T
E
L
K
O
M
N
I
K
A
T
elec
o
m
m
un
ica
t
io
n,
Co
m
pu
t
ing
,
E
lect
ro
nics
a
nd
Co
ntr
o
l
Vo
l.
18
,
No
.
4
,
Octo
b
er
2
0
2
0
,
p
p
.
2
7
4
8
~
275
6
I
SS
N:
1
6
9
3
-
6
9
3
0
,
ac
cr
ed
ited
First Gr
ad
e
b
y
Kem
en
r
is
tek
d
i
k
ti,
Dec
r
ee
No
: 2
1
/E/KPT
/2
0
1
8
DOI
: 1
0
.
1
2
9
2
8
/TE
L
KOM
NI
K
A.
v
1
8
i5
.
1
4
8
9
9
2748
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//jo
u
r
n
a
l.u
a
d
.
a
c.
id
/in
d
ex
.
p
h
p
/TELK
OM
N
I
K
A
Bra
in
-
co
mputer
i
nterf
a
ce
of
focus
a
nd mo
tor ima
g
ery
using
wav
elet
a
nd
rec
urr
ent
neural
networks
E
s
m
er
a
lda
C.
Dj
a
m
a
l
,
Rif
qi D.
P
utr
a
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,
I
n
d
o
n
e
sia
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
e
d
J
u
n
1
9
,
2
0
1
9
R
ev
is
ed
Ap
r
8
,
2
0
2
0
Acc
ep
ted
Ma
y
1
,
2
0
2
0
Bra
in
-
c
o
m
p
u
ter
in
terfa
c
e
is
a
t
e
c
h
n
o
l
o
g
y
th
a
t
a
ll
o
ws
o
p
e
ra
ti
n
g
a
d
e
v
ice
with
o
u
t
i
n
v
o
lv
in
g
m
u
sc
les
a
n
d
s
o
u
n
d
,
b
u
t
d
irec
tl
y
fr
o
m
th
e
b
ra
i
n
th
r
o
u
g
h
th
e
p
r
o
c
e
ss
e
d
e
lec
tri
c
a
l
sig
n
a
ls.
T
h
e
tec
h
n
o
lo
g
y
wo
r
k
s
b
y
c
a
p
t
u
rin
g
e
lec
tri
c
a
l
o
r
m
a
g
n
e
ti
c
sig
n
a
ls
fro
m
t
h
e
b
r
a
in
,
wh
ich
a
re
th
e
n
p
ro
c
e
ss
e
d
to
o
b
tain
in
fo
rm
a
ti
o
n
c
o
n
tain
e
d
t
h
e
re
in
.
Us
u
a
ll
y
,
BCI
u
se
s
i
n
fo
rm
a
t
io
n
fr
o
m
e
lec
tro
e
n
c
e
p
h
a
lo
g
ra
m
(
EE
G
)
sig
n
a
ls
b
a
se
d
o
n
v
a
ri
o
u
s
v
a
riab
les
re
v
iew
e
d
.
Th
is
st
u
d
y
p
ro
p
o
se
d
BCI
to
m
o
v
e
e
x
tern
a
l
d
e
v
ice
s
su
c
h
a
s
a
d
ro
n
e
sim
u
lato
r
b
a
s
e
d
o
n
EE
G
sig
n
a
l
i
n
fo
rm
a
ti
o
n
.
F
r
o
m
t
h
e
E
EG
sig
n
a
l
wa
s
e
x
tra
c
ted
to
g
e
t
m
o
to
r
ima
g
e
ry
(M
I)
a
n
d
fo
c
u
s
v
a
riab
le
u
sin
g
wa
v
e
let
.
Th
e
n
,
th
e
y
we
re
c
las
sified
b
y
re
c
u
rre
n
t
n
e
u
ra
l
n
e
t
wo
rk
s
(RNN
).
In
o
v
e
rc
o
m
i
n
g
th
e
p
ro
b
lem
o
f
v
a
n
ish
i
n
g
m
e
m
o
ry
fr
o
m
RNN
,
wa
s
u
se
d
lo
n
g
sh
o
r
t
-
term
m
e
m
o
ry
(LS
TM
).
Th
e
re
su
lt
s
sh
o
we
d
th
a
t
BCI
u
se
d
wa
v
e
let
,
a
n
d
RNN
c
a
n
d
r
iv
e
e
x
ter
n
a
l
d
e
v
ice
s
o
f
n
o
n
-
train
i
n
g
d
a
ta
wit
h
a
n
a
c
c
u
ra
c
y
o
f
7
9
.
6
%
.
T
h
e
e
x
p
e
ri
m
e
n
t
g
a
v
e
Ad
a
De
lt
a
m
o
d
e
l
is
b
e
tt
e
r
t
h
a
n
th
e
Ad
a
m
m
o
d
e
l
in
term
s
o
f
a
c
c
u
ra
c
y
a
n
d
v
a
l
u
e
lo
ss
e
s.
Wh
e
re
a
s
in
c
o
m
p
u
tati
o
n
a
l
lea
rn
in
g
ti
m
e
,
Ad
a
m
'
s
m
o
d
e
l
is
f
a
ste
r
th
a
n
Ad
a
De
lt
a
'
s m
o
d
e
l.
K
ey
w
o
r
d
s
:
B
r
ain
-
co
m
p
u
ter
in
ter
f
ac
e
E
E
G
s
ig
n
al
Fo
cu
s
Mo
to
r
im
ag
er
y
R
ec
u
r
r
en
t n
eu
r
al
n
etwo
r
k
s
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
.
Djam
al
,
Dep
ar
tm
en
t o
f
I
n
f
o
r
m
atics,
Un
iv
er
s
itas
J
en
d
er
al
Ach
m
ad
Yan
i,
T
er
u
s
an
J
en
d
er
al
Su
d
ir
m
an
C
i
m
ah
i
St.,
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
Hu
m
an
s
in
ev
er
y
d
ay
alwa
y
s
ca
r
r
y
o
u
t
ac
tiv
ities
th
at
in
v
o
lv
e
th
e
m
o
v
em
en
t
o
f
lim
b
s
.
T
h
e
b
r
ain
in
s
tr
u
cts
th
e
r
esu
ltin
g
ac
tio
n
th
r
o
u
g
h
m
o
to
r
n
er
v
es.
E
v
e
r
y
h
u
m
an
ac
tiv
ity
r
eq
u
ir
es
a
f
o
c
u
s
o
n
ca
r
r
y
in
g
o
u
t
ac
tiv
ities
f
o
r
p
ar
ticu
lar
p
u
r
p
o
s
es.
Nev
er
th
eless
,
th
e
f
o
cu
s
ca
n
b
e
i
n
f
lu
en
ce
d
b
y
s
ev
er
al
f
ac
to
r
s
s
u
c
h
as
s
tim
u
latio
n
o
f
s
o
u
n
d
o
r
v
is
io
n
th
at
ca
n
af
f
ec
t
th
e
ac
tiv
ities
b
ein
g
ca
r
r
ie
d
o
u
t.
I
n
m
ea
n
wh
il
e
,
th
e
c
o
m
m
an
d
to
m
o
v
e
a
lim
b
o
cc
u
r
s
o
v
er
a
s
tat
e
o
f
m
in
d
ca
lled
m
o
t
o
r
i
m
ag
er
y
(
MI
)
.
H
o
wev
er
,
m
o
v
in
g
th
e
o
b
ject
ca
n
b
e
ca
r
r
ied
o
u
t
with
o
u
t
in
v
o
lv
e
g
estu
r
es,
m
u
s
cles,
s
o
u
n
d
s
,
an
d
o
th
er
m
o
to
r
f
u
n
ctio
n
s
.
T
h
ese
co
m
m
an
d
s
ar
e
o
b
tain
ed
f
r
o
m
th
e
b
r
ain
th
r
o
u
g
h
in
ter
m
ed
iat
e
d
ev
ices
to
tr
a
n
s
late
b
r
ai
n
c
o
m
m
an
d
s
,
k
n
o
wn
as
th
e
b
r
ain
-
co
m
p
u
ter
in
ter
f
ac
e
(
B
C
I
)
.
T
h
is
s
y
s
tem
ca
n
h
elp
p
e
o
p
le
with
p
h
y
s
ical
d
is
ab
ilit
ie
s
in
m
o
v
in
g
ex
ter
n
al
d
ev
ices.
C
u
r
r
en
tly
,
B
C
I
h
as
b
ee
n
wid
ely
u
s
ed
to
d
r
iv
e
g
am
es
[1
,
2]
,
r
o
b
o
ts
[
3
]
,
to
h
elp
p
o
s
t
-
s
tr
o
k
e
p
atien
ts
[
4
]
a
n
d
n
e
u
r
o
m
u
s
cu
lar
d
is
ab
ilit
ies
[
5
]
.
B
C
I
co
n
s
is
t
s
o
f
th
r
ee
p
ar
ts
,
p
a
r
ticu
lar
ly
co
m
m
a
n
d
in
p
u
t,
in
te
r
m
ed
iate
d
ev
ice,
a
n
d
c
o
m
m
an
d
co
n
tr
ol
.
B
C
I
u
s
u
ally
u
s
es
s
tan
d
ar
d
to
o
l
s
s
u
ch
as
elec
tr
o
en
ce
p
h
alo
g
r
a
m
(
EEG
)
s
ig
n
als to
t
r
an
s
late
b
r
ain
co
m
m
a
n
d
s
[
6
]
.
T
h
e
E
E
G
s
ig
n
a
l
is
b
io
elec
tr
ic
in
th
e
b
r
ai
n
th
at
is
ca
p
tu
r
e
d
o
n
th
e
s
u
r
f
ac
e
o
f
th
e
s
ca
lp
.
E
E
G
s
ig
n
al
h
as
a
lo
w
am
p
litu
d
e
,
n
o
n
-
s
tatio
n
ar
y
,
an
d
co
m
p
licated
p
atter
n
s
.
T
h
e
s
ig
n
al
co
n
s
is
ts
o
f
f
r
eq
u
en
c
y
co
m
p
o
n
en
ts
s
u
ch
as
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
B
r
a
in
-
co
mp
u
ter in
terf
a
ce
o
f fo
cu
s
a
n
d
mo
t
o
r
ima
g
ery
u
s
in
g
w
a
ve
let
a
n
d
… (
E
s
mera
ld
a
C
.
Dja
ma
l
)
2749
Alp
h
a
wav
es
(
8
-
1
3
Hz)
,
B
eta
wav
es
(
1
4
-
3
0
Hz)
,
T
h
eta
wa
v
es
(
4
-
7
Hz)
,
Delta
wav
es
(
0
.
5
-
3
Hz)
,
an
d
Gam
m
a
wav
es
(
>
3
0
Hz)
.
E
E
G
s
ig
n
al
id
en
tifie
d
m
e
n
tal
task
v
a
r
i
ab
les
as
ap
p
r
o
p
r
iate
ac
tio
n
s
o
n
a
c
o
m
p
u
ter
[
7
]
.
T
h
er
ef
o
r
e,
th
e
r
ep
r
esen
tatio
n
o
f
E
E
G
s
ig
n
als
in
to
th
e
f
r
eq
u
en
c
y
d
o
m
ain
is
v
er
y
co
n
s
id
er
ate
f
o
r
g
ettin
g
co
n
d
itio
n
s
p
ar
ticu
lar
th
o
u
g
h
ts
.
So
m
e
p
r
ev
io
u
s
s
t
u
d
ies
in
ex
tr
ac
tin
g
E
E
G
s
ig
n
als
u
s
in
g
wav
elet
tr
an
s
f
o
r
m
[8
-
11]
in
B
C
I
.
