T
E
L
K
O
MN
I
K
A
T
el
eco
m
m
u
n
i
ca
t
i
o
n
,
C
o
m
p
u
t
i
n
g
,
E
l
ect
ro
n
i
cs
a
n
d
C
o
n
t
ro
l
Vo
l
.
1
9
, N
o
.
5
,
O
ct
o
b
er
202
1
, p
p
.
1588
~
1593
I
S
S
N
:
1693
-
6
9
3
0
,
accr
ed
i
t
ed
F
i
r
s
t
G
r
ad
e
by K
e
m
e
nr
i
s
t
e
kdi
kt
i
,
D
e
c
r
e
e
N
o:
21/
E
/
K
P
T
/
2018
D
O
I
:
10.
12928/
T
E
L
K
O
M
N
I
K
A
.
v1
9
i
5
.
19850
1588
Jou
r
n
al
h
om
e
page
:
ht
t
p:
/
/
j
our
nal
.
uad
.
ac
.
i
d/
i
nde
x
.
php/
T
E
L
K
O
M
N
I
K
A
C
l
a
s
s
i
f
i
ca
t
i
o
n o
f
E
E
G
s
i
g
na
l
s
f
o
r f
a
ci
a
l
ex
p
res
s
i
o
n a
nd
m
ot
or
ex
ec
ut
i
o
n w
i
t
h
d
e
ep l
ea
r
ni
ng
A
reej
H
a
m
eed
A
l
-
A
nb
a
r
y
1
,
Sa
l
i
h A
l
-
Q
ar
aaw
i
2
1
De
pa
r
tm
e
n
t of
c
om
m
un
ic
a
t
io
n,
C
ol
le
ge
of
I
nf
or
m
a
ti
on
En
gi
ne
e
r
i
ng,
A
l
-
Na
hr
a
i
n
Un
ive
r
si
ty,
B
a
g
hda
d
,
Iraq
2
De
pa
r
tm
e
n
t of
M
e
d
i
c
a
l
I
n
str
um
e
n
t
s En
gi
ne
e
r
i
ng Te
c
h
n
iqu
e
s
,
Al
-
H
us
sa
i
n
Un
ive
r
si
ty
C
ol
le
ge
,
Ka
r
ba
la
,
Iraq
A
rt
i
cl
e I
n
f
o
AB
S
T
RACT
A
r
tic
le
h
is
to
r
y
:
R
ecei
v
ed
J
a
n 29,
2021
R
ev
i
s
ed
J
un 16,
2021
A
ccep
t
ed
J
un 28,
2021
R
e
c
e
nt
ly,
a
lg
or
i
thm
s of
m
a
c
hine
le
a
r
ni
ng a
r
e
wi
de
l
y
u
se
d wi
th t
he
f
ie
l
d
of
e
le
c
tr
oe
nc
e
pha
lo
gr
a
p
hy (
E
EG
)
br
a
in
-
c
o
mp
u
t
er i
n
t
e
rfaces
(BCI).
T
h
e
pr
e
pr
oc
e
s
si
ng s
ta
ge
f
or
t
he
EEG
s
ig
na
l
s i
s pe
r
f
or
m
e
d by a
pp
ly
in
g th
e
pr
i
nc
i
ple
c
om
p
one
nt
a
na
ly
si
s (
P
C
A)
a
lg
or
it
hm
t
o e
xtr
a
c
t t
he
im
p
or
ta
n
t
f
e
a
tur
e
s a
nd
r
e
duc
in
g t
he
da
ta
r
e
du
nda
nc
y.
A m
ode
l
f
or
c
la
s
sif
yi
ng
EEG
,
tim
e
se
r
ie
s,
si
gna
ls f
or
f
a
c
ia
l
e
xpr
e
s
si
on a
nd
som
e
m
o
to
r
e
xe
c
u
ti
on pr
oc
e
sse
s
ha
d be
e
n de
s
ig
ne
d.
A
ne
ur
a
l
ne
t
wor
k of
t
hr
e
e
h
id
d
e
n la
ye
r
s
wi
th
de
e
p
le
a
r
n
in
g
c
la
ss
if
ie
r
ha
d be
e
n us
e
d i
n
thi
s
w
or
k.
Da
ta
of
f
our
dif
f
e
r
e
nt su
bje
c
ts
we
r
e
c
ol
le
c
te
d by u
si
ng a
14
c
ha
nne
ls Em
ot
iv
EP
OC
+
de
vic
e
.
EEG
da
ta
se
t
sa
m
p
le
s
in
c
l
ud
in
g
te
n
a
c
t
io
n
c
la
s
se
s
f
or
the
f
a
c
ia
l
e
x
pr
e
s
sio
n
a
nd
s
om
e
m
ot
or
e
xe
c
u
ti
on m
ove
m
e
nt
s a
r
e
r
e
c
or
de
d.
A c
la
ss
if
ic
a
t
io
n r
e
s
u
lts w
it
h a
c
c
ur
a
c
y
r
a
nge
(
91.
25
-
95.
7
5%
)
f
or
the
c
ol
le
c
te
d sa
m
p
le
s we
r
e
obt
a
i
ne
d wi
th r
e
s
pe
c
t
to: num
be
r
of
sa
m
p
le
s f
or
e
a
c
h c
la
ss,
t
ota
l num
be
r
of
EEG
da
ta
se
t sa
m
ple
s
a
nd ty
pe
of
a
c
ti
va
t
io
n
f
u
nc
t
io
n
w
it
hi
n the
hi
dde
n
a
nd the
ou
tp
ut la
ye
r
ne
ur
on
s.
A ti
m
e
se
r
ie
s
EEG
s
ig
na
l
wa
s
ta
ke
n a
s s
ig
na
l v
a
lue
s n
ot a
s
im
a
ge
or
his
to
gr
a
m
,
a
na
ly
se
d a
nd c
la
s
sif
ie
d
wi
th
de
e
p
le
a
r
n
in
g t
o ob
ta
i
n t
he
sa
ti
sf
ie
d
r
e
su
lt
s of
a
c
c
ur
a
c
y.
Ke
y
wo
r
d
s
:
BCI
D
eep
l
ear
n
i
n
g
EEG
N
u
er
al
n
et
w
o
r
k
P
CA
T
his
is
a
n
o
pe
n
ac
c
e
s
s
ar
tic
le
u
nde
r
the
CC
B
Y
-
SA
lic
e
n
se
.
C
or
r
e
s
pon
di
n
g A
u
t
h
or
:
S
a
lih
A
l
-
Q
ar
aaw
i
D
e
pa
r
t
m
e
nt
of
M
e
di
c
a
l
I
ns
t
r
um
e
nt
s
E
ngi
ne
e
r
i
ng
T
e
c
hni
que
s
Al
-
H
us
s
a
i
n U
ni
ve
r
s
i
t
y
C
ol
l
e
ge
K
ar
b
al
a
,
Ira
q
E
ma
il:
d
r
s
a
lih
a
lq
a
r
a
a
w
i@
g
ma
il.
c
o
m; p
r
o
f
.
d
r
.
s
a
lih
@
h
u
c
ir
a
q
i.
e
d
u
.
iq
1.
I
NT
RO
DUC
T
I
O
N
I
t
i
s
w
e
l
l
know
n t
ha
t
,
t
he
s
ys
t
e
m
w
hi
c
h c
onne
c
t
s
hum
a
n br
a
i
n
s
i
gna
l
s
w
i
t
h a
ppl
i
a
nc
e
s
or
de
vi
c
e
s
w
i
t
hout
r
e
qui
r
i
ng
of
a
ny phys
i
c
a
l
c
ont
a
c
t
i
s
c
a
l
l
e
d
br
a
i
n
-
co
m
p
u
t
er
i
n
t
er
f
aces
(
B
C
I
)
.
I
t
h
as
b
een
s
een
as
a n
ew
w
a
y f
or
c
om
m
uni
c
a
t
i
on,
w
he
r
e
t
he
b
r
a
in
a
c
tiv
ity
h
as
b
een
u
s
ed
as
a
r
efl
ect
ed
f
o
r
m
b
y
el
ect
r
i
c
b
r
ai
n
s
i
g
n
al
s
t
o
m
an
ag
e ex
t
er
n
al
s
y
s
t
em
s
u
ch
as
c
om
put
e
r
s
,
w
he
e
l
c
ha
i
r
s
,
s
w
i
t
c
he
s
,
or
ne
ur
o
pr
os
t
he
t
i
c
e
xt
e
ns
i
ons
[
1
]
-
[
6]
.
