I
nte
rna
t
io
na
l J
o
urna
l o
f
E
lect
rica
l a
nd
Co
m
p
ute
r
E
ng
in
ee
ring
(
I
J
E
CE
)
Vo
l.
11
,
No
.
1
,
Feb
r
u
ar
y
2021
,
p
p
.
424
~
433
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v
1
1
i
1
.
p
p
4
2
4
-
433
424
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ec
e.
ia
esco
r
e.
co
m
Bio
-
sig
na
ls
c
o
m
p
ress
io
n using
a
uto
-
e
nco
der
Su
nil
ku
m
a
r
K
.
N
.
,
Sh
iv
a
s
ha
n
k
a
r
,
K
esh
a
v
a
m
urt
hy
S
ri
V
e
n
k
a
tes
h
w
a
r
a
Co
ll
e
g
e
o
f
En
g
in
e
e
rin
g
,
In
d
i
a
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ju
l
18
,
2
0
1
9
R
ev
i
s
ed
J
u
n
28
,
20
20
A
cc
ep
ted
A
u
g
6
,
2
0
20
L
a
tes
t
d
e
v
e
lo
p
m
e
n
ts
in
w
e
a
ra
b
le
d
e
v
ice
s
p
e
r
m
it
s
u
n
-
d
a
m
a
g
e
a
b
le
a
n
d
c
h
e
a
p
e
st
w
a
y
f
o
r
g
a
th
e
rin
g
o
f
m
e
d
ica
l
d
a
ta
su
c
h
a
s
b
io
-
sig
n
a
ls
li
k
e
EC
G
,
Re
sp
iratio
n
,
Blo
o
d
p
re
ss
u
re
e
tc.
G
a
th
e
rin
g
a
n
d
a
n
a
ly
sis
o
f
v
a
rio
u
s
b
io
m
a
rk
e
rs
a
re
c
o
n
sid
e
re
d
to
p
ro
v
id
e
a
n
ti
c
ip
a
to
ry
h
e
a
lt
h
c
a
re
th
ro
u
g
h
c
u
sto
m
ize
d
a
p
p
li
c
a
ti
o
n
s
f
o
r
m
e
d
ica
l
p
u
rp
o
se
.
W
e
a
ra
b
le
d
e
v
ice
s
w
il
l
re
ly
o
n
siz
e
,
re
so
u
rc
e
s
a
n
d
b
a
tt
e
ry
c
a
p
a
c
it
y
;
we
n
e
e
d
a
n
o
v
e
l
a
l
g
o
rit
h
m
t
o
ro
b
u
stly
c
o
n
tro
l
m
e
m
o
r
y
a
n
d
th
e
e
n
e
rg
y
o
f
th
e
d
e
v
ice
.
T
h
e
ra
p
id
g
ro
w
th
o
f
th
e
tec
h
n
o
lo
g
y
h
a
s
led
to
n
u
m
e
r
o
u
s
a
u
t
o
e
n
c
o
d
e
rs
th
a
t
g
u
a
ra
n
tee
th
e
re
su
lt
s
b
y
e
x
trac
ti
n
g
f
e
a
tu
re
se
lec
ti
o
n
f
ro
m
ti
m
e
a
n
d
f
re
q
u
e
n
c
y
d
o
m
a
in
in
a
n
e
ff
ici
e
n
t
w
a
y
.
T
h
e
m
a
in
a
im
is
to
trai
n
t
h
e
h
id
d
e
n
lay
e
r
to
re
c
o
n
stru
c
t
th
e
d
a
ta
sim
il
a
r
to
th
a
t
o
f
in
p
u
t.
I
n
th
e
p
re
v
io
u
s
w
o
rk
s,
to
a
c
c
o
m
p
li
sh
th
e
c
o
m
p
re
ss
io
n
a
ll
fe
a
tu
re
s
we
re
n
e
e
d
e
d
b
u
t
in
o
u
r
p
r
o
p
o
se
d
f
ra
m
e
w
o
rk
b
io
-
sig
n
a
ls
c
o
m
p
re
ss
io
n
u
sin
g
a
u
to
-
e
n
c
o
d
e
r
(BCA
E)
w
il
l
p
e
rf
o
r
m
tas
k
b
y
tak
in
g
o
n
ly
i
m
p
o
rtan
t
f
e
a
tu
re
s
a
n
d
c
o
m
p
re
ss
it
.
B
y
d
o
in
g
th
is
it
c
a
n
re
d
u
c
e
p
o
w
e
r
c
o
n
su
m
p
ti
o
n
a
t
th
e
so
u
r
c
e
e
n
d
a
n
d
h
e
n
c
e
in
c
re
a
se
s
b
a
tt
e
ry
li
fe
.
T
h
e
p
e
rf
o
r
m
a
n
c
e
o
f
th
e
re
su
lt
c
o
m
p
a
riso
n
is
d
o
n
e
f
o
r
th
e
3
p
a
ra
m
e
ters
c
o
m
p
re
ss
io
n
ra
ti
o
,
re
c
o
n
stru
c
ti
o
n
e
rr
o
r
a
n
d
p
o
w
e
r
c
o
n
su
m
p
ti
o
n
.
Ou
r
p
ro
p
o
se
d
w
o
rk
o
u
tp
e
rf
o
rm
s
w
it
h
re
sp
e
c
t
to
th
e
S
URF
m
e
th
o
d
.
K
ey
w
o
r
d
s
:
Au
to
-
e
n
co
d
er
B
io
-
m
ed
ical
s
i
g
n
als
Featu
r
e
s
elec
t
io
n
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
:
Su
n
il
k
u
m
ar
K
.
N
.
,
Sri
Ven
k
ates
h
w
ar
a
C
o
lle
g
e
o
f
E
n
g
i
n
ee
r
i
n
g
,
B
an
g
alo
r
e
,
I
n
d
ia
.
E
m
ail:
s
u
n
il
k
u
m
ar
k
n
2
6
@
g
m
a
il.c
o
m
1.
I
NT
RO
D
UCT
I
O
N
I
o
T
tech
n
o
lo
g
y
o
f
f
er
s
d
ev
i
ce
s
to
s
en
s
e
t
h
e
p
r
ac
tical
s
u
r
r
o
u
n
d
in
g
an
d
ef
f
o
r
tles
s
l
y
co
m
b
i
n
e
th
e
co
llected
in
f
o
r
m
atio
n
in
to
r
ef
in
ed
ap
p
licatio
n
s
t
h
at
p
er
m
it
s
f
o
r
s
i
g
n
if
ica
n
t
e
n
h
a
n
ce
m
en
ts
o
f
ac
tiv
i
ties
o
f
th
e
p
er
s
o
n
.
T
h
e
m
ai
n
ai
m
o
f
t
h
is
p
ap
er
is
s
e
n
s
i
n
g
t
h
e
h
u
m
a
n
’
s
b
eh
a
v
io
r
[
1
]
v
ia
I
o
T
d
ev
ice
th
at
is
w
ea
r
ab
le
lik
e
c
h
e
s
t
s
tr
ap
s
,
s
m
ar
t
w
atc
h
es,
w
r
i
s
tb
an
d
s
,
etc.
t
h
at
ca
n
b
e
u
t
ilized
to
p
o
in
t
-
o
u
t
th
e
s
p
ec
if
ic
h
ea
l
th
an
d
th
e
r
eq
u
ir
ed
f
it
n
e
s
s
o
f
th
e
clie
n
ts
[
2
]
.
W
ea
r
ab
le
d
ev
ices
ca
n
b
e
co
m
b
i
n
ed
in
to
b
o
d
y
s
e
n
s
o
r
n
et
w
o
r
k
s
t
h
at
ar
e
w
ir
ele
s
s
l
y
co
n
n
ec
ted
to
b
r
in
g
r
ep
o
r
ts
o
f
t
h
e
m
ed
ical
u
p
-
to
-
d
ate
t
h
o
u
g
h
in
te
r
n
et.
T
h
er
ef
o
r
e,
allo
w
in
g
an
ticip
atio
n
,
ea
r
l
y
e
x
a
m
in
e
a
n
d
s
p
ec
ial
ca
r
e.
A
n
y
w
a
y
s
,
as
t
h
e
y
ar
e
n
ee
d
ed
to
b
e
tin
y
an
d
le
s
s
w
ei
g
h
t,
t
h
e
y
ar
e
also
li
m
i
ted
w
i
th
t
h
e
r
eso
u
r
ce
s
s
u
c
h
as p
o
w
er
,
ca
p
ab
ilit
y
o
f
t
h
e
tr
an
s
itio
n
a
n
d
m
e
m
o
r
y
.
I
n
th
is
p
ap
er
,
w
e
i
m
p
le
m
e
n
t
n
e
w
s
o
l
u
tio
n
f
o
r
d
ata
p
r
o
ce
s
s
in
g
f
o
r
th
e
en
d
u
r
an
ce
m
o
n
ito
r
in
g
o
f
E
C
G
s
ig
n
al
s
.
T
h
ese
m
ed
ical
r
ec
o
r
d
s
ar
e
b
asicall
y
s
i
m
p
le
to
m
ea
s
u
r
e,
b
u
t
s
i
m
u
ltan
eo
u
s
l
y
,
ex
c
ep
tio
n
all
y
v
alu
ab
le
f
o
r
th
e
alr
ea
d
y
m
e
n
tio
n
ed
d
eter
m
i
n
atio
n
.
W
e
ass
u
m
e
th
e
ac
h
iev
e
m
e
n
t
o
f
t
h
o
s
e
k
in
d
o
f
s
ig
n
als
v
ia
s
m
all
w
ea
r
ab
le
m
ac
h
in
e
s
[
3
]
ar
e
w
o
r
r
ied
ab
o
u
t
th
e
ex
te
n
d
i
n
g
th
e
d
u
r
ab
ilit
y
o
f
t
h
e
b
atter
y
o
f
th
o
s
e
d
e
v
ices
v
i
a
co
m
p
r
es
s
io
n
o
f
lo
s
s
y
s
i
g
n
als.
W
e
ass
u
m
e
t
h
e
co
n
d
itio
n
s
w
h
er
e
tr
an
s
m
i
s
s
io
n
o
f
E
C
G
s
ig
n
a
ls
w
ir
eles
s
l
y
to
s
o
m
e
s
p
ec
i
f
ic
ac
ce
s
s
p
o
in
t
i
s
n
ee
d
ed
,
s
o
th
at
t
h
e
s
i
g
n
als
ca
n
b
e
p
r
eser
v
ed
o
n
th
e
s
er
v
er
an
d
ca
n
b
e
u
s
ed
b
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
B
io
-
s
ig
n
a
ls
co
mp
r
ess
io
n
u
s
in
g
a
u
to
-
e
n
co
d
er (
S
u
n
ilku
ma
r
K
.
N
.
)
425
th
e
cl
in
icia
n
s
t
h
r
o
u
g
h
o
u
t
t
h
e
w
o
r
ld
.
T
h
e
ad
v
a
n
ce
d
w
ea
r
ab
le
I
o
T
d
ev
ices
w
h
er
e
th
e
en
er
g
y
an
d
m
e
m
o
r
y
ar
e
th
e
m
ai
n
is
s
u
es
w
h
ile
co
llecti
n
g
th
e
h
i
g
h
-
d
i
m
e
n
s
io
n
al
d
ata
s
u
c
h
as
h
ea
r
t
-
r
ate,
le
v
el
o
f
t
h
e
o
x
y
g
e
n
,
an
d
o
t
h
er
b
io
-
in
f
o
r
m
atio
n
b
y
t
h
e
I
o
T
d
ev
ice
a
n
d
t
h
o
s
e
lar
g
e
d
ata
ar
e
t
r
an
s
m
itted
to
th
e
u
n
it
s
w
h
ic
h
ar
e
m
o
r
e
ca
p
ab
le
o
f
h
an
d
li
n
g
a
n
d
p
ictu
r
i
n
g
-
(
r
ep
r
esen
ti
n
g
)
t
h
o
s
e
b
io
-
i
n
f
o
r
m
atio
n
al
d
ata.
As
t
h
e
d
ata
co
llecte
d
b
y
th
e
I
o
T
d
ev
ices
ar
e
v
er
y
lar
g
e,
w
e
n
ee
d
an
a
lg
o
r
ith
m
,
w
h
ic
h
i
s
v
er
y
e
f
f
ici
en
t
i
n
co
m
p
r
es
s
i
n
g
th
e
d
ata
an
d
tr
an
s
m
i
tti
n
g
to
th
e
m
o
r
e
ca
p
ab
le
u
n
it
s
,
w
h
ic
h
d
o
es
th
e
ex
ec
u
tio
n
o
f
h
a
n
d
li
n
g
a
n
d
p
ictu
r
i
n
g
o
f
d
ata
ac
c
u
r
atel
y
a
n
d
e
f
f
icien
tl
y
as
w
el
l.
A
lt
h
o
u
g
h
an
au
to
-
e
n
c
o
d
er
an
d
i
ts
v
ar
ian
ts
ar
e
u
s
ed
to
co
m
p
r
ess
t
h
e
d
ata
w
h
er
e
t
h
e
o
u
tp
u
t
s
h
o
u
ld
b
e
al
m
o
s
t
s
i
m
ilar
to
t
h
e
i
n
p
u
t.
B
u
t,
t
h
er
e
w
a
s
a
m
aj
o
r
li
m
itatio
n
,
it
u
s
es
m
o
s
t
o
f
t
h
e
h
i
g
h
-
lev
el
f
ea
tu
r
es
t
h
at
ar
e
r
ed
u
n
d
an
t to
u
s
e
w
h
ile
r
ep
r
esen
ti
n
g
t
h
e
o
u
tp
u
t in
t
h
e
tr
ai
n
in
g
d
ata
w
h
ic
h
i
s
n
o
t r
elate
d
w
ith
th
e
p
atter
n
.