W
av
elets
ca
n
f
ilter
s
ig
n
als
f
r
o
m
p
r
ec
is
e
f
r
eq
u
en
cy
co
m
p
o
n
e
n
t
s
.
So
th
at
m
et
h
o
d
is
p
r
o
p
er
f
o
r
n
o
n
-
s
tatio
n
ar
y
s
ig
n
als
s
u
ch
as
E
E
G
.
Ho
wev
er
,
wi
th
a
s
h
o
r
t
tim
e
s
eg
m
e
n
tatio
n
,
i
t
is
p
o
s
s
ib
le
to
u
s
e
o
th
er
m
et
h
o
d
s
s
u
ch
as
f
ast
Fo
u
r
ier
tr
a
n
s
f
o
r
m
(
FF
T
)
f
o
r
ex
tr
ac
tin
g
n
o
n
-
s
tatio
n
a
r
y
s
ig
n
als
[
1
2
]
.
FF
T
h
as
ad
v
an
tag
es
in
ter
m
s
o
f
c
o
m
p
u
tatio
n
al
s
p
ee
d
,
alth
o
u
g
h
wav
elet
is
u
s
u
ally
m
o
r
e
ac
c
u
r
ate.
W
av
el
et
ex
tr
ac
tio
n
ca
n
in
cr
ea
s
e
ac
c
u
r
ac
y
b
y
3
.
6
% a
n
d
ac
ce
ler
ate
d
etec
tio
n
tim
e
b
y
0
.
0
0
3
s
ec
o
n
d
s
.
T
h
e
last
s
tu
d
y
o
b
tain
ed
9
3
.
6
%
ac
cu
r
ac
y
o
f
tr
ain
in
g
d
ata
,
a
n
d
9
0
% o
f
n
o
n
-
tr
ai
n
in
g
d
ata
[
1
3
]
.
So
m
e
v
ar
iab
les
th
at
af
f
ec
t
ed
E
E
G
s
ig
n
als
ar
e
d
eter
m
in
a
n
ts
o
f
class
if
icatio
n
in
p
r
ev
io
u
s
s
tu
d
ies,
s
u
ch
as
em
o
tio
n
s
[
1
4
]
,
d
is
o
r
d
er
[
1
5
]
a
n
d
f
o
cu
s
[
1
6
]
.
Me
an
wh
ile,
u
s
u
ally
,
th
e
E
E
G
s
ig
n
a
l
v
ar
iab
le
tr
an
s
lated
in
B
C
I
is
f
o
cu
s
[
1
6
]
,
atten
tio
n
lev
el
[
1
7
]
,
r
elax
[
1
]
,
e
m
o
tio
n
[
1
8
,
1
9
]
,
h
an
d
g
r
asp
in
g
im
ag
i
n
atio
n
[
2
0
]
an
d
h
y
b
r
i
d
o
f
m
o
to
r
im
ag
er
y
a
n
d
s
p
ee
ch
im
ag
e
r
y
[
2
1
]
.
Usu
ally
,
t
h
e
s
tu
d
ies
u
s
e
d
o
n
e
c
h
ar
ac
t
er
is
tic
v
ar
iab
le
in
th
e
class
if
icatio
n
p
r
o
ce
s
s
.
Pre
v
io
u
s
s
tu
d
ies
u
s
ed
B
C
I
to
m
o
v
e
ch
ar
ac
ter
s
in
ar
ca
d
e
g
am
es
b
ased
o
n
f
o
cu
s
f
ee
d
b
ac
k
[
2
2
]
,
co
n
t
r
o
ls
f
o
r
c
o
m
p
u
ter
a
p
p
licatio
n
s
,
o
r
ac
tio
n
o
n
im
ag
in
ed
c
o
n
d
itio
n
s
o
f
th
e
m
in
d
[
6
]
a
n
d
wh
ee
lch
air
r
o
b
o
tic
[
2
3
]
.
I
n
p
atter
n
r
ec
o
g
n
itio
n
,
af
te
r
ex
tr
ac
tio
n
f
ea
tu
r
es,
th
en
in
to
th
e
class
if
icatio
n
s
y
s
tem
.
I
n
B
C
I
a
p
p
licatio
n
,
th
e
p
r
ev
i
o
u
s
r
esear
ch
u
s
ed
s
o
m
e
m
eth
o
d
s
s
u
ch
as
lear
n
i
n
g
v
ec
to
r
q
u
an
tizatio
n
(
L
VQ)
[
1
]
,
r
ec
u
r
r
e
n
t
n
eu
r
al
n
etwo
r
k
s
(
R
NN)
[
2
4
]
,
an
d
co
n
v
o
lu
tio
n
al
n
eu
r
al
n
etwo
r
k
s
(
C
NN)
[
2
5
]
.
T
h
e
r
e
wa
s
u
s
in
g
C
NN
to
B
C
I
g
am
e
co
n
tr
o
l
[
2
6
]
.
Me
a
n
wh
ile,
tim
e
s
er
ies
ca
s
es
o
f
ten
u
s
e
R
NN,
wh
ich
f
ac
ilit
ates
th
e
c
o
n
n
ec
tio
n
o
f
s
eq
u
e
n
tial
d
ata
with
p
ast tim
e
[
1
9
]
.
I
n
p
r
ev
io
u
s
s
tu
d
ies u
s
in
g
R
NN
to
r
ec
o
g
n
ize
em
o
tio
n
s
f
r
o
m
E
E
G
s
ig
n
al
s
with
an
ac
cu
r
ac
y
r
ate
o
f
8
7
%
[
1
4
]
.
T
h
is
r
esear
c
h
p
r
o
p
o
s
ed
th
e
B
C
I
m
o
d
el
to
d
r
iv
e
th
e
d
r
o
n
e
s
im
u
lat
o
r
f
r
o
m
th
e
f
o
c
u
s
s
tate
an
d
m
o
to
r
im
a
g
er
y
.
M
o
d
els
d
ev
el
o
p
ed
u
s
in
g
wav
elet
an
d
r
ec
u
r
r
en
t
n
eu
r
al
n
etwo
r
k
s
(
R
NNs).
T
h
e
d
r
o
n
e
s
im
u
lato
r
is
d
esig
n
ed
to
b
e
d
r
iv
en
b
y
a
n
im
ag
er
y
m
o
to
r
i
n
to
f
o
u
r
,
p
ar
ticu
lar
ly
"f
o
r
war
d
",
"r
ig
h
t",
"lef
t",
an
d
"sil
en
t".
B
esid
es,
th
e
s
im
u
lato
r
ac
tio
n
ad
d
ed
f
o
c
u
s
f
ac
to
r
(
two
class
es:
f
o
cu
s
o
r
n
o
t
f
o
cu
s
)
,
wh
ich
is
d
escr
ib
ed
as
th
e
r
o
tatio
n
s
p
ee
d
.
So
th
at
e
ig
h
t c
lass
es a
r
e
o
b
tain
ed
.
2.
RE
S
E
ARCH
M
E
T
H
O
D
T
h
is
r
esear
ch
p
r
o
p
o
s
ed
B
C
I
to
d
r
iv
e
th
e
d
r
o
n
e
s
im
u
lato
r
t
h
r
o
u
g
h
E
E
G
s
ig
n
als
u
s
in
g
wav
elet
an
d
R
NN,
as
s
h
o
wn
in
Fig
u
r
e
1
.
T
h
e
s
y
s
tem
u
s
ed
v
ar
iab
le
MI
an
d
Fo
c
u
s
th
at
wer
e
p
r
o
ce
s
s
ed
b
y
s
im
u
ltan
eo
u
s
.
So
th
at
ar
e
eig
h
t
class
es
o
f
b
o
th
v
ar
iab
les.
T
h
e
m
o
d
el
u
s
ed
d
ata
s
et
with
em
o
tiv
ep
o
ch
E
E
G
r
ec
o
r
d
in
g
as
s
h
o
wn
in
Fig
u
r
e
2
.
E
E
G
S
i
g
n
a
l
s
P
r
a
p
r
o
c
e
s
s
i
n
g
S
e
g
m
e
n
t
a
t
i
o
n
W
a
v
e
l
e
t
E
x
t
r
a
c
t
i
o
n
R
e
c
u
r
r
e
n
t
N
e
u
r
a
l
N
e
t
w
o
r
k
s
I
d
e
n
t
i
f
i
c
a
t
i
o
n
1
s
t
L
S
T
M
L
a
y
e
r
(
R
e
l
u
)
D
r
o
p
o
u
t
L
a
y
e
r
2
n
d
L
S
T
M
L
a
y
e
r
(
S
i
g
m
o
i
d
)
D
e
n
s
e
L
a
y
e
r
(
S
i
g
m
o
i
d
)
W
e
i
g
h
t
R
e
c
u
r
r
e
n
t
N
e
u
r
a
l
N
e
t
w
o
r
k
s
T
r
a
i
n
i
n
g
E
E
G
S
i
g
n
a
l
P
r
a
p
r
o
c
e
s
s
i
n
g
S
e
g
m
e
n
t
a
t
i
o
n
W
a
v
e
l
e
t
E
x
t
r
a
c
t
i
o
n
C
o
n
c
e
n
t
r
a
t
i
o
n
T
h
i
n
g
k
i
n
g
F
o
r
w
a
r
d
T
r
a
i
n
i
n
g
D
a
t
a
6
4
0
0
s
e
t
L
S
T
M
L
a
y
e
r
1
(
R
e
l
u
)
D
r
o
p
o
u
t
L
a
y
e
r
L
S
T
M
L
a
y
e
r
2
(
S
i
g
m
o
i
d
)
D
e
n
s
e
L
a
y
e
r
(
S
i
g
m
o
i
d
)
O
u
t
p
u
t
I
n
p
u
t
D
a
t
a
T
r
a
i
n
i
n
g
E
E
G
S
i
g
n
a
l
2
5
6
d
a
t
a
p
o
i
n
t
s
e
a
c
h
s
e
g
m
e
n
t
e
v
e
r
y
c
h
a
n
n
e
l
2
5
6
d
a
t
a
p
o
i
n
t
s
e
a
c
h
s
e
g
m
e
n
t
e
v
e
r
y
c
h
a
n
n
e
l
C
o
n
c
e
n
t
r
a
t
i
o
n
T
h
i
n
g
k
i
n
g
T
u
r
n
R
i
g
h
t
C
o
n
c
e
n
t
r
a
t
i
o
n
T
h
i
n
g
k
i
n
g
T
u
r
n
L
e
f
t
N
o
t
C
o
n
c
e
n
t
r
a
t
i
o
n
T
h
i
n
g
k
i
n
g
F
o
r
w
a
r
d
N
o
t
C
o
n
c
e
n
t
r
a
t
I
o
n
T
h
i
n
k
i
n
g
I
d
l
e
N
o
t
C
o
n
c
e
n
t
r
a
t
i
o
n
T
h
i
n
g
k
i
n
g
T
u
r
n
R
i
g
h
t
N
o
t
C
o
n
c
e
n
t
r
a
t
i
o
n
T
h
i
n
g
k
i
n
g
T
u
r
n
L
e
f
t
C
o
n
c
e
n
t
r
a
t
i
o
n
T
h
i
n
g
k
i
n
g
I
d
l
e
Fig
u
r
e
1
.
B
C
I
b
ased
o
n
f
o
c
u
s
an
d
MI
v
a
r
iab
le
o
f
E
E
G
s
ig
n
a
l
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
,
Vo
l.
18
,
No
.