E
l
ect
r
o
en
cep
h
al
o
g
r
ap
h
y
(
E
E
G
)
i
s
t
h
e p
r
o
ces
s
o
f
f
et
ch
i
n
g
t
h
e el
ect
r
i
cal
b
r
ai
n
’
s
s
i
g
n
al
s
a
nd r
e
c
or
di
ng
t
he
m
,
s
o
t
he
a
c
t
i
vi
t
y of
hum
a
n c
a
n be
a
na
l
yz
e
d m
a
ki
ng t
he
r
e
a
l
pr
oc
e
s
s
i
ng of
t
he
br
a
i
n c
l
e
a
r
t
o
t
he
us
e
r
.
E
l
ect
r
o
d
es
ar
e p
u
t
o
n
t
h
e
h
u
m
an
s
cal
p
,
i
n
an
eas
y
w
ay
,
t
o
co
l
l
ect
b
r
ai
n
’
s
el
ect
r
i
cal
s
i
g
n
al
s
.
A
n
E
E
G
s
i
g
n
al
i
s
b
a
n
d
limite
d
i
n f
r
e
que
nc
y (
0.
1
-
60 H
z
)
,
E
E
G
s
i
gna
l
s
a
r
e
m
ode
l
e
d
a
nd c
l
a
s
s
i
f
i
e
d i
nt
o
f
i
ve
t
ype
s
:
(
t
h
et
a,
d
el
t
a
,
b
et
a,
al
p
h
a,
an
d
g
am
m
a
w
av
es
)
,
w
h
i
ch
ar
e
r
es
p
o
n
s
i
b
l
e t
o
cap
t
u
r
e d
i
f
f
er
en
t
as
s
o
ci
at
ed
b
r
ai
n
act
i
v
i
t
i
e
s
i
n
s
i
d
e
t
h
e b
r
ai
n
[7
]
,
[
8]
.
E
E
G
s
i
gna
l
s
c
ont
a
i
n a
hi
gh
r
e
du
nda
nc
y i
n t
he
c
ol
l
e
c
t
e
d da
t
a
,
s
o t
he
i
m
por
t
a
nt
s
t
a
g
e
be
f
or
e
be
i
ng
c
la
s
s
if
y
in
g
th
o
s
e
s
ig
n
a
ls
,
is
f
e
a
tu
r
e
e
x
tr
a
c
ti
o
n
s
ta
g
e
.
I
n
f
a
c
t,
a
f
e
a
tu
r
e
i
llu
s
tr
a
te
s
a
d
is
tin
c
tiv
e
a
ttr
ib
u
te
,
i
d
en
t
i
f
i
ab
l
e m
eas
u
r
e,
an
d
f
u
n
ct
i
o
n
al
el
em
en
t
g
et
t
i
n
g
f
r
o
m
a s
eg
m
en
t
o
f
s
am
p
l
es
.
F
eat
u
r
e ex
t
r
act
i
o
n
u
s
ed
t
o
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
el
eco
m
m
u
n
C
om
put
E
l
C
ont
r
o
l
C
l
as
s
i
f
i
c
at
i
on of
E
E
G
s
i
gnal
s
f
or
f
ac
i
al
e
x
pr
e
s
s
i
on
and
m
ot
or
execu
t
i
o
n
w
i
t
h
….
(
A
r
eej
H
a
m
eed
A
l
-
A
nbar
y
)
1589
ma
in
ta
in
th
e
s
ig
n
if
ic
a
n
t in
f
o
r
ma
tio
n
in
th
e
s
ig
n
a
l a
n
d
min
imiz
in
g
th
e
ir
lo
s
t a
s
mu
c
h
a
s
p
o
s
s
ib
le
,
a
s
w
e
ll a
s
to
s
i
m
pl
i
f
y t
he
ne
e
de
d r
e
s
our
c
e
s
f
or
de
s
c
r
i
bi
ng t
he
hu
ge
a
m
ount
of
da
t
a
i
n a
n
a
c
c
u
r
a
te
ma
n
n
e
r
.
S
o
,
th
is
w
ill
l
ead
to
a
s
imp
le
imp
le
me
n
ta
tio
n
th
a
t
r
e
d
u
c
e
s
th
e
p
r
o
c
e
s
s
in
g
c
o
s
t f
o
r
th
e
in
f
o
r
ma
tio
n
,
a
n
d
e
limin
a
te
s
th
e
ne
e
d f
or
da
t
a
c
om
pr
e
s
s
i
on [
9
]
-
[
14]
.
I
n t
h
i
s
w
or
k,
pr
i
n
c
i
pl
e
c
om
pone
nt
a
na
l
ys
i
s
(
P
C
A
)
m
e
t
hod i
s
u
s
e
d f
or
uns
upe
r
vi
s
e
d f
eat
u
r
e ex
t
r
act
i
o
n
p
r
o
ces
s
.
T
h
i
s
m
et
h
o
d
i
s
a d
es
cr
i
p
t
i
v
e s
t
at
i
s
t
i
cal
t
ech
n
i
q
u
e w
h
i
ch
d
es
c
r
i
b
es
t
h
e
d
i
f
f
er
en
ces
b
et
w
een
t
h
e s
am
p
l
es
o
f
t
h
e d
at
as
et
a
n
d
t
h
e m
o
s
t
co
r
r
el
at
ed
s
am
p
l
es
.
P
C
A
d
et
ect
s
t
h
e
pr
i
nc
i
pl
e
c
om
pone
nt
of
da
t
a
s
e
t
of
t
he
s
i
gna
l
,
s
o i
t
w
i
l
l
pe
r
f
or
m
t
he
di
m
e
ns
i
on
r
e
duc
t
i
on of
t
he
da
t
a
[
15]
.
A
lg
o
r
ith
ms
f
o
r
c
la
s
s
if
y
in
g
E
E
G
-
b
as
ed
B
C
I
s
w
er
e cl
as
s
i
f
i
ed
i
n
t
o
f
o
u
r
m
ai
n
cl
as
s
es
:
m
at
r
i
x
a
nd t
e
ns
or
,
ad
ap
t
i
v
e,
d
eep
l
ear
n
i
n
g
,
an
d
t
r
an
s
f
er
l
ear
n
i
n
g
cl
as
s
i
f
i
er
s
as
w
el
l
as
a
f
ew
o
t
h
er
d
i
v
er
s
e
cl
as
s
i
f
i
er
s
[2
]
,
[
1
2]
,
[
16
]
-
[
2
0
]
.
I
n
E
E
G
r
es
ear
ch
es
,
m
ach
i
n
e l
ear
n
i
n
g
h
ad
b
een
u
s
ed
t
o
d
i
s
co
v
er
t
h
e
r
el
at
ed
in
f
o
r
ma
tio
n
f
o
r
ne
ur
oi
m
a
gi
ngý a
nd
ne
ur
a
l
c
l
a
s
s
i
f
i
c
a
t
i
on.
T
he
a
dva
nc
e
s
i
n
m
a
c
hi
ne
l
e
a
r
ni
ng
a
nd t
he
a
va
i
l
a
bi
l
i
t
y of
huge
E
E
G
da
t
a
s
e
t
s
l
e
d t
o de
e
p l
e
a
r
ni
ng de
pl
oym
e
nt
i
n a
na
l
yz
i
ng E
E
G
s
i
gna
l
s
a
nd i
n t
he
f
i
e
l
d
of
unde
r
s
t
a
ndi
ng
br
a
i
n f
unc
t
i
ona
l
i
t
y
by
de
f
i
ni
ng
c
ol
l
e
c
t
e
d i
nf
or
m
a
t
i
on
i
ns
i
de
i
t
[
6
]
,
[
21
]
-
[
24]
.
T
h
e u
s
e o
f
de
e
p l
e
a
r
ni
ng w
i
t
h
E
E
G
a
ppl
i
c
a
t
i
ons
i
n ge
ne
r
a
,
l
f
e
l
l
i
nt
o
f
i
ve
g
r
oups
:
m
ot
or
i
m
a
ge
r
y
,
e
m
ot
i
on
r
e
c
o
gni
t
i
on,
m
e
nt
a
l
t
a
s
k w
or
kl
oa
d,
s
e
i
z
ur
e
obs
e
r
va
t
i
on,
ev
en
t
r
el
at
ed
p
o
t
en
t
i
al
(E
R
P
)
t
as
k
s
d
et
ect
i
o
n
,
an
d
s
l
eep
s
t
at
es
r
e
c
or
di
ng
[
2
5]
.