T
h
er
ef
o
r
e,
to
s
o
lv
e
th
is
p
r
o
b
lem
,
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
co
n
s
is
t
s
o
f
b
o
t
h
t
h
e
p
ar
ts
o
f
th
e
al
g
o
r
ith
m
to
co
m
p
r
es
s
th
e
d
ata
w
h
er
e
s
i
m
u
ltan
eo
u
s
l
y
t
h
e
s
elec
tio
n
o
f
i
m
p
o
r
tan
t
d
ata
is
p
e
r
f
o
r
m
ed
to
s
elec
t
th
e
d
ata
f
r
o
m
th
e
lear
n
ed
h
id
d
en
-
la
y
er
s
i
n
o
n
e
g
r
o
u
p
an
d
t
h
e
d
ata,
w
h
ic
h
ar
e
n
o
t
r
elev
an
t
i
n
an
o
t
h
er
g
r
o
u
p
.
C
o
n
cu
r
r
en
tl
y
,
th
e
e
x
ec
u
t
io
n
o
f
en
co
d
i
n
g
is
d
o
n
e
to
th
e
h
id
d
en
la
y
er
’
s
s
elec
ted
d
ata,
w
h
ic
h
ar
e
i
m
p
o
r
ta
n
t.
W
e
ar
e
p
r
o
p
o
s
in
g
an
al
g
o
r
ith
m
w
h
er
e
s
elec
tio
n
o
f
i
m
p
o
r
ta
n
t
d
ata
f
r
o
m
in
p
u
t
i
s
d
o
n
e
i
n
t
h
e
h
id
d
en
la
y
er
,
w
h
ich
i
s
co
r
r
elate
d
to
ea
ch
o
th
er
,
a
n
d
n
e
g
lecti
n
g
t
h
e
r
e
m
ai
n
in
g
d
ata
o
f
t
h
e
i
n
p
u
t
,
w
h
ic
h
ar
e
n
o
t
m
u
c
h
u
s
e
f
u
l
o
r
n
o
n
-
co
r
r
elate
d
.
Her
e
w
e
ar
e
tak
i
n
g
th
e
b
io
-
s
i
g
n
al
d
ata
b
y
a
W
ea
r
ab
le
I
o
T
d
ev
ice
to
p
er
f
o
r
m
t
h
e
co
m
p
r
ess
io
n
o
f
th
e
d
ata,
o
u
r
alg
o
r
ith
m
h
a
s
t
w
o
p
ar
ts
:
a.
Her
e,
th
e
n
ec
ess
ar
y
i
n
f
o
r
m
a
tio
n
(
Data
)
f
r
o
m
th
e
o
v
er
all
in
p
u
t
d
ata
is
ch
o
s
en
f
o
r
th
e
o
p
er
atio
n
o
f
th
e
au
to
-
en
co
d
in
g
p
r
o
ce
s
s
an
d
th
e
r
est
all
d
ata
th
at
ar
e
n
o
t
n
ec
es
s
ar
y
ar
e
ig
n
o
r
ed
b
y
w
h
ic
h
f
e
w
o
f
th
e
u
n
w
a
n
ted
d
ata
ca
n
b
e
eli
m
in
ated
.
b.
On
t
h
e
s
elec
ted
d
ata,
t
h
e
n
o
n
-
lin
ea
r
d
ata
i
s
p
er
f
o
r
m
ed
w
it
h
t
h
e
u
n
s
u
p
er
v
is
ed
lear
n
i
n
g
u
s
in
g
a
u
to
-
e
n
co
d
er
.
W
e
d
em
o
n
s
tr
ate
o
u
r
w
o
r
k
a
s
f
o
llo
w
s
:
(
i)
Firs
t
w
e
co
n
v
er
t
th
e
s
en
s
o
r
d
ata
in
to
a
h
id
d
en
-
la
y
er
s
b
y
m
ac
h
in
e
-
lear
n
in
g
pr
o
ce
s
s
.
(
ii)
T
h
en
o
u
r
alg
o
r
ith
m
is
u
s
ed
t
o
s
ep
ar
ate
th
e
i
m
p
o
r
tan
t
an
d
r
elev
an
t
d
ata
i
n
o
n
e
g
r
o
u
p
a
n
d
n
o
n
-
r
ele
v
an
t
d
ata
in
to
an
o
t
h
er
g
r
o
u
p
.
(
iii)
I
n
p
ar
allel
to
th
e
ab
o
v
e
p
o
i
n
t,
en
co
d
i
n
g
o
f
th
e
r
elev
a
n
t d
ata
is
p
er
f
o
r
m
ed
w
h
ic
h
is
i
n
th
e
h
id
d
en
-
la
y
er
s
.
T
h
e
f
o
llo
w
i
n
g
p
ap
er
is
ar
r
an
g
ed
as
m
e
n
tio
n
ed
b
elo
w
.
I
n
s
ec
tio
n
2
,
d
escr
ib
es
s
tu
d
y
in
g
ab
o
u
t
th
e
r
elate
d
w
o
r
k
s
b
ased
o
n
t
h
e
s
e
lectio
n
o
f
f
ea
t
u
r
es
f
r
o
m
th
e
in
p
u
t
d
ata
a
n
d
p
er
f
o
r
m
i
n
g
th
e
co
m
p
r
es
s
io
n
u
s
i
n
g
au
to
-
en
co
d
er
tech
n
iq
u
e,
Sectio
n
3
p
r
o
v
id
es
an
o
v
er
v
i
e
w
o
f
t
h
e
p
r
o
p
o
s
ed
w
o
r
k
a
n
d
s
ec
tio
n
4
p
r
o
v
id
es
th
e
s
i
m
u
la
ted
r
esu
lts
co
m
p
ar
ed
w
i
th
th
e
o
th
er
co
m
p
r
e
s
s
io
n
ap
p
r
o
ac
h
SUR
F.
Ou
r
p
r
o
p
o
s
ed
w
o
r
k
an
d
at
las
t
w
e
w
i
ll b
e
p
r
o
v
in
g
t
h
at
o
u
r
r
es
u
lts
ar
e
b
etter
th
a
n
o
th
er
ex
is
ti
n
g
m
et
h
o
d
s
.
2.
L
I
T
E
R
AT
U
RE
SU
RVE
Y
T
h
er
e
is
a
co
m
p
r
es
s
io
n
al
g
o
r
ith
m
f
o
r
t
h
e
s
i
g
n
a
ls
o
f
th
e
E
C
G,
w
h
ic
h
is
d
i
v
id
ed
in
to
t
h
r
ee
i
m
p
o
r
tan
t
m
et
h
o
d
s
:
P
ar
a
m
eter
e
x
tr
ac
tio
n
,
T
r
an
s
f
o
r
m
atio
n
an
d
th
e
d
i
r
ec
t
m
et
h
o
d
s
.
Fo
r
o
u
r
al
g
o
r
ith
m
t
h
er
e
ar
e
t
w
o
i
m
p
o
r
tan
t
co
n
d
itio
n
:
1
)
s
elec
tin
g
n
ec
es
s
ar
y
d
ata
f
r
o
m
t
h
e
h
id
d
en
la
y
er
s
t
h
at
ar
e
tr
ain
ed
.
2
)
Pro
ce
s
s
in
g
o
f
th
e
s
elec
ted
i
m
p
o
r
tan
t d
ata
u
s
i
n
g
a
u
to
-
e
n
co
d
er
.
2
.
1
.
Select
io
n o
f
t
he
i
m
po
rt
a
nt
da
t
a
f
ro
m
t
he
t
ra
ined hi
d
den la
y
er
s
Her
e
s
elec
tio
n
o
f
t
h
e
i
m
p
o
r
tan
t
d
ata
f
r
o
m
t
h
e
co
m
p
lete
tr
ai
n
ed
d
ataset
h
id
d
en
la
y
er
s
i
s
g
r
o
u
p
ed
in
to
o
n
e
s
et
an
d
th
e
d
ata
w
h
ic
h
ar
e
n
o
n
-
r
elev
a
n
t
is
g
r
o
u
p
ed
in
to
a
n
o
th
er
d
ata
s
et.
W
e
ar
e
u
s
in
g
th
e
f
ea
tu
r
e
s
elec
tio
n
m
et
h
o
d
to
g
r
o
u
p
t
h
e
r
elev
a
n
t
a
n
d
n
o
n
-
r
ele
v
an
t
d
ata
in
o
u
r
al
g
o
r
ith
m
.
I
n
[
1
]
,
an
“Au
to
m
a
tic
f
ea
t
u
r
e
ex
tr
ac
tio
n
(
FE
)
an
d
th
e
s
e
lectio
n
f
o
r
r
elate
d
d
atasets
a
n
d
th
e
co
n
d
itio
n
m
o
n
ito
r
in
g
.
”
W
h
er
ea
s
,
th
e
co
m
b
in
a
tio
n
o
f
m
eth
o
d
s
f
o
r
FE
an
d
s
elec
tio
n
h
as
b
ee
n
p
r
o
p
o
s
ed
s
u
itab
le
f
o
r
ex
tr
ac
ti
n
g
t
h
e
h
u
g
e
r
elate
d
f
ea
t
u
r
es
f
o
r
m
o
n
ito
r
i
n
g
th
e
m
ac
h
in
e
co
n
d
itio
n
a
n
d
t
h
e
r
elev
an
t
ap
p
licatio
n
s
f
r
o
m
th
e
f
r
eq
u
en
c
y
d
o
m
ai
n
,
ti
m
e
d
o
m
ai
n
,
t
i
m
e
-
f
r
eq
u
e
n
c
y
d
o
m
ain
a
n
d
s
tatis
tical
s
h
ar
i
n
g
o
f
th
e
d
i
m
e
n
s
io
n
v
alu
e
s
[
1
]
.
T
h
e
ap
p
r
o
ac
h
is
f
u
ll
y
au
to
m
ated
an
d
s
u
itab
le
f
o
r
m
u
ltip
le
co
n
d
itio
n
m
o
n
ito
r
i
n
g
t
ask
s
s
u
c
h
as
v
ib
r
atio
n
a
n
d
p
r
o
ce
s
s
s
en
s
o
r
-
b
ased
an
al
y
s
is
.
T
h
is
ad
ap
tab
ilit
y
is
ex
p
lain
ed
b
y
a
n
al
y
zi
n
g
th
e
t
w
o
m
eth
o
d
o
f
t
h
e
m
o
n
ito
r
in
g
d
ataset
s
f
r
o
m
th
e
s
e
lf
-
e
x
a
m
in
a
tio
n
a
n
d
t
h
e
v
ar
io
u
s
f
r
ee
l
y
p
r
ese
n
t
ti
m
e
s
er
i
es c
las
s
i
f
icatio
n
o
p
er
atio
n
s
.
U
s
u
all
y
t
h
e
co
n
d
itio
n
m
o
n
ito
r
i
n
g
is
tr
ai
n
ed
in
t
h
r
ee
s
tep
s
,
n
a
m
el
y
f
ea
tu
r
e
e
x
tr
ac
ti
o
n
(
FE)
f
r
o
m
r
a
w
d
ata,
f
ea
t
u
r
e
s
elec
tio
n
(
F
S)
o
f
th
e
m
o
s
t
r
elev
a
n
t
f
ea
t
u
r
es
a
n
d
f
a
u
lt
clas
s
i
f
icatio
n
u
s
in
g
m
ac
h
in
e
lear
n
i
n
g
.
FE
a
n
d
F
S
th
er
eb
y
r
ep
r
esen
t
a
g
r
ad
u
al
d
ata
d
i
m
en
s
io
n
al
it
y
r
ed
u
ctio
n
th
at
i
s
n
ec
es
s
ar
y
t
o
p
r
ev
en
t
th
e
clas
s
i
f
ier
to
s
u
f
f
er
f
r
o
m
o
v
er
f
itti
n
g
an
d
th
e
“
c
u
r
s
e
o
f
d
i
m
en
s
io
n
al
it
y
”
[
2
]
,
w
h
ic
h
m
a
k
es
m
ac
h
in
e
lear
n
in
g
in
e
f
f
ec
ti
v
e,
if
th
e
n
u
m
b
er
o
f
f
ea
t
u
r
es
i
s
h
ig
h
.
T
h
u
s
,
h
ig
h
f
ea
tu
r
e
q
u
a
lit
y
is
t
h
e
p
r
ec
o
n
d
itio
n
f
o
r
g
o
o
d
p
r
e
d
ictio
n
r
esu
lt
s
.
I
n
th
e
f
ield
o
f
f
ea
tu
r
e
ex
tr
ac
tio
n
,
al
g
o
r
ith
m
s
ca
n
b
e
g
r
o
u
p
ed
in
to
o
n
e
o
f
t
h
e
f
o
llo
win
g
ca
te
g
o
r
ies:
a.
P
iece
w
is
e
ap
p
r
o
x
i
m
a
tio
n
[
3
,
4
]
.
b.
T
im
e
d
o
m
ai
n
tr
an
s
f
o
r
m
atio
n
s
P
C
A
[
5
]
)
c.
Fre
q
u
en
c
y
d
o
m
ai
n
tr
an
s
f
o
r
m
a
tio
n
s
(
e.
g
.
Fo
u
r
ier
tr
an
s
f
o
r
m
at
io
n
[
6
]
)
d.
T
im
e
-
f
r
eq
u
e
n
c
y
tr
an
s
f
o
r
m
atio
n
s
(
e.
g
.
W
av
elet
tr
a
n
s
f
o
r
m
at
io
n
[
7
]
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
1
:
4
2
4
-
433
426
e.
Statis
t
ical
f
ea
t
u
r
es (
e.
g
.
[
8
,
9
]
)
E
ac
h
o
f
th
e
s
e
ap
p
r
o
ac
h
es
w
o
r
k
s
u
n
d
er
ce
r
tain
cir
cu
m
s
ta
n
ce
s
a
n
d
ca
n
e
x
tr
ac
t
i
n
f
o
r
m
atio
n
u
n
o
b
tai
n
ab
le
f
o
r
th
e
o
th
er
s
.
T
h
is
a
n
d
th
e
i
m
p
o
s
s
ib
ilit
y
to
a
u
to
m
at
icall
y
(
i.e
.
w
it
h
o
u
t
d
et
ailed
k
n
o
w
led
g
e
o
f
th
e
r
esp
ec
tiv
e
s
y
s
te
m
)
d
eter
m
i
n
e
th
e
b
est
m
et
h
o
d
in
ad
v
an
ce
m
o
tiv
ate
s
th
e
b
asic
id
ea
o
f
th
e
ap
p
r
o
ac
h
p
r
esen
ted
in
t
h
i
s
p
ap
er
.