5
,
Octo
b
e
r
2
0
2
0
:
2
7
4
8
-
275
6
2750
I
m
a
g
i
n
e
M
o
v
i
n
g
G
o
F
o
r
w
a
r
d
I
m
a
g
i
n
e
M
o
v
i
n
g
T
u
r
n
L
e
f
t
I
m
a
g
i
n
e
M
o
v
i
n
g
T
u
r
n
R
i
g
h
t
I
m
a
g
i
n
e
M
o
t
i
o
n
l
e
s
s
I
n
s
t
r
u
c
t
i
o
n
I
n
s
t
r
u
c
t
i
o
n
I
n
s
t
r
u
c
t
i
o
n
I
n
s
t
r
u
c
t
i
o
n
0
1
0
3
0
4
0
6
0
7
0
9
0
1
0
0
1
2
0
S
e
c
o
n
d
s
V
i
s
u
a
l
I
n
s
t
r
u
c
t
i
o
n
O
n
e
C
a
t
e
g
o
r
y
(
3
0
S
e
c
o
n
d
s
)
I
n
s
t
r
u
c
t
i
o
n
S
e
g
m
e
n
1
S
e
g
m
e
n
2
S
e
g
m
e
n
3
S
e
g
m
e
n
4
S
e
g
m
e
n
5
S
e
g
m
e
n
6
S
e
g
m
e
n
7
S
e
g
m
e
n
8
S
e
g
m
e
n
9
S
e
g
m
e
n
1
0
0
1
0
1
2
1
4
1
6
1
8
2
0
2
2
2
4
2
6
2
8
3
0
S
e
c
o
n
d
s
Fig
u
r
e
2
.
R
ec
o
r
d
in
g
s
ce
n
ar
io
2
.
2.
Wa
v
elet
ex
t
ra
ct
io
n
T
h
e
wav
elet
tr
an
s
f
o
r
m
ca
n
ex
t
r
ac
t th
e
n
ee
d
ed
s
ig
n
al
co
m
p
o
n
en
ts
,
h
en
ce
r
ed
u
cin
g
th
e
n
u
m
b
er
o
f
d
ata
with
o
u
t
lo
s
in
g
im
p
o
r
tan
t
in
f
o
r
m
atio
n
.
B
esid
es,
th
is
m
eth
o
d
is
s
u
itab
le
f
o
r
n
o
n
-
s
tatio
n
ar
y
s
ig
n
als.
T
h
e
o
u
tp
u
t
o
f
th
e
wav
elet
tr
a
n
s
f
o
r
m
in
to
a
tim
e
-
d
o
m
ain
allo
ws its
ap
p
li
ca
tio
n
as a
p
r
e
-
m
o
d
el
[
6
]
.
W
av
elet
tr
an
s
f
o
r
m
atio
n
h
as
two
m
ain
p
r
o
ce
s
s
es,
s
p
ec
if
ica
lly
d
ec
o
m
p
o
s
itio
n
,
th
at
ex
tr
ac
t
a
s
ig
n
al
in
t
o
a
s
p
ec
if
ic
f
r
eq
u
e
n
cy
a
n
d
r
ec
o
n
s
tr
u
ctio
n
th
at
r
ec
o
m
b
in
e
ex
tr
ac
ted
s
ig
n
als
in
to
th
eir
o
r
ig
in
al
f
o
r
m
[
2
7
]
.
W
av
elet
w
o
r
k
s
in
a
co
n
v
o
lu
ti
o
n
s
ig
n
al
with
m
o
th
er
w
av
elet
.
Var
io
u
s
f
o
r
m
s
o
f
wav
elets
u
s
ed
f
o
r
E
E
G
s
ig
n
al
ex
tr
ac
tio
n
f
r
o
m
p
r
ev
io
u
s
s
tu
d
ies,
s
u
ch
a
s
Dau
b
ec
h
ies
Haa
r
an
d
Sy
m
m
let
.
T
h
e
r
esear
ch
e
r
s
d
id
n
o
t
s
p
ec
if
ically
m
en
tio
n
th
e
b
asic
s
h
ap
e
o
f
th
e
wav
elet
th
at
g
iv
es
g
o
o
d
ac
cu
r
ac
y
.
B
u
t
in
g
en
er
al,
th
e
asy
m
m
etr
ic
Dau
b
ec
h
ies
[
7
]
,
co
m
b
in
e
o
f
Dau
b
ec
h
ies
an
d
Sy
m
let
[
2
8
]
an
d
Sy
m
m
let
[
2
9
]
.
B
o
th
f
o
r
m
s
ar
e
c
o
m
p
ati
b
le
wit
h
E
E
G
s
ig
n
al
ch
ar
ac
ter
is
tics
.
On
e
ty
p
e
o
f
wav
elet
tr
an
s
f
o
r
m
is
wav
elet
p
ac
k
et
d
ec
o
m
p
o
s
itio
n
(
W
PD)
.
W
av
elet
p
ac
k
ets
ar
e
lin
ea
r
co
m
b
in
atio
n
s
o
f
wav
elet
f
u
n
ctio
n
s
[
9
]
.
A
wav
e
let
f
u
n
ctio
n
h
as
th
r
ee
in
d
ices,
j:
in
d
ex
s
ca
le
(
in
teg
er
)
,
k
:
t
r
an
s
latio
n
co
ef
f
icien
t
,
n
: o
s
cillatio
n
p
ar
am
eter
a
n
d
t i
s
tim
e
as (
1
)
.
,
=
2
/
2
(
2
−
)
(
1
)
T
h
e
wav
elet
p
ac
k
et
f
u
n
ctio
n
s
ar
e
a
s
ca
lin
g
f
u
n
ctio
n
)
(
t
an
d
t
h
e
m
o
th
er
wav
elet
f
u
n
ctio
n
)
(
t
.
W
av
elet
p
ac
k
et
f
u
n
ctio
n
s
with
h
ig
h
er
f
ilter
ar
e:
0
,
0
2
=
√
2
∑
ℎ
(
)
1
,
2
(
2
−
)
(
2
)
0
,
0
2
+
1
=
√
2
∑
(
)
1
,
2
(
2
−
)
(
3
)
T
h
e
f
ac
to
r
h
(
k
)
an
d
g
(
k
)
in
d
ic
ate
q
u
ad
r
atu
r
e
m
ir
r
o
r
e
x
tr
ac
tio
n
[
3
0
]
.
T
h
e
v
al
u
e
(
h
(
k
)
an
d
g
(
k
)
r
elate
d
to
th
e
s
ca
lin
g
f
u
n
ctio
n
an
d
th
e
m
o
th
er
wav
elet
f
u
n
ctio
n
.
T
h
e
in
n
er
p
r
o
d
u
ct
s
ig
n
al
f
(
t)
w
ith
wav
elet
p
ac
k
et
f
u
n
ctio
n
s
in
a
r
an
g
e
o
f
t sh
o
w
(
4
)
:
,
=
(
)
*
,
=
∑
(
)
,
(
2
−
)
(
4
)
Fo
r
o
r
ig
in
al
s
ig
n
al
S,
th
e
lef
t
-
s
id
e
is
o
b
tain
ed
in
lo
w
p
ass
f
i
lter
h
(
k
)
as
an
a
p
p
r
o
x
im
atio
n
co
ef
f
icien
t
an
d
th
e
r
ig
h
t
s
id
e
as
h
ig
h
p
ass
f
ilter
g
(
k
)
o
r
d
etail.
I
n
(
6
)
s
h
o
wed
th
e
s
ca
le,
tr
an
s
latio
n
,
an
d
o
s
cillatio
n
v
alu
es
.
I
n
(
4
)
,
th
e
s
ig
n
al
ca
n
b
e
d
ec
o
m
p
o
s
ed
in
t
o
a
s
ca
le
f
ac
to
r
j
in
a
p
a
r
ticu
lar
f
r
eq
u
e
n
cy
,
eith
er
h
ig
h
o
r
l
o
w.
I
n
th
is
s
tu
d
y
,
u
s
in
g
th
e
s
tan
d
ar
d
f
o
r
m
Dau
b
ec
h
ies
4
,
wh
ich
co
n
s
is
ts
o
f
f
o
u
r
lo
w
-
p
ass
f
ilter
co
ef
f
icien
ts
[
2
9
]
.
W
av
elets
d
ec
o
m
p
o
s
e
s
ig
n
als in
to
s
p
ec
if
i
c
f
r
eq
u
en
cy
r
an
g
es,
s
u
ch
as
d
e
lta,
alp
h
a,
b
eta,
th
eta,
an
d
g
am
m
a
wa
v
es,
s
u
ch
as
Fig
u
r
e
3
.
2
.
3
.
Rec
urre
nt
neura
l net
wo
rk
s
R
NN
is
o
n
e
m
eth
o
d
u
s
ed
in
D
ee
p
L
ea
r
n
in
g
f
o
r
s
eq
u
e
n
tial
d
ata
[
3
1
]
,
b
y
lo
o
p
in
g
to
s
to
r
e
i
n
f
o
r
m
atio
n
f
r
o
m
th
e
p
ast
[
3
2
]
.
T
h
is
co
n
f
i
g
u
r
atio
n
is
s
h
o
wn
in
Fig
u
r
e
4
.
R
NN
i
s
ac
tiv
ated
with
a
f
u
n
ctio
n
s
u
ch
as si
g
m
o
id
as
Fig
u
r
e
5
.
R
NN
h
as
t
h
e
p
r
o
b
lem
s
o
f
s
h
o
r
t
m
em
o
r
y
,
s
o
it
n
ee
d
s
co
n
tr
o
l
to
f
o
r
g
et
s
o
m
e
p
ar
ts
t
h
r
o
u
g
h
o
u
t
th
e
g
ate.
So
m
e
o
f
th
e
m
eth
o
d
s
ar
e
g
ated
r
ec
u
r
r
en
t
u
n
it
(
GR
U)
,
b
ac
k
p
r
o
p
ag
atio
n
th
r
o
u
g
h
tim
e
(
B
PTT
)
,
an
d
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
L
STM
)
.
T
h
is
r
esear
ch
u
s
ed
t
h
e
L
STM
g
ate
to
o
v
e
r
co
m
e
s
h
o
r
t
-
ter
m
m
em
o
r
y
p
r
o
b
lem
s
o
r
o
f
ten
ca
lled
v
a
n
is
h
in
g
g
r
ad
ien
t
[
1
4
]
, w
h
ich
h
as a
in
c
r
ea
s
e
in
ca
p
ab
ilit
y
f
r
o
m
a
s
in
g
le
lay
er
[
3
3
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
B
r
a
in
-
co
mp
u
ter in
terf
a
ce
o
f fo
cu
s
a
n
d
mo
t
o
r
ima
g
ery
u
s
in
g
w
a
ve
let
a
n
d
… (
E
s
mera
ld
a
C
.
Dja
ma
l
)
2751
1
-
6
4
H
z
1
-
3
2
H
z
3
3
-
6
4
H
z
1
-
1
6
H
z
1
-
8
H
z
9
-
1
6
H
z
1
7
-
3
2
H
z
1
7
-
2
4
H
z
2
5
-
3
2
H
z
9
-
1
2
H
z
1
3
-
1
6
H
z
1
-
4
H
z
5
-
8
H
z
1
3
-
1
4
H
z
1
5
-
1
6
H
z
5
-
6
H
z
7
-
8
H
z
2
5
-
2
8
H
z
2
9
-
3
2
H
z
2
9
-
3
0
H
z
3
1
-
3
2
H
z
B
e
t
a
A
l
f
a
/
M
u
T
e
t
a
A
D
A
A
A
A
A
A
A
A
A
A
D
A
D
D
A
D
D
D
A
A
D
A
A
A
D
A
D
A
A
D
D
A
A
D
D
D
D
A
A
A
D
A
A
A
A
D
D
A
D
D
D
A
A
A
D
A
A
A
D
D
A
A
D
D
A
A
A
D
D
D
G
a
m
m
a
3
3
-
4
8
H
z
D
D
D
A
4
9
-
6
4
H
z
3
3
-
4
0
H
z
D
A
D
4
1
-
4
8
H
z
D
A
A
Fig
u
r
e
3
.