2.
R
ES
EA
R
C
H
M
ETH
O
D
T
he
w
or
k i
n
t
hi
s
pa
pe
r
f
oc
us
e
s
on E
E
G
s
i
gna
l
f
e
a
t
ur
e
s
t
o i
de
nt
i
f
y
t
he
E
E
G
s
i
gna
l
s
f
or
f
a
c
i
a
l
e
x
p
r
e
ssi
o
n
s (
F
E
s)
a
nd s
om
e
m
ot
or
e
xe
c
ut
i
on a
c
t
i
o
ns
.
F
E
s
i
nc
l
ude
:
s
ur
pr
i
s
e
,
s
m
i
l
e
,
l
e
f
t
w
i
nk,
r
i
ght
w
i
nk,
a
nd
m
out
h ope
ne
d.
W
hi
l
e
,
m
ot
or
e
xe
c
ut
i
on a
c
t
i
ons
i
n
c
l
ude
: r
ig
h
t
h
a
n
d
li
f
tin
g
,
le
f
t h
a
n
d
lif
t
in
g
,
r
i
ght
r
o
t
a
t
i
ng of
he
a
d,
l
e
f
t
r
ot
a
t
i
ng o
f
he
a
d ,
a
n
d
c
la
p
p
in
g
.
A
ll th
e
s
e
s
ig
n
a
l
s
f
ir
s
t
co
l
l
ect
ed
b
y
E
mo
tiv
E
P
O
C
+
14
ch
an
n
el
m
o
b
i
l
e b
r
ai
n
w
ear
h
ead
s
et
,
an
d
f
et
ch
ed
b
y
t
h
e
l
i
cen
s
ed
s
o
f
t
w
ar
e
of
E
m
ot
i
v
P
r
o
w
i
t
h
pyt
hon
e
nvi
r
on
m
e
nt
.
A
m
ode
l
f
or
c
l
a
s
s
i
f
yi
ng t
hos
e
s
i
gna
l
s
ha
d be
e
n
de
s
i
gne
d.
F
i
gur
e
1 s
how
s
t
he
r
e
s
e
a
r
c
h m
e
t
hodol
o
gy bl
oc
k
di
a
gr
a
m
.
T
he
de
t
a
i
l
o
f
e
a
c
h s
t
e
p w
i
l
l
be
e
xpl
a
i
ne
d
i
n t
he
ne
xt
s
ubs
e
c
t
i
ons
.
F
i
gur
e
1
.
R
es
ear
ch
m
e
t
hodol
ogy
b
l
oc
k
d
i
ag
r
am
2.
1.
D
a
t
a
co
l
l
ect
i
o
n
T
he
f
i
r
s
t
s
t
a
ge
of
r
e
s
e
a
r
c
h m
e
t
hodol
ogy
be
gi
ns
w
i
t
h c
ol
l
e
c
t
i
ng
da
t
a
s
e
t
s
a
m
pl
e
s
by us
i
ng
E
m
o
t
i
v
E
p
o
c+ h
ead
s
e
t
de
vi
c
e
w
i
t
h
14
c
ha
nne
l
s
e
xt
e
nde
d
a
r
ound t
he
he
a
d.
T
he
da
t
a
w
a
s
c
ol
l
e
c
t
e
d
f
r
om
f
our
s
ubj
e
c
t
s
w
ith
d
i
f
f
er
en
t
ag
es
(
1
0
-
50 y
ear
s
)
,
m
al
es
an
d
f
em
al
es
w
h
i
l
e t
h
ey
d
o
i
n
g
t
h
e r
eq
u
i
r
ed
f
aci
al
ex
p
r
es
s
i
o
n
s
an
d
t
h
e
m
ot
or
e
xe
c
ut
i
on a
c
t
i
ons
.
T
he
E
E
G
s
i
gna
l
s
w
e
r
e
r
e
c
or
de
d by t
he
m
ont
hl
y
l
i
c
e
ns
e
d E
m
ot
i
v s
of
t
w
a
r
e
(
E
m
ot
i
v
Pr
o
)
an
d
s
av
ed
as
ex
cel
f
i
l
es
(
.
cs
v
f
i
l
es
)
t
o
b
e
u
s
ed
l
at
er
i
n
t
r
ai
n
i
ng
t
he
ne
ur
a
l
ne
t
w
or
k
w
i
t
hi
n
pyt
hon
e
nvi
r
onm
e
nt
.
dur
i
ng
t
he
r
e
c
or
di
ng p
r
oc
e
s
s
a
bout
6487
E
E
G
s
am
p
l
es
w
er
e co
l
l
ect
ed
.
T
ab
l
e
1
s
h
o
w
s
s
o
m
e
s
a
m
pl
e
s
of
t
he
c
ol
l
e
c
t
e
d E
E
G
da
t
a
f
or
l
i
f
t
i
ng
l
e
f
t
h
a
nd f
or
one
s
ubj
e
c
t
.
2.
2.
D
a
t
a p
r
e
-
p
ro
ces
s
i
n
g
T
h
is
s
ta
g
e
is
th
e
a
r
tif
a
c
ts
r
e
mo
v
a
l
o
f
E
E
G
s
ig
n
a
ls
,
w
h
ic
h
is
d
o
in
g
b
y
th
e
E
mo
tiv
h
e
a
d
s
e
t its
e
lf
,
w
h
er
e t
h
e d
at
a
i
s
r
e
c
or
de
d di
r
e
c
t
l
y a
s
i
t
i
s
r
e
c
e
i
ve
d f
r
om
t
he
he
a
ds
e
t
.
T
he
r
e
i
s
a
good a
m
ount
of
s
i
gna
l
pr
oc
e
s
s
i
ng a
nd
f
ilte
r
in
g
in
t
h
e h
ead
s
et
t
o
r
em
o
v
e
ar
t
i
f
act
s
an
d
h
ar
m
o
n
i
c
f
r
eq
u
en
ci
es
.
S
o
,
t
h
e
s
i
g
n
al
s
ap
p
ear
cl
ean
w
h
en
w
e
ga
i
ne
d a
good c
ont
a
c
t
qua
l
i
t
y.
T
he
s
i
gna
l
s
ha
d be
e
n s
a
m
pl
e
d a
t
2048 H
z
s
a
m
pl
i
ng f
r
e
que
nc
y,
a
nd t
he
n
a
ppl
i
e
d t
o a
dua
l
not
c
h
f
i
l
t
e
r
a
t
50
H
z
a
nd 60 H
z
a
s
w
e
l
l
a
s
a l
o
w
p
as
s
f
i
l
t
er
at
6
4
H
z cu
t
o
f
f
f
r
e
que
nc
y.
F
i
na
l
l
y,
t
he
da
t
a
w
a
s
f
i
l
t
e
r
e
d dow
n t
o
1
28 or
256
H
z
.
2.
3.
F
ea
t
u
re ex
t
ra
ct
i
o
n
I
n t
hi
s
s
t
a
ge
,
t
he
obt
a
i
ne
d p
r
e
pr
oc
e
s
s
e
d da
t
a
f
r
om
E
mo
tiv
he
a
ds
e
t
i
s
pr
oc
e
s
s
e
d w
i
t
h P
C
A
a
l
gor
i
t
hm
t
o
imp
r
o
v
e
th
e
c
la
s
s
if
ie
r
'
s
a
c
c
ur
a
c
y.
P
C
A
i
s
a
t
e
c
h
ni
que
us
e
d f
or
r
e
duc
t
i
on of
di
m
e
ns
i
ona
l
i
t
y of
t
he
l
a
r
ge
da
t
a
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1693
-
6930
T
E
L
KOM
NI
KA
T
el
eco
m
m
u
n
C
om
put
E
l
C
ont
r
o
l
,
Vo
l
.
1
9
, N
o
.
5
,
O
ct
o
b
er
2021
:
1588
-
1593
1590
s
e
t
s
.
T
hi
s
c
a
n be
a
c
hi
e
ve
d by c
onve
r
t
i
ng t
he
huge
s
e
t
of
va
r
i
a
bl
e
s
i
nt
o a
s
m
a
l
l
e
r
one
w
hi
c
h c
ont
a
i
ns
m
os
t
of
t
he
i
nf
or
m
a
t
i
on i
n
t
he
l
a
r
ge
s
e
t
[
15]
,
[
26]
.
W
e
ha
ve
6487
sa
m
pl
e
s
f
r
om
e
a
c
h one
of
t
he
14 c
ha
nne
l
s
of
t
he
he
a
ds
e
t
.