I
n
[
1
0
]
,
th
e
y
h
av
e
in
tr
o
d
u
ce
d
“
A
N
o
v
el
m
er
it
f
o
r
th
e
s
elec
tio
n
o
f
f
ea
t
u
r
e
o
n
t
h
e
b
asi
s
o
f
r
o
u
g
h
s
et
th
eo
r
y
(
R
ST
)
”,
w
h
ic
h
is
m
o
ti
v
ated
b
y
th
e
m
er
it
o
f
th
e
co
r
r
elatio
n
-
d
ep
en
d
en
t
an
d
k
n
o
w
n
as
th
e
r
o
u
g
h
s
et
-
b
ased
m
er
it
is
i
m
p
le
m
e
n
ted
u
n
d
er
r
o
u
g
h
s
et
im
m
ed
iate
r
ed
u
ce
s
alg
o
r
it
h
m
t
o
ch
o
o
s
e
a
im
p
o
r
tan
t
s
u
b
s
et
o
f
t
h
e
f
ea
tu
r
es.
I
n
[
1
1
]
,
th
e
R
ST
w
a
s
d
ef
i
n
ed
f
o
r
th
e
an
al
y
s
i
s
o
f
d
ata
in
th
e
r
ec
o
g
n
itio
n
p
atter
n
,
m
ac
h
in
e
lear
n
i
n
g
an
d
t
h
e
d
ata
m
i
n
i
n
g
ar
ea
s
.
R
S
is
s
tr
o
n
g
to
o
l
to
id
en
tify
a
r
elate
d
r
elatio
n
a
m
o
n
g
s
t
t
h
e
cla
s
s
an
d
th
e
at
tr
ib
u
tes
u
n
d
er
th
e
d
ataset.
Hen
ce
,
u
tili
za
t
io
n
o
f
R
S
a
s
th
e
FS
s
tep
to
id
en
ti
f
y
a
d
ec
r
e
m
en
t
o
f
th
e
f
ea
tu
r
e
s
f
o
r
a
r
o
b
u
s
t
a
n
d
f
ast
ca
teg
o
r
izatio
n
o
f
t
h
e
d
atas
ets.
T
h
is
p
ap
er
g
iv
i
n
g
atte
n
tio
n
o
n
F
S
i
n
DN
A
o
f
th
e
m
icr
o
ar
r
ay
s
(
also
k
n
o
w
n
as
g
e
n
e
e
x
p
r
ess
io
n
)
d
ata
s
et
s
[
1
2
,
1
3
]
.
T
h
e
m
icr
o
ar
r
a
y
d
atasets
(
MD
s
)
h
ad
th
e
f
ea
t
u
r
es
o
f
r
ea
l
-
v
a
lu
e
s
a
m
e
a
s
t
h
e
r
ea
l
-
wo
r
ld
d
atase
ts
.
A
MD
s
is
a
s
tr
o
n
g
ex
a
m
p
le
o
f
t
h
e
lar
g
e
d
i
m
en
s
io
n
al
it
y
.
MD
s
co
n
s
i
s
ts
o
f
th
e
s
e
v
er
al
f
ea
t
u
r
es
a
n
d
f
e
w
n
u
m
b
er
s
o
f
t
h
e
s
a
m
p
les.
M
o
r
eo
v
er
,
s
h
ar
in
g
o
f
th
e
clas
s
es i
n
b
in
ar
y
M
Ds is
u
n
eq
u
al
a
n
d
s
in
ce
t
h
e
y
ar
e
th
e
d
atasets
o
f
clas
s
-
i
m
b
ala
n
ce
d
.
2
.
2
.
P
er
f
o
rm
ing
t
he
o
pera
t
io
n o
f
a
uto
enco
der
o
n t
he
s
elec
t
ed
da
t
a
f
ro
m
t
he
co
m
plet
e
da
t
a
I
n
p
ap
er
[
1
4
]
,
th
ey
i
n
tr
o
d
u
ce
d
a
“
s
e
m
i
-
s
u
p
er
v
i
s
ed
-
s
ta
ck
ed
-
lab
el
-
co
n
s
is
te
n
t
-
a
u
to
-
e
n
c
o
d
er
”
f
o
r
re
co
n
s
tr
u
ctio
n
a
n
d
in
v
e
s
tig
a
ti
o
n
o
f
B
io
-
s
i
g
n
a
ls
.
Old
r
e
s
ea
r
ch
h
as
g
i
v
en
r
ec
o
n
s
tr
u
ctio
n
a
n
d
ca
teg
o
r
izi
n
g
as
t
w
o
d
if
f
er
en
t
i
s
s
u
es.
Fo
r
th
e
p
u
r
p
o
s
e
o
f
tele
-
m
o
n
ito
r
in
g
,
th
e
m
et
h
o
d
o
lo
g
y
o
f
r
ec
o
n
s
tr
u
c
tio
n
o
f
b
io
-
s
i
g
n
al
s
ar
e
h
u
g
el
y
d
ep
en
d
in
g
o
n
t
h
e
co
m
p
r
ess
ed
-
s
e
n
s
i
n
g
(
C
S);
t
h
e
y
ar
e
d
esig
n
ed
ap
p
r
o
ac
h
es
w
h
er
e
th
e
f
o
r
m
u
latio
n
o
f
th
e
r
ec
o
n
s
tr
u
ctio
n
is
d
ep
en
d
ed
o
n
s
o
m
e
p
r
ed
ictio
n
ab
o
u
t
t
h
e
s
ig
n
al.
C
S
d
ep
en
d
e
n
t
p
r
o
j
ec
t
s
o
lu
tio
n
s
,
a
p
ar
t
o
f
th
e
g
a
th
er
ed
s
i
g
n
al
(
s
u
ch
a
s
o
n
e
s
ec
o
n
d
)
o
n
a
m
atr
i
x
t
h
at
is
r
an
d
o
m
(
s
p
ar
s
e
b
in
ar
y
,
Gau
s
s
i
an
an
d
b
in
o
m
ial)
s
o
th
at,
t
h
e
p
r
o
j
ec
ted
d
ata’
s
s
ize
is
les
s
w
h
e
n
co
m
p
ar
ed
w
it
h
t
h
e
s
ize
o
f
s
a
m
p
le
s
t
h
at
w
a
s
t
ak
en
in
o
n
e
s
ec
o
n
d
[
1
5
,
16
]
.
A
s
C
S
n
ee
d
s
o
n
l
y
a
p
r
o
d
u
ct
o
f
m
atr
i
x
-
v
ec
to
r
,
it
co
s
ts
le
s
s
f
o
r
co
m
p
u
tatio
n
.
T
h
er
e
ar
e
m
a
n
y
r
e
s
ea
r
ch
s
tu
d
ie
s
,
w
h
ic
h
i
m
p
le
m
e
n
t
ef
f
ec
tiv
e
co
m
p
u
tatio
n
h
ar
d
w
ar
e
an
d
en
er
g
y
e
f
f
ec
ti
v
e
f
o
r
s
a
m
e
[
1
7
-
2
1
]
.
T
h
e
d
at
a
th
at
i
s
co
m
p
r
es
s
ed
w
ir
eles
s
l
y
s
e
n
t
to
b
ase
s
ta
tio
n
.
A
t
b
ase
s
tat
io
n
,
r
ec
o
n
s
tr
u
ctio
n
with
t
h
e
h
elp
o
f
C
S
m
et
h
o
d
o
lo
g
y
is
d
o
n
e
f
o
r
later
m
o
n
i
to
r
in
g
an
d
e
x
a
m
in
at
io
n
[
2
2
,
2
3
]
.
T
h
er
e
ca
n
b
e
m
an
y
t
y
p
e
s
o
f
g
e
n
er
al
C
S
m
et
h
o
d
o
lo
g
y
f
o
r
r
e
-
co
n
s
tr
u
cti
o
n
;
[
1
7
,
24
,
2
5
]
u
s
e
t
y
p
es
o
f
s
p
ar
s
e
-
b
a
y
esia
n
lear
n
in
g
(
SB
)
[
2
6
,
27
,
28
]
.
Ma
n
y
o
f
t
h
e
p
ap
er
s
u
til
ize
t
h
e
t
y
p
i
ca
l
C
S
m
et
h
o
d
s
f
o
r
r
ec
o
v
er
y
in
f
e
w
o
f
t
h
e
o
p
er
atio
n
s
,
t
h
e
r
ec
o
n
s
tr
u
ctio
n
o
f
th
e
s
i
g
n
als
w
a
s
d
o
n
e
w
ith
t
h
e
h
elp
o
f
in
tr
a
an
d
in
ter
-
ch
a
n
n
el
co
r
r
elatio
n
s
[
29
,
3
0
]
.
I
n
th
i
s
ar
ticle,
t
h
e
y
i
n
tr
o
d
u
ce
a
n
e
w
ar
g
u
m
en
ta
tiv
e
g
r
ap
h
,
s
u
r
r
o
u
n
d
in
g
s
tr
u
ct
u
r
e
f
o
r
d
ata
o
f
th
e
g
r
ap
h
.
T
h
e
s
tr
u
ct
u
r
e
tr
an
s
lates
th
e
t
o
p
o
lo
g
ical
f
r
a
m
e
w
o
r
k
an
d
c
o
n
ten
t
o
f
t
h
e
n
o
d
e
in
a
g
r
a
p
h
f
o
r
co
m
p
r
es
s
ed
d
em
o
n
s
tr
atio
n
,
o
n
w
h
ich
t
h
e
t
r
ain
in
g
o
f
t
h
e
d
ec
o
d
er
is
d
o
n
e
to
r
ec
o
n
s
tr
u
ct
th
e
s
tr
u
ct
u
r
e
o
f
th
e
g
r
ap
h
[
31
,
3
2
].
Gr
ap
h
s
ar
e
i
m
p
o
r
tan
t
to
o
ls
to
r
ec
o
g
n
ize
an
d
b
u
ild
co
m
p
le
x
r
elatio
n
s
b
et
w
ee
n
t
h
e
d
ata.
I
n
d
i
f
f
er
e
n
t
t
y
p
es
o
f
g
r
ap
h
ap
p
licatio
n
s
co
n
s
is
tin
g
p
r
o
tein
-
p
r
o
tein
co
m
m
u
n
icat
i
o
n
n
et
w
o
r
k
,
citatio
n
n
et
w
o
r
k
an
d
s
o
cial
m
ed
ia
n
et
w
o
r
k
s
,
d
ata
o
f
th
e
an
al
y
zin
g
g
r
ap
h
is
e
s
s
e
n
tial
i
n
d
if
f
er
en
t
d
ata
m
i
n
i
n
g
o
p
er
atio
n
s
co
n
s
i
s
tin
g
clas
s
i
f
icatio
n
o
f
t
h
e
n
o
d
es,
p
r
ed
ictio
n
o
f
t
h
e
lin
k
s
an
d
clu
s
ter
i
n
g
o
f
th
e
n
o
d
es
[
3
2,
33
]
.
A
n
y
w
a
y
s
,
t
h
e
p
a
r
alleliza
b
ilit
y
th
a
t i
s
lo
w
,
co
m
p
le
x
it
y
o
f
h
i
g
h
co
m
p
u
tatio
n
s
,
an
d
i
n
ca
p
ab
ilit
y
o
f
m
ac
h
in
e
-
le
ar
n
i
n
g
(
ML
)
tec
h
n
iq
u
e
to
t
h
e
d
ata
o
f
th
e
g
r
ap
h
m
ad
e
th
e
o
p
er
atio
n
o
f
a
n
al
y
tic
o
f
t
h
e
g
r
ap
h
m
o
r
e
d
if
f
ic
u
lt.
A
l
g
o
r
it
h
m
s
th
at
ar
e
b
ased
o
n
f
ac
to
r
izatio
n
o
f
t
h
e
m
atr
ix
.
F
o
r
e
x
a
m
p
le:
HOP
E
,
Gr
aRep
,
M
-
NM
F
[
1
5
,
29
]
p
r
e
-
p
r
o
ce
s
s
th
e
f
r
a
m
e
w
o
r
k
o
f
th
e
g
r
ap
h
in
to
an
ad
j
ac
en
cy
m
a
tr
ix
an
d
o
b
tain
th
e
e
m
b
ed
d
in
g
b
y
d
is
in
teg
r
ati
n
g
th
e
ad
j
ac
en
c
y
m
atr
i
x
.
L
atel
y
,
it
is
r
ep
r
esen
ted
t
h
at
s
ev
er
al
al
g
o
r
ith
m
s
o
f
p
r
o
b
ab
ili
s
tic
ar
e
eq
u
al
to
f
ac
to
r
izatio
n
o
f
m
atr
ix
m
eth
o
d
s
.
Dee
p
lear
n
in
g
m
et
h
o
d
s
s
p
ec
if
icall
y
a
u
to
-
e
n
co
d
er
d
ep
en
d
en
t
te
ch
n
iq
u
es,
ar
e
also
m
ai
n
l
y
r
esear
c
h
ed
f
o
r
e
m
b
ed
d
in
g
o
f
t
h
e
g
r
ap
h
.
T
h
e
alg
o
r
ith
m
o
f
MG
A
E
u
s
e
s
a
r
el
eg
ated
o
n
e
-
la
y
er
a
u
to
-
e
n
co
d
er
to
lear
n
clu
s
ter
i
n
g
r
ep
r
esen
tatio
n
[
34
,
3
5
]
.
An
a
u
to
-
e
n
co
d
er
th
at
is
s
tan
d
ar
d
ized
b
y
g
r
ap
h
w
as
p
r
o
p
o
s
ed
b
y
Yu
e
t
a
l
.
ai
m
in
g
to
s
e
lec
t
th
e
g
r
ap
h
to
g
u
id
e
th
e
o
p
er
atio
n
s
o
f
en
c
o
d
in
g
an
d
d
ec
o
d
in
g
.