W
av
elet
m
u
ltil
ev
el
Fig
u
r
e
4
.
R
ec
u
r
r
en
t n
e
u
r
al
n
et
wo
r
k
ar
ch
it
ec
tu
r
e
Fig
u
r
e
5
.
L
STM
ce
ll
ar
c
h
itectu
r
e
T
h
e
L
STM
n
etwo
r
k
co
n
s
is
ts
o
f
m
o
d
u
les
with
r
e
p
etitiv
e
p
r
o
c
ess
in
g
,
as
in
Fig
u
r
e
4
.
Me
m
o
r
y
in
L
STM
is
ca
lled
ce
lls
th
at
tak
e
in
p
u
t
f
r
o
m
th
e
p
r
ev
i
o
u
s
s
tate
(
h
t
-
1
)
an
d
cu
r
r
en
t
in
p
u
t
(
x
t
)
.
T
h
e
co
llec
tio
n
o
f
ce
lls
d
ec
id
e
s
wh
at
will
b
e
s
to
r
e
d
in
m
em
o
r
y
an
d
wh
at
will
b
e
r
em
o
v
ed
f
r
o
m
m
em
o
r
y
.
L
STM
c
o
m
b
in
es
th
e
p
r
ev
io
u
s
s
tate,
cu
r
r
en
t
m
e
m
o
r
y
,
an
d
in
p
u
t.
L
STM
h
as
th
r
ee
g
ates,
p
ar
ticu
la
r
ly
th
e
f
o
r
g
et
g
ate,
to
d
ete
r
m
in
e
wh
ich
elim
in
atin
g
in
f
o
r
m
ati
o
n
f
r
o
m
t
h
e
ce
ll
u
s
in
g
th
e
s
ig
m
o
i
d
lay
e
r
[
3
4
]
.
I
n
(
5
)
with
th
e
ac
tiv
atio
n
f
u
n
ctio
n
u
s
ed
is
th
e
r
elea
s
e
s
h
o
wn
in
(
8
)
.
=
(
.
[
ℎ
−
1
,
]
+
)
(
5
)
(
)
=
ma
x
(
0
,
)
(
6
)
T
h
e
s
ec
o
n
d
g
ate
is
th
e
in
p
u
t
g
ate
(
i)
,
wh
ich
o
f
th
e
s
ig
m
o
i
d
lay
er
(
σ
)
will
b
e
u
p
d
ated
,
a
n
d
tan
h
o
f
th
e
lay
er
will
b
e
f
o
r
m
u
lated
as
a
v
ec
to
r
o
f
th
e
u
p
d
ate
d
v
al
u
e.
I
t
ca
n
b
e
s
ee
n
at
(
7
)
an
d
(
8
)
wh
er
e
x
t
is
in
p
u
t
f
o
r
ea
ch
cu
r
r
e
n
t step
tim
e.
At
th
is
lay
er
,
a
v
ec
to
r
o
f
u
p
d
ated
v
alu
es will b
e
p
r
o
d
u
ce
d
[
3
5
]
.
=
(
.
[
ℎ
−
1
,
]
+
)
(
7
)
̃
=
ta
n
h
(
.
[
ℎ
−
1
,
]
+
)
(
8
)
T
h
en
th
e
ce
lls
o
f
(
7
)
,
(
8)
will b
e
u
p
d
ated
u
s
in
g
(
9
).
=
∗
−
1
+
∗
̃
(
9
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
,
Vo
l.
18
,
No
.
5
,
Octo
b
e
r
2
0
2
0
:
2
7
4
8
-
275
6
2752
Fin
ally
,
th
e
o
u
tp
u
t
g
a
te
will
b
e
ca
lcu
lated
b
ased
o
n
ce
ll
u
p
d
at
es,
an
d
th
e
s
ig
m
o
id
la
y
er
lo
o
k
s
lik
e
(
1
0
)
a
n
d
(
1
1
)
.
=
(
.
[
ℎ
−
1
,
]
+
)
(
10
)
ℎ
=
∗
ta
n
h
(
)
(
1
1)
wh
er
e
is
th
e
s
ig
m
o
i
d
ac
tiv
ati
o
n
f
u
n
ctio
n
,
an
d
tan
h
as
t
h
e
t
an
h
ac
tiv
atio
n
f
u
n
ctio
n
u
s
ed
f
o
r
t
h
e
r
esu
lts
o
f
m
u
ltip
ly
in
g
th
e
weig
h
t o
f
ea
c
h
g
ate,
n
am
ely
W
f
, W
t
,
W
c
,
W
o
with
in
p
u
t v
alu
es a
n
d
ad
d
ed
b
ias in
clu
d
in
g
b
f,
b
i
,
b
c
,
b
o
.
Gate
u
s
ed
is
g
ate
in
p
u
t
i
t
,
f
o
r
g
et
f
t
,
,
an
d
o
u
tp
u
t
o
t
.
E
ac
h
p
as
s
in
g
g
ate
will
b
e
s
ea
r
ch
ed
f
o
r
h
id
d
en
s
tate
ca
n
d
id
ates
̃
o
b
tain
e
d
f
r
o
m
th
e
g
ate
ca
lc
u
latio
n
with
th
e
cu
r
r
en
t
h
id
d
en
s
tate
f
u
r
th
er
m
o
r
e
,
th
e
p
r
e
v
io
u
s
l
y
h
id
d
en
s
tate
C
t
-
1
to
p
r
o
d
u
ce
th
e
latest h
id
d
en
s
tate,
wh
ich
is
u
s
ed
as th
e
o
u
tp
u
t o
f
t
h
e
h
id
d
e
n
lay
er
.
I
n
th
e
id
en
tific
atio
n
lay
er
,
th
e
r
e
is
an
ef
f
o
r
t
to
m
i
n
im
ize
th
e
d
if
f
er
en
ce
b
etwe
en
th
e
tar
g
et
o
u
tp
u
t
an
d
th
e
o
u
tp
u
t
o
f
co
m
p
u
tatio
n
al
r
e
s
u
lts
.
Ob
jectiv
e
f
u
n
ctio
n
s
ar
e
o
f
ten
r
ef
e
r
r
ed
to
as
co
s
t
f
u
n
ctio
n
s
o
r
lo
s
s
f
u
n
ctio
n
s
so
-
ca
lled
"lo
s
s
".
On
e
o
f
th
e
lo
s
s
f
u
n
ctio
n
s
th
at
ca
n
b
e
u
s
ed
is
cr
o
s
s
-
en
tr
o
p
y
,
as
in
(
1
2
)
.
W
h
e
r
e
lo
s
s
is
a
d
is
tan
ce
,
S
is
th
e
r
esu
lt
o
f
th
e
ac
tiv
ati
o
n
f
u
n
ctio
n
,
an
d
L
is
th
e
tar
g
et
o
f
ea
c
h
class
lab
el.
A
lo
s
s
f
u
n
ctio
n
is
u
s
ed
t
o
m
ea
s
u
r
e
co
n
v
er
g
en
ce
in
th
e
le
ar
n
in
g
p
r
o
ce
s
s
.
(
)
(
)
−
=
i
i
i
S
L
L
S
L
o
s
s
l
o
g
,
(
1
2
)
I
n
m
ac
h
in
e
lear
n
in
g
s
u
ch
R
N
N,
it
is
ess
en
tial
to
s
et
in
p
u
t
f
ea
tu
r
es.
T
h
is
r
esear
ch
u
s
ed
MI
an
d
f
o
cu
s
v
ar
iab
le,
s
o
alp
h
a,
m
u
,
b
eta,
a
n
d
g
am
m
a
(
3
2
-
4
0
Hz)
wav
es
r
elate
th
at.
B
ased
o
n
Fig
u
r
e
3
,
we
g
o
t
t
h
e
wav
es o
f
f
o
u
r
c
h
an
n
els,
as
s
h
o
wn
in
T
ab
le
1
.
T
h
e
B
C
I
wo
r
k
s
ev
e
r
y
two
s
ec
o
n
d
s
.
W
h
ile
th
e
R
NN
co
n
f
ig
u
r
atio
n
is
as
s
h
o
wn
in
T
ab
le
2
.
I
n
th
e
f
ir
s
t
m
o
d
el,
th
e
n
u
m
b
er
o
f
n
e
u
r
o
n
s
o
f
th
e
m
o
d
el
is
f
ac
ed
with
th
e
s
am
e
am
o
u
n
t
as
th
e
in
p
u
t
v
ec
to
r
with
a
two
-
d
i
m
en
s
io
n
al
s
h
ap
e
th
at
a
p
p
lies
th
e
r
etu
r
n
s
eq
u
en
ce
t
o
th
e
s
ec
o
n
d
L
STM
m
o
d
el.
T
h
e
L
STM
m
o
d
el
h
as
a
o
n
e
-
d
im
en
s
io
n
al
v
ec
to
r
,
wh
ich
r
es
u
lts
in
th
e
d
en
s
e
lay
er
,
wh
ich
h
as
eig
h
t
n
e
u
r
o
n
s
,
ac
co
r
d
in
g
to
th
e
n
u
m
b
e
r
o
f
class
es p
r
o
d
u
ce
d
.
T
ab
le
1
.
R
NN
f
ea
tu
r
es
o
f
B
C
I
No
C
o
m
p
o
n
e
n
t
N
u
mb
e
r
o
f
p
o
i
n
t
s
D
e
scri
p
t
i
o
n
A
l
l
c
h
a
n
n
e
l
1
A
l
p
h
a
,
B
e
t
a
,
G
a
mm
a
(
9
-
4
0
H
z
)
12
8
F
o
u
r
-
c
h
a
n
n
e
l
5
1
2
2
M
u
(
9
-
1
4
H
z
)
24
F
C
5
a
n
d
F
C
6
o
n
l
y
48
To
t
a
l
5
6
0
T
ab
le
2
.
Ar
c
h
itectu
r
e
m
o
d
el
r
ec
u
r
r
en
t
n
eu
r
al
n
etwo
r
k
s
M
o
d
e
l
N
e
u
r
o
n
O
u
t
p
u
t
s
h
a
p
e
LSTM
5
6
0
1
,
5
6
0
D
r
o
p
o
u
t
0
.
2
1
,
5
6
0
D
e
n
se
8
8
3.
RE
SU
L
T
S
A
ND
AN
AL
Y
SI
S
T
h
e
ex
p
er
im
en
t
was
ca
r
r
ied
o
u
t
b
y
co
m
p
ar
in
g
th
e
ef
f
e
ct
o
f
u
s
in
g
wav
elet
as
th
e
e
x
tr
ac
tio
n
o
f
alp
h
a,
m
u
,
b
eta,
an
d
g
a
m
m
a
wav
es.
E
x
p
er
im
en
ts
wer
e
also
ca
r
r
ied
o
u
t
o
n
th
e
m
o
d
el
in
ter
m
s
o
f
p
r
o
v
id
in
g
th
e
h
ig
h
est
ac
cu
r
ac
y
an
d
co
n
s
id
er
i
n
g
th
e
co
m
p
u
tatio
n
al
tim
e
o
f
lear
n
i
n
g
.