T
o
i
m
pl
e
m
e
nt
P
C
A
,
t
he
m
e
a
n va
l
ue
s
m
us
t
be
c
om
put
e
d
f
i
r
s
t
l
y
,
s
o
t
ha
t
w
e
c
a
n c
om
put
e
t
he
s
ta
n
d
a
r
d
iz
a
tio
n
(
Z
)
o
f
th
e
in
itia
l
v
a
lu
e
s
o
f
th
e
d
a
ta
s
e
t,
a
s
in
(
1
)
,
to
t
r
a
n
s
f
o
r
m a
ll
th
e
v
a
r
ia
b
le
s
to
t
h
e
s
a
me
r
an
g
e
[
26]
.
=
V
al
u
e
−
S
t
an
d
ar
d
d
e
v
ia
t
io
n
(1
)
T
ab
l
e
1
.
EE
G
d
at
as
et
s
s
a
mp
le
s
e
x
a
mp
le
f
o
r
le
f
t
h
a
n
d
lif
tin
g
f
r
o
m
E
mo
tiv
h
e
a
d
s
e
t
E
E
G.
A
F3
E
E
G.
F7
E
E
G.
F3
E
E
G.
F
C5
E
E
G.
T7
E
E
G.
P7
E
E
G.
O1
E
E
G.
O2
E
E
G.
P8
E
E
G.
T8
E
E
G.
F
C6
E
E
G.
F4
E
E
G.
F8
E
E
G.
A
F4
4627.3
08
6175.
769
7376.
923
7018.4
61
7112.
82
3223.
461
3223.
846
1274.
103
3862.
051
629.7
436
8023.9
74
7521.
795
8334.
743
3344.3
59
7502.4
36
5615.
513
6324.
231
6095.5
13
1287.
821
7299.
872
7261.
41
6439.
615
1462.
692
7645.
513
3923.9
74
3256.
539
107.6
923
6799.6
15
5662.1
8
5518.
461
4492.
308
3957.4
36
5409.
359
4291.
667
715.6
411
4300.
897
3987.
821
6205.
385
7224.1
03
7947.
18
3203.
974
4669.1
03
5591.5
66
5528.
655
4493.
29
4027.3
43
5410.
139
4219.
964
723.8
898
4265.
287
3929.
946
6159.
411
7237.6
89
7829.
861
3138.
312
4696.3
39
5520.9
52
5538.
849
4494.
271
4097.2
5
5410.
919
4148.
262
732.1
386
4229.
675
3872.
07
6113.
437
7251.2
76
7712.
542
3072.
649
4723.5
76
5450.3
38
5549.
042
4495.
254
4167.1
58
5411.
699
4076.
56
740.3
873
4194.
064
3814.
195
6067.
463
7264.8
64
7595.
224
3006.
986
4750.8
13
5379.7
25
5559.
236
4496.
236
4237.0
65
5412.
479
4004.
858
748.6
361
4158.
453
3756.
32
6021.
489
7278.4
51
7477.
905
2941.
323
4778.0
5
5309.1
11
5569.
43
4497.
218
4306.9
72
5413.
26
3933.
156
756.8
849
4122.
842
3698.
445
5975.
516
7292.0
38
7360.
586
2875.
66
4805.2
87
5238.4
97
5579.
624
4498.
2
4376.8
79
5414.
04
3861.
454
765.1
337
4087.
231
3640.
57
5929.
542
7305.6
25
7243.
268
2809.
997
4832.5
23
5167.8
83
5589.
817
4499.
182
4446.7
87
5414.
82
3789.
752
773.3
824
4051.
62
3582.
695
5883.
568
7319.2
12
7125.
949
2744.
334
4859.7
6
5097.2
7
5600.
011
4500.
164
4516.6
94
5415.
6
3718.
05
781.6
312
4016.
009
3524.
82
5837.
594
7332.7
99
7008.
631
2678.
672
4886.9
97
5026.6
56
5610.
205
4501.
146
4586.6
01
5416.
38
3646.
347
789.8
8
3980.
398
3466.
945
5791.
62
7346.3
86
6891.
312
2613.
009
4914.2
33
4956.0
42
5620.
398
4502.
127
4656.5
08
5417.
16
3574.
646
798.1
287
3944.
787
3409.
07
5745.
646
7359.9
73
6773.
994
2547.
346
4941.4
7
4885.4
28
5630.
592
4503.
11
4726.4
16
5417.
94
3502.
943
806.3
775
3909.
176
3351.
195
5699.
673
7373.5
6
6656.
675
2481.
683
4968.7
07
4814.8
14
5640.
786
4504.
092
4796.3
23
5418.
721
3431.
241
814.6
263
3873.
565
3293.
32
5653.
699
7387.1
47
6539.
356
2416.
02
4995.9
44
T
h
e
s
eco
n
d
s
t
e
p of
P
C
A
i
s
t
o c
om
put
e
t
he
c
ova
r
i
a
nc
e
m
a
t
r
i
x,
t
o
c
he
c
k i
f
t
he
r
e
i
s
a
ny r
e
l
a
t
i
ons
hi
p or
co
r
r
el
at
i
o
n
b
et
w
een
t
h
e v
ar
i
ab
l
es
o
f
t
h
e
d
at
as
et
t
o
r
ed
u
ce
t
h
e i
n
f
o
r
m
at
i
o
n
r
ed
u
n
d
an
cy
as
m
u
ch
as
p
o
s
s
i
b
l
e.
F
i
r
s
t
o
f
al
l
,
t
h
e co
v
ar
i
an
ce b
et
w
een
al
l
p
o
t
en
t
i
al
p
ai
r
s
o
f
th
e
in
itia
l d
a
ta
s
e
t v
a
r
ia
b
le
s
w
a
s
c
o
mp
u
te
d
u
s
in
g
(
2
)
,
in
o
r
d
e
r
to
in
s
tr
u
c
t
th
e
e
n
tr
ie
s
o
f
th
e
c
o
v
a
r
ia
n
c
e
ma
tr
ix
,
w
h
ic
h
is
a
p
×
p
s
y
mme
tr
ic
ma
tr
ix
.
[
,
]
=
∑
−
=
1
(
2)
w
h
er
e
;
m
ean
s
t
h
e m
ean
v
al
u
e o
f
v
ar
i
ab
l
e X
p i
s
t
he
di
m
e
ns
i
on’
s
num
be
r
T
he
t
hi
r
d s
t
e
p of
P
C
A
i
s
t
o c
om
put
e
t
he
e
i
ge
nve
c
t
or
s
a
nd e
i
ge
nva
l
ue
s
f
or
t
he
da
t
a
s
e
t
va
l
ue
s
,
i
n or
de
r
t
o l
oc
a
t
e
t
he
i
r
pr
i
nc
i
pa
l
c
om
pone
nt
s
.
T
he
pr
i
nc
i
pa
l
c
om
pone
nt
s
a
r
e
t
he
ne
w
unc
or
r
e
l
a
t
e
d
va
r
i
a
bl
e
s
a
nd ha
ve
t
he
m
os
t
of
i
n
f
or
m
a
t
i
on
a
bout
t
he
da
t
a
s
e
t
i
s
c
om
pr
e
s
s
e
d i
n t
he
f
i
r
s
t
c
om
pone
nt
s
a
nd
i
t
gr
a
dua
l
l
y
de
s
c
e
nds
.
T
he
f
our
t
h s
t
e
p i
s
t
o f
i
nd t
he
f
e
a
t
ur
e
ve
c
t
or
,
w
hi
c
h
i
s
r
e
pr
e
s
e
nt
e
d by m
a
t
r
i
x w
i
t
h c
ol
um
ns
of
e
i
ge
nve
c
t
or
s
f
or
t
he
r
e
qui
r
e
d c
om
pone
n
t
f
r
om
t
he
p
r
e
vi
ous
s
t
e
p.
T
hi
s
w
i
l
l
l
e
a
d t
o ke
e
p onl
y k c
om
pone
nt
s
(
e
i
ge
nve
c
t
or
s
)
i
ns
t
e
a
d of
t
he
t
ot
a
l
num
be
r
o
f
t
he
m
(
p)
.