A
n
y
w
a
y
s
,
it
is
s
till
h
ar
d
to
lear
n
w
i
th
s
ev
er
al
n
o
n
-
r
elate
d
p
atter
n
s
i
n
d
ata,
an
d
p
r
ese
n
t
au
to
-
e
n
co
d
er
d
if
f
er
en
ce
is
s
ti
ll
n
o
t
i
n
clu
d
ed
t
h
e
u
n
its
,
w
h
i
ch
ar
e
h
id
d
en
in
to
2
s
ep
ar
ate
p
ar
ts
,
f
ir
s
t is
w
o
r
k
-
r
elate
d
an
d
th
e
s
ec
o
n
d
o
n
e
i
s
wo
r
k
n
o
n
-
r
elate
d
.
3.
P
RO
P
O
SE
D
WO
RK
I
n
th
i
s
s
ec
tio
n
w
e
w
il
l
b
e
ex
p
lain
i
n
g
w
h
at
is
t
h
e
ai
m
o
f
o
u
r
al
g
o
r
ith
m
an
d
h
o
w
w
i
ll
th
at
b
e
p
er
f
o
r
m
ed
,
f
o
llo
w
in
g
b
y
t
h
e
co
m
b
i
n
ed
m
o
d
el
o
f
s
elec
ti
n
g
i
m
p
o
r
tan
t
d
ata
w
h
ich
ar
e
r
el
ev
an
t
a
n
d
ap
p
l
y
i
n
g
au
to
-
e
n
co
d
er
in
t
h
e
s
elec
ted
d
ata
o
f
h
id
d
en
la
y
er
s
to
co
m
p
r
es
s
th
e
d
ata.
A
f
ter
th
at
w
e
s
tu
d
y
ab
o
u
t
t
h
e
t
w
o
alg
o
r
ith
m
s
o
n
w
h
ic
h
o
u
r
p
r
o
p
o
s
ed
alg
o
r
ith
m
i
s
b
ased
.
An
d
also
,
i
n
d
etail
o
u
r
ar
ch
ite
ctu
r
e
is
ex
p
lai
n
ed
.
T
h
e
m
ai
n
co
n
ce
p
t
o
f
a
u
to
-
e
n
c
o
d
er
is
to
ar
r
an
g
e
th
e
d
ata
f
r
o
m
t
h
e
n
o
n
-
li
n
ea
r
e
n
co
d
er
in
to
th
e
h
id
d
en
la
y
er
s
an
d
th
e
h
id
d
en
u
n
it
s
ar
e
u
tili
z
ed
as th
e
n
e
w
d
ata
ar
r
an
g
e
m
e
n
ts
:
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
B
io
-
s
ig
n
a
ls
co
mp
r
ess
io
n
u
s
in
g
a
u
to
-
e
n
co
d
er (
S
u
n
ilku
ma
r
K
.
N
.
)
427
=
(
+
)
;
̂
=
(
+
)
(
1
)
Her
e,
∈
is
th
e
ar
r
a
n
g
e
m
e
n
t
o
f
t
h
e
h
id
d
en
u
n
it
s
an
d
̂
∈
is
as
s
u
m
ed
as
a
r
eb
u
i
lt
o
f
th
e
g
i
v
en
in
p
u
t
∈
.
T
h
e
s
et
o
f
t
h
e
p
ar
a
m
eter
s
co
n
s
is
t
s
o
f
w
e
ig
h
t
m
atr
ice
s
∈
×
,
∈
×
an
d
th
e
v
ec
to
r
s
w
h
ic
h
ar
e
o
f
f
s
et
ar
e
∈
an
d
∈
w
h
o
s
e
s
izes
ar
e
an
d
r
esp
ec
tiv
el
y
.
is
d
ef
in
ed
as
n
o
n
lin
ea
r
ac
tiv
atio
n
-
f
u
n
c
tio
n
.
T
h
e
Neu
r
al
Netw
o
r
k
w
h
ich
h
as
s
a
m
e
i
n
p
u
t a
n
d
tar
g
et
d
ata
is
n
o
th
in
g
b
u
t a
n
au
to
-
e
n
co
d
er
w
h
o
co
n
s
is
t o
f
o
n
l
y
o
n
e
h
id
d
en
la
y
er
,
s
u
c
h
t
h
at
min
,
,
,
1
2
∑
‖
−
̂
‖
=
1
2
2
(
2
)
Her
e,
n
is
th
e
d
ata’
s
s
ize
o
f
th
e
s
a
m
p
le,
is
th
e
r
esu
lt
af
ter
r
e
-
b
u
ild
i
n
g
it
an
d
̂
is
th
e
tar
g
et
d
ata
to
w
h
ic
h
th
e
r
eb
u
ilt
d
ata
i
s
co
m
p
ar
e
d
.
A
b
etter
ar
r
an
g
e
m
e
n
t
o
f
th
e
h
id
d
en
la
y
er
d
ata
ca
n
b
e
g
ain
ed
w
it
h
th
e
s
k
ill
s
to
r
eb
u
ild
th
e
d
ata.
As
w
e
s
t
u
d
ied
b
ef
o
r
e,
all
h
id
d
en
u
n
its
,
w
h
ich
ar
e
at
h
i
g
h
-
lev
el,
h
elp
i
n
f
i
n
d
in
g
t
h
e
ess
en
tial
d
ata
f
r
o
m
t
h
e
g
iv
e
n
i
n
p
u
t
d
ata
wh
ile
r
eb
u
ild
i
n
g
t
h
e
d
ata
o
r
ar
r
an
g
i
n
g
th
e
s
elec
ted
i
m
p
o
r
ta
n
t
d
ata.
A
n
y
w
a
y
s
,
th
e
i
m
p
o
r
tan
ce
o
f
t
h
ese
u
n
it
s
i
s
n
o
t
s
a
m
e
as
t
h
e
i
m
p
o
r
t
an
ce
o
f
o
u
r
class
i
f
icatio
n
w
o
r
k
.
Fo
r
in
s
tan
ce
,
s
o
m
e
d
ata
ar
e
u
s
ed
to
b
u
i
ld
th
e
b
ac
k
g
r
o
u
n
d
i
m
a
g
e
i
n
th
e
o
b
j
ec
t
i
m
ag
e,
w
h
ich
is
n
o
t
r
eq
u
i
r
ed
.
T
h
o
s
e
ty
p
e
o
f
d
ata
ar
e
k
n
o
w
n
as
ir
r
elev
a
n
t
d
ata
o
r
less
im
p
o
r
tan
t
d
ata,
w
h
ic
h
ar
e
u
n
w
a
n
ted
f
o
r
o
u
r
f
ea
tu
r
es,
w
h
ich
h
av
e
b
ee
n
tr
ain
ed
n
e
w
l
y
.
I
n
p
ar
allel
,
m
ea
n
w
h
ile
t
h
e
o
ld
m
et
h
o
d
,
w
h
ic
h
i
s
n
o
t
s
u
p
er
v
is
ed
,
h
ad
a
ca
p
ac
it
y
w
i
th
s
o
m
e
li
m
it
s
to
r
ec
o
n
s
tr
u
ct
th
e
i
n
p
u
t
d
is
tr
ib
u
tio
n
m
ar
g
in
al
l
y
f
o
r
th
e
ai
m
-
s
u
p
er
v
is
ed
p
r
o
ce
s
s
es.
O
n
t
h
e
h
id
d
en
u
n
it
s
,
f
e
w
ex
i
s
ti
n
g
w
o
r
k
s
m
ad
e
th
e
l
ab
el
in
f
o
r
m
at
io
n
w
it
h
th
e
h
elp
o
f
So
f
t
-
Ma
x
L
a
y
er
b
y
“
So
ch
e
r
et
a
l
.
”
in
th
e
y
ea
r
2
0
1
1
.
Ma
k
in
g
co
u
n
t
o
f
th
e
r
ec
en
t
ap
p
r
o
x
i
m
atio
n
r
eg
ar
d
in
g
p
r
o
ce
s
s
u
n
its
,
w
h
ich
ar
e
n
o
t
r
elev
an
t,
it
is
n
o
t
o
k
a
y
o
r
also
to
d
ec
lar
e
all
th
e
h
id
d
en
u
n
it
s
,
as
th
e
y
ar
e
u
s
ele
s
s
.
So
w
e
h
a
v
e
t
w
o
o
u
tco
m
es:
1
)
o
n
e
is
to
s
elec
t
th
e
i
m
p
o
r
tan
t
d
ata
r
eq
u
ir
ed
o
u
t
o
f
ir
r
elev
a
n
t
d
ata
w
it
h
th
e
h
elp
o
f
Feat
u
r
e
s
elec
tio
n
ap
p
r
o
ac
h
,
an
d
2
)
th
e
s
elec
ted
i
m
p
o
r
tan
t
d
ata
o
n
l
y
h
as
to
co
m
b
in
ed
an
d
s
e
n
t
b
y
th
e
a
u
to
-
e
n
co
d
er
.
W
ith
th
e
a
b
o
v
e
s
t
u
d
y
,
w
e
i
m
p
le
m
e
n
t
o
u
r
co
m
b
i
n
ed
s
elec
tio
n
o
f
i
m
p
o
r
tan
t
d
ata
an
d
co
m
p
r
es
s
i
n
g
t
h
e
s
elec
ted
d
ata
u
s
i
n
g
au
to
-
en
co
d
er
alg
o
r
ith
m
in
an
i
n
te
g
r
ated
s
tr
u
ct
u
r
e.
3
.
1
.
B
io
-
s
ig
na
ls
co
m
pres
s
io
n
us
ing
a
uto
-
enco
der
(
B
CAE
)
I
n
th
i
s
p
ar
t,
w
e
i
m
p
le
m
e
n
t o
u
r
alg
o
r
ith
m
b
y
co
m
b
i
n
in
g
b
o
th
th
e
co
n
ce
p
t o
f
f
ea
t
u
r
e
s
e
lectio
n
to
s
ele
c
t
th
e
i
m
p
o
r
tan
t
d
ata
a
n
d
au
to
-
en
co
d
er
to
co
m
p
r
ess
th
e
d
ata
w
i
th
i
n
a
s
in
g
le
al
g
o
r
it
h
m
.
E
x
p
licitl
y
,
w
e
ap
p
l
y
f
ea
t
u
r
e
s
elec
tio
n
ap
p
r
o
ac
h
o
n
th
e
u
n
it
s
o
f
h
id
d
en
la
y
er
d
ata.
A
s
s
u
m
e
∈
2
×
is
th
e
s
et
o
f
tr
ain
i
n
g
ele
m
e
n
ts
,
w
h
er
e
2
is
th
e
s
ize
o
f
th
e
v
i
s
u
al
d
escr
ip
to
r
an
d
n
is
t
h
e
a
m
o
u
n
t o
f
s
a
m
p
les o
f
t
h
e
d
ata.
mi
n
,
,
,
,
1
2
‖
−
(
(
)
)
‖
2
+
10
5
(
,
(
)
)
(
3
)
W
h
er
e
(
+
)
is
a
s
s
i
g
n
ed
a
s
(
)
an
d
(
(
)
+
)
is
ass
i
g
n
ed
a
s
(
(
)
)
an
d
an
d
ar
e
ea
ch
iter
atio
n
’
s
d
ata
r
ep
r
o
d
u
ctio
n
o
f
co
lu
m
n
o
f
an
d
r
esp
ec
tiv
el
y
.
T
h
e
s
ta
n
d
ar
d
ized
f
o
r
m
o
f
s
elec
tio
n
o
f
i
m
p
o
r
tan
t
d
ata
is
g
i
v
en
as
(
,
(
)
)
,
alo
n
g
w
it
h
i.e
.
m
a
tr
ix
co
n
s
is
t
in
g
tr
ai
n
ed
f
e
atu
r
e
s
elec
tio
n
d
ata
p
er
f
o
r
m
i
n
g
o
n
(
)
w
h
ic
h
is
r
ep
r
esen
ted
as
h
id
d
en
u
n
its
.
E
x
p
licitl
y
,
in
,
th
e
v
ec
to
r
o
f
ℎ
co
lu
m
n
is
r
ep
r
esen
ted
as
∈
w
h
i
ch
is
r
ep
r
esen
ted
as,
=
[
0
,
…
,
0
,
1
,
0
,
…
,
0
]
’
(
4
)
Her
e,
th
e
h
id
d
en
u
n
it
s
’
a
m
o
u
n
t
is
d
en
o
ted
as
an
d
r
ep
r
esen
t
s
w
h
ic
h
co
lu
m
n
v
ec
to
r
ch
o
o
s
es
th
e
-
t
h
u
n
it
f
r
o
m
t
h
e
s
u
b
co
llectio
n
s
o
f
t
h
e
f
ea
t
u
r
es
s
e
lecte
d
∈
.
T
h
en
th
e
m
eth
o
d
o
f
f
ea
tu
r
e
s
elec
tin
g
ca
n
b
e
r
ep
r
esen
ted
as
th
e
r
ea
l
s
et
o
f
f
ea
t
u
r
es
(
)
,
lo
o
k
in
g
f
o
r
th
e
Q
-
m
atr
ix
to
ch
o
o
s
e
t
h
e
n
e
w
f
ea
t
u
r
es
an
d
co
llec
t
in
to
a
s
et
=
′
(
)
th
at
o
p
ti
m
izes t
h
e
r
e
latab
le
cr
iter
io
n
(
,
(
)
)
.
Gen
er
all
y
,
o
p
er
atio
n
f
o
r
s
el
ec
tio
n
o
f
d
ata
th
at
ar
e
n
ec
ess
ar
y
ca
n
b
e
ta
k
es
p
lace
i
n
3
w
a
y
s
:
u
n
s
u
p
er
v
is
ed
,
s
u
p
er
v
i
s
ed
,
s
e
m
i
-
s
u
p
er
v
i
s
ed
.
C
o
m
in
g
to
s
u
p
er
v
is
ed
m
e
th
o
d
s
,
m
o
s
t
o
f
th
e
ti
m
e
s
i
t
co
n
s
is
t
o
f
ir
r
elev
an
t
i
n
f
o
r
m
atio
n
w
h
ile
s
elec
ti
n
g
t
h
e
n
ec
e
s
s
ar
y
f
ea
t
u
r
es
s
u
c
h
as
f
is
h
er
s
co
r
e
th
at
w
as
i
n
tr
o
d
u
ce
d
in
th
e
y
ea
r
o
f
2
0
1
2
,
b
y
Han
,
L
i
an
d
G
u
.