I
d
e
n
tific
atio
n
in
v
o
lv
es
ei
g
h
t
class
es,
n
am
ely
"Fo
r
war
d
",
"Ri
g
h
t",
"L
ef
t",
an
d
"Silen
t"
ea
ch
in
f
o
cu
s
an
d
n
o
t.
I
n
u
s
in
g
B
C
I
,
p
er
f
o
r
m
an
ce
d
ep
en
d
s
o
n
tr
an
s
latin
g
v
ar
iab
les
f
r
o
m
th
e
E
E
G
s
ig
n
al
b
ein
g
r
ev
iewe
d
.
I
d
en
tific
atio
n
p
er
f
o
r
m
an
ce
is
v
er
y
d
ep
en
d
en
t
o
n
th
e
u
s
e
o
f
e
x
tr
ac
tio
n
m
eth
o
d
s
.
T
h
er
ef
o
r
e,
test
in
g
b
eg
i
n
s
with
wav
elet
p
er
f
o
r
m
an
ce
.
3
.
1
.
Wa
v
elet
ex
t
ra
ct
io
n
W
av
elet
ex
tr
ac
tio
n
u
s
es Da
u
b
ec
h
ies 4
at
9
-
40
Hz,
wh
ic
h
h
as
b
ee
n
n
o
r
m
alize
d
as in
Fig
u
r
e
6
.
W
av
elet
ex
tr
ac
tio
n
is
s
h
o
wn
in
b
lu
e
co
m
p
ar
ed
to
t
h
e
o
r
ig
i
n
al
s
ig
n
al
u
s
in
g
o
r
an
g
e.
T
h
e
E
E
G
s
ig
n
al
af
ter
g
o
in
g
th
r
o
u
g
h
wav
elet
ex
tr
ac
tio
n
is
m
o
r
e
s
ta
b
le
b
ec
au
s
e
it
is
ad
ju
s
ted
to
t
h
e
wav
es
wh
ich
ar
e
in
t
h
e
f
r
e
q
u
en
cy
r
an
g
e
.
T
h
en
elim
in
ate
u
n
u
s
ed
wav
es
an
d
s
ig
n
al
n
o
is
e.
T
h
e
r
esu
lts
o
f
e
ac
h
ch
an
n
el
ar
e
s
to
r
ed
s
eq
u
en
tially
in
to
in
p
u
t
v
ec
t
o
r
s
.
3
.
2
.
Co
m
pa
re
d bet
wee
n o
ptim
iza
t
io
n
m
o
del
T
h
is
s
tu
d
y
u
s
ed
th
r
ee
-
weig
h
t
c
o
r
r
ec
tio
n
m
o
d
els
th
at
ar
e
ad
ap
t
iv
e
m
o
m
en
t
esti
m
atio
n
(
Ad
am
)
,
ad
ap
tiv
e
lear
n
in
g
esti
m
atio
n
(
Ad
aDe
lta
)
,
an
d
s
to
ch
asti
c
g
r
ad
ien
t
d
esc
en
t
(
SGD)
.
W
e
ex
p
er
ien
ce
o
p
t
im
izer
m
o
d
els
an
d
op
tim
al
lear
n
in
g
p
ar
am
eter
s
th
at
h
ig
h
er
ac
c
u
r
ac
y
an
d
s
h
o
r
t
est
tim
e
co
m
p
u
tin
g
.
Ad
am
h
a
s
a
f
ast
co
n
v
er
g
en
c
e
p
r
o
p
er
t
y
,
b
u
t
it
is
o
n
ly
u
n
s
tab
l
e
d
u
e
to
v
er
y
r
ap
id
er
r
o
r
r
e
d
u
ctio
n
.
B
esid
es
th
e
o
p
tim
izer
m
o
d
el,
we
co
m
p
ar
e
d
u
s
in
g
wav
elet
an
d
with
o
u
t
wa
v
elet,
as
in
F
ig
u
r
e
7
o
f
Acc
u
r
ac
y
an
d
Fig
u
r
e
8
o
f
lo
s
s
es
v
alu
e
.
T
h
er
e
a
r
e
th
r
ee
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
B
r
a
in
-
co
mp
u
ter in
terf
a
ce
o
f fo
cu
s
a
n
d
mo
t
o
r
ima
g
ery
u
s
in
g
w
a
ve
let
a
n
d
… (
E
s
mera
ld
a
C
.
Dja
ma
l
)
2753
m
o
d
els.
Ad
am
a
n
d
Ad
aDe
lta
ar
e
co
n
v
er
g
e
n
t
o
f
1
0
0
ep
o
c
h
lear
n
in
g
,
ex
ce
p
t
f
o
r
th
e
S
GD
m
o
d
el.
So
th
at
1
0
0
ep
o
c
h
s
ar
e
o
p
tim
al
en
o
u
g
h
,
e
x
ce
p
t
with
SGD
with
5
0
0
e
p
o
ch
ad
d
itio
n
.
E
ac
h
c
o
l
o
r
in
Fig
u
r
e
7
an
d
Fig
u
r
e
8
in
d
icate
s
th
e
test
in
g
o
f
tr
ain
in
g
d
ata
with
wav
elet
(
b
lu
e)
,
tr
ain
in
g
d
ata
with
o
u
t
wav
elet
(
g
r
ee
n
)
,
non
-
tr
ai
n
in
g
/v
alid
atio
n
d
ata
with
wav
elet
(
o
r
an
g
e)
,
an
d
v
alid
atio
n
d
ata
with
o
u
t
wav
elet
(
r
ed
)
.
Fro
m
th
e
th
r
e
e
m
o
d
els
s
h
o
wn
in
Fig
u
r
e
7
th
a
t
u
s
in
g
wav
elet
ca
n
in
c
r
ea
s
e
ac
cu
r
ac
y
an
d
r
e
d
u
ce
c
o
m
p
u
tin
g
tim
e.
T
h
e
e
x
ac
t
v
alu
es
o
f
th
e
th
r
ee
m
o
d
els
a
r
e
s
h
o
wn
in
T
ab
le
3
.
L
ik
ewise,
th
e
v
alu
e
o
f
L
o
s
s
es
f
r
o
m
u
s
in
g
wav
elets
f
o
r
th
e
th
r
ee
m
o
d
els
g
en
er
ally
d
ec
r
ea
s
es.
T
h
is
r
esu
lt
to
ld
th
at
wav
elet
c
o
u
ld
im
p
r
o
v
e
ac
c
u
r
ac
y
b
y
r
e
d
u
cin
g
th
e
n
o
n
-
s
tatio
n
ar
y
p
r
o
p
e
r
ties
o
f
E
E
G
s
ig
n
als.
Fig
u
r
e
6
.
W
av
elet
ex
tr
ac
tio
n
(
a)
(
b
)
(
c)
Fig
u
r
e
7
.
Acc
u
r
ac
y
o
f
th
e
o
p
ti
m
izer
m
o
d
el
;
(
a
)
Ad
am
,
(
b
)
A
d
aDe
lta
(
c)
SGD
T
h
e
ac
cu
r
ac
y
o
f
th
e
t
h
r
ee
m
o
d
els
is
r
elativ
ely
th
e
s
am
e,
m
a
in
ly
b
etwe
en
7
6
-
8
0
%
f
o
r
v
ali
d
atio
n
d
ata
wh
ile
1
0
0
%
f
o
r
tr
ain
in
g
d
ata
.
E
v
en
s
o
,
th
e
h
ig
h
est
Ad
aD
elta
m
o
d
el
is
7
9
.
8
1
%.
T
h
e
e
x
citin
g
th
in
g
is
th
at
th
e
Ad
am
m
o
d
el
q
u
ic
k
ly
co
r
r
ec
ts
weig
h
ts
,
wh
ich
ca
u
s
e
s
ac
cu
r
ac
y
to
in
cr
ea
s
e
r
ap
id
l
y
an
d
lo
s
s
es
to
d
ec
r
ea
s
e
at
th
e
b
eg
i
n
n
in
g
o
f
th
e
iter
ati
o
n
.
Ho
wev
e
r
,
c
o
n
d
itio
n
s
o
f
s
m
all
f
lu
ctu
atio
n
s
co
n
tin
u
e
at
th
e
en
d
o
f
th
e
ep
o
ch
.
W
h
ile
th
e
Ad
aDe
lta
m
o
d
el
ten
d
s
to
b
e
s
tab
le
at
th
e
en
d
o
f
th
e
ep
o
ch
,
it a
c
h
iev
es lo
n
g
e
r
th
a
n
th
e
Ad
am
m
o
d
el.
B
u
t
it
is
u
n
d
er
s
to
o
d
th
at
th
e
w
eig
h
t
co
r
r
ec
tio
n
m
et
h
o
d
o
f
th
e
Ad
am
m
o
d
el
ten
d
s
to
ju
m
p
li
k
e
a
b
all
t
h
at
r
o
lls
ea
s
ily
.
B
esid
es,
th
e
SGD
m
o
d
el
h
ad
alm
o
s
t
n
o
r
ip
p
les o
f
in
s
tab
ilit
y
d
u
r
in
g
th
e
tr
ain
in
g
.
B
u
t
th
e
d
is
ad
v
a
n
tag
es
r
eq
u
ir
e
lo
n
g
er
i
ter
atio
n
s
.
E
v
en
in
th
e
5
0
0
th
e
p
o
ch
,
th
e
ac
cu
r
a
cy
is
s
till
in
cr
ea
s
in
g
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
,
Vo
l.
18
,
No
.
5
,
Octo
b
e
r
2
0
2
0
:
2
7
4
8
-
275
6
2754
(
a)
(
b
)
(
c)
Fig
u
r
e
8
.
L
o
s
s
es
o
f
th
e
o
p
tim
i
ze
r
m
o
d
el:
(
a
)
Ad
am
,
(
b
)
A
d
a
Delta
(
c)
SGD
T
ab
le
3
.
C
o
m
p
a
r
is
o
n
o
f
lo
s
s
an
d
ac
cu
r
ac
y
u
s
in
g
Ad
am
,
Ad
a
Delta
,
an
d
SGD
m
o
d
el
M
o
d
e
l
e
p
o
c
h
s
W
a
v
e
l
e
t
W
i
t
h
o
u
t
W
a
v
e
l
e
t
Tr
a
i
n
D
a
t
a
V
a
l
i
d
a
t
i
o
n
D
a
t
a
Tr
a
i
n
D
a
t
a
V
a
l
i
d
a
t
i
o
n
D
a
t
a
A
c
c
u
r
a
c
y
Lo
ss
A
c
c
u
r
a
c
y
Lo
ss
Ac
c
u
r
a
c
y
Lo
ss
A
c
c
u
r
a
c
y
Lo
ss
A
d
a
m
1
0
0
9
8
,
5
0
,
0
4
2
2
7
8
,
8
4
1
,
0
7
6
7
1
0
0
,
0
0
,
0
0
0
3
7
6
.
9
2
1
,
4
2
1
8
A
d
a
D
e
l
t
a
1
0
0
1
0
0
,
0
0
,
0
0
0
5
7
9
,
8
1
1
,
3
5
6
1
1
0
0
,
0
0
,
0
0
0
2
7
4
,
0
4
1
,
3
2
4
0
S
G
D
1
0
0
1
7
.
7
2
.
0
5
1
2
1
7
.
3
1
2
.
0
1
5
8
3
2
.
0
5
1
.
9
6
2
2
2
2
.
1
2
2
.
0
3
0
4
S
G
D
5
0
0
1
0
0
,
0
0
.
