I
nte
rna
t
io
na
l J
o
urna
l o
f
Adv
a
nces in Applie
d Science
s
(
I
J
AAS)
Vo
l
.
1
5
,
No
. 1
,
Ma
r
ch
20
26
,
p
p
.
19
7
~
20
7
I
SS
N:
2252
-
8
8
1
4
,
DOI
:
1
0
.
1
1
5
9
1
/ijaas
.
v15.
i
1
.
pp
19
7
-
20
7
197
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
a
a
s
.
ia
esco
r
e.
co
m
Im
pro
v
ed seizure
det
ec
tion usin
g
o
ptimized
t
ime
seq
uence
ba
sed deep l
ea
rni
ng
f
ra
mewo
rk
P
us
pa
nja
li M
a
llik
1,
2
,
Aj
it
K
um
a
r
Na
y
a
k
3
,
Sa
t
y
a
pra
ka
s
h Swa
in
4
1
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
I
TER
,
S
O
A
D
e
e
m
e
d
t
o
b
e
U
n
i
v
e
r
s
i
t
y
,
O
d
i
sh
a
,
I
n
d
i
a
2
S
h
a
i
l
a
b
a
l
a
W
o
me
n
’
s Au
t
o
n
o
m
o
u
s
C
o
l
l
e
g
e
,
C
u
t
t
a
c
k
,
I
n
d
i
a
3
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
I
n
f
o
r
ma
t
i
o
n
T
e
c
h
n
o
l
o
g
y
,
I
TER
,
S
O
A
D
e
e
me
d
t
o
b
e
U
n
i
v
e
r
s
i
t
y
,
B
h
u
b
a
n
e
sw
a
r
,
I
n
d
i
a
4
D
e
p
a
r
t
me
n
t
o
f
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
a
n
d
E
n
g
i
n
e
e
r
i
n
g
,
I
n
st
i
t
u
t
e
o
f
M
a
n
a
g
e
me
n
t
a
n
d
I
n
f
o
r
mat
i
o
n
Te
c
h
n
o
l
o
g
y
(
I
M
I
T)
,
C
u
t
t
a
c
k
,
B
P
U
T
,
O
d
i
s
h
a
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
u
n
7
,
2
0
2
5
R
ev
is
ed
No
v
1
8
,
2
0
2
5
Acc
ep
ted
J
an
1
,
2
0
2
6
Ep
il
e
p
s
y
d
ise
a
se
o
rig
i
n
a
tes
d
u
e
t
o
t
h
e
p
re
se
n
c
e
o
f
d
iso
r
d
e
re
d
n
e
u
ro
n
s
,
a
n
d
e
p
il
e
p
sy
d
e
tec
ti
o
n
sta
n
d
s
a
s
a
c
h
a
ll
e
n
g
in
g
tas
k
f
o
r
n
e
u
ro
l
o
g
ists
.
Wi
th
re
c
e
n
t
a
d
v
a
n
c
e
s,
e
lec
tro
e
n
c
e
p
h
a
lo
g
ra
p
h
y
(
EEG
)
-
b
a
se
d
a
n
a
ly
sis
is
i
n
c
re
a
sin
g
ly
su
p
p
o
rted
b
y
d
e
e
p
lea
rn
i
n
g
a
n
d
m
e
tah
e
u
risti
c
o
p
ti
m
iza
ti
o
n
a
p
p
r
o
a
c
h
e
s
in
o
rd
e
r
to
imp
r
o
v
e
t
h
e
te
st
re
su
lt
s
.
Th
is
e
x
p
e
rime
n
t
u
se
s
a
c
o
n
v
o
l
u
ti
o
n
a
l
n
e
u
ra
l
n
e
tw
o
rk
(CNN
)
m
o
d
e
l
h
y
b
rid
ize
d
with
b
i
d
irec
ti
o
n
a
l
lo
n
g
sh
o
rt
-
term
m
e
m
o
ry
(BiL
S
TM
)
.
CNN
lev
e
ra
g
e
s
th
e
wo
rk
with
im
p
ro
v
e
d
fe
a
tu
re
e
x
trac
ti
o
n
c
u
m
c
las
sifica
ti
o
n
s
u
p
p
o
rts
,
a
n
d
BiL
S
T
M
k
e
e
p
s
th
e
ti
m
e
se
q
u
e
n
c
e
o
f
d
a
ta
in
b
o
t
h
th
e
f
o
r
wa
rd
a
n
d
b
a
c
k
wa
rd
d
irec
ti
o
n
f
o
r
imp
ro
v
i
n
g
sig
n
a
l
m
a
p
p
i
n
g
p
u
r
p
o
se
s
.
T
o
re
d
u
c
e
th
e
c
o
m
p
u
tatio
n
a
l
o
v
e
r
h
e
a
d
a
n
d
imp
ro
v
e
e
x
e
c
u
ti
o
n
a
c
c
u
ra
c
y
,
a
h
y
b
ri
d
o
p
t
imiz
a
ti
o
n
a
lg
o
rit
h
m
c
a
ll
e
d
se
c
re
tary
b
ird
o
p
ti
m
iza
ti
o
n
a
l
g
o
r
it
h
m
(
S
BOA
)
is
u
se
d
to
fin
e
-
tu
n
e
t
h
e
e
x
e
c
u
ti
o
n
.
Ke
y
c
las
sifica
ti
o
n
p
a
r
a
m
e
ters
su
c
h
a
s
a
c
c
u
ra
c
y
,
se
n
siti
v
it
y
,
a
n
d
sp
e
c
ifi
c
it
y
re
flec
t
t
h
e
m
o
d
e
l’s
s
tro
n
g
p
re
d
icti
v
e
c
a
p
a
b
il
it
y
,
wit
h
a
c
c
u
ra
c
y
re
a
c
h
in
g
u
p
t
o
9
8
.
4
9
%
.
T
h
e
p
r
o
p
o
se
d
m
e
th
o
d
d
e
m
o
n
stra
tes
th
e
p
o
ten
ti
a
l
fo
r
h
ig
h
-
p
e
rfo
rm
a
n
c
e
EE
G
-
b
a
se
d
se
i
z
u
re
d
e
tec
ti
o
n
,
p
a
v
in
g
th
e
wa
y
fo
r
fu
tu
r
e
in
teg
ra
ti
o
n
wit
h
e
d
g
e
c
o
m
p
u
ti
n
g
d
e
v
ice
s
to
su
p
p
o
rt
re
m
o
te
c
li
n
ica
l
d
iag
n
o
stics
a
n
d
c
o
n
ti
n
u
o
u
s m
o
n
it
o
rin
g
i
n
re
a
l
-
wo
rld
h
e
a
lt
h
c
a
re
a
p
p
li
c
a
ti
o
n
s.
K
ey
w
o
r
d
s
:
B
iLST
M
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
Me
tah
eu
r
is
tic
o
p
tim
izatio
n
Secr
etar
y
b
ir
d
o
p
tim
izatio
n
alg
o
r
ith
m
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
:
Saty
ap
r
ak
ash
Swain
Dep
ar
tm
en
t o
f
C
o
m
p
u
ter
Scie
n
ce
an
d
E
n
g
in
ee
r
in
g
I
n
s
titu
te
o
f
Ma
n
ag
e
m
en
t a
n
d
I
n
f
o
r
m
atio
n
T
ec
h
n
o
lo
g
y
(
I
M
I
T
)
,
C
u
ttack
,
B
PUT
Od
is
h
a,
C
u
ttack
,
I
n
d
ia
E
m
ail:
s
aty
aim
it@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
As
ep
ilep
s
y
is
a
life
-
d
is
tr
ess
in
g
d
is
ea
s
e,
th
e
m
o
s
t
ad
v
is
ab
le
s
tep
is
to
d
etec
t
th
e
p
r
esen
ce
o
f
s
eizu
r
e
s
ig
n
als
in
o
r
d
er
to
p
r
o
v
id
e
clin
ical
s
u
g
g
esti
o
n
s
to
s
av
e
th
e
li
v
es
o
f
t
h
e
n
eu
r
o
-
d
is
o
r
d
e
r
p
a
tien
ts
.
Pre
s
en
tly
,
a
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
etwo
r
k
(
C
NN)
is
co
n
s
id
er
ed
as
th
e
ad
v
an
ce
d
class
if
icatio
n
tech
n
i
q
u
e
in
th
e
f
ield
o
f
ep
ilep
s
y
d
etec
tio
n
,
wh
ic
h
co
n
s
is
ts
o
f
o
n
e
in
p
u
t
la
y
er
,
o
n
e
o
u
tp
u
t
la
y
er
,
an
d
m
o
r
e
th
an
o
n
e
co
n
v
o
l
u
tio
n
al
lay
er
,
wh
er
e
th
e
in
p
u
t
lay
er
is
co
n
n
e
cted
with
t
h
e
f
i
r
s
t
co
n
v
o
lu
tio
n
al
lay
e
r
an
d
th
e
o
u
tp
u
t
lay
er
is
c
o
n
n
ec
te
d
to
th
e
last
co
n
v
o
l
u
tio
n
al
la
y
er
[
1
]
.
T
h
is
C
NN
co
m
p
r
is
es
m
illi
o
n
s
o
f
n
e
u
r
o
n
s
,
an
d
ea
ch
n
eu
r
o
n
is
m
ath
em
atica
lly
ex
p
r
ess
ed
as:
=
(
+
)
,
‘
y
’
is
th
e
o
u
tp
u
t
p
ar
a
m
eter
,
‘
x
’
is
th
e
in
p
u
t
p
ar
am
ete
r
,
‘
W
’
is
th
e
weig
h
t
m
atr
i
x
,
‘
b
’
is
th
e
b
ias
v
alu
e
,
an
d
‘
f
’
is
th
e
ac
tiv
a
tio
n
f
u
n
ctio
n
[
2
]
.
Als
o
,
m
etah
eu
r
is
tic
alg
o
r
ith
m
s
ar
e
em
er
g
in
g
as
th
e
m
o
s
t
p
o
wer
f
u
l
an
d
p
o
p
u
lar
ly
u
s
ed
a
lg
o
r
ith
m
s
,
wh
ich
ar
e
in
teg
r
ated
with
m
ac
h
in
e
lear
n
in
g
alg
o
r
ith
m
s
[
3
]
.
T
h
e
r
e
ar
e
d
i
f
f
er
en
t
ty
p
es
o
f
m
e
tah
eu
r
is
tic
alg
o
r
ith
m
s
,
s
u
ch
as
n
atu
r
e
-
in
s
p
ir
ed
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
8
1
4
I
n
t J Ad
v
Ap
p
l Sci
,
Vo
l
.
1
5
,
No
.
1
,
Ma
r
c
h
2
0
2
6
:
197
-
2
0
7
198
b
io
-
in
s
p
ir
ed
,
s
war
m
-
in
s
p
ir
e
d
,
an
d
tr
ajec
to
r
y
-
b
ased
alg
o
r
it
h
m
s
[
4
]
.
T
h
er
e
a
r
e
d
if
f
er
en
t
ty
p
es
o
f
n
eu
r
al
d
is
o
r
d
er
s
an
d
Par
k
in
s
o
n
’
s
d
is
ea
s
e
is
o
n
e
am
o
n
g
t
h
em
ch
a
r
ac
ter
ized
b
y
s
h
o
win
g
s
lo
w
m
o
v
em
en
t,
p
o
s
tu
r
al
in
s
tab
ilit
y
,
r
ig
id
it
y
r
ef
lectin
g
t
h
e
d
e
g
en
er
ativ
e
co
n
d
itio
n
o
f
t
h
e
b
r
ain
.
Alth
o
u
g
h
t
h
e
tr
ea
tm
e
n
t
o
f
th
is
d
is
ea
s
e
is
a
tim
e
s
p
an
f
ac
to
r
,
s
till
th
en
t
h
e
p
er
f
ec
t
an
aly
s
is
o
f
th
e
d
is
ea
s
e
will
b
e
co
n
s
id
er
ed
as
a
co
r
e
p
ar
t
o
f
clin
ical
p
r
o
ce
d
u
r
es
.
T
ill
y
et,
th
e
elec
t
r
o
en
ce
p
h
alo
g
r
a
p
h
y
(
EEG
)
test
an
d
m
a
g
n
etic
r
eso
n
an
ce
im
ag
in
g
(
MRI
)
o
f
th
e
s
ca
lp
ar
e
u
s
ef
u
l
to
d
ia
g
n
o
s
e
th
e
s
y
m
p
to
m
s
o
f
Par
k
in
s
o
n
’
s
d
is
ea
s
e
.
I
t
r
eq
u
ir
es
th
e
co
r
r
ec
t
s
eizu
r
e
s
ig
n
al
an
aly
s
is
,
an
d
p
r
ef
e
r
ab
ly
,
th
e
b
est
an
aly
s
is
is
p
er
f
o
r
m
ed
with
d
ee
p
lear
n
i
n
g
alg
o
r
ith
m
s
lik
e
s
u
b
-
b
an
d
an
aly
s
is
with
g
ated
r
ec
u
r
r
en
t
u
n
it
(
GR
U)
[
5
]
an
d
m
u
ltis
ca
le
C
NN
[
6
]
.