I
n
th
e
y
ea
r
o
f
2
0
0
5
,
Ni
y
o
g
i,
He
a
n
d
C
ai
i
n
tr
o
d
u
ce
d
lap
lacia
n
s
co
r
e
f
o
r
u
n
s
u
p
er
v
is
ed
m
e
th
o
d
.
Un
s
u
p
e
r
v
is
ed
m
et
h
o
d
s
is
as
s
e
m
b
les
o
n
l
y
t
h
e
r
elev
a
n
t
a
n
d
n
ee
d
ed
d
ata
w
h
ile
s
elec
tin
g
m
ai
n
f
ea
t
u
r
es.
T
r
y
in
g
to
w
o
r
k
w
it
h
v
ar
io
u
s
ca
s
es
i
n
t
h
e
ac
tu
al
w
o
r
ld
,
w
e
s
i
m
p
lify
t
h
e
(
,
(
)
)
in
s
i
m
p
lest
m
et
h
o
d
.
E
x
p
licitl
y
,
w
e
i
n
tr
o
d
u
ce
th
e
r
eg
u
lar
izer
f
o
r
th
e
s
elec
tio
n
o
f
f
ea
tu
r
es i
n
g
en
er
al
as
:
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
1
:
4
2
4
-
433
428
(
,
(
)
)
=
(
′
(
)
′
(
)
)
(
′
(
)
′
(
)
)
(
5
)
T
h
at
o
f
f
er
s
a
b
asic
m
et
h
o
d
to
ad
j
u
s
t
w
it
h
v
ar
io
u
s
ca
s
es
b
y
s
elec
ti
n
g
an
d
in
v
ar
io
u
s
m
et
h
o
d
s
.
O
u
r
p
r
o
p
o
s
ed
s
y
s
te
m
f
o
llo
w
s
t
h
e
f
i
s
h
er
s
c
o
r
e
s
u
p
er
v
i
s
ed
ap
p
r
o
ac
h
in
o
r
d
er
to
co
llect
th
e
in
f
o
r
m
at
io
n
i
n
s
u
p
er
v
is
ed
m
an
n
er
.
Fo
r
f
is
h
er
s
co
r
e
,
t
w
o
g
r
ap
h
s
an
d
w
h
ic
h
ar
e
u
n
d
ir
e
cted
ar
e
b
u
ilt
u
s
i
n
g
p
r
o
v
id
ed
d
ata
(
h
er
e,
w
e
tak
e
h
elp
o
f
t
h
e
i
n
p
u
t
d
ata
P
f
r
o
m
t
h
e
o
r
ig
i
n
al
o
n
e
to
k
e
ep
th
e
g
eo
m
e
tr
ical
f
r
a
m
e
w
o
r
k
s
a
f
e
w
h
ile
s
elec
ti
n
g
t
h
e
f
ea
tu
r
es),
w
h
ich
co
r
r
esp
o
n
d
in
g
l
y
e
f
f
ec
ts
t
h
e
af
f
i
n
it
y
r
elatio
n
s
h
ip
o
f
th
e
b
et
w
ee
n
-
clas
s
a
n
d
i
n
s
id
e
-
clas
s
(
i
n
tr
o
d
u
ce
d
b
y
-
Yan
et
a
l.
i
n
t
h
e
y
ea
r
-
2
0
0
7
)
.
I
n
th
e
s
a
m
e
m
a
n
n
er
,
t
w
o
w
eig
h
ted
m
atr
ice
s
an
d
ar
e
g
en
er
ated
to
ca
teg
o
r
ies
b
o
th
th
e
g
r
ap
h
s
.
T
h
u
s
,
w
e
g
et
th
e
L
ap
lacia
n
m
atr
ices
d
en
o
ted
as
=
−
,
h
er
e
is
th
e
′
d
ia
g
o
n
al
m
atr
ices,
in
t
h
e
s
a
m
e
w
a
y
f
o
r
an
d
.
B
y
eq
u
ati
n
g
b
elo
w
i
s
s
u
e
o
f
o
p
ti
m
izat
io
n
,
w
e
ca
n
g
et
t
h
e
s
e
lectio
n
o
f
f
ea
t
u
r
e
m
atr
i
x
Q
th
a
t g
e
n
er
ate
s
th
e
s
u
b
s
et
o
f
th
e
f
ea
t
u
r
es
w
it
h
th
e
lo
w
cr
iter
io
n
s
co
r
e:
=
min
(
′
(
)
′
(
)
)
(
′
(
)
′
(
)
)
(
6
)
Un
w
a
n
ted
l
y
,
th
er
e
h
a
s
n
o
t
b
ee
n
a
d
ir
ec
t
p
r
o
b
le
m
to
r
eso
l
v
e
t
h
e
g
i
v
e
n
ab
o
v
e
tr
ac
e
-
r
at
io
is
s
u
es
,
b
ec
au
s
e
o
f
t
h
e
ab
s
e
n
ce
o
f
cl
o
s
ed
-
f
o
r
m
r
eso
l
u
tio
n
s
.
T
h
er
ef
o
r
e,
in
p
lace
o
f
s
tr
ai
g
h
tf
o
r
war
d
ly
d
ea
lin
g
w
it
h
tr
ac
e
-
r
atio
is
s
u
e
s
.
Ma
n
y
j
o
b
s
d
esire
s
to
ch
a
n
g
e
to
s
i
m
ilar
tr
ac
e
-
v
ar
io
u
s
i
s
s
u
es
to
g
ai
n
a
g
lo
b
al
r
eso
lu
tio
n
f
o
r
o
p
tim
izatio
n
(
P
r
o
p
o
s
ed
b
y
Ni
e
et
a
l.
in
y
ea
r
-
2
0
0
8
)
.
A
s
s
u
m
e
s
tan
d
ar
d
s
co
r
e
f
o
r
s
u
b
s
et
-
le
v
el
is
(
,
(
)
)
in
th
e
(
5
)
g
o
es u
p
to
th
e
g
lo
b
al
lo
w
e
s
t
∗
s
atis
f
y
i
n
g,
∗
=
min
(
′
(
)
′
(
)
)
(
′
(
)
′
(
)
)
(
7
)
T
h
is
is
to
s
h
o
w
t
h
at,
⟹
(
′
(
)
′
(
)
)
(
′
(
)
′
(
)
)
≥
∗
,
∀
⟹
(
′
(
)
(
−
∗
)
′
(
)
)
≥
0
,
∀
⟹
min
(
′
(
)
(
−
∗
)
′
(
)
)
=
0
T
o
th
is
s
ta
g
e,
w
e
ca
n
r
ep
r
esen
t th
e
f
u
n
ctio
n
o
f
w
h
ile
u
s
in
g
o
th
er
s
as c
o
n
s
tan
t a
s
:
(
)
=
min
(
′
(
)
(
−
)
′
(
)
)
(
8
)
T
h
u
s
,
lo
o
k
in
g
f
o
r
th
e
g
lo
b
al
o
p
tim
al
ca
n
b
e
tr
an
s
f
o
r
m
ed
in
to
eq
u
atin
g
th
e
eq
u
atio
n
’
s
r
o
o
t
(
)
=
0
,
w
h
ich
i
s
tr
ac
e
-
v
ar
ian
ce
is
s
u
e.
See
th
at
(
)
is
a
f
u
n
ctio
n
,
w
h
ich
i
n
cr
ea
s
e
s
m
o
n
o
to
n
ic
all
y
.
B
y
p
r
o
p
o
s
in
g
t
h
e
tr
ac
e
v
a
r
ian
ce
o
p
ti
m
izatio
n
is
s
u
e,
w
h
ic
h
is
ill
u
s
tr
ated
ab
o
v
e
in
th
e
(
8
)
w
i
th
i
n
th
e
au
to
-
e
n
co
d
er
’
s
h
id
d
en
la
y
er
u
p
d
atin
g
,
w
e
eq
u
ate
o
u
r
f
i
n
al
ai
m
f
u
n
ctio
n
as:
min
,
,
,
,
,
=
1
2
‖
−
(
(
)
)
‖
2
+
2
(
′
(
)
(
−
)
′
(
)
)
(
9
)
Her
e,
is
th
e
p
ar
a
m
eter
u
s
ed
to
b
alan
ce
w
h
ic
h
is
p
lace
d
b
et
w
ee
n
th
e
s
elec
tio
n
o
f
t
h
e
f
ea
t
u
r
e
an
d
au
to
-
en
co
d
er
ter
m
.
Den
o
tes
t
h
e
r
atio
o
f
tr
ac
e’
s
s
co
r
e
af
ter
g
ai
n
i
n
g
th
e
o
p
ti
m
izatio
n
Q
an
d
h
er
e
Q
i
s
th
e
tr
ac
e
r
atio
o
f
o
p
ti
m
izatio
n
o
f
r
ec
en
t is
s
u
e.
3
.
2
.
E
nh
a
nce
m
e
nt
Fo
r
m
u
la
(
9
)
is
b
it
d
i
f
f
icu
l
t
to
r
eso
lv
e
b
ec
au
s
e
o
f
it
s
co
m
p
le
x
an
d
n
o
n
-
li
n
ea
r
it
y
o
f
t
h
e
d
ec
o
d
er
s
an
d
th
e
en
co
d
er
s
,
th
er
ef
o
r
e
an
o
th
er
w
a
y
f
o
r
en
h
an
c
in
g
is
u
s
e
d
to
r
en
e
w
t
h
e
p
ar
am
e
ter
s
o
f
th
e
au
to
-
e
n
co
d
er
,
,
,
an
d
v
ar
iab
le
o
f
s
elec
tio
n
o
f
th
e
f
ea
t
u
r
e
Q
an
d
also
ea
ch
ti
m
e.
E
x
p
licitl
y
,
w
e
r
eso
lv
e
th
e
o
p
ti
m
iza
tio
n
w
it
h
2
s
u
b
-
i
s
s
u
es.
Firs
t
is
lear
n
in
g
t
h
e
s
c
o
r
e
o
f
th
e
s
elec
tio
n
o
f
t
h
e
f
ea
tu
r
e
an
d
t
h
e
s
ec
o
n
d
o
n
e
is
th
e
o
p
ti
m
izatio
n
o
f
th
e
au
to
-
e
n
co
d
er
w
h
ich
i
s
r
eg
u
lar
i
ze
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
B
io
-
s
ig
n
a
ls
co
mp
r
ess
io
n
u
s
in
g
a
u
to
-
e
n
co
d
er (
S
u
n
ilku
ma
r
K
.
N
.
)
429
3
.
3
.
L
ea
rning
t
he
s
co
re
o
f
t
h
e
s
elec
t
io
n o
f
t
he
f
ea
t
ures
W
h
en
t
h
e
au
to
-
en
co
d
er
’
s
p
ar
am
eter
s
ar
e
m
ad
e
co
n
s
tan
t,
w
e
ca
n
en
h
an
ce
t
h
e
s
elec
tio
n
o
f
t
h
e
f
ea
t
u
r
e
s
s
co
r
e
an
d
m
atr
ix
o
f
s
elec
tio
n
o
f
t
h
e
f
ea
t
u
r
e
Q
i
n
a
n
o
ld
m
et
h
o
d
o
f
tr
ac
e
r
atio
.
E
x
p
lic
itl
y
,
w
e
u
s
e
tr
ac
e
v
ar
ian
ce
eq
u
a
tio
n
=
min
(
′
(
)
(
−
)
′
(
)
)
(
1
0
)
Ass
u
m
e
is
th
e
o
u
tco
m
e
af
ter
o
p
tim
izatio
n
in
th
e
k
-
t
h
iter
atio
n
o
f
o
p
tim
iza
tio
n
,
th
er
e
f
o
r
e
is
ev
alu
a
ted
b
y
=
(
′
(
)
′
(
)
)
(
′
(
)
′
(
)
)
(
1
1
)
T
h
u
s
,
w
e
ca
n
g
et
(
)
as
(
)
=
(
+
1
′
(
)
(
−
)
′
(
)
+
1
)
(
1
2
)
w
h
er
e
+
1
m
a
y
b
e
e
v
al
u
ated
p
r
o
f
icien
tl
y
w
it
h
r
esp
ec
t
to
s
co
r
e
r
an
k
o
f
ev
er
y
s
i
n
g
le
f
ea
t
u
r
e.
T
h
e
eq
u
atio
n
’
s
r
o
o
t
(
)
=
0
an
d
th
e
e
n
h
an
ce
d
s
o
l
u
tio
n
f
o
r
(
6
)
ca
n
b
e
g
ai
n
ed
v
ia
it
s
r
ep
ea
tin
g
p
r
o
ce
s
s
.
See
t
h
at
is
r
en
e
w
ed
as
th
e
en
h
an
ce
d
s
co
r
e
w
h
ich
is
g
lo
b
al
f
o
r
th
e
cr
iter
io
n
o
f
th
e
s
elec
tio
n
o
f
th
e
f
ea
t
u
r
e
,
an
d
o
p
e
r
ates
as
a
p
ar
am
eter
i
n
s
u
cc
ee
d
in
g
iter
a
tio
n
o
f
au
to
-
e
n
co
d
er
.
A
lg
o
r
it
h
m
1
ex
p
lai
n
s
th
e
s
o
l
u
tio
n
o
f
en
h
a
n
ce
m
en
t.
Alg
o
rit
h
m
1
:
E
n
h
an
ce
m
e
n
t
f
o
r
th
e
is
s
u
e
o
f
T
r
ac
e
-
R
atio
.
Input:
Trained
features
of
the
hidden
layer
(
)
,
re
le
va
nt
da
ta
se
le
ct
ed
fr
o
m
th
e
gi
ve
n
input data
, matrices
and
.
Initialize:
as
si
gn
Q
wi
th
I,
wh
er
e
I
pr
es
en
ts
th
e
id
en
ti
ca
l
ma
tr
ix
a
nd
it
be
lo
ng
s
to
×
,
with Equation (11),
=
10
−
9
,
=
0
,
=
10
3
IF
≤
and non
-
converge
do
Calculate
score
of
every
single
feature
of
-
th
it
er
at
io
n
wi
th
th
e
he
lp
of
Eq
ua
ti
on
(1
1)
by assigning
=
[
0
,
…
,
0
,
1
,
0
,
…
,
0
]
’
Ar
range the features according to their score in incremental order.