0
3
4
9
7
7
,
8
2
0
.
9
2
8
1
1
0
0
,
0
0.
0
0
5
2
7
7
.
8
8
1
.
1
6
1
2
Fro
m
v
ar
io
u
s
ex
p
e
r
im
en
ts
,
s
h
o
wed
th
at
R
NN
an
d
wav
elet
c
o
u
ld
b
e
u
s
ed
to
s
u
p
p
o
r
t
B
C
I
u
s
in
g
MI
a
n
d
Fo
cu
s
v
ar
iab
les
with
an
ac
cu
r
ac
y
o
f
alm
o
s
t
8
0
%
n
o
n
-
tr
ain
i
n
g
d
ata.
T
h
e
f
u
tu
r
e
ex
p
er
im
e
n
t
n
ee
d
s
to
lo
o
k
o
u
t
th
e
co
n
f
ig
u
r
atio
n
o
f
th
e
in
p
u
t
f
e
atu
r
es
an
d
ch
a
n
n
el
u
s
ag
e
o
f
th
e
E
E
G
s
ig
n
al.
So
th
at
ca
n
im
p
r
o
v
e
ac
c
u
r
ac
y
.
T
h
is
r
esear
ch
g
av
e
th
e
d
u
r
atio
n
o
f
co
m
p
u
tatio
n
al
lear
n
in
g
u
s
in
g
Ad
a
m
,
Ad
aDe
lta
,
a
n
d
SGD
o
p
tim
izatio
n
with
s
ev
er
al
co
n
f
ig
u
r
atio
n
s
.
A
c
o
m
p
ar
is
o
n
o
f
th
e
len
g
th
o
f
tim
e
f
r
o
m
th
e
t
h
r
ee
op
tim
izatio
n
s
ca
n
b
e
s
e
en
in
T
ab
le
4
with
s
ev
er
al
co
n
f
ig
u
r
atio
n
s
in
1
0
0
ep
o
ch
s
.
T
ab
le
4
.
C
o
m
p
u
tin
g
tim
e
o
f
1
0
0
ep
o
ch
s
M
o
d
e
l
M
e
t
h
o
d
s
Le
a
r
n
i
n
g
t
i
me
(
sec
o
n
d
)
A
d
a
m
W
i
t
h
W
a
v
e
l
e
t
5
1
6
W
i
t
h
o
u
t
W
a
v
e
l
e
t
5
4
0
A
d
a
D
e
l
t
a
W
i
t
h
W
a
v
e
l
e
t
6
0
6
W
i
t
h
o
u
t
W
a
v
e
l
e
t
6
4
2
S
G
D
W
i
t
h
W
a
v
e
l
e
t
3
7
0
W
i
t
h
o
u
t
W
a
v
e
l
e
t
3
8
0
4.
CO
NCLU
SI
O
N
T
h
is
r
esear
ch
s
h
o
wed
th
at
b
r
a
in
-
co
m
p
u
ter
in
ter
f
ac
e
c
o
u
ld
u
s
e
m
o
to
r
im
ag
er
y
an
d
f
o
cu
s
v
ar
iab
le
o
f
E
E
G
s
ig
n
al
to
m
o
v
e
d
r
o
n
e
s
im
u
lato
r
.
Nev
e
r
th
eless
,
th
e
em
p
h
asis
is
r
ea
l
-
tim
e
ac
tio
n
with
o
t
h
er
co
m
p
u
tin
g
ti
m
e
ap
p
licatio
n
s
th
at
ca
n
b
e
u
s
ed
.
Pro
p
o
s
ed
m
eth
o
d
s
u
s
in
g
R
NN
an
d
wav
elet
co
u
l
d
s
u
p
p
o
r
t
B
C
I
with
MI
an
d
f
o
cu
s
v
ar
iab
les
with
an
ac
c
u
r
ac
y
o
f
alm
o
s
t
8
0
%
n
o
n
-
tr
ain
i
n
g
d
ata.
T
h
e
r
esear
c
h
g
a
v
e
th
at
th
e
u
s
e
o
f
wa
v
elet
as
a
p
r
e
-
p
r
o
ce
s
s
ca
n
im
p
r
o
v
e
ac
cu
r
ac
y
,
lead
to
s
tab
ilit
y
,
an
d
r
ed
u
ce
th
e
tr
ain
in
g
d
ata
co
m
p
u
tatio
n
tim
e.
T
h
is
r
esu
lt
is
co
n
s
is
ten
t
with
th
e
h
y
p
o
th
esis
r
ef
er
r
in
g
t
o
p
r
e
v
io
u
s
r
esear
c
h
th
at
wav
elet
is
s
u
itab
le
f
o
r
n
o
n
-
s
tatio
n
ar
y
s
ig
n
als
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
B
r
a
in
-
co
mp
u
ter in
terf
a
ce
o
f fo
cu
s
a
n
d
mo
t
o
r
ima
g
ery
u
s
in
g
w
a
ve
let
a
n
d
… (
E
s
mera
ld
a
C
.
Dja
ma
l
)
2755
s
u
ch
as
E
E
G
s
ig
n
als.
I
n
u
s
in
g
R
NN,
it
i
s
n
ec
ess
ar
y
to
p
ay
atten
tio
n
to
th
e
o
p
tim
izatio
n
m
o
d
el
f
o
r
th
e
co
r
r
ec
tio
n
o
f
weig
h
ts
an
d
th
e
n
u
m
b
er
o
f
ep
o
ch
s
u
s
ed
.
Ad
am
'
s
m
o
d
el
r
ea
ch
es
asy
m
p
to
tically
f
aster
,
b
u
t
s
till
f
lu
ctu
ates
at
th
e
e
n
d
o
f
th
e
e
p
o
ch
,
s
o
it r
e
q
u
ir
es th
e
r
ig
h
t
n
u
m
b
er
o
f
iter
atio
n
s
.
T
h
e
SGD
m
o
d
el
is
q
u
ieter
in
p
er
f
o
r
m
an
ce
b
u
t
r
eq
u
ir
es
f
ar
m
o
r
e
ep
o
c
h
s
.
W
h
ile
th
e
Ad
aDe
lta
m
o
d
el
ad
o
p
ts
b
o
th
m
o
d
els,
it
p
r
o
v
id
es
s
tab
le
an
d
h
ig
h
ac
cu
r
ac
y
.
T
h
e
n
ex
t
th
in
g
to
lo
o
k
o
u
t
f
o
r
is
th
e
co
n
f
ig
u
r
a
tio
n
o
f
th
e
in
p
u
t
f
ea
t
u
r
es
an
d
ch
an
n
el
u
s
ag
e
o
f
th
e
E
E
G
s
ig
n
al.
ACK
NO
WL
E
DG
E
M
E
NT
S
T
h
e
r
esear
ch
was
f
u
n
d
e
d
b
y
"PT
UPT
–
Pen
elitian
T
er
ap
an
Un
g
g
u
lan
Per
g
u
r
u
an
T
in
g
g
i"
f
r
o
m
th
e
Min
is
tr
y
o
f
R
esear
ch
T
ec
h
n
o
lo
g
y
an
d
Hig
h
er
E
d
u
ca
tio
n
,
R
ep
u
b
lik
I
n
d
o
n
esia.
RE
F
E
R
E
NC
E
S
[1
]
E.
C.
Dja
m
a
l,
M
.
Y.
Ab
d
u
ll
a
h
,
a
n
d
F
.
Re
n
a
l
d
i,
“
Bra
in
Co
m
p
u
ter
In
terfa
c
e
G
a
m
e
Co
n
tro
ll
i
n
g
Us
in
g
F
a
st
F
o
u
r
ier
Tran
sfo
rm
a
n
d
Lea
rn
in
g
Ve
c
to
r
Qu
a
n
t
iza
ti
o
n
,
”
J
o
u
rn
a
l
o
f
T
e
lec
o
mm
u
n
ica
ti
o
n
,
El
e
c
tro
n
ic
a
n
d
C
o
m
p
u
te
r
En
g
i
n
e
e
rin
g
(J
T
EC)
,
v
o
l
.
9
,
n
o
.
2
-
5
,
p
p
.
7
1
-
7
4
,
2
0
1
7
.
[2
]
M
.
Va
n
Vliet,
A.
Ro
b
b
e
n
,
N.
Ch
u
m
e
rin
,
e
t
a
l.
,
“
De
sig
n
in
g
a
b
ra
in
-
c
o
m
p
u
ter
i
n
terfa
c
e
c
o
n
tr
o
ll
e
d
v
i
d
e
o
-
g
a
m
e
u
si
n
g
c
o
n
su
m
e
r
g
ra
d
e
EE
G
h
a
rd
wa
re
,
”
2
0
1
2
IS
S
NIP
Bi
o
sig
n
a
ls
a
n
d
Bi
o
ro
b
o
ti
c
s
Co
n
fer
e
n
c
e
:
B
io
sig
n
a
l
s
a
n
d
R
o
b
o
ti
c
s
fo
r B
e
tt
e
r a
n
d
S
a
fer
L
ivin
g
,
B
RC
2
0
1
2
,
2
0
1
2
.
[3
]
S
.
Ra
m
e
sh
,
M
.
G
.
Kri
sh
n
a
,
a
n
d
M
.
Na
k
irek
a
n
ti
,
“
Bra
in
C
o
m
p
u
ter
I
n
terfa
c
e
S
y
ste
m
fo
r
M
i
n
d
C
o
n
tr
o
l
led
Ro
b
o
t
u
sin
g
Blu
e
to
o
th
,
”
I
n
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
C
o
mp
u
ter
Ap
p
li
c
a
ti
o
n
,
v
o
l.
1
0
4
,
n
o
.
1
5
,
p
p
.
2
0
-
2
3
,
2
0
1
4
.
[4
]
B.
Vá
rk
u
ti
,
e
t
a
l.
,
“
Re
stin
g
S
tate
Ch
a
n
g
e
s
in
F
u
n
c
ti
o
n
a
l
Co
n
n
e
c
ti
v
it
y
C
o
rre
late
Wi
t
h
M
o
v
e
m
e
n
t
Re
c
o
v
e
ry
fo
r
BCI
a
n
d
Ro
b
o
t
-
As
siste
d
Up
p
e
r
-
Ex
tre
m
it
y
Train
i
n
g
Afte
r
S
tr
o
k
e
,
”
Ne
u
ro
re
h
a
b
i
li
ta
ti
o
n
a
n
d
Ne
u
r
a
l
Rep
a
ir
,
v
o
l
.
2
7
,
n
o
.
1
,
p
p
.
5
3
-
6
2
,
2
0
1
3
.
[5
]
D.
J.
M
c
F
a
rlan
d
a
n
d
J
.
R.
W
o
l
p
a
w,
“
EE
G
-
Ba
se
d
Bra
in
-
Co
m
p
u
ter
In
terfa
c
e
s,”
Cu
rr
e
n
t
Op
in
i
o
n
in
Bi
o
me
d
ica
l
En
g
i
n
e
e
rin
g
,
v
o
l.
4
,
n
o
.
De
c
,
p
p
.
1
9
4
-
2
0
0
,
2
0
1
7
.
[6
]
E.
C.
Dja
m
a
l
a
n
d
S
u
p
ri
jan
to
,
“
Re
c
o
g
n
i
ti
o
n
o
f
El
e
c
tro
e
n
c
e
p
h
a
l
o
g
ra
m
S
ig
n
a
l
P
a
tt
e
r
n
a
g
a
in
st
S
o
u
n
d
S
t
i
m
u
latio
n
u
si
n
g
S
p
e
c
tral
o
f
Wa
v
e
let,
”
T
e
n
c
o
n
2
0
1
1
,
p
p
.