T
he
f
i
na
l
s
t
e
p of
P
C
A
i
s
t
he
r
e
f
o
r
m
a
t
i
on o
f
t
he
or
i
gi
na
l
da
t
a
s
e
t
a
xi
s
t
o
t
he
a
xi
s
of
t
he
s
e
l
e
c
t
e
d pr
i
nc
i
pa
l
c
om
pone
nt
s
,
by
m
ul
t
i
pl
yi
n
g t
he
t
r
a
ns
pos
e
of
f
e
a
t
ur
e
ve
c
t
or
a
s
i
n
(3
):
F
i
n
al
d
at
as
et
= F
eat
u
r
eV
ect
o
r
T
* Z
T
(
3)
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
el
eco
m
m
u
n
C
om
put
E
l
C
ont
r
o
l
C
l
as
s
i
f
i
c
at
i
on of
E
E
G
s
i
gnal
s
f
or
f
ac
i
al
e
x
pr
e
s
s
i
on
and
m
ot
or
execu
t
i
o
n
w
i
t
h
….
(
A
r
eej
H
a
m
eed
A
l
-
A
nbar
y
)
1591
2.
4
.
C
l
as
s
i
f
i
c
at
i
on
m
od
e
l
d
e
ve
l
op
m
e
n
t
I
n t
hi
s
wo
r
k
,
a
ne
ur
a
l
ne
t
w
or
k
w
i
t
h
de
e
p l
e
a
r
ni
ng
w
a
s
bui
l
t
t
o
c
l
a
s
s
i
f
y t
he
E
E
G
s
i
gna
l
s
f
or
t
he
t
e
n
a
c
t
i
ons
i
nc
l
udi
ng f
a
c
i
a
l
e
xpr
e
s
s
i
on a
nd m
ot
or
e
xe
c
ut
i
on.
T
he
m
a
i
n f
a
c
i
l
i
t
y o
f
a
ppl
yi
ng de
e
p
l
e
a
r
ni
ng
m
ech
an
i
s
m
i
s
t
h
at
,
i
t
o
f
t
en
co
n
t
i
n
u
es
t
o
i
m
p
r
o
v
e as
t
h
e s
i
ze o
f
t
h
e d
at
as
et
i
n
cr
eas
es
.
T
h
i
s
t
as
k
w
as
imp
le
me
n
te
d
w
ith
s
p
id
e
r
3
.
3
.
1
\
P
y
th
o
n
e
n
v
ir
o
n
me
n
t b
y
imp
o
r
tin
g
K
e
r
a
s
lib
r
a
r
ie
s
,
w
h
i
ch
i
s
a
d
eep
l
ear
n
i
n
g
A
P
I
w
r
i
t
t
e
n i
n
P
yt
hon.
A
S
e
que
nt
i
a
l
m
ode
l
,
w
hi
c
h i
s
a
l
i
ne
a
r
s
t
a
c
k of
l
a
ye
r
s
,
w
i
t
h 3 h
i
dde
n l
a
ye
r
s
w
hi
c
h
c
ont
a
i
n (
1024,
512
a
nd 256
)
ne
ur
ons
r
e
s
pe
c
t
i
ve
l
y w
a
s
bui
l
t
,
w
i
t
h
a
c
t
i
va
t
i
on
f
unc
t
i
on
of
t
ype
t
a
nh(
X
)
.
T
he
out
put
l
a
ye
r
c
ons
i
s
t
s
of
10 out
put
ne
ur
ons
w
i
t
h a
c
t
i
va
t
i
on f
unc
t
i
on of
t
ype
s
of
t
m
a
x(
X
)
.
F
i
gu
r
e
2 s
how
s
t
he
s
e
que
nt
i
a
l
m
ode
l
of
t
he
w
or
k.
F
i
gur
e
2.
S
e
que
nt
i
a
l
m
ode
l
r
e
pr
e
s
e
nt
a
t
i
on
2.
5.
P
ref
o
r
m
a
n
ce
e
val
u
at
i
o
n
T
he
c
ol
l
e
c
t
e
d da
t
a
s
e
t
s
a
m
pl
e
s
a
r
e
di
vi
de
d i
nt
o t
w
o gr
oups
:
80%
t
r
a
i
ni
ng da
t
a
s
e
t
a
nd 20%
t
e
s
t
i
n
g
d
at
as
et
t
o
co
n
s
t
r
u
ct
t
h
e s
eq
u
en
t
i
al
m
o
d
el
o
f
t
h
e cl
as
s
i
f
i
cat
i
o
n
t
o
b
e t
es
t
ed
.
T
h
e p
er
f
o
r
m
an
ce i
s
ev
al
u
at
ed
i
n
each
ep
o
ch
w
ith
r
e
s
p
e
c
t to
tw
o
p
a
r
a
me
te
r
s
:
l
o
s
s
-
v
al
u
es
an
d
accu
r
acy
o
f
t
h
e cl
as
s
i
f
i
cat
i
o
n
.
A
ccu
r
acy
cal
cu
l
at
es
t
h
e
pe
r
c
e
nt
a
ge
of
pr
e
di
c
t
e
d
va
l
ue
s
(
yP
r
e
d)
t
ha
t
m
a
t
c
h
w
i
t
h a
c
t
ua
l
va
l
ue
s
(
yT
r
ue
)
.
W
he
n
r
unni
ng
t
he
m
ode
l
,
i
m
p
o
r
t
an
t
p
ar
am
et
er
s
ef
f
ect
m
u
s
t
b
e o
b
s
er
v
ed
s
i
n
ce t
h
ey
s
i
g
n
i
f
i
can
t
l
y
af
f
ect
t
h
e accu
r
acy
an
d
t
h
e
pr
oc
e
s
s
i
ng t
i
m
e
of
t
he
c
la
s
s
if
ic
a
tio
n
p
r
o
ces
s
.
T
h
e
p
ar
am
et
er
s
i
n
cl
u
d
e:
n
u
m
b
er
o
f
s
am
p
l
es
f
o
r
each
cl
as
s
,
to
ta
l
num
be
r
of
s
a
m
pl
e
s
,
a
nd t
he
t
ype
of
t
he
a
c
t
i
va
t
i
on f
unc
t
i
on a
ppl
i
e
d w
i
t
hi
n t
he
hi
dde
n a
nd out
put
l
a
ye
r
s
ne
ur
ons
.
W
he
n us
i
ng a
n e
qu
al
n
u
m
b
er
o
f
s
am
p
l
es
f
o
r
each
cl
as
s
,
t
h
i
s
w
i
l
l
g
i
v
e b
et
t
er
cl
as
s
i
f
i
cat
i
o
n
accu
r
acy
t
ha
n t
hos
e
w
i
t
h a
r
a
ndom
num
be
r
of
s
a
m
pl
e
s
pe
r
c
l
a
s
s
a
s
w
e
l
l
a
s
t
o t
he
obvi
ous
r
e
duc
t
i
on i
n t
he
nu
m
be
r
of
e
poc
hs
r
e
qui
r
e
d
t
o
t
r
a
i
n
t
he
ne
ur
a
l
ne
t
w
or
k,
a
nd
h
e
nc
e
t
he
ove
r
a
l
l
p
r
oc
e
s
s
i
ng t
i
m
e
w
i
l
l
be
r
e
duc
e
d,
a
s
s
how
n
i
n F
i
gur
e
3.
T
h
e t
o
t
al
n
u
m
b
er
o
f
s
am
p
l
es
i
s
t
h
e s
i
ze o
f
t
h
e co
l
l
ect
ed
s
am
p
l
es
,
as
t
h
i
s
s
i
ze i
n
cr
ea
s
es
t
h
e d
eep
le
a
r
n
in
g
w
ill
g
i
v
e a b
et
t
er
cl
as
s
i
f
i
cat
i
o
n
r
es
u
l
t
s
b
u
t
t
h
i
s
i
n
cr
eas
e can
n
o
t
b
e co
n
t
i
n
u
ed
s
i
n
ce t
h
e p
r
o
ces
s
i
n
g
time
w
ill
b
e
i
n
cr
eas
ed
as
w
el
l
a
s
to
th
e
s
ta
b
ility
o
f
th
e
a
c
c
u
r
a
c
y
r
e
s
u
lts
to
a
s
p
e
c
if
ic
v
a
lu
e
.
F
in
a
lly
,
th
e
r
e
a
r
e
m
a
ny t
ype
s
of
a
c
t
i
va
t
i
on
f
unc
t
i
ons
s
uc
h a
s
:
s
i
gm
o
i
d,
r
e
l
u
,
s
of
t
m
a
x
,
t
a
nh
a
nd e
xpone
nt
i
a
l
a
c
t
i
va
t
i
on
f
unc
t
i
on,
s
o
af
t
er
imp
le
m
e
nt
i
ng t
hos
e
t
ype
s
w
i
t
hi
n t
he
hi
dde
n l
a
ye
r
's
ne
ur
ons
.