Choose the highest
feature to renew
∈
×
Evaluate
with the help of Equation (11)
Convergence constrains is checked:
‖
−
‖
<
Start again from
IF
condition.
End of IF
Output:
is
th
e
ma
tr
ix
of
se
le
ct
io
n
of
th
e
f
ea
tu
re
an
d
is
th
e
en
ha
nc
em
en
t
sc
or
e
globally.
3
.
4
.
E
nh
a
nce
m
e
nt
o
f
t
he
a
u
to
-
enco
der
w
hich is
re
g
ula
rize
d
W
h
en
Q
an
d
ar
e
m
ad
e
co
n
s
tan
t,
w
e
ca
n
u
s
e
s
to
ch
ast
ic
-
s
u
b
-
g
r
ad
ien
t
-
d
esce
n
t
tech
n
iq
u
e
to
g
e
t
,
,
,
.
T
h
e
s
lo
p
es
o
f
th
e
ai
m
i
n
g
f
u
n
ctio
n
in
(
9
)
ac
c
o
r
d
in
g
to
th
e
p
ar
am
eter
s
o
f
th
e
d
ec
o
d
in
g
is
ca
lcu
lated
as
g
iv
en
b
elo
w
:
=
(
−
(
(
)
)
)
⨀
(
(
)
)
′
(
)
,
=
(
−
(
(
)
)
)
⨀
(
(
)
)
=
,
=
(
′
+
′
(
)
(
−
)
)
⊙
(
)
′
,
=
(
′
+
′
(
)
(
−
)
)
⊙
(
)
.
T
h
en
,
,
,
,
ca
n
b
e
r
en
e
w
ed
w
it
h
t
h
e
h
elp
o
f
t
h
e
al
g
o
r
ith
m
o
f
g
r
ad
ien
t
d
escen
t
a
s
g
iv
e
n
b
elo
w
:
=
−
,
=
−
,
=
−
,
=
−
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
1
:
4
2
4
-
433
430
Her
e,
th
e
r
ate
o
f
lear
n
i
n
g
is
d
en
o
ted
as
.
an
d
ar
e
th
e
co
lu
m
n
m
ea
n
o
f
an
d
r
esp
ec
tiv
el
y
.
T
o
ad
d
,
th
e
t
w
o
p
ar
t
o
f
th
e
is
s
u
e
s
g
i
v
en
ab
o
v
e
co
u
ld
b
e
r
en
ew
ed
r
ec
u
r
s
i
v
el
y
.
A
l
g
o
r
ith
m
2
ex
p
lai
n
s
th
e
en
h
a
n
ce
m
e
n
t i
n
d
etail.
Alg
o
rit
h
m
2:
R
eso
l
v
i
n
g
t
h
e
I
s
s
u
e
i
n
(
9
)
Input:
parameters
, Learning Data P, Layer_size, number of features selected
<
.
Initialize:
=
50
,
=
0
,
=
10
−
7
.
While
≤
and non
-
converge
do
Keep others constant a
nd renew
and
with the help of Equation (10).
Make
as constant and renew
1
,
2
1
,
2
with the help of Equation (13).
Check for the constraints of convergence:
‖
−
‖
∞
<
.
End
Output:
,
,
,
,
(
=
′
(
+
)
)
Ca
n
be
ut
il
iz
ed
as
an
in
p
ut
ne
xt
fo
r
ou
r
al
go
ri
th
m
,
to
fo
rm
th
e
fr
am
ew
or
k
of stack.
Examine:
the representation of the new features can be done as:
=
′
(
+
)
4.
E
XP
E
R
I
M
E
NT
A
L
RE
SUL
T
AND
ANA
L
YS
I
S
W
e
r
ep
r
esen
t
th
e
q
u
a
n
titati
v
e
o
u
tco
m
es
to
esti
m
a
te
th
e
ef
f
icien
c
y
o
f
m
o
d
i
f
ied
au
to
-
en
co
d
er
s
in
th
e
b
io
m
ed
ical
s
ig
n
al
co
m
p
r
e
s
s
io
n
.
MI
MI
C
-
I
I
[
2
1
]
,
th
e
d
atab
ase
ca
n
b
e
av
ailab
le
i
n
2
f
o
r
m
s
.
I
n
1
st
f
o
r
m
,
th
e
in
ter
ested
r
esear
ch
er
s
ca
n
g
et
th
e
te
x
t
v
er
s
io
n
o
f
f
lat
-
f
ile
o
f
m
ed
ical
d
atab
ase
an
d
also
th
e
s
c
h
e
m
e
o
f
ass
o
ciate
d
d
atab
ase
w
h
ich
p
e
r
m
its
to
r
ec
r
ea
te
th
e
d
atab
ase
u
tili
z
in
g
t
h
eir
tec
h
n
iq
u
e
o
f
c
h
o
ice.
I
n
2
nd
f
o
r
m
,
atten
ti
v
e
r
esear
ch
er
s
ca
n
o
b
t
ain
ac
ce
s
s
to
d
atab
ase
v
ia
s
e
cu
r
e
w
it
h
p
ass
w
o
r
d
w
eb
s
er
v
ices.
T
h
e
d
atab
ase
r
esear
ch
er
s
n
ee
d
th
e
co
n
s
u
m
er
s
to
ex
p
lain
th
e
m
s
el
v
es
w
it
h
th
e
h
elp
o
f
d
atab
ase
la
y
o
u
t
to
th
e
q
u
er
ies
o
f
p
r
o
g
r
a
m
d
atab
ase
u
tili
z
in
g
t
h
e
SQL
(
Q
u
er
y
o
u
tp
u
t
ca
n
b
e
ex
p
o
r
ted
to
co
m
m
a
-
s
ep
ar
ated
files
to
b
e
an
al
y
ze
d
o
f
fl
i
n
e)
.
T
h
e
q
u
er
y
o
u
tp
u
t
is
tr
an
s
p
o
r
ted
to
th
e
f
ile
s
as
co
m
m
a
-
s
ep
ar
ated
to
b
e
ex
a
m
in
ed
o
f
f
li
n
e
u
til
izin
g
s
tatis
t
ical
o
r
an
o
th
er
s
o
f
t
w
ar
e.
T
h
e
d
ata
is
p
r
o
ce
s
s
in
g
an
d
ac
ce
s
s
in
g
f
r
o
m
t
h
e
MI
MI
C
-
I
I
,
w
h
ic
h
is
v
er
y
co
m
p
le
x
.
I
t
ca
n
b
e
h
i
g
h
l
y
s
u
g
g
ested
t
h
at
t
h
e
s
t
u
d
ies
ar
e
b
as
ed
o
n
th
e
MI
MI
C
-
I
I
d
atab
ase
th
at
ca
n
b
e
d
ir
ec
ted
as
co
llab
o
r
ativ
e
ef
f
o
r
ts
,
w
h
i
ch
co
n
tai
n
s
ta
tis
t
ical,
r
elatio
n
al
an
d
cli
n
ical
d
atab
ase
e
x
p
er
tis
e.
A
t
t
h
e
e
n
d
,
w
e
u
t
ilized
P
h
y
s
io
n
et
MI
MI
C
I
I
d
atab
ase
[
2
1
]
th
at
co
n
s
i
s
ts
i
n
f
o
r
m
atio
n
tr
ac
es
f
o
r
t
h
e
E
C
G.
A
d
d
itio
n
a
ll
y
,
w
e
ass
u
m
ed
o
u
r
E
C
G
tr
ac
es,
w
h
ich
d
ev
elo
p
ed
v
ia
th
e
Z
ep
h
y
r
’
s
B
io
,
h
ar
n
e
s
s
3
h
ea
r
t
r
ate
m
o
n
ito
r
.
W
e
r
ep
r
esen
t
th
e
ch
o
s
e
n
o
u
tc
o
m
e
s
w
h
ic
h
is
g
o
tte
n
f
r
o
m
t
h
e
to
tal
o
f
1
6
9
8
5
lik
e
s
eq
u
en
c
es
f
r
o
m
3
v
ar
io
u
s
P
h
y
s
io
n
et
o
f
p
atien
ts
a
s
w
e
ll
as
1
4
0
0
f
o
r
o
u
r
m
eth
od
o
f
E
C
G
m
ea
s
u
r
e
m
e
n
ts
a
n
d
th
e
tes
t
is
p
ick
ed
s
o
as
to
co
v
er
th
e
i
m
p
o
r
tan
t r
an
g
e
o
f
s
itu
atio
n
s
.
Fig
u
r
e
1
r
ep
r
esen
ts
th
e
ac
tu
al
s
ig
n
al
g
r
ap
h
s
an
d
r
ec
r
ea
ted
t
h
e
w
a
v
ef
o
r
m
o
f
E
C
G
s
ig
n
al
w
it
h
tr
ac
e
1
0
0
.
A
s
w
e
ca
n
s
ee
t
h
e
g
r
ap
h
,
th
e
ac
t
u
al
s
i
g
n
a
l
w
as
r
ec
r
ea
ted
s
u
cc
es
s
f
u
l
l
y
w
it
h
t
h
e
h
elp
o
f
p
r
o
p
o
s
ed
m
o
d
el
co
m
p
r
es
s
io
n
.
T
h
e
m
o
d
i
f
ied
au
to
en
co
d
er
(
MA
E
)
w
as
wo
r
k
ed
in
r
ec
r
ea
tin
g
all
R
i
n
ac
tu
al
s
i
g
n
al
w
it
h
th
e
h
elp
o
f
a
m
p
lit
u
d
e
d
if
f
er
en
ce
.
1
7
f
r
ag
m
en
ts
w
it
h
tr
ac
e
1
0
0
w
er
e
u
til
ized
in
test
p
h
ase.
Fig
u
r
e
2
r
ep
r
esen
ts
t
h
e
ac
tu
a
l
s
ig
n
al
g
r
ap
h
a
n
d
r
ec
r
ea
ted
th
e
E
C
G
s
ig
n
al
w
a
v
e
f
o
r
m
w
it
h
tr
ac
e
1
0
0
.
W
h
er
ea
s
,
w
e
ca
n
ce
e
t
h
at
t
h
e
g
r
ap
h
,
th
a
t
ac
tu
al
s
i
g
n
a
l
w
as
r
ec
r
ea
ted
s
u
cc
ess
f
u
ll
y
at
r
ec
ei
v
er
w
it
h
t
h
e
h
e
lp
o
f
p
r
o
p
o
s
ed
m
o
d
el
co
m
p
r
ess
io
n
.
T
h
e
M
A
E
w
as
s
u
cc
es
s
f
u
l
i
n
r
ec
r
ea
tin
g
th
e
ac
tu
a
l
s
ig
n
al
w
i
th
t
h
e
h
elp
o
f
lit
tle
d
if
f
er
e
n
ce
an
d
as t
h
e
h
id
d
en
n
o
d
es
w
er
e
m
a
x
i
m
izi
n
g
t
h
e
r
ec
o
n
s
tr
u
ct
io
n
ac
cu
r
ac
y
also
i
n
cr
ea
s
es.
Fig
u
r
e
3
r
ep
r
esen
ts
t
h
e
ac
tu
a
l
s
ig
n
al
g
r
ap
h
a
n
d
r
ec
r
ea
ted
th
e
E
C
G
s
ig
n
al
w
a
v
e
f
o
r
m
w
it
h
tr
ac
e
1
0
0
.
Fro
m
th
e
g
r
ap
h
,
w
e
ca
n
s
ee
th
e
d
ev
elo
p
m
en
t
i
n
th
e
r
ec
o
n
s
tr
u
cted
s
ig
n
al
w
it
h
th
e
ac
tu
al
s
ig
n
al.
W
ith
m
ax
i
m
ize
s
t
h
e
h
id
d
en
n
o
d
es
th
e
ac
c
u
r
ac
y
o
f
r
ec
o
n
s
tr
u
cti
n
g
h
a
s
i
m
p
r
o
v
ed
w
h
e
n
co
m
p
ar
ed
w
i
th
h
id
d
en
n
o
d
es
2
an
d
4
.
Fig
u
r
e
4
s
h
o
ws
th
e
i
m
p
r
o
v
e
m
en
t
in
t
h
e
r
ec
o
n
s
tr
u
cte
d
s
ig
n
al
o
f
E
C
G
s
ig
n
al
at
r
ec
eiv
er
w
i
th
th
at
o
f
ac
t
u
al
s
i
g
n
a
l.
I
n
th
e
a
b
o
v
e
f
ig
u
r
e
u
s
es
2
4
h
id
d
en
n
o
d
es
f
o
r
tr
ain
in
g
an
d
r
ec
o
n
s
t
r
u
ctin
g
t
h
e
s
i
g
n
al.
T
h
e
r
ec
o
n
s
tr
u
cted
s
ig
n
al
i
s
h
a
v
in
g
n
eg
l
ig
ib
le
a
m
p
lit
u
d
e
d
if
f
er
en
ce
w
ith
r
esp
ec
t to
o
r
ig
i
n
al
s
ig
n
al.
Fig
u
r
e
5
r
ep
r
esen
ts
th
e
ac
tu
a
l
s
ig
n
al
g
r
ap
h
a
n
d
t
h
e
r
ec
r
ea
ted
th
e
w
a
v
ef
o
r
m
o
f
a
n
E
C
G
s
i
g
n
al
w
it
h
tr
ac
e
1
0
0
.
A
s
w
e
ca
n
s
ee
f
r
o
m
th
e
g
r
ap
h
,
t
h
e
ac
tu
al
s
i
g
n
a
l
w
as
s
u
cc
es
s
f
u
l
l
y
r
ec
r
ea
ted
w
it
h
t
h
e
p
r
o
p
o
s
e
d
m
o
d
el
o
f
co
m
p
r
ess
io
n
f
o
r
h
id
d
en
n
o
d
es
4
8
.