So
m
etim
es,
th
er
e
ar
e
also
th
e
p
o
s
s
ib
ilit
ies
o
f
ar
tifa
cts
in
ter
m
ix
ed
wit
h
th
e
s
ig
n
als
,
d
u
e
to
wh
ich
th
e
co
r
r
e
ct
s
eizu
r
e
p
r
e
d
ictio
n
a
n
d
d
etec
tio
n
b
ec
o
m
e
to
u
g
h
,
an
d
th
e
p
er
f
o
r
m
an
ce
ac
c
u
r
ac
y
b
ec
o
m
es
lo
wer
.
I
n
th
is
ca
s
e,
t
h
e
d
ee
p
lear
n
in
g
m
o
d
els
m
u
s
t
b
e
in
teg
r
ated
with
lear
n
ab
le
ca
p
ac
ity
,
a
n
d
th
e
m
o
d
el
is
ca
lled
a
lear
n
ab
le
an
d
e
x
p
lain
ab
le
wav
elet
n
e
u
r
al
n
et
wo
r
k
[
7
]
,
[
8
]
.
W
h
en
th
e
E
E
G
s
ig
n
al
is
r
ec
o
r
d
ed
,
i
t
ap
p
ea
r
s
ir
r
e
g
u
lar
a
n
d
non
-
s
m
o
o
th
f
o
r
w
h
ic
h
it
b
ec
o
m
es
d
if
f
icu
lt
to
m
e
asu
r
e
s
eizu
r
e
f
r
eq
u
en
c
y
[
9
]
.
A
lig
h
t
weig
h
t c
o
n
v
o
lu
tio
n
tr
an
s
f
o
r
m
er
(
L
C
T
)
is
p
r
o
p
o
s
ed
[
1
0
]
f
o
r
cr
o
s
s
-
p
atien
t seizu
r
e
d
ete
ctio
n
,
wh
ich
p
r
o
d
u
ce
s
s
m
o
o
th
n
ess
in
s
eizu
r
e
s
ig
n
als
.
Dep
r
ess
io
n
is
a
s
p
ec
ial
ca
s
e
o
f
a
s
eizu
r
e
s
y
m
p
t
o
m
th
at
m
ay
h
ap
p
en
d
u
e
to
p
er
s
o
n
al
p
r
o
b
l
em
s
,
an
d
d
ee
p
d
ep
r
ess
io
n
af
f
ec
ts
th
e
f
r
ee
th
in
k
in
g
o
f
a
p
e
r
s
o
n
.
T
h
e
u
s
ef
u
l
to
o
ls
em
p
lo
y
ed
f
o
r
d
ep
r
ess
io
n
d
etec
tio
n
ar
e
an
en
co
d
er
f
o
r
d
ata
co
m
p
r
ess
io
n
,
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
L
STM
)
f
o
r
ex
p
r
ess
in
g
th
e
t
em
p
o
r
al
v
ib
r
atio
n
s
o
f
th
e
r
ec
o
r
d
ed
s
ig
n
al
,
a
n
d
an
atten
tio
n
m
ec
h
a
n
is
m
to
in
tr
o
d
u
ce
p
ar
allelis
m
am
o
n
g
th
e
co
m
p
r
ess
ed
in
f
o
r
m
atio
n
[
1
1
]
.
B
u
t
if
th
e
s
ig
n
al
is
co
m
p
lex
an
d
n
o
is
y
in
n
atu
r
e
,
th
en
t
h
e
R
iem
an
n
ian
s
p
ec
tr
a
l
clu
s
ter
in
g
[
1
2
]
m
et
h
o
d
is
f
o
llo
wed
to
id
e
n
tify
t
h
e
o
u
tlier
s
[
1
3
]
.
T
h
er
e
a
r
e
s
o
m
e
s
p
ec
ial
c
ases
wh
er
e
d
ataset
p
r
iv
ac
y
is
m
ai
n
tain
ed
[
1
4
]
alo
n
g
with
s
eizu
r
e
d
etec
tio
n
f
r
o
m
th
e
p
r
iv
ac
y
p
o
in
t
o
f
v
iew
o
f
p
atien
t
in
f
o
r
m
atio
n
.
So
m
e
ty
p
ical
n
e
u
r
al
n
etwo
r
k
s
ar
e
d
ev
is
ed
f
o
r
s
p
ec
if
ic
d
i
s
ea
s
e
s
,
lik
e
Alzh
eim
er
’
s
d
is
ea
s
e
d
etec
tio
n
[
1
5
]
p
er
f
o
r
m
ed
with
Ad
az
d
-
Net
[
1
6
]
a
n
d
an
au
t
o
m
ate
d
d
ee
p
n
eu
r
al
n
etwo
r
k
[
1
7
]
m
o
d
el
.
W
an
g
et
a
l.
[
1
8
]
p
r
o
p
o
s
ed
a
h
y
b
r
id
m
o
d
el
u
s
in
g
a
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
(
SVM)
an
d
k
er
n
el
s
p
ar
s
e
r
ep
r
esen
tatio
n
class
if
icatio
n
(
KSR
C
)
;
W
u
et
a
l.
[
1
9
]
p
r
o
p
o
s
ed
a
s
p
atial
f
ea
tu
r
e
f
u
s
ed
co
n
v
o
lu
t
io
n
al
n
e
two
r
k
(
SC
Net)
f
o
r
E
E
G
p
ath
o
lo
g
y
d
etec
tio
n
.
T
o
o
v
e
r
co
m
e
t
h
e
p
r
o
ce
s
s
in
g
co
m
p
lex
ity
,
m
etah
eu
r
is
tic
o
p
tim
izatio
n
alg
o
r
ith
m
s
a
r
e
in
teg
r
ated
with
n
eu
r
al
n
etwo
r
k
s
.
T
h
is
p
ap
er
u
s
es
th
e
s
ec
r
etar
y
b
ir
d
o
p
tim
i
za
tio
n
alg
o
r
it
h
m
(
SB
OA)
to
f
in
e
-
tu
n
e
th
e
m
o
d
el
an
d
to
r
e
d
u
ce
th
e
o
p
er
atio
n
al
o
v
er
h
ea
d
[
2
0
]
.
Dee
p
lear
n
i
n
g
with
s
eq
u
en
t
ial
ar
r
an
g
e
m
en
t
[
2
1
]
in
teg
r
atin
g
with
L
STM
[
2
2
]
e
n
s
u
r
es
d
ata
d
ep
e
n
d
en
cies
o
v
e
r
tim
e
,
an
d
also
r
ea
l
-
tim
e
-
b
ased
d
ee
p
lear
n
in
g
m
o
d
els
[
2
3
]
,
[
2
4
]
en
s
u
r
e
E
E
G
d
etec
tio
n
in
a
cr
u
cial
tim
e
f
r
am
e
[
2
5
]
.
Mo
s
t
o
f
th
e
wo
r
k
i
m
p
lem
en
ts
E
E
G
d
etec
tio
n
b
y
in
teg
r
atin
g
d
ee
p
lear
n
in
g
a
n
d
a
b
id
ir
ec
tio
n
al
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
B
iLST
M)
m
o
d
el
to
an
aly
ze
s
p
atial
r
elatio
n
s
h
ip
am
o
n
g
E
E
G
s
ig
n
a
l b
y
C
NN
,
an
d
th
e
n
tem
p
o
r
al
an
aly
s
is
b
y
u
s
in
g
B
iLST
M
[
2
6
]
–
[
2
8
]
.
All
th
e
m
en
tio
n
ed
r
ev
iew
ar
ti
cles
en
h
an
ce
th
e
ac
cu
r
ate
E
E
G
d
etec
tio
n
by
in
teg
r
atin
g
o
p
tim
izatio
n
tech
n
iq
u
es,
f
ea
tu
r
e
ex
tr
ac
tio
n
m
eth
o
d
s
,
a
n
d
s
tatis
tical
ap
p
r
o
ac
h
es
to
m
ac
h
in
e
in
tellig
en
ce
.
Alth
o
u
g
h
no
t
ab
le
im
p
r
o
v
em
e
n
ts
ar
e
alr
ea
d
y
p
l
ac
ed
in
s
eizu
r
e
d
etec
tio
n
,
s
ev
er
al
s
ig
n
if
ican
t
lim
itatio
n
s
s
till
ex
is
t
.
Nu
m
er
o
u
s
tech
n
iq
u
es
in
v
o
lv
e
in
t
r
icate
p
r
ep
r
o
ce
s
s
in
g
,
wh
ich
h
am
p
er
s
th
eir
p
r
ac
tical
u
s
e
in
r
ea
l
-
ti
m
e
s
y
s
tem
s
.
So
m
e
m
o
d
els ar
e
test
ed
o
n
ly
o
n
s
p
ec
if
ic
d
atasets
,
lim
itin
g
th
eir
g
en
er
aliza
b
ilit
y
ac
r
o
s
s
d
if
f
er
en
t E
E
G
s
ig
n
als.
T
o
ad
d
r
ess
th
e
lim
itatio
n
s
o
f
ea
r
lier
s
eizu
r
e
d
etec
tio
n
m
o
d
els,
we
p
r
o
p
o
s
e
an
e
n
h
an
ce
d
ap
p
r
o
ac
h
th
at
co
m
b
i
n
es
a
C
NN
an
d
B
iLST
M
to
p
e
r
f
o
r
m
th
e
r
o
b
u
s
t
f
ea
tu
r
e
ex
tr
ac
tio
n
,
class
if
icatio
n
,
an
d
p
er
tain
in
g
o
f
tim
e
s
er
i
es
d
ata
elem
en
ts
.
Fu
r
th
er
,
it
u
s
es
an
im
p
r
o
v
ed
m
etah
eu
r
is
tic
o
p
tim
izatio
n
te
ch
n
iq
u
e,
SB
OA
,
to
f
in
e
-
tu
n
e
th
e
m
o
d
el
a
n
d
to
r
ed
u
ce
th
e
p
r
o
ce
s
s
in
g
co
m
p
lex
ity
.
T
h
is
h
y
b
r
id
m
eth
o
d
f
i
n
e
-
tu
n
es
th
e
m
o
d
el
p
ar
am
eter
s
ef
f
icien
tly
,
r
ed
u
cin
g
tr
ain
in
g
lo
s
s
an
d
tim
e
.
Ou
r
f
in
al
r
esu
lt,
ac
h
iev
ed
at
1
0
0
ep
o
ch
s
,
s
h
o
ws
an
ac
cu
r
ac
y
o
f
9
8
.
4
9
%,
s
en
s
itiv
it
y
o
f
9
6
.
0
5
%,
s
p
ec
if
icity
o
f
9
7
.
0
3
%,
Ma
tth
ews
co
r
r
elatio
n
co
ef
f
icien
t
(
MCC
)
o
f
9
7
.
0
1
%,
a
n
d
ar
e
a
u
n
d
er
th
e
c
u
r
v
e
(
AUC
)
o
f
0
.
9
7
.
T
h
e
r
es
t
o
f
o
u
r
ar
ticle
is
elab
o
r
ated
in
s
ec
tio
n
s
2
to
4
.
Sect
io
n
2
is
th
e
c
o
r
e
p
ar
t
o
f
t
h
e
e
x
p
er
im
en
t
ex
p
lain
s
ab
o
u
t
th
e
m
eth
o
d
u
s
ed
with
th
e
s
u
b
-
s
ec
tio
n
s
o
f
2
.
1
th
at
elab
o
r
ates
f
lo
w
o
f
wo
r
k
,
2
.
2
elab
o
r
ates
th
e
f
ea
tu
r
es
o
f
ex
p
er
im
en
tal
B
o
n
n
E
E
G
d
ataset,
2
.
3
s
tates
th
e
t
ec
h
n
iq
u
es
o
f
E
E
G
d
ata
p
r
e
-
p
r
o
ce
s
s
in
g
,
2
.
4
s
tates
d
ata
au
g
m
en
tatio
n
,
2
.
5
s
tates
ab
o
u
t
t
h
e
v
o
tin
g
m
o
d
els
,
2
.
6
p
r
esen
ts
th
e
B
iLST
M
n
etwo
r
k
,
an
d
2
.
7
p
r
esen
ts
th
e
o
p
tim
ized
SB
OA
alg
o
r
ith
m
.
Sectio
n
3
p
r
esen
ts
ex
p
er
im
en
tal
r
esu
lts
an
d
d
is
cu
s
s
io
n
s
,
an
d
s
ec
tio
n
4
s
tates
th
e
co
n
clu
s
io
n
a
n
d
f
u
tu
r
e
s
co
p
e
o
f
th
e
p
ap
er
.
2.
M
E
T
H
O
D
2
.
1
.
