n
d
e
ra
l
Ac
h
m
a
d
Ya
n
i.
F
o
r
m
o
re
th
a
n
2
0
y
e
a
rs,
sh
e
h
a
s
re
se
a
rc
h
e
d
EE
G
sig
n
a
l
p
ro
c
e
ss
in
g
a
n
d
m
a
c
h
in
e
lea
rn
in
g
.
S
h
e
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
e
sm
e
ra
ld
a
.
c
o
n
tes
sa
@le
c
tu
re
.
u
n
jan
i.
a
c
.
id
.
Re
v
a
n
Vi
o
E
n
d
r
ia
n
sy
a
h
is
a
n
u
n
d
e
rg
ra
d
u
a
te
stu
d
e
n
t
in
I
n
fo
rm
a
ti
c
s
fro
m
Un
iv
e
rsitas
Je
n
d
e
ra
l
Ac
h
m
a
d
Ya
n
i.
His
th
e
sis
fo
c
u
se
d
o
n
e
m
o
ti
o
n
id
e
n
ti
fica
ti
o
n
fro
m
EE
G
sig
n
a
ls
u
si
n
g
a
tran
sf
o
rm
e
r.
He
is
n
o
w
a
m
a
c
h
in
e
lea
rn
in
g
re
se
a
rc
h
e
r
a
n
d
sy
ste
m
d
e
v
e
lo
p
e
r.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
re
v
a
n
v
io
2
2
@if
.
u
n
jan
i.
a
c
.
i
d
.
Da
sw
a
r
a
Dja
j
a
sa
sm
ita
is
a
n
e
u
ro
l
o
g
ist
a
n
d
lec
tu
re
r
a
t
t
h
e
M
e
d
ica
l
Ed
u
c
a
ti
o
n
P
ro
g
ra
m
,
F
a
c
u
lt
y
o
f
M
e
d
icin
e
a
t
Un
iv
e
rsitas
Je
n
d
e
ra
l
Ac
h
m
a
d
Ya
n
i
.
His
re
se
a
rc
h
in
tere
sts
a
re
stro
k
e
a
n
d
p
o
st
-
str
o
k
e
m
a
n
a
g
e
m
e
n
t,
in
c
l
u
d
i
n
g
si
g
n
a
l
p
ro
c
e
ss
in
g
.
He
c
a
n
b
e
c
o
n
tac
ted
a
t
e
m
a
il
:
d
a
sw
a
ra
d
j@g
m
a
il
.
c
o
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.