T
h
e
m
o
d
i
f
ied
au
to
e
n
co
d
er
m
o
d
el
s
u
cc
ee
d
ed
in
r
ec
o
n
s
tr
u
ct
in
g
w
it
h
t
h
at
o
f
ac
t
u
al
s
i
g
n
al
w
it
h
ze
r
o
am
p
lit
u
d
e
d
if
f
er
en
ce
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
B
io
-
s
ig
n
a
ls
co
mp
r
ess
io
n
u
s
in
g
a
u
to
-
e
n
co
d
er (
S
u
n
ilku
ma
r
K
.
N
.
)
431
Fig
u
r
e
1
.
Featu
r
e
g
r
ap
h
at
h
id
d
en
n
o
d
es =
2
Fig
u
r
e
2
.
Featu
r
e
g
r
ap
h
at
hi
d
d
en
n
o
d
es =
4
Fig
u
r
e
3
.
Featu
r
e
g
r
ap
h
at
h
id
d
en
n
o
d
es =
1
2
Fig
u
r
e
4
.
Featu
r
e
g
r
ap
h
at
h
id
d
en
n
o
d
es =
2
4
Fig
u
r
e
5
.
Featu
r
e
g
r
ap
h
at
h
id
d
en
n
o
d
es =
4
8
Fig
u
r
e
6
s
h
o
w
s
t
h
e
r
o
o
t
m
ea
n
s
q
u
ar
e
er
r
o
r
v
s
en
er
g
y
co
n
s
u
m
p
tio
n
(
j
o
u
le
p
er
b
it)
u
s
i
n
g
o
u
r
p
r
o
p
o
s
ed
m
o
d
el
a
n
d
SU
R
F
[
24
]
,
w
h
er
e
o
u
r
m
o
d
if
ied
a
u
to
-
en
co
d
er
g
o
t
less
R
MSE
w
.
r
.
t
SUR
F
[
2
4
]
ap
p
r
o
ac
h
.
Fig
u
r
e
7
s
h
o
w
s
th
e
R
M
SE
(
%)
w
it
h
r
e
s
p
ec
t
to
C
o
m
p
r
es
s
io
n
E
f
f
icie
n
c
y
w
h
er
e
o
u
r
m
o
d
el
h
as
p
er
f
o
r
m
ed
s
i
g
n
i
f
ica
n
tl
y
w
ell
w
.
r
.
t
SU
R
F
[
24
]
ap
p
r
o
ac
h
.
Fig
u
r
e
8
r
ep
r
esen
ts
t
h
e
R
MSE
w
it
h
r
eg
ar
d
s
to
th
e
v
ar
io
u
s
n
u
m
b
er
o
f
th
e
Hid
d
en
No
d
es,
w
h
er
e
o
u
r
m
o
d
el
h
as p
er
f
o
r
m
ed
s
ig
n
i
f
ica
n
tl
y
w
ell
w
.
r
.
t SU
R
F [
24
]
ap
p
r
o
ac
h
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8708
I
n
t J
E
lec
&
C
o
m
p
E
n
g
,
Vo
l.
11
,
No
.
1
,
Feb
r
u
ar
y
2
0
2
1
:
4
2
4
-
433
432
Fig
u
r
e
6
.
R
MSE
v
s
e
n
er
g
y
c
o
n
s
u
m
p
t
io
n
Fig
u
r
e
7
.
R
MSE
v
s
c
o
m
p
r
ess
i
o
n
e
f
f
icie
n
c
y
Fig
u
r
e
8
.
R
MSE
w
it
h
r
esp
ec
t t
o
d
if
f
er
en
t
n
u
m
b
er
o
f
h
id
d
en
n
o
d
es
5.
CO
NCLU
SI
O
N
I
n
th
i
s
r
esear
ch
w
o
r
k
,
w
e
h
av
e
p
r
esen
ted
B
io
s
ig
n
a
l
co
mp
r
ess
io
n
u
s
in
g
a
u
to
en
co
d
e
r
s
(
B
C
A
E
)
co
m
p
r
es
s
io
n
al
g
o
r
it
h
m
f
o
r
wea
r
ab
le
f
itn
e
s
s
m
o
n
ito
r
s
u
s
i
n
g
I
o
T
.
W
e
p
r
o
p
o
s
ed
th
e
E
C
G
-
b
ased
b
io
m
etr
ic
s
y
s
te
m
w
h
er
ea
s
t
h
e
less
er
d
i
m
en
s
io
n
a
l
o
f
n
o
n
-
li
n
ea
r
r
ep
r
esen
tat
io
n
s
o
f
t
h
e
h
ea
r
tb
ea
t
te
m
p
lates
ar
e
lear
n
ed
th
r
o
u
g
h
t
h
e
d
ee
p
m
o
d
if
ied
Au
to
en
co
d
er
.
T
h
is
s
tu
d
y
also
r
ep
r
esen
ts
t
h
e
ef
f
icie
n
t
s
i
g
n
a
l
o
f
E
C
G
co
m
p
r
es
s
io
n
m
et
h
o
d
w
h
ic
h
i
s
u
tili
ze
d
i
n
b
io
m
ed
ical
f
ield
i.e
.
e
-
h
ea
l
th
ap
p
licatio
n
s
,
Ho
lter
s
y
s
te
m
s
an
d
tele
m
etr
y
.
T
h
e
p
r
o
p
o
s
ed
m
eth
o
d
o
f
co
m
p
r
ess
io
n
i
n
te
g
r
ate
s
th
e
ar
ch
ite
ctu
r
e
o
f
M
A
E
,
w
h
ic
h
h
av
e
r
e
ce
n
tl
y
b
ec
o
m
e
v
er
y
p
o
p
u
lar
in
t
h
e
m
ac
h
in
e
lear
n
i
n
g
(
ML
)
f
ie
ld
.
T
h
er
ef
o
r
e,
it
c
an
b
e
p
o
s
s
ib
le
to
g
et
t
h
e
d
ata
in
th
e
d
ee
p
la
y
er
s
ab
o
u
t
th
e
ac
tu
a
l
i
n
p
u
t
s
i
g
n
al
,
co
n
s
is
tin
g
lo
w
er
lev
el
to
h
ig
h
er
lev
el
o
f
f
ea
t
u
r
es.
I
n
t
h
is
s
t
u
d
y
,
th
e
d
ee
p
s
tr
u
ct
u
r
e
o
f
M
A
E
is
p
r
o
p
o
s
ed
,
2
0
0
0
×
1
E
C
G
s
i
g
n
als
w
er
e
r
ec
r
ea
ted
s
u
cc
es
s
f
u
ll
y
b
y
s
h
o
w
i
n
g
o
n
l
y
62
×
1
d
im
e
n
s
io
n
.
T
h
e
ap
p
licatio
n
o
f
co
m
p
r
eh
e
n
s
i
v
e
w
as
m
ad
e
o
n
th
e
4
8
0
0
f
r
ag
m
e
n
ts
o
f
2
0
0
0
s
a
m
p
les
f
r
o
m
th
e
4
8
p
atie
n
ts
in
th
e
d
atase
t
o
f
MI
T
-
B
I
H
ar
r
h
y
th
m
ia.
W
ith
t
h
e
h
elp
o
f
p
r
o
p
o
s
ed
m
o
d
el,
th
e
i
m
p
o
r
ta
n
t
co
m
p
r
es
s
io
n
o
u
tco
m
es
w
er
e
g
ai
n
ed
w
i
th
t
h
e
r
ate
o
f
C
R
3
2
.
2
5
an
d
a
n
a
v
er
ag
e
r
ate
o
f
P
R
D
2
.
7
3
%.
T
h
e
p
r
o
p
o
s
ed
m
et
h
o
d
o
lo
g
y
a
ls
o
h
as
th
e
s
tr
u
ct
u
r
e
w
h
ich
c
an
b
e
u
til
ized
s
af
e
l
y
to
tr
a
n
s
f
er
th
e
b
io
m
ed
ical
in
f
o
r
m
atio
n
to
r
e
m
o
te
lo
ca
tio
n
s
s
ec
u
r
el
y
.
RE
F
E
R
E
NC
E
S
[1
]
P
.
P
a
tel
a
n
d
A
.
T
h
a
k
k
a
r,
“
T
h
e
u
p
su
rg
e
o
f
d
e
e
p
lea
rn
in
g
f
o
r
c
o
m
p
u
ter
v
isio
n
a
p
p
li
c
a
ti
o
n
s
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
C
o
mp
u
ter
En
g
in
e
e
rin
g
(
IJ
ECE
)
,
v
o
l.
1
0
,
n
o
.
1
,
p
p
.
5
3
8
-
5
4
8
,
2
0
2
0
.
[2
]
S.
Na
re
jo
,
e
t
a
l.
,
“
EE
G
Ba
se
d
Ey
e
S
tate
Clas
si
f
ic
a
ti
o
n
u
si
n
g
De
e
p
Be
li
e
f
N
e
t
w
o
rk
a
n
d
S
tac
k
e
d
A
u
to
En
c
o
d
e
r
,”
In
ter
n
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
E
lec
trica
l
a
n
d
C
o
mp
u
ter
En
g
in
e
e
rin
g
(
IJ
ECE
)
,
v
o
l.
6
,
n
o
.
6
,
p
p
.
3
1
3
1
-
3
1
4
1
,
2
0
1
6
.
[3
]
P
.
Re
im
a
n
n
a
n
d
A
.
S
c
h
ü
tze
,
“
F
ir
e
d
e
tec
ti
o
n
in
c
o
a
l
m
in
e
s
b
a
se
d
o
n
se
m
ico
n
d
u
c
t
o
r
g
a
s
se
n
so
rs,”
S
e
n
s
or
Rev
iew
,
v
o
l.
3
2
,
n
o
.
1
,
p
p
.
4
7
-
5
8
,
2
0
1
2
.
[4
]
F
.
Ya
s
m
e
e
n
,
e
t
a
l.
,
“
De
te
c
ti
o
n
o
f
Re
a
l
T
i
m
e
QRSCo
m
p
lex
Us
in
g
Wav
e
letT
r
a
n
sf
o
r
m
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
C
o
mp
u
ter
En
g
in
e
e
ri
n
g
(
IJ
ECE
)
,
v
o
l.
8
,
n
o.
5
,
p
p
.
2
8
5
7
-
2
8
6
3
,
2
0
1
8
.
[5
]
S
.
W
o
ld
,
e
t
a
l.
,
“
P
ri
n
c
ip
a
l
c
o
m
p
o
n
e
n
t
a
n
a
ly
sis,”
Ch
e
mo
m
e
trics
a
n
d
In
tell
ig
e
n
t
L
a
b
o
r
a
to
ry
S
y
st
e
ms
,
v
o
l.
2
,
n
o
.
1
-
3
,
p
p
.
3
7
-
5
2
,
1
9
8
7
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J
E
lec
&
C
o
m
p
E
n
g
I
SS
N:
2
0
8
8
-
8708
B
io
-
s
ig
n
a
ls
co
mp
r
ess
io
n
u
s
in
g
a
u
to
-
e
n
co
d
er (
S
u
n
ilku
ma
r
K
.
N
.
)
433
[6
]
F
.
M
o
rc
h
e
n
,
“
T
i
m
e
s
e
ries
fe
a
t
u
re
e
x
trac
ti
o
n
f
o
r
d
a
ta
m
in
in
g
u
sin
g
DWT
a
n
d
DFT
,
”
Tec
h
.
Re
p
.
No
.
3
3
,
De
p
.
M
a
th
m
a
ti
c
s Co
m
p
u
t.
S
c
i.
P
h
il
ip
p
s
-
Un
iv
e
rsity
M
a
rb
u
rg
,
p
p
.
1
-
31
,
2
0
0
3
.
[7
]
R.
Ya
n
,
e
t
a
l
.
,
“
W
a
v
e
lets
f
o
r
f
a
u
lt
d
iag
n
o
sis
o
f
ro
tary
m
a
c
h
in
e
s:
A
re
v
ie
w
w
it
h
a
p
p
li
c
a
ti
o
n
s,”
S
ig
n
a
l
Pro
c
e
ss
in
g
,
v
o
l.
9
6
,
p
p
.
1
-
1
5
,
2
0
1
4
.
[8
]
N.
He
l
w
ig
,
e
t
a
l.
,
“
Co
n
d
it
i
o
n
m
o
n
it
o
rin
g
o
f
a
c
o
m
p
lex
h
y
d
ra
u
li
c
s
y
ste
m
u
sin
g
m
u
lt
iv
a
riate
sta
ti
stics
,
”
in
2
0
1
5
IEE
E
In
ter
n
a
t
io
n
a
l
I
n
stru
me
n
t
a
ti
o
n
a
n
d
M
e
a
su
re
me
n
t
T
e
c
h
n
o
l
o
g
y
C
o
n
f
e
re
n
c
e
(
I2
M
T
C)
Pro
c
e
e
d
in
g
s
,
p
p
.
2
1
0
-
2
1
5
,
2
0
1
5
.
[9
]
S
.
S
.
Ko
h
l
i,
e
t
a
l.
,
“
Hilb
e
rt
T
ra
n
s
fo
rm
Ba
se
d
A
d
a
p
ti
v
e
EC
G
R
-
P
e
a
k
De
te
c
ti
o
n
T
e
c
h
n
iq
u
e
,
”
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
C
o
mp
u
ter
En
g
in
e
e
rin
g
(
IJ
ECE
)
,
v
ol
.
2
,
n
o
.
5
,
p
p
.
6
3
9
-
6
4
3
,
2
0
1
2
.
[1
0
]
M
.
M
o
h
tas
h
a
m
i
a
n
d
M
.
Ef
tek
h
a
r
i,
“
Us
in
g
a
n
o
v
e
l
m
e
rit
f
o
r
f
e
a
tu
re
se
lec
ti
o
n
b
a
se
d
o
n
r
o
u
g
h
se
t
th
e
o
ry
,
”
2
0
1
8
6
t
h
Ira
n
ia
n
J
o
in
t
Co
n
g
re
ss
o
n
Fu
zz
y
a
n
d
In
tell
ig
e
n
t
S
y
ste
ms
(
CFIS
)
,
K
e
r
m
a
n
,
p
p
.
6
8
-
70
,
2
0
1
8
.
[1
1
]
Z.