F
l
o
w
o
f
wo
r
k
T
h
is
p
r
o
p
o
s
ed
m
o
d
el
f
o
r
E
E
G
d
etec
tio
n
,
d
ep
icted
in
Fig
u
r
e
1
,
is
a
f
r
am
ewo
r
k
t
o
o
p
tim
ize
th
e
test
ac
cu
r
ac
y
an
d
to
r
ed
u
ce
th
e
p
r
o
ce
s
s
in
g
co
m
p
lex
ity
.
At
f
ir
s
t,
it
ap
p
lies
th
e
s
eq
u
en
ce
o
f
th
e
p
r
e
-
p
r
o
ce
s
s
in
g
p
ip
elin
e
to
im
p
r
o
v
e
d
ata
q
u
al
ity
,
f
o
llo
wed
b
y
d
ata
au
g
m
en
tatio
n
f
o
r
im
p
r
o
v
in
g
th
e
m
o
d
el’
s
g
en
er
aliza
tio
n
.
T
h
en
it
tr
ies f
o
r
d
ee
p
lear
n
in
g
m
o
d
els f
o
r
class
if
icatio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ad
v
Ap
p
l Sci
I
SS
N:
2252
-
8
8
1
4
I
mp
r
o
ve
d
s
eizu
r
e
d
etec
tio
n
u
s
in
g
o
p
timiz
ed
time
s
eq
u
en
ce
b
a
s
ed
d
ee
p
lea
r
n
in
g
…
(
P
u
s
p
a
n
j
a
li Ma
llik)
199
2
.
2
.
Clini
ca
l da
t
a
s
et
s
T
h
e
B
o
n
n
U
n
i
v
e
r
s
i
t
y
d
a
ta
s
et
,
w
h
i
c
h
i
s
c
o
ll
e
c
t
e
d
f
r
o
m
P
h
y
s
i
o
n
e
t
,
i
s
a
p
u
b
l
i
c
l
y
a
v
a
il
a
b
l
e
E
E
G
d
a
t
a
b
a
s
e
c
e
n
t
e
r
[
2
1
]
.
I
t
i
s
a
m
u
l
t
i
cl
a
s
s
E
E
G
s
i
g
n
al
d
a
t
a
b
as
e
.
T
h
e
r
e
ar
e
f
i
v
e
c
l
as
s
es
p
r
e
s
e
n
t
i
n
t
h
e
d
a
t
a
s
et
s
,
cl
a
s
s
A
t
o
c
l
a
s
s
E
,
a
n
d
e
ac
h
d
a
t
as
e
t
c
o
n
ta
i
n
s
1
0
0
t
x
t
f
il
es
w
it
h
4
,
0
9
6
s
a
m
p
l
e
s
i
n
A
SC
I
I
f
o
r
m
a
t
.
C
l
as
se
s
A
t
o
D
c
o
n
t
a
i
n
n
o
r
m
a
l
s
i
g
n
a
l
a
n
d
c
l
as
s
E
c
o
n
t
a
i
n
s
s
e
i
z
u
r
e
s
i
g
n
a
l
s
.
T
h
e
r
e
a
r
e
1
0
0
c
h
a
n
n
e
l
s
p
r
e
s
e
n
t
in
e
a
c
h
d
a
t
a
s
et
.
F
o
r
c
o
l
l
e
c
ti
n
g
E
E
G
s
i
g
n
a
l
,
t
h
e
e
l
ec
t
r
o
d
e
s
a
r
e
k
e
p
t
o
v
e
r
t
h
e
h
e
a
d
o
f
t
h
e
p
a
t
i
e
n
t
a
n
d
i
n
s
i
d
e
i
n
t
r
a
c
r
a
n
i
a
l
r
e
g
i
o
n
s
o
f
t
h
e
h
e
a
d
f
o
r
a
t
i
m
e
p
e
r
i
o
d
o
f
2
3
.
6
s
e
c
o
n
d
s
.
F
i
g
u
r
e
2
d
e
p
i
c
t
s
t
h
e
s
am
p
l
e
i
m
a
g
e
s
c
o
l
l
e
ct
e
d
f
r
o
m
E
E
G
p
a
t
i
e
n
ts
.
Fig
u
r
e
1
.
Pro
p
o
s
ed
m
o
d
el
d
es
cr
ib
in
g
th
e
f
lo
w
o
f
th
e
wo
r
k
Fig
u
r
e
2
.
B
o
n
n
m
u
lticlas
s
E
E
G
s
ig
n
al
im
ag
es
2
.
3
.
Da
t
a
pre
-
pro
ce
s
s
ing
Data
p
r
e
-
p
r
o
ce
s
s
in
g
is
a
p
ip
elin
e
o
f
co
n
v
er
s
io
n
an
d
f
iltr
atio
n
tech
n
iq
u
es
th
at
co
m
es
af
ter
d
ata
co
llectio
n
to
b
o
o
s
t
th
e
ac
cu
r
a
cy
an
d
c
o
n
s
is
ten
cy
o
f
th
e
test
r
esu
lt
.
I
n
itially
,
th
e
im
ag
es
wer
e
co
n
v
er
te
d
to
g
r
ay
s
ca
le,
f
o
llo
wed
b
y
n
o
is
e
r
ed
u
ctio
n
u
s
in
g
Gau
s
s
i
an
an
d
m
ed
ian
f
ilter
in
g
.
C
o
n
tr
ast
was
th
en
en
h
an
ce
d
u
s
in
g
co
n
tr
ast
lim
ited
a
d
o
p
tiv
e
h
is
to
g
r
am
eq
u
aliza
tio
n
(
C
L
AHE
)
to
h
ig
h
lig
h
t
im
p
o
r
tan
t
f
ea
tu
r
es
.
T
h
e
im
ag
es
wer
e
r
esized
to
a
s
tan
d
ar
d
d
im
en
s
io
n
,
an
d
m
in
-
m
ax
n
o
r
m
aliza
tio
n
was
ap
p
lied
to
s
ca
le
p
ix
el
v
alu
es
u
n
if
o
r
m
ly
,
en
s
u
r
i
n
g
r
ea
d
in
ess
f
o
r
ac
cu
r
ate
an
d
e
f
f
icien
t a
n
al
y
s
is
.
Fig
u
r
e
3
s
h
o
ws th
e
p
r
e
-
p
r
o
ce
s
s
ed
im
ag
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
8
1
4
I
n
t J Ad
v
Ap
p
l Sci
,
Vo
l
.
1
5
,
No
.
1
,
Ma
r
c
h
2
0
2
6
:
197
-
2
0
7
200
2
.
4
.
Da
t
a
a
ug
m
ent
a
t
io
n
T
o
en
h
an
ce
t
h
e
v
a
r
iab
ilit
y
o
f
E
E
G
im
ag
e
s
am
p
les
a
n
d
am
elio
r
ate
th
e
m
o
d
el’
s
g
e
n
er
aliza
tio
n
ca
p
ab
ilit
y
,
a
c
o
m
p
r
e
h
en
s
iv
e
d
ata
au
g
m
e
n
tatio
n
s
tr
ateg
y
was
em
p
lo
y
ed
.
T
h
is
p
r
o
ce
s
s
h
elp
s
to
m
itig
ate
o
v
er
f
itti
n
g
b
y
in
tr
o
d
u
cin
g
d
i
v
er
s
e
v
is
u
al
p
atter
n
s
in
to
th
e
tr
ain
in
g
d
ataset.
T
h
e
ap
p
lied
tr
an
s
f
o
r
m
atio
n
s
in
clu
d
ed
im
ag
e
r
o
tatio
n
,
h
o
r
i
zo
n
tal
an
d
v
er
tical
f
lip
p
in
g
,
r
esizin
g
,
co
lo
r
jitt
er
in
g
,
th
e
ad
d
itio
n
o
f
Gau
s
s
ian
n
o
is
e,
an
d
in
ten
s
ity
m
o
d
i
f
icatio
n
s
.
T
h
ese
au
g
m
e
n
tatio
n
s
s
im
u
late
r
ea
lis
tic
alter
atio
n
s
th
at
c
o
u
ld
o
cc
u
r
in
E
E
G
im
ag
in
g
c
o
n
d
itio
n
s
,
th
er
e
b
y
en
ab
lin
g
th
e
m
o
d
el
to
lear
n
m
o
r
e
r
o
b
u
s
t
a
n
d
g
en
e
r
alize
d
r
e
p
r
esen
tatio
n
s
.
E
x
am
p
les o
f
th
es
e
a
u
g
m
en
te
d
im
ag
es a
r
e
illu
s
tr
ated
in
Fig
u
r
e
4
.
Fig
u
r
e
3
.
Pre
-
p
r
o
ce
s
s
ed
im
ag
e
o
f
E
E
G
s
ig
n
al
Fig
u
r
e
4
.
Data
au
g
m
en
tatio
n
im
ag
es
2
.
5
.
Vo
t
ing
mo
del
Af
ter
th
e
d
ata
au
g
m
en
tatio
n
p
h
ase,
th
e
en
s
em
b
le
lear
n
in
g
m
eth
o
d
is
ap
p
lied
to
th
e
s
eizu
r
e
s
ig
n
al
to
g
et
th
e
o
p
tim
u
m
r
esu
lt
.
Her
e
,
th
e
en
s
em
b
le
lear
n
in
g
tech
n
iq
u
e
u
s
es
D
-
Net,
R
esNet,
G
-
Net
,
an
d
C
NN
as
th
e
v
o
tin
g
class
if
ier
,
an
d
th
e
r
esu
lt
in
g
o
u
tp
u
t
is
th
e
co
m
b
in
ed
m
ajo
r
ity
o
f
ea
ch
o
f
th
e
v
o
t
in
g
class
if
ier
s
.
T
h
e
1D
-
C
NN
u
s
ed
h
er
e
is
a
s
eq
u
e
n
tial
C
NN
b
u
il
t
u
p
with
lay
er
s
s
u
ch
as
C
o
n
v
1
D,
Ma
x
Po
o
lin
g
1
D,
f
latten
,
d
en
s
e
,
an
d
d
r
o
p
o
u
t
lay
er
s
.
T
h
e
C
o
n
v
1
D
lay
er
u
s
es
6
4
f
ilter
s
,
k
er
n
el
s
ize
is
3
,
ac
tiv
atio
n
f
u
n
ctio
n
u
s
ed
is
“
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
eL
U)
,
”
an
d
it
p
r
o
ce
s
s
es
E
E
G
s
ig
n
als
in
s
eq
u
e
n
tial
o
r
d
er
.
I
n
th
e
Ma
x
Po
o
lin
g
1
D
lay
er
,
C
NN
u
s
es
a
win
d
o
w
o
f
p
o
o
l
s
ize
2
.
T
h
e
f
latten
ed
lay
er
r
e
-
ar
r
an
g
es
th
e
ex
tr
ac
ted
f
ea
tu
r
es
an
d
th
e
n
f
latten
s
th
e
ch
an
n
el
d
im
en
s
io
n
.
T
h
e
d
en
s
e
lay
er
u
s
es
2
5
6
n
eu
r
o
n
s
,
an
d
th
e
ac
tiv
atio
n
f
u
n
ctio
n
u
s
ed
h
e
r
e
is
R
eL
U
.
R
eL
U
p
r
o
ce
s
se
s
th
e
s
ig
n
als
an
d
h
el
p
s
th
e
C
NN
to
lear
n
n
o
n
lin
ea
r
p
atter
n
s
.
T
h
e
d
r
o
p
o
u
t
lay
er
u
s
es
r
eg
u
lar
izatio
n
tech
n
iq
u
e
f
o
r
p
r
ev
e
n
tin
g
o
v
er
f
itti
n
g
in
C
NN
.
I
t
h
as
a
d
r
o
p
o
u
t
r
ate
o
f
0
.
5
.
T
h
is
C
NN
u
s
es
d
en
s
e
l
ay
er
s
an
d
d
r
o
p
o
u
t
lay
er
alter
n
ativ
ely
.
T
h
e
r
esu
lt
is
s
p
ec
if
ied
in
T
ab
le
1
.
Fig
u
r
e
5
d
ep
icts
th
e
s
tep
s
o
f
s
elec
tin
g
th
e
id
ea
l
v
o
tin
g
m
o
d
el
an
d
T
ab
le
1
s
to
r
es th
e
ac
cu
r
ac
y
,
l
o
s
s
an
d
v
al
-
l
o
s
s
v
alu
es o
f
th
e
ch
o
s
en
v
o
tin
g
m
o
d
els.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ad
v
Ap
p
l Sci
I
SS
N:
2252
-
8
8
1
4
I
mp
r
o
ve
d
s
eizu
r
e
d
etec
tio
n
u
s
in
g
o
p
timiz
ed
time
s
eq
u
en
ce
b
a
s
ed
d
ee
p
lea
r
n
in
g
…
(
P
u
s
p
a
n
j
a
li Ma
llik)
201
T
ab
le
1
.