T
he
m
os
t
a
c
c
e
pt
a
bl
e
a
c
c
ur
a
c
y l
e
ve
l
w
a
s
obt
a
i
ne
d w
he
n us
i
ng t
a
nh(
x)
a
c
t
i
va
t
i
on f
unc
t
i
on
,
w
hi
l
e
t
he
s
of
t
m
a
x(
x)
w
a
s
us
e
d w
i
t
hi
n t
he
out
p
ut
l
a
ye
r
s
ne
ur
ons
.
R
oot
m
ean
s
q
u
ar
e
(
R
M
S
)
o
p
timiz
e
r
w
a
s
u
s
e
d
to
min
imiz
e
th
e
e
r
r
o
r
w
h
ile
le
a
r
n
in
g
th
e
n
e
u
r
a
l
ne
t
w
or
k.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
1693
-
6930
T
E
L
KOM
NI
KA
T
el
eco
m
m
u
n
C
om
put
E
l
C
ont
r
o
l
,
Vo
l
.
1
9
, N
o
.
5
,
O
ct
o
b
er
2021
:
1588
-
1593
1592
F
i
gur
e
3
.
C
l
as
s
i
f
i
cat
i
o
n
accu
r
acy
l
ev
el
s
3.
R
ES
U
LTS
AND DI
S
CUS
I
O
N
F
i
r
s
t
l
y,
E
E
G
s
i
gna
l
s
c
l
a
s
s
i
f
i
c
a
t
i
on o
f
10
c
l
a
s
s
e
s
of
f
a
c
i
a
l
e
xpr
e
s
s
i
ons
a
nd m
ot
o
r
e
xe
c
ut
i
ons
a
c
t
i
ons
w
a
s
i
m
p
l
em
en
t
ed
f
o
r
f
o
u
r
s
u
b
j
ect
s
.
T
h
e p
er
f
o
r
m
an
ce o
f
t
h
e
cl
as
s
i
f
i
cat
i
o
n
m
o
d
el
w
as
ev
al
u
at
ed
,
as
m
en
t
i
o
n
ed
i
n
t
he
pr
e
vi
ous
s
e
c
t
i
on.
T
he
t
r
a
i
ni
ng
a
c
c
ur
a
c
y
i
s
r
a
ngi
ng f
r
om
(
91.
25
%
t
o 95
.
75%
)
,
a
nd t
he
be
s
t
r
e
s
ul
t
s
w
e
r
e
obt
a
i
ne
d w
he
n t
r
a
i
ni
ng t
he
m
ode
l
w
i
t
h 100 s
a
m
pl
e
s
/
c
l
a
s
s
w
i
t
h 973 t
ot
a
l
num
be
r
of
s
a
m
pl
e
s
.
T
he
s
e
r
e
s
ul
t
s
w
i
l
l
be
us
e
d i
n
t
he
f
ut
ur
e
w
o
r
k w
i
t
h
m
a
ny a
ppl
i
c
a
t
i
ons
s
uc
h a
s
bi
ndi
ng t
hos
e
c
l
a
s
s
e
s
w
i
t
h
s
p
eci
f
i
c t
en
s
es
o
r
w
or
ds
i
n or
de
r
t
o he
l
p t
he
s
pe
e
c
hl
e
s
s
pe
r
s
ons
t
o r
e
pr
e
s
e
nt
t
he
i
r
t
hought
s
,
s
o t
he
m
a
i
n
goa
l
of
t
hi
s
p
a
pe
r
i
s
t
o
de
s
i
gn a
s
i
m
pl
e
E
E
G
c
l
a
s
s
i
f
i
e
r
,
t
o be
u
t
i
l
i
z
e
d f
o
r
he
l
pi
ng
t
he
s
pe
e
c
hl
e
s
s
pe
r
s
ons
,
s
o t
ha
t
gi
vi
ng t
he
m
t
he
a
b
ility
t
o r
e
pr
e
s
e
nt
t
he
i
r
i
nt
e
nde
d t
hought
s
.
4.
CO
NCL
US
I
O
NS
I
n t
hi
s
pa
pe
r
,
t
e
n c
l
a
s
s
e
s
of
E
E
G
t
i
m
e
s
e
r
i
e
s
s
i
gna
l
va
l
ue
s
w
e
r
e
c
l
a
s
s
i
f
i
e
d by bui
l
di
ng de
e
p ne
ur
a
l
ne
t
w
or
k a
nd
imp
le
me
n
tin
g
d
eep
l
ear
n
i
n
g
t
ech
n
i
q
u
es
.
A
s
p
eci
al
i
zed
d
at
as
et
s
am
p
l
es
w
a
s
r
eco
r
d
e
d
.
I
n
t
h
e
o
f
f
lin
e
t
r
a
i
ni
ng,
t
he
c
l
a
s
s
i
f
i
c
a
t
i
on a
c
c
ur
a
c
y
r
e
s
ul
t
s
r
e
a
c
he
d t
o
95
.
75%
w
i
t
h
m
i
ni
m
i
z
i
ng c
os
t
of
c
om
put
a
t
i
on
a
nd s
t
or
a
ge
r
e
qui
r
e
m
e
nt
s
by
a
ppl
yi
ng on
l
y t
he
P
C
A
a
l
gor
i
t
hm
on
E
E
G
da
t
a
s
e
t
s
i
gna
l
s
va
l
ue
s
w
i
t
hout
a
ny
o
t
h
er
f
i
l
t
e
r
i
n
g
as
w
el
l
as
t
o
f
eed
t
h
e d
eep
n
ue
r
a
l
ne
t
w
or
k w
i
t
h
E
E
G
s
i
gna
l
va
l
ue
s
not
a
s
i
m
a
ge
o
r
hi
s
t
ogr
a
m
.
R
EF
ER
EN
C
ES
[
1]
A.
H.
Al
-
a
n
ba
r
y a
n
d S
a
li
h Al
-
Qa
r
a
a
w
i,
“
A S
ur
ve
y of
Ee
g S
i
gna
ls P
r
e
p
os
se
s
si
ng a
n
d C
la
s
sif
ic
a
t
io
n f
or
I
m
a
g
in
e
d
S
pe
e
c
h A
pp
lic
a
ti
on,
”
I
n
te
r
na
ti
on
al J
ou
rn
al o
f
I
n
nov
at
io
n En
gi
ne
e
r
in
g a
nd Sc
ie
nc
e
Re
se
arc
h
,
vo
l.
4,
no.
3,
p
p.
1
-
9
,
M
a
y 20
20.
[2
]
X.
Hua
ng
e
t al.
,
“
M
ul
ti
-
m
o
da
l
e
m
o
ti
on a
na
l
ys
is
f
r
om
f
a
c
ia
l e
xpr
e
ss
io
ns
a
n
d e
le
c
tr
oe
nc
e
p
ha
l
ogr
a
m
,
”
C
o
mpu
t
e
r
Vis
io
n
a
nd
I
ma
ge
Un
de
r
st
an
di
ng
,
v
ol.
1
47,
p
p.
11
4
-
1
24,
201
6,
do
i:
1
0.
1
01
6/
j.
c
v
iu.
2
01
5.
0
9.
01
5.
[3
]
A.
N.
B
e
lka
c
e
m
,
D.
S
h
in,
H.
Ka
m
ba
r
a
,
N.
Y
os
him
ur
a
,
a
nd
Y.
Ko
ike
,
“
O
nl
ine
c
la
s
sif
ic
a
t
io
n a
l
gor
it
hm
f
o
r
e
y
e
-
mo
v
emen
t
-
ba
se
d c
om
m
u
ni
c
a
t
io
n sys
te
m
s u
si
ng t
wo te
m
p
or
a
l E
EG
se
nsor
s,
”
Bi
ome
dic
al Si
gn
al P
roc
e
s
sin
g
and C
on
tr
ol
,
vol.
1
6,
pp.
4
0
-
4
7,
20
15,
d
oi
:1
0.
10
16
/j.
bs
p
c
.
201
4.
1
0.
00
5.
[4
]
J.
Jin,
I
.
Da
l
y,
Y.
Z
ha
n
g,
X.
W
a
n
g,
a
nd A.
C
ic
hoc
ki,
“
An o
pt
im
iz
e
d ER
P
br
a
in
-
c
om
p
ute
r
in
te
r
f
a
c
e
ba
se
d o
n f
a
c
ia
l
e
xpr
e
ss
io
n
c
ha
n
ge
s,
”
J
.