7
6
7
-
7
7
1
,
2
0
1
1
.
[7
]
M
.
H.
Alo
m
a
ri,
A
.
Ab
u
b
a
k
e
r,
A.
Tu
ra
n
i,
e
t
a
l.
,
“
EE
G
M
o
u
se
:
A
M
a
c
h
in
e
Lea
rn
i
n
g
-
Ba
se
d
Bra
in
C
o
m
p
u
ter
I
n
terfa
c
e
,
”
(IJ
ACS
A)
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Ad
v
a
n
c
e
d
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
Ap
p
li
c
a
ti
o
n
s
,
v
o
l.
5
,
n
o
.
4
,
p
p
.
1
9
3
-
1
9
8
,
2
0
1
4
.
[8
]
S
.
C
h
a
u
d
h
a
ry
,
S
.
Tara
n
,
V.
Ba
jaj,
a
n
d
S
.
S
iu
l
y
,
“
A
flex
i
b
le
a
n
a
ly
ti
c
wa
v
e
let
tran
sf
o
rm
b
a
se
d
a
p
p
r
o
a
c
h
f
o
r
m
o
to
r
-
ima
g
e
ry
tas
k
s
c
las
sifica
ti
o
n
in
BCI
a
p
p
li
c
a
ti
o
n
s,”
Co
m
p
u
ter
M
e
th
o
d
s
a
n
d
Pr
o
g
r
a
ms
in
Bi
o
me
d
icin
e
,
v
o
l.
1
5
,
n
o
.
4
5
,
2
0
2
0
.
[9
]
H.
G
ö
k
su
,
“
BCI
o
rien
ted
E
EG
a
n
a
ly
sis u
si
n
g
l
o
g
e
n
e
rg
y
e
n
tro
p
y
o
f
wa
v
e
let
p
a
c
k
e
ts,”
Bi
o
me
d
ica
l
S
ig
n
a
l
Pro
c
e
ss
in
g
a
n
d
Co
n
tro
l
,
v
o
l.
4
4
,
p
p
.
1
0
1
-
1
0
9
,
2
0
1
8
.
[1
0
]
T.
Ng
u
y
e
n
,
A.
Kh
o
sra
v
i,
D.
Cre
ig
h
to
n
,
a
n
d
S
.
Na
h
a
v
a
n
d
i,
“
EE
G
sig
n
a
l
c
las
sifica
ti
o
n
fo
r
BCI
a
p
p
l
ica
ti
o
n
s
b
y
wa
v
e
lets
a
n
d
i
n
terv
a
l
t
y
p
e
-
2
fu
z
z
y
lo
g
ic sy
ste
m
s,”
Exp
e
rt S
y
ste
m W
it
h
Ap
p
l
i
c
a
ti
o
n
s
,
v
o
l.
4
2
,
n
o
.
9
,
p
p
.
4
3
7
0
-
4
3
8
0
,
2
0
1
5
.
[1
1
]
A.
Kh
a
laf,
E.
S
e
jd
ic,
a
n
d
M
.
Ak
c
a
k
a
y
a
,
“
Co
m
m
o
n
sp
a
ti
a
l
p
a
tt
e
rn
a
n
d
wa
v
e
let
d
e
c
o
m
p
o
siti
o
n
fo
r
m
o
to
r
ima
g
e
ry
EEG
-
fTCD
b
ra
in
-
c
o
m
p
u
ter i
n
ter
fa
c
e
,
”
J
o
u
rn
a
l
o
f
Ne
u
r
o
sc
ien
c
e
M
e
th
o
d
s
,
v
o
l.
3
2
0
,
n
o
.
Ap
ril
,
p
p
.
9
8
-
1
0
6
,
2
0
1
9
.
[1
2
]
F
.
An
sa
ri,
D.
R
.
E
d
la,
S
.
Do
d
ia,
a
n
d
V.
K
u
p
p
il
i,
“
Bra
in
-
C
o
m
p
u
t
e
r
In
terfa
c
e
fo
r
Wh
e
e
lch
a
ir
C
o
n
t
ro
l
O
p
e
ra
ti
o
n
s:
An
Ap
p
r
o
a
c
h
b
a
se
d
o
n
F
a
st
F
o
u
rier
Tran
sfo
rm
a
n
d
On
-
Li
n
e
S
e
q
u
e
n
ti
a
l
E
x
trem
e
Lea
rn
in
g
M
a
c
h
in
e
,
”
Cli
n
ic
a
l
Ep
id
e
mi
o
lo
g
y
a
n
d
Gl
o
b
a
l
He
a
l
th
,
v
o
l
.
7
,
n
o
.
3
,
p
p
.
2
7
4
-
2
7
8
,
2
0
1
8
.
[1
3
]
S
.
H.
F
a
irclo
u
g
h
,
“
BCI
a
n
d
p
h
y
sio
lo
g
ica
l
c
o
m
p
u
ti
n
g
fo
r
c
o
m
p
u
t
e
r
g
a
m
e
s:
Diffe
re
n
c
e
s,
sim
il
a
rit
ies
&
in
t
u
it
iv
e
c
o
n
tro
l,
”
Pro
c
e
e
d
i
n
g
s
o
f
CHI’0
8
,
n
o
.
1
,
p
p
.
1
-
6
,
2
0
0
8
.
[1
4
]
S
.
Al
h
a
g
ry
,
A.
A.
F
a
h
m
y
,
a
n
d
R.
A.
El
-
Kh
o
rib
i,
“
Emo
ti
o
n
Re
c
o
g
n
i
ti
o
n
b
a
se
d
o
n
EE
G
u
si
n
g
LS
T
M
R
e
c
u
rre
n
t
Ne
u
ra
l
Ne
two
rk
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Ad
v
a
n
c
e
d
Co
mp
u
ter
S
c
ien
c
e
a
n
d
Ap
p
li
c
a
ti
o
n
s
,
v
o
l.
8
,
n
o
.
1
0
,
p
p
.
3
5
6
-
3
5
8
,
2
0
1
7
.
[1
5
]
J.
R.
W
o
lp
a
w,
N.
Bir
b
a
u
m
e
r,
D.
J.
M
c
F
a
rlan
d
,
e
t
a
l.
,
“
Bra
in
Co
m
p
u
ter
I
n
terfa
c
e
s
fo
r
c
o
m
m
u
n
ica
t
io
n
a
n
d
c
o
n
tr
o
l,
”
Fro
n
ti
e
rs
i
n
Ne
u
ro
sc
ien
c
e
,
v
o
l.
4
,
n
o
.
1
1
3
,
p
p
.
7
6
7
-
7
9
1
,
2
0
0
2
.
[1
6
]
T.
J.
C
h
o
i
,
J.
O.
Kim
,
S
.
M
.
Ji
n
,
a
n
d
G
.
Yo
o
n
,
“
De
term
in
a
ti
o
n
o
f
t
h
e
Co
n
c
e
n
trate
d
S
tate
Us
in
g
M
u
l
ti
p
le
EE
G
Ch
a
n
n
e
ls,”
I
n
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
C
o
mp
u
ter
,
El
e
c
trica
l,
Au
t
o
ma
t
i
o
n
,
Co
n
tro
l
a
n
d
In
f
o
rm
a
ti
o
n
En
g
i
n
e
e
rin
g
,
v
o
l.
8
,
n
o
.
8
,
p
p
.
1
3
5
9
-
1
3
6
2
,
2
0
1
4
.
[1
7
]
K.
G
.
S
m
it
h
a
,
S
.
S
h
e
n
ji
e
,
K
.
P
.
Th
o
m
a
s,
a
n
d
A
.
P
.
Vi
n
o
d
,
“
Tw
o
p
lay
e
r
EE
G
-
ba
se
d
n
e
u
r
o
fe
e
d
b
a
c
k
b
a
ll
g
a
m
e
fo
r
a
tt
e
n
ti
o
n
e
n
h
a
n
c
e
m
e
n
t,
”
in
I
n
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
S
y
ste
ms
,
M
a
n
a
n
d
Cy
b
e
rn
e
ti
c
s
,
2
0
1
4
,
p
p
.
3
1
5
0
-
3
1
5
5
.
[
1
8
]
D
.
I
a
c
o
v
i
e
l
l
o
,
A
.
P
e
t
r
a
c
c
a
,
M
.
S
p
e
z
i
a
l
e
t
t
i
,
a
n
d
G
.
P
l
a
c
i
d
i
,
“
A
r
e
a
l
-
t
i
m
e
c
l
a
s
s
i
f
i
c
a
t
i
o
n
a
l
g
o
r
i
t
h
m
f
o
r
E
E
G
-
b
a
s
e
d
B
C
I
d
r
i
v
e
n
b
y
s
e
l
f
-
i
n
d
u
c
e
d
e
m
o
t
i
o
n
s
,
”
C
o
m
p
u
t
e
r
M
e
t
h
o
d
s
a
n
d
P
r
o
g
r
a
m
s
i
n
B
i
o
m
e
d
i
c
i
n
e
,
v
o
l
.
1
2
2
,
n
o
.
3
,
p
p
.
2
9
3
-
303
,
2
0
1
5
.
[1
9
]
E.
C.
Dja
m
a
l,
H.
F
a
d
h
il
a
h
,
A.
N
a
jmu
rro
k
h
m
a
n
,
A.
Wu
lan
d
a
ri,
a
n
d
F
.
Re
n
a
ld
i
,
“
Emo
ti
o
n
b
ra
in
-
c
o
m
p
u
ter
in
terfa
c
e
u
sin
g
wa
v
e
let
a
n
d
re
c
u
rre
n
t
n
e
u
r
a
l
n
e
two
rk
s,”
I
n
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
A
d
v
a
n
c
e
s
in
In
telli
g
e
n
t
In
f
o
rm
a
ti
c
s
,
v
o
l.
6
,
n
o
.
1
,
p
p
.
1
-
1
2
,
2
0
2
0
.
[2
0
]
E.
C.
Dj
a
m
a
l,
S
u
p
ri
jan
t
o
,
a
n
d
S
.
J.
S
e
ti
a
d
i,
“
Clas
sifica
ti
o
n
o
f
EE
G
-
Ba
s
e
d
Ha
n
d
G
ra
sp
in
g
Im
a
g
in
a
ti
o
n
Us
in
g
Au
to
re
g
re
ss
iv
e
a
n
d
Ne
u
ra
l
Ne
two
rk
s,”
J
u
rn
a
l
T
e
k
n
o
lo
g
i
,
v
o
l.
7
8
,
n
o
.
6
-
6
,
p
p
.
1
0
5
-
1
1
0
,
2
0
1
6
.
[2
1
]
L.
Wan
g
,
X.
L
iu
,
Z.
Li
a
n
g
,
Z.
Ya
n
g
,
a
n
d
X.
Hu
,
“
An
a
l
y
sis a
n
d
c
las
sifica
ti
o
n
o
f
h
y
b
ri
d
BCI
b
a
se
d
o
n
m
o
to
r
ima
g
e
ry
a
n
d
sp
e
e
c
h
ima
g
e
ry
,
”
M
e
a
s
u
re
me
n
t
,
v
o
l
.
1
4
7
,
p
.
1
0
6
8
4
2
,
2
0
1
9
.
[2
2
]
A.
F
i
n
k
e
,
A.
Len
h
a
r
d
t,
a
n
d
H.