Vo
tin
g
class
if
ier
p
er
f
o
r
m
an
ce
r
esu
lt
La
y
e
r
O
u
t
p
u
t
C
o
n
v
1
D
(
1
8
,
6
4
)
M
a
x
P
o
o
l
i
n
g
1
D
(
9
,
6
4
)
C
o
n
v
1
D
(
7
,
1
2
8
)
M
a
x
P
o
o
l
i
n
g
1
D
(
3
,
1
2
8
)
F
l
a
t
t
e
n
(
3
8
4
)
D
e
n
se
(
2
5
6
)
D
r
o
p
o
u
t
(
2
5
6
)
D
e
n
se
(
1
2
8
)
D
r
o
p
o
u
t
(
1
2
8
)
D
e
n
se
(
1
)
Fig
u
r
e
5
.
Step
s
o
f
ch
o
o
s
in
g
b
e
s
t
v
o
tin
g
m
o
d
el
as C
NN
S
u
cc
ess
f
u
lly
r
u
n
th
e
m
o
d
els
lis
ted
in
T
ab
le
2
,
an
d
af
ter
co
m
p
letio
n
,
it
is
f
o
u
n
d
th
at
th
e
r
esu
lt
o
b
tain
ed
i
n
C
NN
m
o
d
el
s
u
p
e
r
s
ed
es
th
e
r
esu
lt
s
with
D
-
Net,
R
esNet
,
an
d
G
-
Net.
T
h
e
ac
cu
r
ate
r
ate
o
f
C
NN
is
8
9
.
0
8
,
lo
s
s
v
alu
e
is
0
.
1
6
,
an
d
v
al
-
lo
s
s
v
alu
e
is
0
.
1
5
.
T
h
en
,
test
th
e
o
th
er
p
er
f
o
r
m
an
ce
m
etr
ics
to
p
r
o
v
e
th
e
s
u
p
er
io
r
ity
o
f
C
NN
m
o
d
el
am
o
n
g
D
-
Net,
R
esNet
,
an
d
G
-
Ne
t
m
o
d
els.
T
h
e
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
es
in
ter
m
s
o
f
ac
cu
r
ac
y
,
s
en
s
itiv
ity
,
s
p
ec
if
ici
ty
,
an
d
AUC s
co
r
es a
r
e
l
is
ted
in
T
ab
le
3
.
T
ab
le
2
.
Acc
u
r
ac
y
,
lo
s
s
,
an
d
v
al
-
lo
s
s
o
f
v
o
tin
g
m
o
d
els
D
-
N
e
t
R
e
sN
e
t
G
-
N
e
t
C
N
N
A
c
y
Lo
ss
V
-
l
o
ss
A
c
y
Lo
ss
V
-
l
o
ss
A
c
y
Lo
ss
V
-
l
o
ss
A
c
y
Lo
ss
V
-
l
o
ss
Te
st
1
7
5
.
0
2
0
.
2
5
0
.
3
3
7
9
.
0
2
0
.
1
7
0
.
3
2
8
0
.
0
1
0
.
2
7
0
.
3
4
8
9
.
0
1
0
.
2
6
0
.
3
6
Te
st
2
7
5
.
0
4
0
.
2
4
0
.
2
1
7
9
.
0
2
0
.
1
6
0
.
3
2
8
0
.
0
2
0
.
2
8
0
.
2
4
8
9
.
0
3
0
.
2
1
0
.
3
2
Te
st
3
7
5
.
0
6
0
.
2
2
0
.
2
5
7
9
.
0
4
0
.
2
3
0
.
3
1
8
0
.
0
5
0
.
2
5
0
.
3
3
8
9
.
0
4
0
.
1
8
0
.
2
5
Te
st
4
7
5
.
0
6
0
.
1
8
0
.
2
3
7
9
.
0
5
0
.
2
8
0
.
3
1
8
0
.
0
6
0
.
2
1
0
.
2
9
8
9
.
0
6
0
.
1
9
0
.
2
9
Te
st
5
7
5
.
8
2
0
.
1
7
0
.
2
1
7
9
.
0
6
0
.
2
2
0
.
2
9
8
0
.
0
7
0
.
1
9
0
.
2
8
8
9
.
0
8
0
.
1
6
0
.
1
5
T
ab
le
3
.
Per
f
o
r
m
an
ce
m
etr
ics o
f
p
r
e
f
er
r
ed
m
o
d
els
M
o
d
e
l
s
A
c
c
u
r
a
c
y
(
%)
S
e
n
s
i
t
i
v
i
t
y
(
%)
S
p
e
c
i
f
i
c
i
t
y
(
%)
M
C
C
AUC
D
-
N
e
t
7
5
.
8
2
7
8
.
6
2
7
6
.
0
9
7
8
.
0
3
0
.
7
7
R
e
sN
e
t
7
9
.
0
6
7
9
.
1
9
7
9
.
2
8
8
0
.
3
2
0
.
7
9
G
-
N
e
t
8
0
.
0
7
8
0
.
0
9
8
1
.
0
3
8
0
.
0
4
0
.
8
0
C
N
N
8
9
.
0
8
8
9
.
0
1
9
0
.
0
1
8
9
.
0
9
0
.
8
9
2
.
6
.
B
idi
re
ct
io
na
l
lo
ng
s
ho
rt
-
t
er
m
m
emo
ry
A
f
t
e
r
t
h
e
s
i
g
n
al
s
a
r
e
e
x
t
r
a
ct
e
d
b
y
v
o
t
i
n
g
m
o
d
e
l
s
,
i
t
r
e
q
u
i
r
e
s
an
a
p
p
r
o
p
r
i
a
t
e
o
r
d
e
r
i
n
g
o
f
r
e
p
r
e
s
e
n
t
a
ti
o
n
i
s
r
e
q
u
i
r
e
d
.
H
e
r
e
,
t
h
e
e
x
t
r
ac
t
e
d
E
E
G
s
i
g
n
a
ls
u
n
d
e
r
w
e
n
t
t
h
r
o
u
g
h
a
t
i
m
e
s
e
r
i
es
s
e
q
u
e
n
t
i
a
l
r
ep
r
e
s
e
n
t
a
ti
o
n
o
f
d
a
ta
o
r
d
e
r
i
n
g
,
c
a
l
l
e
d
a
s
B
i
L
S
T
M
,
w
h
i
c
h
c
a
n
s
u
c
c
e
s
s
f
u
l
l
y
h
a
n
d
l
e
p
e
r
s
i
s
t
e
n
t
d
e
p
e
n
d
e
n
c
i
es
in
t
i
m
e
s
e
q
u
e
n
c
i
n
g
d
a
t
a
[
2
6
]
a
s
s
p
e
c
i
f
i
e
d
i
n
T
a
b
l
e
4
,
a
n
d
F
i
g
u
r
e
6
d
e
p
i
c
t
s
a
r
c
h
i
te
c
t
u
r
e
o
f
B
i
L
S
T
M
m
o
d
e
l
.
B
i
L
S
T
M
i
n
t
e
g
r
a
te
s
t
h
e
s
i
g
n
a
l
f
l
o
w
o
f
t
w
o
L
S
T
M
S
,
o
n
e
i
n
f
o
r
w
a
r
d
d
i
r
e
c
t
i
o
n
a
n
d
o
t
h
er
i
n
r
e
v
e
r
s
e
d
i
r
e
ct
i
o
n
,
w
it
h
a
s
in
g
l
e
o
u
t
p
u
t
l
a
y
e
r
.
T
ab
le
4
.
Dep
e
n
d
en
cies in
tim
e
s
eq
u
en
cin
g
d
ata
La
y
e
r
O
u
t
p
u
t
I
n
p
u
t
(
1
7
8
,
1
)
D
e
n
se
(
1
7
8
,
3
2
)
B
i
d
i
r
e
c
t
i
o
n
(
2
5
6
)
D
r
o
p
o
u
t
(
2
5
6
)
B
a
t
c
h
_
n
o
r
m
a
l
i
z
a
t
i
o
n
(
2
5
6
)
D
e
n
se
(
6
4
)
D
r
o
p
o
u
t
(
6
4
)
B
a
t
c
h
_
n
o
r
m
a
l
i
z
a
t
i
o
n
(
6
4
)
D
e
n
se
(
2
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
8
1
4
I
n
t J Ad
v
Ap
p
l Sci
,
Vo
l
.
1
5
,
No
.
1
,
Ma
r
c
h
2
0
2
6
:
197
-
2
0
7
202
Fig
u
r
e
6
.
B
iLST
M
ar
ch
itectu
r
e
2
.
7
.
O
pti
m
ized
SB
O
A
a
lg
o
rit
hm
Secr
etar
y
b
ir
d
is
an
Af
r
ica
n
s
tr
ik
in
g
b
ir
d
wh
ic
h
lo
o
k
s
s
i
m
ilar
to
th
e
ea
g
le
b
ir
d
,
h
a
v
in
g
win
g
s
o
f
g
r
ey
-
b
r
o
wn
f
ea
t
h
er
s
,
ch
est
is
wh
ite
in
co
lo
r
,
an
d
b
elie
p
ar
t
is
d
ee
p
b
lack
in
co
lo
r
.
T
h
e
m
o
s
t
p
ec
u
liar
ch
ar
ac
ter
is
tic
o
f
s
ec
r
etar
y
b
ir
d
is
its
h
u
n
tin
g
s
ty
le
an
d
ev
ad
i
n
g
s
ty
le
wh
ile
it
tr
a
v
els
th
r
o
u
g
h
g
r
ass
lan
d
s
.
T
h
ese
two
n
atu
r
es
ar
e
m
ath
em
atica
lly
f
o
r
m
u
lated
i
n
th
e
f
o
r
m
o
f
ex
p
lo
r
atio
n
a
n
d
ex
p
lo
itatio
n
.
W
h
ile
it
p
r
ay
s
th
e
s
n
ak
es
in
th
e
g
r
ass
lan
d
s
,
its
h
au
n
tin
g
s
ty
le
is
d
ef
i
n
ed
as
e
x
p
lo
r
atio
n
p
h
ase
in
th
e
g
lo
b
a
l
s
ea
r
ch
s
p
ac
e
,
an
d
wh
ile
it
escap
es
f
r
o
m
t
h
e
r
o
d
e
n
ts
,
it
is
ca
lled
as
ex
p
lo
itatio
n
in
th
e
lo
ca
l
s
ea
r
ch
s
p
ac
e
.
B
o
th
o
f
th
e
ex
p
lo
r
atio
n
an
d
e
x
p
lo
itatio
n
ar
e
th
e
m
aj
o
r
co
n
tr
ib
u
tio
n
s
o
f
t
h
e
s
ec
r
etar
y
b
ir
d
'
s
ch
ar
ac
ter
is
tics
to
s
o
lv
i
n
g
th
e
o
p
tim
izatio
n
p
r
o
b
lem
s
i
n
th
e
s
ea
r
ch
s
p
ac
e
.
Ou
r
p
r
o
p
o
s
ed
m
o
d
el
u
s
es
th
is
o
p
tim
ize
d
alg
o
r
ith
m
to
r
ed
u
ce
th
e
test
in
g
co
m
p
lex
ity
a
n
d
to
im
p
r
o
v
e
th
e
ac
cu
r
ac
y
r
ate
,
as sh
o
wn
in
Al
g
o
r
ith
m
1
.
Alg
o
r
itm
1
.
C
NN
-
B
iLST
M
-
S
B
OA
m
o
d
el
I
n
p
u
t:
th
e
r
aw
E
E
G
s
ig
n
al
Ou
tp
u
t:
ex
tr
ac
ted
s
eizu
r
e
E
E
G
s
ig
n
al
an
d
n
o
r
m
al
E
E
G
s
ig
n
al
L
is
t N
// T
h
e
n
u
m
b
er
o
f
f
ea
tu
r
es
=
1
∶
{
E
x
tr
ac
t th
e
f
ea
tu
r
es with
PC
A
Sto
r
e
th
e
tim
e
s
er
ies s
eq
u
en
ce
o
f
d
ata
u
s
in
g
B
iLST
M
C
las
s
if
y
th
e
f
ea
tu
r
es with
C
NN
Op
tim
ize
th
e
p
er
f
o
r
m
an
ce
u
s
in
g
SB
OA
}
Fin
d
th
e
r
esu
lt
2
.
7
.
1
.
M
a
t
hem
a
t
ic
a
l
m
o
delli
ng
T
h
e
m
ath
em
atica
l m
o
d
ellin
g
o
f
th
e
SB
OA
is
p
r
esen
ted
as f
o
llo
ws:
i)
I
n
itial
p
h
ase
: t
h
e
r
an
d
o
m
i
n
itializatio
n
o
f
th
e
p
o
s
itio
n
o
f
s
ec
r
etar
y
b
ir
d
is
in
(
1
)
.
,
=
+
×
(
−
)
,
=
1
,
2
,
.
.
.
,
,
=
1
,
2
,
.
.
.