J
our
na
l
of N
e
u
ra
l En
gi
ne
e
r
ing
,
v
ol.
11,
n
o.
3,
pp.
1
-
12,
20
14,
d
oi
:
10.
1
08
8/
17
41
-
256
0/
11
/3
/0
36
00
4.
[5
]
D.
R
.
Ed
la
,
M
.
F
.
A
nsa
r
i,
N.
C
ha
u
dha
r
y,
a
n
d S
.
D
od
i
a
,
“
C
la
ss
if
i
c
a
t
io
n of
F
a
c
ia
l
E
xpr
e
ss
io
ns f
r
om
EEG
s
ig
na
l
s
usi
ng
W
a
ve
le
t P
a
c
ke
t Tr
a
nsf
or
m
a
nd S
VM
f
or
W
he
e
lc
ha
ir
C
o
ntr
ol
Ope
r
a
ti
on
s,
”
P
roc
e
di
a C
o
mp
ute
r Sc
ie
nc
e
,
vol.
1
32,
no.
I
c
c
i
ds,
p
p.
14
67
-
14
76,
2
01
8,
do
i:
1
0.
1
01
6/
j
.
pr
oc
s.
20
18.
0
5.
0
81.
[6
]
A.
Al
-
Na
f
ja
n,
M
.
Ho
sn
y,
Y.
Al
-
O
ha
l
i,
a
nd A.
A
l
-
Wa
bi
l,
“
R
e
vie
w a
nd c
la
ss
if
ic
a
ti
on of
e
m
o
ti
on r
e
c
og
ni
ti
on ba
se
d
on EEG
br
a
in
-
c
o
m
p
u
t
e
r
i
n
t
e
r
f
a
c
e
s
y
s
t
e
m
r
e
s
e
a
r
c
h: A sys
te
m
a
tic
r
e
vie
w,
”
Ap
pl
ie
d Sc
ie
nc
e
s
,
vo
l.
7,
n
o.
1
2,
De
c
.
201
7,
do
i:
10.
3
39
0/
a
p
p7
12
12
39.
[7
]
Z.
Ga
o
e
t al
., “
E
E
G
-
B
a
se
d S
pa
t
io
-
Te
m
p
or
a
l C
on
vo
lu
ti
ona
l Ne
ur
a
l Ne
tw
or
k f
or
Dr
ive
r
F
a
ti
gue
Eva
lua
ti
on,
”
I
E
EE
tra
ns
ac
t
io
ns
o
n ne
ur
al
ne
tw
o
rk
s
an
d le
ar
ni
ng
s
y
s
t
e
m
s,
v
ol.
30,
no.
9,
pp.
27
55
-
27
63,
2
019
,
doi
: 1
0.
11
09
/
TN
NL
S
.
201
8.
28
86
41
4.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
el
eco
m
m
u
n
C
om
put
E
l
C
ont
r
o
l
C
l
as
s
i
f
i
c
at
i
on of
E
E
G
s
i
gnal
s
f
or
f
ac
i
al
e
x
pr
e
s
s
i
on
and
m
ot
or
execu
t
i
o
n
w
i
t
h
….
(
A
r
eej
H
a
m
eed
A
l
-
A
nbar
y
)
1593
[8
]
D.
Khe
ir
a
a
n
d M
.
B
e
la
dg
ha
m
,
“
P
e
r
f
or
m
a
nc
e
of
c
ha
nn
e
l
se
le
c
t
io
n u
se
d f
or
M
ul
ti
-
c
la
s
s E
EG
si
gna
l c
la
ss
if
ic
a
ti
on
of
m
oto
r
im
a
ge
r
y,
”
I
n
do
ne
s
ia
n J
ou
rn
al
of E
le
c
tr
ic
a
l
Eng
ine
e
r
i
ng
an
d C
om
pu
te
r Sc
ie
nc
e
,
vo
l.
1
5,
no
.
3,
pp.
13
05
-
1
31
2,
S
e
p.
20
19,
d
oi
: 10.
11
59
1/
ije
e
c
s.
v1
5.
i
3.
[9
]
A
. S
. A
l
-
F
a
ho
um
a
n
d A.
A.
Al
-
F
r
ai
h
at
,
“
M
e
th
od
s of
EEG
S
ig
na
l F
e
a
t
ur
e
s
Ex
tr
a
c
t
io
n Us
in
g L
ine
a
r
A
na
l
ys
i
s in
F
r
e
que
nc
y a
n
d T
im
e
-
F
r
e
q
ue
nc
y D
om
a
i
ns,
”
I
S
RN
N
e
ur
osc
ie
nc
e
,
v
ol.
2
01
4,
pp.
1
-
7,
2
0
14,
doi
: 1
0.
11
55
/2
01
4/
73
02
18.
[
10]
A.
T.
S
o
ha
i
b,
S
.
Q
ur
e
s
hi,
J.
Ha
ge
l
bä
c
k,
O.
H
ilb
or
n,
a
n
d
P
.
Je
r
č
ić
,
“
E
va
l
ua
t
in
g c
la
ss
if
ie
r
s f
or
e
m
o
ti
on r
e
c
o
gn
it
io
n
usi
ng EEG
,
”
C
on
fe
re
nc
e
:
I
n
te
r
na
ti
on
al C
o
nfe
re
n
c
e
on
Au
gme
nte
d C
og
ni
ti
on
,
v
ol.
802
7 L
NAI
,
2
01
3,
pp.
4
92
-
501
,
doi
: 1
0.
10
07
/9
78
-
3
-
6
42
-
39
45
4
-
6_
53.
[
11]
H.
Xu
a
n
d K.
N.
P
la
ta
n
io
ti
s,
“
E
EG
-
ba
se
d a
f
f
e
c
t s
ta
te
s
c
la
s
sif
ic
a
t
io
n u
si
ng
De
e
p
B
e
l
ie
f
Ne
tw
or
k
s,
”
Di
gi
ta
l
M
e
d
ia
I
nd
ust
ry
Ac
ad
im
ic
F
or
um (
D
MI
A
F)
p
roc
e
di
ng
,
20
16
,
pp
.
148
-
1
53
,
do
i: 1
0.
1
10
9/
DM
I
AF
.
2
01
6.
75
74
92
1.
[
12]
M.
Z.
A
l
-
F
a
i
z
a
nd A.
A.
Al
-
Ha
m
a
da
ni
,
“
I
m
p
le
m
e
n
ta
t
i
on of
EEG
S
i
gna
l
P
r
oc
e
s
si
ng
a
nd
De
c
o
di
ng
fo
r
Tw
o
-
C
l
a
ss
M
otor
I
m
a
ge
r
y Da
ta
,”
B
io
me
d
ic
a
l E
ng
ine
e
ri
ng
-
Ap
pl
ic
ati
on
s,
B
as
is a
nd
C
om
mu
nic
at
io
ns
J
o
ur
na
l
,
vo
l.
31,
n
o.
4,
pp.
1
-
10,
2
01
9,
do
i: 1
0.
4
01
5/S
1
01
62
37
21
95
00
28
5.
[
13]
N.
Ja
tu
pa
i
bo
on,
S
.
P
a
n
-
N
g
u
m
,
a
n
d
P
.
I
s
r
a
s
e
n
a
,
“
R
e
a
l
-
tim
e
EEG
-
ba
se
d ha
pp
ine
ss de
te
c
t
io
n sy
ste
m
,
”
Sc
ie
n
t
ifi
c
Wo
rld J
ou
rn
al
,
vol.
2
01
3,
20
13,
d
oi
: 1
0.
11
55
/2
01
3/
61
8
649.
[
14]
M.
Z.
Al
-
F
a
i
z
a
nd A.
A.
Al
-
Ha
m
a
da
ni,
“
Ana
ly
si
s a
n
d I
m
ple
m
e
n
ta
t
io
n of
B
r
a
in Wa
ve
s F
e
a
tur
e
E
xtr
a
c
ti
on
a
nd
C
la
ss
if
i
c
a
t
io
n to C
on
tr
o
l R
ob
ot
ic
Ha
nd,
”
I
r
aq
i J
our
n
al of I
n
fo
rm
at
io
n &
C
ommu
nic
at
io
ns T
e
c
h
no
lo
gy
,
vol
.
1,
no.
3,
pp.
3
1
-
41,
2
01
9,
do
i:
1
0.
3
19
87
/i
jic
t.