Rit
ter,
“
Th
e
M
i
n
d
G
a
m
e
:
A
P
3
0
0
-
b
a
se
d
b
ra
in
-
c
o
m
p
u
ter
i
n
terfa
c
e
g
a
m
e
,
”
Ne
u
ra
l
Ne
two
rk
s
,
v
o
l
.
2
2
,
n
o
.
9
,
p
p
.
1
3
2
9
-
1
3
3
3
,
2
0
0
9
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
,
Vo
l.
18
,
No
.
5
,
Octo
b
e
r
2
0
2
0
:
2
7
4
8
-
275
6
2756
[2
3
]
T.
I.
Vo
z
n
e
n
k
o
,
E
.
V C
h
e
p
i
n
,
I
.
V
o
z
n
e
n
k
o
,
e
t
a
l,
,
“
Th
e
C
o
n
tro
l
S
y
st
e
m
Ba
se
d
Ba
se
d
o
n
Ex
ten
d
e
d
BC
I
fo
r
a
Ro
b
o
t
ic
Wh
e
e
lch
a
ir,
”
in
Pro
c
e
d
i
a
Co
m
p
u
ter
S
c
ien
c
e
,
2
0
1
8
,
v
o
l.
1
2
3
,
p
p
.
5
2
2
-
5
2
7
.
[2
4
]
T.
Ho
sm
a
n
,
e
t
a
l
.
,
“
BCI
d
e
c
o
d
e
r
p
e
rfo
rm
a
n
c
e
c
o
m
p
a
riso
n
o
f
a
n
L
S
TM
re
c
u
rre
n
t
n
e
u
ra
l
n
e
two
r
k
a
n
d
a
Ka
lma
n
fil
ter
in
re
tr
o
sp
e
c
ti
v
e
sim
u
lati
o
n
,
”
2
0
1
9
9
t
h
In
ter
n
a
ti
o
n
a
l
I
EE
E/
E
M
BS
C
o
n
fer
e
n
c
e
o
n
Ne
u
r
a
l
En
g
i
n
e
e
rin
g
(NER
)
,
p
p
.
1
0
6
6
-
1
0
7
1
,
2
0
1
9
.
[2
5
]
J.
Li
u
,
Y.
C
h
e
n
g
,
a
n
d
W.
Zh
a
n
g
,
“
De
e
p
lea
rn
in
g
EE
G
re
sp
o
n
se
re
p
r
e
se
n
tatio
n
f
o
r
b
ra
in
c
o
m
p
u
ter
in
te
rfa
c
e
,
”
Ch
in
e
se
Co
n
tro
l
Co
n
fer
e
n
c
e
,
p
p
.
3
5
1
8
-
3
5
2
3
,
2
0
1
5
.
[2
6
]
S
.
Um
a
r,
M
.
Alsu
laim
a
n
,
a
n
d
G
.
M
u
h
a
m
m
a
d
,
“
De
e
p
Lea
rn
in
g
fo
r
EE
G
m
o
to
r
ima
g
e
ry
c
las
sifica
ti
o
n
b
a
se
d
o
n
m
u
lt
i
-
lay
e
r
CNN
s fea
tu
re
fu
sio
n
,
”
Fu
tu
re
Ge
n
e
ra
ti
o
n
C
o
mp
u
ter
S
y
ste
ms
,
v
o
l
.
1
0
1
,
p
p
.
5
4
2
-
5
5
4
,
2
0
1
9
.
[2
7
]
E.
A.
M
o
h
a
m
e
d
,
M
.
Z
.
B.
Yu
s
o
f
f,
N.
K.
S
e
lma
n
,
a
n
d
A.
S
.
M
a
li
k
,
“
E
n
h
a
n
c
i
n
g
EE
G
S
ig
n
a
ls
i
n
B
ra
in
Co
m
p
u
ter
In
terfa
c
e
Us
in
g
Wav
e
let
Tran
sfo
r
m
,
”
in
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
I
n
fo
rm
a
t
io
n
a
n
d
El
e
c
tro
n
ics
En
g
i
n
e
e
rin
g
,
v
o
l.
4
,
n
o
.
3
,
p
p
.
2
3
4
-
2
3
8
,
2
0
1
4
.
[2
8
]
C.
Ro
d
rig
u
e
s,
P
.
P
.
Re
b
o
u
ç
a
s,
E.
P
e
ix
o
t
o
,
A.
K.
N,
V.
H
u
g
o
,
a
n
d
C
.
D
e
Al
b
u
q
u
e
rq
u
e
,
“
Clas
sifica
ti
o
n
o
f
EE
G
sig
n
a
ls
to
d
e
tec
t
a
lco
h
o
li
sm
u
sin
g
m
a
c
h
i
n
e
lea
rn
in
g
tec
h
n
iq
u
e
s,”
Pa
tt
e
rn
Rec
o
g
n
it
io
n
L
e
tt
e
rs
,
v
o
l.
1
2
5
,
p
p
.
1
4
0
-
1
4
9
,
2
0
1
9
.
[2
9
]
E.
C.
Dja
m
a
l
a
n
d
P
.
L
o
d
a
y
a
,
“
EE
G
Ba
se
d
Emo
ti
o
n
M
o
n
it
o
rin
g
Us
i
n
g
Wav
e
let
a
n
d
Lea
rn
i
n
g
Ve
c
to
r
Qu
a
n
ti
z
a
ti
o
n
,
”
2
0
1
7
4
t
h
In
ter
n
a
ti
o
n
a
l
Co
n
fer
e
n
c
e
o
n
El
e
c
trica
l
En
g
i
n
e
e
rin
g
,
Co
mp
u
ter
S
c
ien
c
e
a
n
d
In
f
o
rm
a
ti
c
s
(EE
CS
I
2
0
1
7
)
,
p
p
.
1
9
-
21
,
2
0
1
7
.
[3
0
]
A.
N.
Ak
a
n
s
u
a
n
d
R.
A.
Ha
d
d
a
d
,
“
M
u
lt
ires
o
lu
t
io
n
sig
n
a
l
d
e
c
o
m
p
o
siti
o
n
,
”
Aca
d
e
mic
Pre
ss
,
S
e
c
o
n
d
.
,
n
o
.
1
,
T.
R
.
Hs
in
g
,
E
d
.
Ac
a
d
e
m
ic P
re
ss
,
p
p
.
2
2
-
25
,
2
0
0
1
.
[3
1
]
Y.
S
h
e
n
,
H.
Lu
,
a
n
d
J.
Jia
,
“
Clas
sifica
ti
o
n
o
f
M
o
to
r
Im
a
g
e
ry
E
E
G
S
ig
n
a
ls
with
De
e
p
Lea
rn
in
g
M
o
d
e
ls,”
L
e
c
tu
re
No
tes
in
C
o
mp
u
ter
S
c
ien
c
e
b
o
o
k
se
rie
s
,
v
o
l.
1
0
5
5
9
,
p
p
.
1
8
1
-
1
9
0
,
2
0
1
7
.
[3
2
]
J.
-
S
.
Wan
g
,
Y.
-
T.
Ya
n
g
,
C.
-
Y
.
Hs
u
,
a
n
d
Y.
Hs
u
,
“
A Rec
u
rre
n
t
Ne
u
r
a
l
S
lee
p
-
S
tag
e
Clas
sifier Usin
g
E
n
e
rg
y
F
e
a
tu
re
s
o
f
EE
G
S
ig
n
a
ls,”
Ne
u
ro
c
o
m
p
u
ti
n
g
,
v
o
l.
1
0
4
,
p
p
.
1
0
5
-
1
1
4
,
2
0
1
3
.
[3
3
]
A.
M
a
z
u
m
d
e
r,
A.
Ra
k
sh
it
,
a
n
d
D.
N.
Ti
b
a
re
wa
la,
“
A
Ba
c
k
-
P
ro
p
a
g
a
ti
o
n
Th
r
o
u
g
h
Ti
m
e
Ba
se
d
Re
c
u
rre
n
t
Ne
u
ra
l
Ne
two
rk
Ap
p
ro
a
c
h
f
o
r
Clas
sifica
t
io
n
o
f
C
o
g
n
it
i
v
e
EE
G
S
tate
s,”
2
0
1
5
IEE
E
In
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
o
n
E
n
g
in
e
e
rin
g
a
n
d
T
e
c
h
n
o
l
o
g
y
,
n
o
.
M
a
rc
h
,
p
p
.
1
-
5
,
2
0
1
5
.
[3
4
]
J.
Zh
o
u
,
M
.
M
e
n
g
,
Y.
G
a
o
,
Y.
M
a
,
a
n
d
Q.
Z
h
a
n
g
,
“
Clas
sifica
ti
o
n
o
f
M
o
to
r
Im
a
g
e
ry
EE
G
Us
in
g
W
a
v
e
let
En
v
e
l
o
p
e
An
a
ly
sis a
n
d
LS
T
M
Ne
two
rk
s
,
”
2
0
1
8
C
h
in
e
se
Co
n
tro
l
An
d
De
c
isio
n
C
o
n
fer
e
n
c
e
(CCDC)
,
p
p
.
5
6
0
0
-
5
6
0
5
,
2
0
1
8
.
[3
5
]
Y.
Li
,
J.
H
u
a
n
g
,
H.
Z
h
o
u
,
a
n
d
N.
Z
h
o
n
g
,
“
Hu
m
a
n
Emo
ti
o
n
Re
c
o
g
n
it
io
n
wi
th
El
e
c
tro
e
n
c
e
p
h
a
lo
g
ra
p
h
i
c
M
u
lt
i
d
ime
n
sio
n
a
l
F
e
a
tu
re
s
b
y
H
y
b
rid
De
e
p
Ne
u
ra
l
Ne
two
rk
s,”
A
p
p
li
e
d
S
c
ien
c
e
s
,
v
o
l.
7
,
n
o
.
1
0
,
p
.
1
0
6
0
,
2
0
1
7
.
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
re
e
i
n
En
g
in
e
e
rin
g
P
h
y
sic
s
fro
m
In
stit
u
t
Te
k
n
o
lo
g
i
Ba
n
d
u
n
g
in
1
9
9
4
,
a
M
a
ste
r
's
d
e
g
re
e
in
In
str
u
m
e
n
t
a
n
d
C
o
n
tr
o
l
fro
m
In
stit
u
t
Tek
n
o
l
o
g
i
Ba
n
d
u
n
g
in
1
9
9
8
.
S
in
c
e
Ph
.
D.
d
isse
rtatio
n
u
n
t
il
n
o
w,
re
se
a
rc
h
o
n
EE
G
c
las
sifica
ti
o
n
a
n
d
fi
n
ish
e
d
d
o
c
to
ra
l
p
ro
g
ra
m
fr
o
m
I
n
stit
u
t
Tek
n
o
lo
g
i
Ba
n
d
u
n
g
in
2
0
0
5
.
S
h
e
is a l
e
c
tu
re
r
o
f
In
f
o
rm
a
ti
c
s De
p
a
rtme
n
t,
Un
i
v
e
rsitas
Je
n
d
e
ra
l
Ac
h
m
a
d
Ya
n
i.
Rifq
i
Da
n
i
a
P
u
tr
a
re
c
e
iv
e
d
h
is
b
a
c
h
e
lo
r'
s
d
e
g
re
e
in
In
f
o
rm
a
ti
c
s
d
e
p
a
rtme
n
t
fro
m
Un
iv
e
rsitas
Je
n
d
e
ra
l
Ac
h
m
a
d
Ya
n
i
i
n
2
0
1
9
.
E
-
m
a
il
:
d
a
n
ia
p
u
tra
2
4
@g
m
a
il
.
c
o
m
Evaluation Warning : The document was created with Spire.PDF for Python.