,
(
1
)
W
h
er
e
is
th
e
p
o
s
itio
n
o
f
th
e
ℎ
Secr
etar
y
b
ir
d
an
d
ar
e
th
e
u
p
p
er
a
n
d
l
o
wer
b
o
u
n
d
s
,
a
n
d
r
i
s
a
r
an
d
o
m
n
u
m
b
er
b
etwe
en
0
a
n
d
1
.
ii)
Hau
n
tin
g
s
tr
ateg
y
o
f
s
ec
r
etar
y
b
ir
d
(
ex
p
lo
r
atio
n
p
h
ase)
:
t
h
is
p
h
ase
is
d
iv
id
e
d
i
n
to
th
r
ee
s
ta
g
es:
s
ea
r
ch
in
g
th
e
p
r
ey
,
in
g
esti
n
g
th
e
p
r
ey
,
a
n
d
attac
k
in
g
th
e
p
r
e
y
.
T
h
e
to
t
al
h
au
n
tin
g
tim
e
is
eq
u
ally
d
iv
id
ed
in
to
tim
e
in
ter
v
als
as in
(
2
)
.
<
1
3
,
1
3
<
<
2
3
an
d
2
3
<
<
.
(
2
)
W
h
er
e
t
is
th
e
cu
r
r
en
t iter
atio
n
,
an
d
T
is
th
e
to
tal
n
u
m
b
er
if
iter
atio
n
.
iii)
Up
d
atin
g
s
ec
r
etar
y
b
ir
d
’
s
p
o
s
itio
n
<
1
3
,
,
1
=
,
+
(
_
1
−
_
2
)
×
1
(
3
)
=
{
,
1
,
1
<
,
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ad
v
Ap
p
l Sci
I
SS
N:
2252
-
8
8
1
4
I
mp
r
o
ve
d
s
eizu
r
e
d
etec
tio
n
u
s
in
g
o
p
timiz
ed
time
s
eq
u
en
ce
b
a
s
ed
d
ee
p
lea
r
n
in
g
…
(
P
u
s
p
a
n
j
a
li Ma
llik)
203
C
o
n
s
u
m
in
g
th
e
p
r
ey
is
m
ath
e
m
atica
lly
ex
p
r
ess
ed
as
(
4
)
-
(
1
0
)
.
=
(
1
,
)
(
4
)
W
h
ile
1
3
<
<
2
3
,
,
1
=
+
(
(
⁄
)
⋀
4
)
×
(
−
0
.
5
)
×
(
−
,
)
(
5
)
=
{
,
1
,
1
<
,
,
(
6
)
W
h
i
le
>
2
3
,
,
1
=
+
(
(
1
−
)
⋀
(
2
×
)
)
×
,
×
(
7
)
=
{
,
1
,
1
<
,
,
(
8
)
=
0
.
5
×
(
)
(
9
)
(
)
=
×
×
|
|
1
(
1
0
)
W
h
er
e
S
is
a
f
ix
ed
c
o
n
s
tan
t
.
T
h
en
is
d
ef
in
ed
as
(
1
1
)
.
=
[
Γ
(
1
+
)
×
(
2
)
Γ
(
1
+
2
)
×
×
2
(
−
1
2
)
]
1
(
1
1
)
iv
)
E
s
ca
p
e
s
tr
ateg
y
(
ex
p
l
o
itatio
n
p
h
ase)
:
t
h
e
escap
e
s
tr
ateg
y
in
tr
o
d
u
ce
s
a
p
e
r
tu
r
b
atio
n
f
ac
to
r
(
1
−
)
2
.
T
h
e
m
ath
em
atica
l f
o
r
m
u
latio
n
o
f
th
e
escap
e
s
tr
ateg
y
is
(
1
2
)
-
(
1
3
)
.
,
2
=
{
1
:
+
(
2
×
−
1
)
×
(
1
−
)
2
×
,
,
<
2
:
,
,
+
2
×
(
−
×
,
,
)
,
(
1
2
)
=
{
,
2
,
2
<
,
,
(
1
3
)
3.
E
XP
E
R
I
M
E
N
T
A
L
RE
SUL
T
S AN
D
D
I
SC
USS
I
O
N
T
h
e
i
m
p
l
e
m
e
n
t
a
ti
o
n
a
n
d
t
r
a
i
n
i
n
g
o
f
t
h
e
p
r
o
p
o
s
e
d
C
N
N
-
B
i
L
S
T
M
-
SB
OA
m
o
d
e
l
w
e
r
e
c
a
r
r
ie
d
o
u
t
o
n
a
h
i
g
h
-
p
e
r
f
o
r
m
a
n
c
e
c
o
m
p
u
t
i
n
g
s
y
s
t
e
m
t
o
e
n
s
u
r
e
e
f
f
i
c
i
e
n
t
e
x
e
cu
t
i
o
n
a
n
d
a
c
c
u
r
a
t
e
c
o
n
v
e
r
g
e
n
c
e
.
T
h
e
s
y
s
t
e
m
w
as
e
q
u
i
p
p
e
d
w
i
t
h
a
n
I
n
t
e
l
C
o
r
e
i
7
1
1
t
h
g
e
n
e
r
a
t
i
o
n
p
r
o
c
es
s
o
r
,
3
2
G
B
o
f
R
A
M
,
a
n
d
a
n
NV
I
D
I
A
G
e
F
o
r
c
e
R
T
X
3
0
8
0
G
P
U
w
i
t
h
1
0
GB
o
f
d
e
d
i
c
at
e
d
m
e
m
o
r
y
f
o
r
a
c
c
e
le
r
a
t
e
d
c
o
m
p
u
t
a
t
i
o
n
.
T
h
e
m
o
d
e
l
w
as
d
e
v
e
l
o
p
e
d
u
s
i
n
g
P
y
t
h
o
n
3
.
9
i
n
a
T
e
n
s
o
r
F
l
o
w
a
n
d
K
e
r
a
s
e
n
v
i
r
o
n
m
e
n
t
,
e
x
e
c
u
t
e
d
o
n
U
b
u
n
t
u
2
0
.
0
4
L
T
S
.
A
l
l
e
x
p
e
r
i
m
e
n
t
s
w
e
r
e
r
u
n
w
i
t
h
c
o
m
p
u
t
e
u
n
i
f
i
e
d
d
e
v
i
c
e
a
r
c
h
i
t
e
c
t
u
r
e
(
C
U
DA
)
a
n
d
C
U
D
A
d
e
e
p
n
e
u
r
a
l
n
e
tw
o
r
k
(
c
u
D
N
N
)
s
u
p
p
o
r
t
to
l
e
v
e
r
a
g
e
G
PU
-
b
a
s
e
d
p
a
r
al
l
e
l
p
r
o
c
e
s
s
i
n
g
a
n
d
r
e
d
u
c
e
t
r
a
i
n
i
n
g
t
im
e
.
T
o
tr
ain
th
e
p
r
o
p
o
s
ed
m
o
d
el
ef
f
ec
tiv
ely
,
we
in
itially
co
llected
a
to
tal
o
f
4
,
0
9
6
E
E
G
i
m
ag
es
.
T
o
en
h
an
ce
th
e
d
at
aset'
s
d
iv
er
s
it
y
an
d
im
p
r
o
v
e
m
o
d
el
g
en
er
al
izatio
n
,
we
ap
p
lied
s
ix
au
g
m
en
tatio
n
tech
n
iq
u
es,
r
esu
ltin
g
in
an
ex
p
a
n
d
ed
d
at
aset
o
f
2
8
,
6
7
2
im
ag
es
.
T
h
is
en
r
ich
ed
d
ataset
was
th
en
d
i
v
id
ed
in
to
tr
ain
in
g
,
test
in
g
,
an
d
v
alid
atio
n
s
ets
u
s
i
n
g
a
7
0
:
1
5
:1
5
r
atio
.
T
h
is
en
s
u
r
es
th
at
th
e
m
o
d
el
is
tr
ain
ed
o
n
a
lar
g
e
p
o
r
tio
n
o
f
th
e
d
ata
wh
ile
also
b
ei
n
g
e
v
alu
ated
an
d
v
alid
ated
o
n
s
ep
ar
ate
s
u
b
s
ets
to
p
r
e
v
en
t
o
v
er
f
itti
n
g
.
T
a
b
le
5
illu
s
tr
ates
th
e
d
ataset
d
is
tr
ib
u
tio
n
ac
r
o
s
s
ea
ch
ca
teg
o
r
y
,
an
d
T
ab
le
6
s
h
o
ws
th
e
p
er
f
o
r
m
an
ce
v
alu
es
o
f
th
e
v
o
tin
g
m
o
d
els with
d
if
f
e
r
en
t t
est d
u
r
atio
n
s
.
T
ab
le
5
.
Sp
litt
in
g
with
r
atio
S
e
t
P
e
r
c
e
n
t
a
g
e
(
%)
N
u
mb
e
r
o
f
i
m
a
g
e
s
Tr
a
i
n
i
n
g
70
2
0
,
0
7
0
Te
st
i
n
g
15
4
,
3
0
1
V
a
l
i
d
a
t
i
o
n
15
4
,
3
0
1
To
t
a
l
1
0
0
2
8
,
6
7
2
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
8
1
4
I
n
t J Ad
v
Ap
p
l Sci
,
Vo
l
.
1
5
,
No
.
1
,
Ma
r
c
h
2
0
2
6
:
197
-
2
0
7
204
T
ab
le
6
.
Acc
u
r
ac
y
,
lo
s
s
,
an
d
v
al
-
lo
s
s
o
f
v
o
tin
g
m
o
d
els
D
-
N
e
t
R
e
sN
e
t
G
-
N
e
t
C
N
N
A
c
c
Lo
ss
V
a
l
-
l
o
ss
A
c
c
Lo
ss
V
a
l
-
l
o
ss
A
c
c
Lo
ss
V
a
l
-
l
o
ss
A
c
c
Lo
s
s
V
a
l
-
l
o
ss
Te
st
1
7
5
.
0
2
0
.
2
5
0
.
3
3
7
9
.
0
2
0
.
1
7
0
.
3
2
8
0
.
0
1
0
.
2
7
0
.
3
4
8
9
.
0
1
0
.
2
6
0
.
3
6
Te
st
2
7
5
.
0
4
0
.
2
4
0
.
2
1
7
9
.
0
2
0
.
1
6
0
.
3
2
8
0
.
0
2
0
.
2
8
0
.
2
4
8
9
.
0
3
0
.
2
1
0
.
3
2
Te
st
3
7
5
.
0
6
0
.
2
2
0
.
2
5
7
9
.
0
4
0
.
2
3
0
.
3
1
8
0
.
0
5
0
.
2
5
0
.
3
3
8
9
.
0
4
0
.
1
8
0
.
2
5
Te
st
4
7
5
.
0
6
0
.
1
8
0
.
2
3
7
9
.
0
5
0
.
2
8
0
.
3
1
8
0
.
0
6
0
.
2
1
0
.
2
9
8
9
.
0
6
0
.
1
9
0
.
2
9
Te
st
5
7
5
.
8
2
0
.
1
7
0
.
2
1
7
9
.
0
6
0
.
2
2
0
.
2
9
8
0
.
0
7
0
.
1
9
0
.
2
8
8
9
.
0
8
0
.
1
6
0
.
1
5
S
u
cc
ess
f
u
lly
r
u
n
th
e
m
o
d
els
lis
ted
in
T
ab
le
6
,
an
d
af
ter
co
m
p
letio
n
,
it
is
f
o
u
n
d
th
at
th
e
r
esu
lt
o
b
tain
ed
i
n
C
NN
m
o
d
el
s
u
p
e
r
s
ed
es
th
e
r
esu
lt
s
with
D
-
Net,
R
esNet
,
an
d
G
-
Net
.
T
h
e
ac
cu
r
ate
r
ate
o
f
C
NN
is
8
9
.
0
8
,
lo
s
s
v
alu
e
is
0
.
1
6
,
an
d
v
al
-
lo
s
s
v
alu
e
is
0
.
1
5
.
T
h
en
test
th
e
o
th
er
p
er
f
o
r
m
an
ce
m
etr
ics
to
p
r
o
v
e
th
e
s
u
p
er
io
r
ity
o
f
C
NN
m
o
d
el
am
o
n
g
D
-
Net,
R
esNet
,
an
d
G
-
Ne
t
m
o
d
els
.
T
h
e
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
es
in
ter
m
s
o
f
ac
cu
r
ac
y
,
s
en
s
itiv
ity
,
s
p
ec
if
ici
ty
,
an
d
AUC s
co
r
es a
r
e
lis
ted
in
T
ab
le
7.
T
ab
le
7
.
Per
f
o
r
m
an
ce
m
ea
s
u
r
e
m
etr
ics o
f
ch
o
s
en
m
o
d
els
M
o
d
e
l
s
A
c
c
u
r
a
c
y
(
%)
S
e
n
s
i
t
i
v
i
t
y
(
%)
S
p
e
c
i
f
i
c
i
t
y
(
%)
M
C
C
A
U
C
s
c
o
r
e
D
-
N
e
t
7
5
.
8
2
7
8
.
6
2
7
6
.
0
9
7
8
.
0
3
0
.
7
7
R
e
sN
e
t
7
9
.