1.
3.
3
5.
[
15]
P
.
Kshir
sa
ga
r
,
“
F
e
a
tur
e
E
xtr
a
c
ti
on of
EEG
S
ig
na
l
s us
i
ng W
a
ve
le
t a
n
d P
r
inc
ipa
l C
om
po
ne
nt a
na
l
ys
is,
”
N
a
ti
on
al
C
onf
e
re
nc
e
on
Re
se
arc
h T
re
n
ds
in
E
le
c
t
ro
nic
s,
C
om
pu
t
e
r Sc
ie
nc
e
&
I
nf
or
ma
ti
on T
e
c
hn
ol
ogy
,
F
e
b
.
20
14
.
[
16]
F
.
L
otte
e
t a
l.
,
“
A r
e
vie
w of
c
la
ss
if
ic
a
ti
on a
l
go
r
i
thm
s f
o
r
EEG
-
ba
se
d br
a
in
-
c
om
p
ute
r
i
nte
r
f
a
c
e
s
: A
10
-
y
e
a
r
u
pda
te
,
”
J
ou
rn
al
of
N
e
ur
al
E
ng
in
e
e
ri
ng
,
vo
l.
15,
n
o.
3,
20
18,
d
oi
:
10.
10
88
/1
74
1
-
25
52
/a
a
b
2f
2.
[
17]
A.
N.
N.
M
.
Y
os
i,
K.
A.
S
ide
k,
H.
S
.
Ya
a
c
ob,
M
.
Ot
hm
a
n,
a
n
d
A.
Z
.
Jus
oh,
“
Em
o
ti
on r
e
c
og
ni
ti
on
us
in
g
e
le
c
tr
oe
nc
e
pha
lo
gr
a
m
s
ig
na
l,
”
I
nd
one
si
an J
ou
rn
al
of
El
e
c
tr
ic
a
l E
ng
ine
e
ri
ng
an
d C
o
mp
ute
r Sc
ie
nc
e
,
vo
l.
15,
no
.
2,
pp.
78
6
-
79
3,
20
19,
d
oi
: 10.
11
59
1/
ije
e
c
s.
v1
5.
i
2.
pp
78
6
-
7
93.
[
18]
P
.
S
z
a
c
h
e
w
i
c
z
,
“
C
l
a
s
s
i
f
i
c
a
t
i
o
n
o
f
M
o
t
o
r
Ima
g
ery
f
o
r Br
ai
n
-
C
om
p
ute
r
I
nte
r
f
a
c
e
s,
”
M
a
ste
r
’
s t
he
s
is,
P
oz
n.
U
ni
ve
r
sit
y
of
Te
c
hn
ol
og
y,
F
a
c
u
lt
y
of
C
om
pu
ti
ng
a
n
d I
nf
or
m
a
t
io
n
S
c
ie
nc
e
,
I
ns
ti
tu
te
of
C
om
pu
ti
ng
S
c
ie
nc
e
M
a
s
te
r
’
s,
p.
50,
201
3.
[
19]
Y.
Z
ha
ng,
S
.
Z
ha
ng,
a
nd
X.
Ji,
“
E
EG
-
ba
se
d c
la
ss
if
ic
a
t
ion
of
e
m
ot
io
ns u
si
ng
e
m
p
ir
ic
a
l
m
o
de
de
c
om
po
si
ti
on
a
nd
a
ut
or
e
gr
e
ss
ive
m
o
de
l,
”
Mu
lt
ime
d.
T
oo
ls Ap
pl
ic
a
tio
ns
,
vo
l.
77,
n
o.
2
0,
pp.
2
66
97
-
26
71
0,
20
18
,
doi
: 1
0.
10
07
/s
11
04
2
-
01
8
-
58
85
-
9
[
20]
M.
Mo
h
ammad
p
o
u
r
,
S
.
M
.
R
.
Ha
she
m
i,
a
n
d N.
Ho
u
shm
a
nd,
“
C
la
ss
if
ic
a
ti
on
of
EEG
-
ba
se
d
e
m
o
ti
on f
or
B
C
I
a
pp
lic
a
ti
on
s,
”
7t
h C
on
fe
re
nc
e
o
n A
rt
if
ic
i
al
I
nte
ll
ige
nc
e
an
d R
ob
ot
ic
s
,
20
17
,
pp.
1
27
-
13
1
,
doi
: 1
0.
11
09
/R
I
OS
.
20
17.
7
95
64
55.
[
21]
J.
L
i,
Z
.
Z
ha
ng,
a
nd H.
He
,
“
Hie
r
a
r
c
hic
a
l
C
on
vo
lu
ti
ona
l Ne
ur
a
l Ne
tw
or
k
s f
or
EEG
-
B
a
se
d Em
ot
io
n R
e
c
og
ni
ti
o
n,
”
C
ogn
it
iv
e
C
o
mp
ut
at
io
n
,
vo
l.
10,
n
o.
2,
pp.
3
68
-
38
0,
20
1
8,
doi
: 1
0.
10
07
/s
12
55
9
-
01
7
-
95
33
-
x.
[
22]
S
.
Alha
gr
y,
A.
Aly,
a
nd
R
.
E
.
Khor
i
bi
,
“
Em
ot
io
n R
e
c
og
ni
ti
on ba
se
d on E
EG
usi
ng L
S
TM
R
e
c
ur
r
e
n
t Ne
ur
a
l
Ne
t
wor
k,
”
I
nte
r
na
ti
on
al J
o
ur
na
l of Adv
anc
e
d C
om
put
e
r Sc
ie
nc
e
an
d Ap
pl
ic
a
ti
on
s
,
vol.
8,
no.
10,
2
017
,
doi
: 1
0.
14
56
9/
ij
a
c
sa
.
2
01
7.
0
81
04
6.
[
23]
A.
C
r
a
ik,
Y.
He
,
a
nd J.
L
.
C
.
Vida
l,
“
De
e
p le
a
r
ni
ng
f
or
e
le
c
tr
o
e
nc
e
pha
lo
gr
a
m
(
EEG
)
c
la
ss
if
ic
a
ti
on ta
s
ks
: A
r
e
vie
w,
”
J
ou
rn
al o
f N
e
u
ra
l En
gi
ne
e
r
in
g
,
vo
l.
16,
n
o.
3,
201
9,
do
i:
10.
1
08
8/
17
41
-
2
55
2/a
b0a
b5.
[
24]
X.
L
i,
X.
Jia
,
G
.
Xun,
a
nd A.
Z
ha
n
g,
“
I
m
pr
o
vi
ng
EEG
f
e
a
tur
e
le
a
r
ni
ng v
ia
sy
nc
hr
on
iz
e
d f
a
c
ia
l v
ide
o
,
”
I
nte
rn
at
io
na
l C
on
fe
re
nc
e
o
n Bi
g D
at
a,
I
EE
E
B
ig
Da
ta
,
pp.
8
43
-
84
8,
20
15,
d
oi
:
10.
11
09
/B
ig
Da
ta
.
20
15.
73
63
83
1
.
[
25]
A.
Tha
r
wa
t,
“
P
r
inc
ipa
l c
om
po
ne
n
t a
na
l
ys
is
-
a
tu
tor
ia
l,
”
I
nte
rn
at
io
na
l
J
o
ur
na
l of Ap
pl
ie
d Pa
tte
rn Re
c
o
gn
it
i
on
,
vol.
3,
n
o.
3,
p
p
.
19
7
-
24
0
,
2
01
6,
do
i: 1
0.
1
50
4/
ija
pr
.
20
16
.
079
73
3.
[2
6
]
S
.
S
iuly a
nd Y.
L
i,
“
De
si
gn
in
g a
r
ob
us
t f
e
a
t
ur
e
e
xtr
a
c
ti
on m
e
t
ho
d ba
se
d o
n op
tim
um
a
ll
oc
a
t
io
n a
n
d pr
i
nc
i
pa
l
c
om
p
one
nt a
na
ly
si
s f
or
e
p
ile
pt
ic
EEG
si
gna
l c
la
ss
if
ic
a
ti
on,
”
C
om
pu
te
r
Me
t
ho
ds a
nd P
ro
gr
am
s i
n Bi
ome
di
c
in
e
J
ou
rn
al
,
v
ol.
1
19,
n
o.
1,
pp.
2
9
-
4
2,
20
15,
d
oi
: 10.
10
16
/j.
c
m
pb.
2
01
5.
0
1.
00
2.
Evaluation Warning : The document was created with Spire.PDF for Python.