0
6
7
9
.
1
9
7
9
.
2
8
8
0
.
3
2
0
.
7
9
G
-
N
e
t
8
0
.
0
7
8
0
.
0
9
8
1
.
0
3
8
0
.
0
4
0
.
8
0
C
N
N
8
9
.
0
8
8
9
.
0
1
9
0
.
0
1
8
9
.
0
9
0
.
8
9
Af
ter
an
aly
zin
g
t
h
e
o
u
tco
m
es
o
f
p
er
f
o
r
m
a
n
ce
m
etr
ics
o
f
T
ab
le
s
6
an
d
7
,
we
f
o
u
n
d
th
at
th
e
C
NN
m
o
d
el
is
th
e
b
est
ch
o
ice
o
f
v
o
tin
g
m
o
d
el
f
o
r
E
E
G
s
ig
n
al
d
et
ec
tio
n
with
th
e
B
o
n
n
E
E
G
d
at
aset
.
Alth
o
u
g
h
th
e
r
esu
lts
o
f
T
ab
le
s
6
an
d
7
ar
e
s
u
f
f
icien
t
to
p
r
o
v
e
th
at
C
N
N
c
an
b
e
p
r
ef
er
r
ed
in
E
E
G
d
etec
tio
n
,
an
d
th
e
r
esu
lts
o
f
C
NN
ar
e
also
m
atch
i
n
g
with
th
e
r
esu
lts
o
f
t
h
e
ex
is
tin
g
m
o
d
els,
it
lack
s
a
s
eq
u
en
tial
tim
e
s
er
ies
r
ep
r
esen
tatio
n
o
f
s
ig
n
als
.
T
o
o
v
er
co
m
e
th
is
,
we
in
teg
r
ate
C
NN
with
B
i
L
STM
,
an
d
th
e
h
y
b
r
id
C
NN
-
B
iLST
M
ea
s
ily
p
r
o
ce
s
s
th
e
lo
n
g
s
eq
u
e
n
ce
o
f
tim
e
s
er
ies d
ata
.
T
h
e
ac
c
u
r
ac
y
r
ate
o
f
C
NN
s
tag
n
ated
with
8
9
.
0
8
%
,
wh
ic
h
n
ee
d
s
to
im
p
r
o
v
e
.
Fo
r
im
p
r
o
v
i
n
g
p
e
r
f
o
r
m
an
ce
m
ea
s
u
r
es
o
f
h
y
b
r
id
C
NN
-
Bi
L
STM
m
o
d
el,
we
co
m
b
in
e
it
with
th
e
o
p
tim
izatio
n
alg
o
r
ith
m
,
SB
OA,
an
d
th
e
o
u
tco
m
es
o
f
p
r
o
p
o
s
ed
C
NN
-
B
iLST
M
-
S
B
OA
i
s
lis
ted
in
T
ab
le
8.
T
ab
le
8
.
Pro
p
o
s
ed
C
NN
-
B
iLS
T
M
-
SB
OA
M
o
d
e
l
A
c
c
u
r
a
c
y
(
%)
S
e
n
s
i
t
i
v
i
t
y
(
%)
S
p
e
c
i
f
i
c
i
t
y
(
%)
M
C
C
AUC
s
c
o
r
e
C
N
N
-
B
i
LS
TM
-
S
B
O
A
(
B
o
n
n
d
a
t
a
s
e
t
a
n
d
A
-
E
c
l
a
ss)
9
8
.
4
9
9
6
.
0
5
9
7
.
0
3
9
7
.
0
1
0
.
9
7
T
o
ev
al
u
ate
th
e
g
e
n
er
aliza
tio
n
ab
ilit
y
o
f
th
e
C
NN
-
B
iLST
M
-
SB
OA
m
o
d
el,
k
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
was
em
p
lo
y
ed
.
T
h
e
d
ataset
was
d
iv
id
ed
in
to
k
s
u
b
s
ets,
an
d
th
e
m
o
d
el
was
tr
ain
ed
k
tim
es
,
ea
ch
tim
e
u
s
in
g
a
d
if
f
er
en
t
s
u
b
s
et
as
th
e
v
alid
at
io
n
s
et
.
T
h
e
r
esu
lt
is
p
r
o
d
u
ce
d
in
T
ab
le
9
.
T
h
is
m
eth
o
d
h
el
p
s
to
en
s
u
r
e
r
o
b
u
s
t
p
er
f
o
r
m
an
ce
an
d
p
r
ev
e
n
ts
o
v
er
f
itti
n
g
.
Fo
r
v
alid
atin
g
th
e
test
in
g
o
u
tco
m
es
o
f
o
u
r
p
r
o
p
o
s
ed
m
o
d
el,
we
co
m
p
ar
e
th
e
p
er
f
o
r
m
a
n
ce
m
ea
s
u
r
es
o
f
ac
cu
r
ac
y
,
s
en
s
itiv
ity
,
s
p
ec
if
icity
,
a
n
d
Ku
l
lb
ac
k
-
L
eib
ler
(
KL
)
d
iv
er
g
en
ce
lo
s
s
f
u
n
c
tio
n
v
al
u
es
o
f
o
u
r
m
o
d
el
with
t
h
e
ex
is
t
in
g
m
o
d
els
,
an
d
th
e
r
esu
lt
f
in
d
in
g
s
ar
e
p
r
esen
ted
in
T
ab
le
1
0
.
T
ab
le
9
.
9
-
f
o
ld
v
alid
atio
n
o
v
er
o
u
r
p
r
o
p
o
s
ed
m
o
d
el
F
o
l
d
A
c
c
u
r
a
c
y
(
%)
S
e
n
s
i
t
i
v
i
t
y
(
%)
S
p
e
c
i
f
i
c
i
t
y
(
%)
AUC
Ti
me
t
a
k
e
n
(
sec
)
2
9
7
.
2
3
9
5
.
1
3
9
6
.
6
1
0
.
9
6
1
5
0
3
9
7
.
7
2
9
5
.
5
1
9
6
.
7
2
0
.
9
6
6
1
8
0
5
9
7
.
9
1
9
5
.
7
2
9
6
.
8
6
0
.
9
6
6
2
1
0
7
9
8
.
1
2
9
5
.
9
1
9
6
.
8
9
0
.
9
6
8
2
4
0
9
9
8
.
4
9
9
6
.
0
5
9
7
.
0
3
0
.
9
7
2
7
0
T
ab
le
1
0
.
Per
f
o
r
m
a
n
ce
co
m
p
a
r
is
o
n
with
ex
is
tin
g
m
o
d
els
M
o
d
e
l
s
A
c
c
u
r
a
c
y
(
%)
S
e
n
s
i
t
i
v
i
t
y
(
%)
S
p
e
c
i
f
i
c
i
t
y
(
%)
A
U
C
sc
o
r
e
K
L
d
i
v
e
r
g
e
n
c
e
l
o
ss
M
u
l
t
i
sc
a
l
e
c
o
n
v
o
l
u
t
i
o
n
a
l
[
6
]
9
2
.
5
9
3
.
1
9
1
.
9
9
2
.
4
0
.
3
9
A
l
e
a
r
n
a
b
l
e
a
n
d
e
x
p
l
a
i
n
a
b
l
e
w
a
v
e
l
e
t
[
7
]
8
9
.
3
8
8
.
4
8
8
.
7
0
.
8
8
0
.
4
8
D
e
e
p
l
e
a
r
n
i
n
g
-
b
a
s
e
d
a
t
t
e
n
t
i
o
n
me
c
h
a
n
i
sm
[
8
]
9
8
.
3
8
-
-
-
0
.
3
3
C
h
a
n
n
e
l
-
w
e
i
g
h
t
e
d
sp
a
t
i
a
l
–
t
e
m
p
o
r
a
l
[
9
]
9
7
.
2
3
-
-
-
0
.
3
2
P
r
o
p
o
se
d
C
N
N
-
B
i
LS
TM
-
S
B
O
A
(
A
-
E
c
l
a
ss)
9
8
.
4
9
9
6
.
0
5
9
7
.
0
3
0
.
9
7
0
.
2
9
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Ad
v
Ap
p
l Sci
I
SS
N:
2252
-
8
8
1
4
I
mp
r
o
ve
d
s
eizu
r
e
d
etec
tio
n
u
s
in
g
o
p
timiz
ed
time
s
eq
u
en
ce
b
a
s
ed
d
ee
p
lea
r
n
in
g
…
(
P
u
s
p
a
n
j
a
li Ma
llik)
205
Her
e
KL
d
iv
er
g
e
n
ce
lo
s
s
f
u
n
ct
io
n
is
u
s
ed
to
co
m
p
ar
e
two
d
at
a
d
is
tr
ib
u
tio
n
s
in
th
e
ca
s
e
o
f
a
s
s
es
s
in
g
d
ataset
an
d
m
o
d
el
d
r
if
t.
Ma
th
e
m
atica
lly
,
it is
ex
p
r
ess
ed
as
(
1
4
)
.
(
|
|
)
=
∑
(
)
×
l
og
(
(
)
(
)
)
(
1
4
)
W
h
e
r
e
X
i
s
t
h
e
p
o
s
s
i
b
l
e
e
v
e
n
t
i
n
t
h
e
p
r
o
b
a
b
i
l
i
t
y
s
e
a
r
c
h
s
p
a
ce
,
P(X)
a
n
d
Q
(
X
)
a
r
e
t
h
e
p
r
o
b
a
b
i
l
i
ti
e
s
o
f
X
i
n
t
h
e
d
i
s
t
r
i
b
u
ti
o
n
o
f
P
a
n
d
Q
,
a
n
d
t
h
e
r
a
t
i
o
(
)
(
)
⁄
r
e
p
r
e
s
e
n
t
s
t
h
e
l
i
k
el
i
h
o
o
d
o
f
e
v
e
n
t
X
a
c
c
o
r
d
i
n
g
t
o
P
c
o
m
p
a
r
e
d
t
o
Q
.
T
h
e
R
O
C
-
A
UC
p
l
o
t
a
n
d
l
o
s
s
-
e
p
o
c
h
p
l
o
t
o
f
t
h
e
p
r
o
p
o
s
e
d
m
o
d
e
l
a
r
e
d
e
p
i
c
t
e
d
i
n
F
i
g
u
r
e
s
7
a
n
d
8
.
Fig
u
r
e
7
.
R
OC
-
AUC p
lo
t o
f
C
NN
-
B
iLST
M
-
SB
OA
m
o
d
el
f
o
r
B
o
n
n
E
E
G
(
A
-
E
)
d
ataset
Fig
u
r
e
8
.
Mo
d
el
lo
s
s
o
f
C
NN
-
B
iLST
M
-
S
B
OA
f
o
r
B
o
n
n
(A
-
E
)
d
ataset
4.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
im
p
r
o
v
es
th
e
E
E
G
d
etec
tio
n
b
y
i
n
teg
r
atin
g
d
ee
p
l
ea
r
n
in
g
with
a
n
o
p
tim
izatio
n
tech
n
iq
u
e
.
T
h
e
ap
p
lied
1
D
s
eq
u
en
tial
C
NN
p
r
o
v
es
its
s
u
p
er
io
r
ity
o
v
e
r
o
th
er
v
o
tin
g
m
o
d
els
with
th
e
test
in
g
ac
cu
r
ac
y
o
f
8
9
.
0
8
%,
s
en
s
itiv
ity
o
f
8
9
.
0
1
%,
s
p
ec
if
icity
o
f
9
0
.
0
1
%,
M
C
C
o
f
8
9
.
0
9
%
,
an
d
AUC
s
co
r
e
o
f
0
.
8
9
.
T
h
e
tim
e
s
eq
u
en
ce
o
r
d
er
in
g
o
f
E
E
G
s
ig
n
als
is
in
teg
r
ated
with
a
B
iLST
M
n
etwo
r
k
to
e
n
h
an
ce
t
h
e
s
eq
u
en
tial
d
ata
p
r
o
ce
s
s
in
g
b
y
ad
d
r
ess
in
g
th
e
v
an
is
h
in
g
g
r
ad
ien
t
p
r
o
b
le
m
.
F
u
r
th
er
to
im
p
r
o
v
e
th
e
test
ac
c
u
r
ac
y
,
to
r
ed
u
ce
th
e
o
v
er
-
f
itti
n
g
p
r
o
b
lem
,
an
d
to
i
n
cr
ea
s
e
th
e
lear
n
i
n
g
r
ate,
th
e
h
y
b
r
id
C
NN
-
B
iLST
M
m
o
d
el
is
co
m
b
in
e
d
with
th
e
ch
o
s
en
SB
OA
o
p
tim
izatio
n
te
ch
n
iq
u
e
,
y
ield
i
n
g
an
ac
cu
r
ac
y
o
f
9
8
.
4
9
%,
s
en
s
itiv
ity
o
f
9
6
.
0
5
%,
s
p
ec
if
icity
o
f
9
7
.
0
3
%,
MCC
v
alu
e
o
f
9
7
.
0
1
%,
R
OC
-
AU
C
v
alu
e
o
f
0
.
9
7
,
an
d
KL
d
iv
er
g
e
n
ce
lo
s
s
v
alu
e
o
f
0
.
2
9
.
T
h
ese
p
er
f
o
r
m
an
ce
m
ea
s
u
r
es
o
f
th
e
p
r
o
p
o
s
ed
C
NN
-
B
iLST
M
-
S
B
OA
p
r
o
v
e
its
ef
f
icien
cy
in
ep
i
lep
s
y
d
etec
tio
n
.
T
h
e
f
u
tu
r
e
wo
r
k
will f
o
cu
s
o
n
im
p
r
o
v
ed
an
o
m
aly
d
etec
tio
n
b
y
s
ep
ar
atin
g
n
o
is
e
an
d
ar
tifa
cts f
r
o
m
th
e
E
E
G
s
ig
n
als
an
d
in
teg
r
atin
g
ed
g
e
d
ev
ices f
o
r
ad
v
a
n
ce
d
clin
ical
d
iag
n
o
s
is
.
F
UNDING
I
NF
O
R
M
A
T
I
O
N
T
h
is
r
esear
ch
is
an
ac
ad
em
ic
wo
r
k
.
No
f
u
n
d
r
ec
eiv
ed
eith
er
f
r
o
m
in
ter
n
al
o
r
ex
ter
n
al
a
g
en
cies
to
ca
r
r
y
o
u
t th
is
wo
r
k
.
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
Pu
s
p
an
jali M
allik
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
Ajit Ku
m
ar
Nay
ak
✓
✓
✓
✓
✓
✓
Saty
ap
r
ak
ash
Swain
✓
✓
✓
✓
✓
✓
✓
✓
✓
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
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2252
-
8
8
1
4
I
n
t J Ad
v
Ap
p
l Sci
,
Vo
l
.
1
5
,
No
.
1
,
Ma
r
c
h
2
0
2
6
:
197
-
2
0
7
206
CO
NF
L
I
C
T
O
F
I
N
T
E
R
E
S
T
ST
A
T
E
M
E
NT
Au
th
o
r
s
s
tate
n
o
co
n
f
lict o
f
in
t
er
est.
DATA AV
AI
L
AB
I
L
I
T
Y
T
h
e
d
ata
th
at
s
u
p
p
o
r
t
th
e
f
i
n
d
in
g
s
o
f
th
is
s
tu
d
y
ar
e
o
p
en
ly
av
ailab
le
in
th
e
B
o
n
n
d
at
a
s
et
at
h
ttp
s
://www.
u
k
b
o
n
n
.
d
e/ep
ilep
to
lo
g
ie/ar
b
eits
g
r
u
p
p
en
/ag
-
leh
n
er
tz
-
n
eu
r
o
p
h
y
s
ik
/d
o
wn
lo
a
d
s
/
.
RE
F
E
R
E
NC
E
S
[
1
]
J.
X
u
,
K
.
Y
a
n
,
Z.
D
e
n
g
,
J.
X
.
Li
u
,
J
.
W
a
n
g
,
a
n
d
S
.
Y
a
n
,
“
EEG
-
b
a
se
d
e
p
i
l
e
p
t
i
c
se
i
z
u
r
e
d
e
t
e
c
t
i
o
n
u
si
n
g
d
e
e
p
l
e
a
r
n
i
n
g
t
e
c
h
n
i
q
u
e
s
:
a
su
r
v
e
y
,
”
N
e
u
r
o
c
o
m
p
u
t
i
n
g
,
v
o
l
.
6
1
0
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
n
e
u
c
o
m.
2
0
2
4
.
1
2
8
6
4
4
.
[
2
]
V
.
P
.
B
a
l
a
m,
“
S
y
s
t
e
m
a
t
i
c
r
e
v
i
e
w
o
f
s
i
n
g
l
e
-
c
h
a
n
n
e
l
EEG
-
b
a
se
d
d
r
o
w
si
n
e
ss
d
e
t
e
c
t
i
o
n
me
t
h
o
d
s,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
I
n
t
e
l
l
i
g
e
n
t
T
ra
n
s
p
o
rt
a
t
i
o
n
S
y
st
e
m
s
,
v
o
l
.
2
5
,
n
o
.
1
1
,
p
p
.
1
5
2
1
0
–
1
5
2
2
8
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
TS.2
0
2
4
.
3
4
4
2
2
4
.
[
3
]
P
.
M
a
l
l
i
k
,
A
.
K
.
N
a
y
a
k
,
S
.
K
.
M
o
h
a
p
a
t
r
a
,
a
n
d
K
.
P
.
S
w
a
i
n
,
“
S
H
-
O
S
P
:
a
h
y
b
r
i
d
a
l
g
o
r
i
t
h
m
u
s
i
n
g
sp
o
t
t
e
d
h
y
e
n
a
o
p
t
i
m
i
z
e
r
e
n
a
b
l
e
d
w
i
t
h
o
p
t
i
ma
l
st
o
c
h
a
s
t
i
c
p
r
o
c
e
ss
f
o
r
e
p
i
l
e
p
t
i
c
se
i
z
u
r
e
d
e
t
e
c
t
i
o
n
,
”
S
N
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
5
,
n
o
.
8
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s4
2
9
7
9
-
0
2
4
-
0
3
4
8
8
-
8.
[
4
]
I
.
A
h
mad
e
t
a
l
.
,
“
A
n
e
f
f
i
c
i
e
n
t
f
e
a
t
u
r
e
sel
e
c
t
i
o
n
a
n
d
e
x
p
l
a
i
n
a
b
l
e
c
l
a
ss
i
f
i
c
a
t
i
o
n
me
t
h
o
d
f
o
r
EEG
-
b
a
se
d
e
p
i
l
e
p
t
i
c
s
e
i
z
u
r
e
d
e
t
e
c
t
i
o
n
,
”
J
o
u
rn
a
l
o
f
I
n
f
o
rm
a
t
i
o
n
S
e
c
u
r
i
t
y
a
n
d
A
p
p
l
i
c
a
t
i
o
n
s
,
v
o
l
.
8
0
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
i
sa
.
2
0
2
3
.
1
0
3
6
5
4
.
[
5
]
N
.
K
h
a
l
i
d
a
n
d
M
.
S
.
E
h
s
a
n
,
“
C
r
i
t
i
c
a
l
a
n
a
l
y
s
i
s
o
f
P
a
r
k
i
n
s
o
n
’
s
d
i
se
a
se
d
e
t
e
c
t
i
o
n
u
si
n
g
E
EG
s
u
b
-
b
a
n
d
s
a
n
d
g
a
t
e
d
r
e
c
u
r
r
e
n
t
u
n
i
t
,
”
En
g
i
n
e
e
ri
n
g
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
a
n
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
,
v
o
l
.
5
9
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
j
e
s
t
c
h
.
2
0
2
4
.
1
0
1
8
5
5
.
[
6
]
L.
Q
i
u
,
J.
Li
,
I
.
Zh
o
n
g
,
W
.
F
e
n
g
,
C
.
Z
h
o
u
,
a
n
d
J.
P
a
n
,
“
A
n
o
v
e
l
EEG
-
b
a
s
e
d
P
a
r
k
i
n
s
o
n
’
s
d
i
se
a
se
d
e
t
e
c
t
i
o
n
mo
d
e
l
u
si
n
g
m
u
l
t
i
s
c
a
l
e
c
o
n
v
o
l
u
t
i
o
n
a
l
p
r
o
t
o
t
y
p
e
n
e
t
w
o
r
k
s,”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
I
n
st
r
u
m
e
n
t
a
t
i
o
n
a
n
d
Me
a
su
r
e
m
e
n
t
,
v
o
l
.
7
3
,
p
p
.
1
–
1
4
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
TI
M
.
2
0
2
4
.
3
3
5
1
2
4
8
.
[
7
]
Y
.
Y
u
,
Y
.
Li
,
Y
.
Z
h
o
u
,
Y
.
W
a
n
g
,
a
n
d
J.
W
a
n
g
,
“
A
l
e
a
r
n
a
b
l
e
a
n
d
e
x
p
l
a
i
n
a
b
l
e
w
a
v
e
l
e
t
n
e
u
r
a
l
n
e
t
w
o
r
k
f
o
r
E
EG
a
r
t
i
f
a
c
t
s
d
e
t
e
c
t
i
o
n
a
n
d
c
l
a
ss
i
f
i
c
a
t
i
o
n
,
”
I
E
EE
T
r
a
n
sa
c
t
i
o
n
s
o
n
N
e
u
r
a
l
S
y
st
e
m
s
a
n
d
R
e
h
a
b
i
l
i
t
a
t
i
o
n
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
3
2
,
p
p
.
3
3
5
8
–
3
3
6
8
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
TN
S
R
E
.
2
0
2
4
.
3
4
5
2
3
1
5
.
[
8
]
C
.
D
i
v
v
a
l
a
a
n
d
M
.
M
i
s
h
r
a
,
“
D
e
e
p
l
e
a
r
n
i
n
g
-
b
a
se
d
a
t
t
e
n
t
i
o
n
mec
h
a
n
i
sm
f
o
r
a
u
t
o
m
a
t
i
c
d
r
o
w
s
i
n
e
ss
d
e
t
e
c
t
i
o
n
u
s
i
n
g
EEG
si
g
n
a
l
,
”
I
EEE
S
e
n
so
r
s Le
t
t
e
rs
,
v
o
l
.
8
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
LS
EN
S
.
2
0
2
4
.
3
3
6
3
7
3
5
.
[
9
]
X
.
L
i
,
J.
Ta
n
g
,
X
.
Li
,
a
n
d
Y
.
Y
a
n
g
,
“
C
W
S
T
R
-
N
e
t
:
a
c
h
a
n
n
e
l
-
w
e
i
g
h
t
e
d
s
p
a
t
i
a
l
–
t
e
mp
o
r
a
l
r
e
si
d
u
a
l
n
e
t
w
o
r
k
b
a
s
e
d
o
n
n
o
n
sm
o
o
t
h
n
o
n
n
e
g
a
t
i
v
e
mat
r
i
x
f
a
c
t
o
r
i
z
a
t
i
o
n
f
o
r
f
a
t
i
g
u
e
d
e
t
e
c
t
i
o
n
u
si
n
g
E
EG
si
g
n
a
l
s,
”
Bi
o
m
e
d
i
c
a
l
S
i
g
n
a
l
Pro
c
e
ss
i
n
g
a
n
d
C
o
n
t
r
o
l
,
v
o
l
.
9
7
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
b
s
p
c
.
2
0
2
4
.
1
0
6
6
8
5
.
[
1
0
]
S
.
R
u
k
h
s
a
r
a
n
d
A
.
K
.
T
i
w
a
r
i
,
“
L
i
g
h
t
w
e
i
g
h
t
c
o
n
v
o
l
u
t
i
o
n
t
r
a
n
sf
o
r
mer
f
o
r
c
r
o
ss
-
p
a
t
i
e
n
t
se
i
z
u
r
e
d
e
t
e
c
t
i
o
n
i
n
m
u
l
t
i
-
c
h
a
n
n
e
l
EEG
si
g
n
a
l
s,
”
C
o
m
p
u
t
e
r
Me
t
h
o
d
s
a
n
d
Pr
o
g
ra
m
s
i
n
B
i
o
m
e
d
i
c
i
n
e
,
v
o
l
.
2
4
2
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
c
m
p
b
.
2
0
2
3
.
1
0
7
8
5
6
.
[
1
1
]
N
.
F
.
A
l
i
,
N
.
A
l
b
a
s
t
a
k
i
,
A
.
N
.
I
.
B
e
l
k
a
c
e
m,
M
.
E.
A
d
e
l
,
a
n
d
M
.
A
t
e
f
,
“
A
l
o
w
-
c
o
m
p
l
e
x
i
t
y
c
o
m
b
i
n
e
d
e
n
c
o
d
e
r
-
LSTM
-
a
t
t
e
n
t
i
o
n
n
e
t
w
o
r
k
s
f
o
r
EEG
-
b
a
s
e
d
d
e
p
r
e
ssi
o
n
d
e
t
e
c
t
i
o
n
,
”
I
EEE
Ac
c
e
ss
,
v
o
l
.
1
2
,
p
p
.
1
2
9
3
9
0
–
1
2
9
4
0
3
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
2
4
.
3
4
3
6
8
9
5
.
[
1
2
]
M
.
S
.
Y
a
m
a
mo
t
o
e
t
a
l
.
,
“
M
o
d
e
l
l
i
n
g
c
o
m
p
l
e
x
E
EG
d
a
t
a
d
i
s
t
r
i
b
u
t
i
o
n
o
n
t
h
e
R
i
e
ma
n
n
i
a
n
ma
n
i
f
o
l
d
t
o
w
a
r
d
o
u
t
l
i
e
r
d
e
t
e
c
t
i
o
n
a
n
d
mu
l
t
i
m
o
d
a
l
c
l
a
ssi
f
i
c
a
t
i
o
n
,
”
I
EEE
T
ra
n
s
a
c
t
i
o
n
s
o
n
B
i
o
m
e
d
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
,
v
o
l
.
7
1
,
n
o
.
2
,
p
p
.
3
7
7
–
3
8
7
,
2
0
2
4
,
d
o
i
:
1
0
.
1
1
0
9
/
TB
M
E
.
2
0
2
3
.
3
2
9
5
7
6
9
.
[
1
3
]
T.
I
sl
a
m,
M
.
B
a
sa
k
,
R
.
I
sl
a
m,
a
n
d
A
.
D
.
R
o
y
,
“
I
n
v
e
s
t
i
g
a
t
i
n
g
p
o
p
u
l
a
t
i
o
n
-
sp
e
c
i
f
i
c
e
p
i
l
e
p
s
y
d
e
t
e
c
t
i
o
n
f
r
o
m
n
o
i
s
y
E
EG
si
g
n
a
l
s
u
s
i
n
g
d
e
e
p
-
l
e
a
r
n
i
n
g
m
o
d
e
l
s
,
”
H
e
l
i
y
o
n
,
v
o
l
.
9
,
n
o
.
1
2
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
h
e
l
i
y
o
n
.
2
0
2
3
.
e
2
2
2
0
8
.
[
1
4
]
C
.
Z
h
a
o
,
R
.
P
e
n
g
,
a
n
d
D
.
W
u
,
“
S
o
u
r
c
e
-
f
r
e
e
d
o
m
a
i
n
a
d
a
p
t
a
t
i
o
n
(
S
F
D
A
)
f
o
r
p
r
i
v
a
c
y
-
p
r
e
s
e
r
v
i
n
g
se
i
z
u
r
e
s
u
b
t
y
p
e
c
l
a
ssi
f
i
c
a
t
i
o
n
,
”
I
EEE
T
r
a
n
s
a
c
t
i
o
n
s
o
n
N
e
u
r
a
l
S
y
st
e
m
s
a
n
d
Re
h
a
b
i
l
i
t
a
t
i
o
n
En
g
i
n
e
e
ri
n
g
,
v
o
l
.
3
1
,
p
p
.
2
3
1
5
–
2
3
2
5
,
2
0
2
3
,
d
o
i
:
1
0
.
1
1
0
9
/
TN
S
R
E
.
2
0
2
3
.
3
2
7
4
5
6
3
.
[
1
5
]
H
.
S
.
O
h
a
l
a
n
d
S
.
M
a
n
t
r
i
,
“
Ex
p
l
o
r
i
n
g
EEG
-
b
a
se
d
b
i
o
mar
k
e
r
s
f
o
r
i
m
p
r
o
v
e
d
e
a
r
l
y
A
l
z
h
e
i
m
e
r
’
s
d
i
se
a
se
d
e
t
e
c
t
i
o
n
:
a
f
e
a
t
u
r
e
-
b
a
s
e
d
a
p
p
r
o
a
c
h
u
t
i
l
i
z
i
n
g
m
a
c
h
i
n
e
l
e
a
r
n
i
n
g
,
”
Me
a
s
u
reme
n
t
:
S
e
n
s
o
rs
,
v
o
l
.
3
6
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
me
a
se
n
.
2
0
2
4
.
1
0
1
4
0
3
.
[
1
6
]
S
.
K
.
K
h
a
r
e
a
n
d
U
.
R
.
A
c
h
a
r
y
a
,
“
A
d
a
z
d
-
N
e
t
:
a
u
t
o
ma
t
e
d
a
d
a
p
t
i
v
e
a
n
d
e
x
p
l
a
i
n
a
b
l
e
A
l
z
h
e
i
mer’
s
d
i
sea
se
d
e
t
e
c
t
i
o
n
sy
s
t
e
m
u
s
i
n
g
EEG
si
g
n
a
l
s,”
K
n
o
w
l
e
d
g
e
-
Ba
s
e
d
S
y
st
e
m
s
,
v
o
l
.
2
7
8
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
k
n
o
sy
s
.
2
0
2
3
.
1
1
0
8
5
8
.
[
1
7
]
M
.
M
.
I
sl
a
m
,
M
.
R
.
I
sl
a
m,
a
n
d
M
.
S
.
I
sl
a
m,
“
A
n
e
f
f
i
c
i
e
n
t
h
u
ma
n
c
o
mp
u
t
e
r
i
n
t
e
r
a
c
t
i
o
n
t
h
r
o
u
g
h
h
a
n
d
g
e
st
u
r
e
u
s
i
n
g
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
N
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
1
,
2
0
2
0
,
d
o
i
:
1
0
.
1
0
0
7
/
s
4
2
9
7
9
-
0
2
0
-
0
0
2
2
3
-
x
.
[
1
8
]
Q
.
W
a
n
g
e
t
a
l
.
,
“
A
h
y
b
r
i
d
S
V
M
a
n
d
k
e
r
n
e
l
f
u
n
c
t
i
o
n
-
b
a
s
e
d
sp
a
r
se
r
e
p
r
e
s
e
n
t
a
t
i
o
n
c
l
a
ssi
f
i
c
a
t
i
o
n
f
o
r
a
u
t
o
ma
t
e
d
e
p
i
l
e
p
sy
d
e
t
e
c
t
i
o
n
i
n
EEG
si
g
n
a
l
s,”
N
e
u
r
o
c
o
m
p
u
t
i
n
g
,
v
o
l
.
5
6
2
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
n
e
u
c
o
m.
2
0
2
3
.
1
2
6
8
7
4
.
[
1
9
]
T.
W
u
,
Y
.
F
a
n
,
Y
.
Z
h
o
n
g
,
X
.
C
h
e
n
g
,
X
.
K
o
n
g
,
a
n
d
I
.
C
h
e
n
,
“
S
C
N
e
t
:
a
s
p
a
t
i
a
l
f
e
a
t
u
r
e
f
u
s
e
d
c
o
n
v
o
l
u
t
i
o
n
a
l
n
e
t
w
o
r
k
f
o
r
m
u
l
t
i
-
c
h
a
n
n
e
l
E
EG
p
a
t
h
o
l
o
g
y
d
e
t
e
c
t
i
o
n
,
”
B
i
o
m
e
d
i
c
a
l
S
i
g
n
a
l
Pr
o
c
e
ssi
n
g
a
n
d
C
o
n
t
ro
l
,
v
o
l
.
8
6
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
b
sp
c
.
2
0
2
3
.
1
0
5
0
5
9
.
[
2
0
]
Y
.
F
u
,
D
.
Li
u
,
J.
C
h
e
n
,
a
n
d
I
.
H
e
,
“
S
e
c
r
e
t
a
r
y
b
i
r
d
o
p
t
i
m
i
z
a
t
i
o
n
a
l
g
o
r
i
t
h
m:
a
n
e
w
m
e
t
a
h
e
u
r
i
st
i
c
f
o
r
s
o
l
v
i
n
g
g
l
o
b
a
l
o
p
t
i
mi
z
a
t
i
o
n
p
r
o
b
l
e
ms,
”
Ar
t
i
f
i
c
i
a
l
I
n
t
e
l
l
i
g
e
n
c
e
R
e
v
i
e
w
,
v
o
l
.
5
7
,
2
0
2
4
,
d
o
i
:
1
0
.
1
0
0
7
/
s
1
0
4
6
2
-
0
2
4
-
1
0
7
2
9
-
y.
[
2
1
]
R
.
G
.
A
n
d
r
z
e
j
a
k
,
K
.
Le
h
n
e
r
t
z
,
F
.
M
o
r
ma
n
n
,
C
.
R
i
e
k
e
,
P
.
D
a
v
i
d
,
a
n
d
C
.
E.
E
l
g
e
r
,
“
I
n
d
i
c
a
t
i
o
n
s
o
f
n
o
n
l
i
n
e
a
r
d
e
t
e
r
mi
n
i
st
i
c
a
n
d
f
i
n
i
t
e
-
d
i
m
e
n
si
o
n
a
l
s
t
r
u
c
t
u
r
e
s
i
n
t
i
m
e
seri
e
s
o
f
b
r
a
i
n
e
l
e
c
t
r
i
c
a
l
a
c
t
i
v
i
t
y
:
d
e
p
e
n
d
e
n
c
e
o
n
r
e
c
o
r
d
i
n
g
r
e
g
i
o
n
a
n
d
b
r
a
i
n
st
a
t
e
,
”
P
h
y
si
c
a
l
Re
v
i
e
w
E
,
v
o
l
.
6
4
,
n
o
.
6
,
2
0
0
1
,
d
o
i
:
1
0
.
1
1
0
3
/
P
h
y
sR
e
v
E.
6
4
.
0
6
1
9
0
7
.
[
2
2
]
A
.
T.
M
o
h
a
n
a
n
d
D
.
V
G
a
i
t
o
n
d
e
,
“
A
d
e
e
p
l
e
a
r
n
i
n
g
b
a
s
e
d
a
p
p
r
o
a
c
h
t
o
r
e
d
u
c
e
d
o
r
d
e
r
m
o
d
e
l
l
i
n
g
f
o
r
t
u
r
b
u
l
e
n
t
f
l
o
w
c
o
n
t
r
o
l
u
si
n
g
LSTM
n
e
u
r
a
l
n
e
t
w
o
r
k
s,
”
2
0
1
8
,
a
rX
i
v
:
1
8
0
4
.
0
9
2
6
9
.
[
2
3
]
M
.
S
h
e
n
,
P
.
W
e
n
,
B
.
S
o
n
g
,
a
n
d
Y
.
Li
,
“
A
n
EEG
b
a
se
d
r
e
a
l
-
t
i
m
e
e
p
i
l
e
p
sy
s
e
i
z
u
r
e
d
e
t
e
c
t
i
o
n
a
p
p
r
o
a
c
h
u
si
n
g
d
i
s
c
r
e
t
e
w
a
v
e
l
e
t
t
r
a
n
sf
o
r
m
a
n
d
ma
c
h
i
n
e
l
e
a
r
n
i
n
g
me
t
h
o
d
s,
”
Bi
o
m
e
d
i
c
a
l
S
i
g
n
a
l
Pr
o
c
e
ss
i
n
g
a
n
d
C
o
n
t
r
o
l
,
v
o
l
.
7
7
,
2
0
2
2
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
b
s
p
c
.
2
0
2
2
.
1
0
3
8
2
0
.
[
2
4
]
L.
S
.
V
i
d
y
a
r
a
t
n
e
a
n
d
K
.
M
.
I
f
t
e
k
h
a
r
u
d
d
i
n
,
“
R
e
a
l
-
t
i
me
e
p
i
l
e
p
t
i
c
se
i
z
u
r
e
d
e
t
e
c
t
i
o
n
u
s
i
n
g
EEG
,
”
I
EE
E
T
r
a
n
s
a
c
t
i
o
n
s
o
n
N
e
u
ra
l
S
y
s
t
e
m
s
a
n
d
Re
h
a
b
i
l
i
t
a
t
i
o
n
En
g
i
n
e
e
r
i
n
g
,
v
o
l
.
2
5
,
p
p
.
2
1
4
6
–
2
1
5
6
,
2
0
1
7
,
d
o
i
:
1
0
.
1
1
0
9
/
TN
S
R
E.
2
0
1
7
.
2
6
9
7
9
2
0
.
[
2
5
]
M
.
A
l
h
u
ssei
n
,
G
.
M
u
h
a
mm
a
d
,
a
n
d
M
.
S
.
H
o
ss
a
i
n
,
“
EEG
p
a
t
h
o
l
o
g
y
d
e
t
e
c
t
i
o
n
b
a
se
d
o
n
d
e
e
p
l
e
a
r
n
i
n
g
,
”
I
EE
E
A
c
c
e
ss
,
v
o
l
.
7
,
p
p
.
2
7
7
8
1
–
2
7
7
8
8
,
2
0
1
9
,
d
o
i
:
1
0
.
1
1
0
9
/
A
C
C
ESS
.
2
0
1
9
.
2
9
0
1
6
7
2
.
[
2
6
]
A
.
B
r
i
t
t
o
e
t
a
l
.
,
“
A
mu
l
t
i
-
d
i
m
e
n
s
i
o
n
a
l
h
y
b
r
i
d
C
N
N
-
B
i
LST
M
f
r
a
mew
o
r
k
f
o
r
e
p
i
l
e
p
t
i
c
se
i
z
u
r
e
d
e
t
e
c
t
i
o
n
u
s
i
n
g
e
l
e
c
t
r
o
e
n
c
e
p
h
a
l
o
g
r
a
m
si
g
n
a
l
scr
u
t
i
n
y
,
”
S
y
st
e
m
s
a
n
d
S
o
f
t
C
o
m
p
u
t
i
n
g
,
v
o
l
.
5
,
2
0
2
3
,
d
o
i
:
1
0
.
1
0
1
6
/
j
.
sas
c
.
2
0
2
3
.
2
0
0
0
6
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.