P
a
w
lak
,
“
Ro
u
g
h
se
ts
,
”
In
t
e
rn
a
t
io
n
a
l
J
o
u
rn
a
l
o
f
P
a
ra
ll
e
l
Pr
o
g
r
a
m
min
g
,
v
o
l
.
1
1
,
n
o
.
5
,
p
p
.
3
4
1
-
3
5
6
,
1
9
8
2
.
[1
2
]
V
.
Bo
l
o
n
-
Ca
n
e
d
o
,
e
t
a
l.
,
“
A
re
v
iew
o
f
m
icro
a
rra
y
d
a
tas
e
ts
a
n
d
a
p
p
li
e
d
f
e
a
tu
re
se
lec
ti
o
n
m
e
th
o
d
s
,
”
In
f
o
rm
a
ti
o
n
S
c
i
e
n
c
e
s
,
v
o
l.
2
8
2
,
p
p
.
1
1
1
-
1
3
5
,
2
0
1
4
.
[1
3
]
Y.
M
a
,
Y.
W
a
n
g
,
J.
Ya
n
g
,
Y.
M
iao
a
n
d
W
.
L
i,
"
Big
He
a
lt
h
A
p
p
li
c
a
ti
o
n
S
y
ste
m
b
a
se
d
o
n
He
a
lt
h
In
tern
e
t
o
f
T
h
in
g
s
a
n
d
Big
Da
ta,"
in
IEE
E
Acc
e
ss
,
v
o
l.
5
,
p
p
.
7
8
8
5
-
7
8
9
7
,
2
0
1
7
.
[1
4
]
L
a
k
sh
m
in
a
ra
y
a
n
a
M
.
a
n
d
M
.
S
a
rv
a
g
y
a
,
“
OFCS
:
Op
ti
m
ize
d
F
ra
m
e
w
o
rk
o
f
Co
m
p
re
ss
iv
e
S
e
n
sin
g
f
o
r
M
e
d
ica
l
Im
a
g
e
s
i
n
Bo
tt
len
e
c
k
Ne
tw
o
rk
C
o
n
d
it
io
n
,
”
I
n
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
El
e
c
trica
l
a
n
d
Co
m
p
u
ter
E
n
g
i
n
e
e
rin
g
(
IJ
ECE
)
,
v
o
l.
8
,
n
o
.
5
,
p
p
.
2
8
2
9
-
2
8
3
8
,
2
0
1
8
.
[1
5
]
S
.
F
a
u
v
e
l
a
n
d
R.
K.
W
a
rd
,
“
An
En
e
rg
y
Eff
i
c
ien
t
Co
m
p
re
ss
e
d
S
e
n
sin
g
F
ra
m
e
w
o
rk
f
o
r
th
e
C
o
m
p
re
ss
io
n
o
f
El
e
c
tro
e
n
c
e
p
h
a
lo
g
ra
m
S
ig
n
a
ls,
”
S
e
n
so
rs
,
v
o
l.
1
4
,
n
o
.
1,
p
p
.
1
4
7
4
-
1
4
9
6
,
2
0
1
4
.
[1
6
]
M
.
M
o
h
sin
a
a
n
d
A
.
M
a
ju
m
d
a
r,
“
Ga
b
o
r
Ba
se
d
A
n
a
l
y
sis
P
rio
r
F
o
rm
u
latio
n
f
o
r
EE
G
S
ig
n
a
l
R
e
c
o
n
stru
c
ti
o
n
,
”
Bi
o
me
d
ica
l
S
i
g
n
a
l
Pr
o
c
e
ss
in
g
a
n
d
Co
n
tro
l
,
v
o
l.
8
,
n
o
.
6
,
p
p
.
9
5
1
-
9
5
5
,
2
0
1
3
.
[1
7
]
F
.
P
a
re
sc
h
i,
e
t
a
l.
,
“
Ha
rd
w
a
re
-
A
l
g
o
rit
h
m
s
Co
-
De
sig
n
a
n
d
Im
p
lem
e
n
tatio
n
o
f
a
n
A
n
a
lo
g
-
to
-
I
n
f
o
rm
a
ti
o
n
Co
n
v
e
rter
f
o
r
Bio
sig
n
a
ls
Ba
se
d
o
n
C
o
m
p
re
ss
e
d
S
e
n
sin
g
,
”
IEE
E
T
r
a
n
s
a
c
ti
o
n
s
o
n
Bi
o
me
d
ica
l
Circ
u
it
s
a
n
d
S
y
ste
ms
,
v
o
l.
1
0
,
no.
1
,
p
p
.
1
4
9
-
1
6
2
,
2
0
1
5
.
[1
8
]
F
.
Ch
e
n
,
e
t
a
l.
,
“
De
sig
n
a
n
d
A
n
a
l
y
si
s
o
f
a
Ha
rd
w
a
re
-
E
ff
icie
n
t
Co
m
p
re
ss
e
d
S
e
n
sin
g
A
rc
h
it
e
c
tu
re
f
o
r
D
a
ta
Co
m
p
re
ss
io
n
in
W
irele
ss
S
e
n
so
rs
,
”
IEE
E
J
o
u
rn
a
l
o
f
S
o
li
d
-
S
t
a
te Ci
rc
u
it
s
,
v
o
l.
47
,
n
o
.
3
,
p
p
.
7
4
4
-
7
5
6
,
2
0
1
2
.
[1
9
]
N.
V
e
rm
a
a
n
d
A
.
Ch
a
n
d
ra
k
a
sa
n
,
“
A
n
u
lt
ra
lo
w
e
n
e
rg
y
1
2
-
b
it
ra
te
-
r
e
so
lu
ti
o
n
sc
a
lab
le
S
A
R
A
DC
f
o
r
w
irele
ss
se
n
so
r
node
,
”
I
EE
E
J
o
u
rn
a
l
o
f
S
o
li
d
-
S
ta
t
e
Circ
u
it
s
,
v
o
l.
4
2
,
n
o
.
6
,
p
p
.
1
1
9
6
-
1
2
0
5
,
2
0
0
7
.
[2
0
]
R.
Ha
rriso
n
a
n
d
C.
C
h
a
rles
,
“
A
lo
w
-
p
o
w
e
r
lo
w
-
n
o
ise
CM
OS
a
m
p
li
f
ier
f
o
r
n
e
u
ra
l
re
c
o
rd
in
g
a
p
p
li
c
a
ti
o
n
s,
”
IEE
E
J
o
u
rn
a
l
o
f
S
o
li
d
-
S
t
a
te
Circ
u
it
s
,
v
o
l.
3
8
,
n
o
.
6
,
p
p
.
9
5
8
-
9
6
5
,
2
0
0
3
.
[2
1
]
Z.
Zh
a
n
g
,
e
t
a
l.
,
“
Co
m
p
re
ss
e
d
se
n
sin
g
f
o
r
e
n
e
rg
y
-
e
ff
icie
n
t
w
irele
ss
tele
m
o
n
it
o
rin
g
o
f
n
o
n
in
v
a
siv
e
f
e
t
a
l
EC
G
v
i
a
b
lo
c
k
sp
a
rse
Ba
y
e
sia
n
lea
rn
in
g
,
”
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Bi
o
me
d
i
c
a
l
En
g
in
e
e
rin
g
,
v
o
l
.
6
0
,
n
o
.
2
,
p
p
.
3
0
0
-
3
0
9
,
2
0
1
3
.
[2
2
]
M
.
H
.
Ka
m
a
l,
e
t
a
l.
,
“
Co
m
p
re
ss
i
v
e
M
u
lt
ich
a
n
n
e
l
C
o
rti
c
a
l
S
ig
n
a
l
Re
c
o
rd
in
g
,
”
2
0
1
3
IEE
E
I
n
ter
n
a
ti
o
n
a
l
C
o
n
fer
e
n
c
e
on
A
c
o
u
stics
,
S
p
e
e
c
h
,
a
n
d
S
ig
n
a
l
P
ro
c
e
ss
in
g
,
2
0
1
3
.
[2
3
]
A
.
M
a
ju
m
d
a
r,
e
t
a
l.
,
“
L
o
w
-
ra
n
k
M
a
tri
x
Re
c
o
v
e
r
y
A
p
p
ro
a
c
h
f
o
r
En
e
rg
y
Eff
i
c
ien
t
EE
G
A
c
q
u
isit
io
n
f
o
r
W
irele
ss
Bo
d
y
A
r
e
a
Ne
t
w
o
rk
,
”
S
e
n
so
rs
,
v
o
l.
1
4
,
n
o
.
9
,
p
p
.
1
5
7
2
9
-
1
5
7
4
8
,
2
0
1
4
.
[2
4
]
Z.
Zh
a
n
g
,
e
t
a
l.
,
“
S
p
a
ti
o
tem
p
o
ra
l
S
p
a
rse
Ba
y
e
sia
n
L
e
a
rn
in
g
w
it
h
A
p
p
li
c
a
ti
o
n
s
t
o
Co
m
p
re
ss
e
d
S
e
n
sin
g
o
f
M
u
lt
ich
a
n
n
e
l
P
h
y
sio
lo
g
ica
l
S
ig
n
a
ls,
”
IEE
E
T
ra
n
s
a
c
ti
o
n
s
o
n
N
e
u
ra
l
S
y
ste
ms
a
n
d
Reh
a
b
il
i
ta
ti
o
n
En
g
i
n
e
e
rin
g
,
v
o
l.
2
2
,
n
o
.
6
,
p
p
.
1
1
8
6
-
1
1
9
7
,
2
0
1
4
.
[2
5
]
Z.
Zh
a
n
g
a
n
d
B.
D.
Ra
o
,
“
Ex
ten
sio
n
o
f
S
BL
A
l
g
o
rit
h
m
s
f
o
r
th
e
Re
c
o
v
e
r
y
o
f
Blo
c
k
S
p
a
rse
S
ig
n
a
ls
with
I
n
tra
-
Blo
c
k
Co
rre
latio
n
,
”
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
S
ig
n
a
l
Pro
c
e
ss
in
g
,
v
o
l.
6
1
,
n
o
.
8
,
p
p
.
2
0
0
9
-
2
0
1
5
,
2
0
1
3
.
[2
6
]
S
.
P
a
n
,
e
t
a
l.
,
“
A
d
v
e
rs
a
riall
y
Re
g
u
lariz
e
d
G
ra
p
h
A
u
to
e
n
c
o
d
e
r
f
o
r
G
ra
p
h
E
m
b
e
d
d
in
g
,
”
Pro
c
e
e
d
in
g
s
o
f
th
e
T
we
n
ty
-
S
e
v
e
n
th
I
n
ter
n
a
t
i
o
n
a
l
J
o
in
t
Co
n
fe
re
n
c
e
o
n
Arti
f
icia
l
I
n
telli
g
e
n
c
e
(
IJ
CAI
-
18)
,
p
p
.
2
6
0
9
-
2
6
1
5
,
2
0
1
8
.
[2
7
]
Z.
Zh
a
n
g
,
T
.
P
.
Ju
n
g
,
S
.
M
a
k
e
ig
,
Z.
P
i
a
n
d
B.
D.
Ra
o
,
"
S
p
a
ti
o
tem
p
o
ra
l
S
p
a
rse
Ba
y
e
sia
n
Lea
rn
in
g
w
i
th
A
p
p
li
c
a
ti
o
n
s
to
Co
m
p
re
ss
e
d
S
e
n
sin
g
o
f
M
u
l
ti
c
h
a
n
n
e
l
P
h
y
sio
lo
g
ica
l
S
ig
n
a
ls
,
"
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Ne
u
r
a
l
S
y
ste
ms
a
n
d
Reh
a
b
il
it
a
ti
o
n
E
n
g
i
n
e
e
rin
g
,
v
o
l
.
2
2
,
n
o
.
6
,
p
p
.
1
1
8
6
-
1
1
9
7
,
2
0
1
4
.
[2
8
]
Jin
k
w
o
n
K
.
,
Ha
n
g
S
.
S
.
,
Kw
a
n
g
s
o
o
S
.
,
M
y
o
u
n
g
h
o
L
.,
“
Ro
b
u
st alg
o
rit
h
m
f
o
r
a
rrh
y
th
m
ia
c
las
si
f
ica
ti
o
n
i
n
ECG
u
sin
g
e
x
tre
m
e
lea
rn
in
g
m
a
c
h
in
e
,”
2
0
1
1
.
[
O
n
li
n
e
]
.
Ava
il
a
b
le: h
tt
p
s://l
i
n
k
.
s
p
rin
g
e
r.c
o
m/
a
rticle
/1
0
.
1
1
8
6
/
1
4
7
5
-
9
2
5
X
-
8
-
3
1
.
[2
9
]
Jin
g
N
.
,
V
icto
ria
G
.
,
S
h
irl
e
y
G
.
,
G
a
e
ll
e
L
.
,
Ch
e
e
T
.
P.
,
“
A
n
o
v
e
l
a
p
p
r
o
a
c
h
to
d
e
tec
t
a
tri
a
l
f
ib
ril
latio
n
e
ff
icie
n
tl
y
a
n
d
a
c
c
u
ra
tel
y
f
ro
m
4
8
h
o
u
rs
o
f
EC
G
d
a
ta
,”
J
o
u
rn
a
l
o
f
El
e
c
tro
n
ics
a
n
d
Co
mm
u
n
ic
a
ti
o
n
En
g
in
e
e
rin
g
R
e
se
a
rc
h
,
v
o
l.
2
,
n
o
.
1
,
p
p
.
01
-
06
,
2
0
1
4
.
[3
0
]
J.
W
.
Ch
o
i,
B.
S
h
im
,
Y.
Din
g
,
B
.
Ra
o
,
a
n
d
D.
I.
Kim
,
“
Co
m
p
re
ss
e
d
se
n
sin
g
f
o
r
w
irele
ss
c
o
m
m
u
n
ica
ti
o
n
s:
Us
e
f
u
l
ti
p
s a
n
d
tri
c
k
s
,
”
IEE
E
Co
mm
u
n
.
S
u
rv
e
y
s T
u
ts.,
v
o
l.
1
9
,
n
o
.
3
,
p
p
.
1
5
2
7
–
1
5
5
0
,
2
0
1
7
.
[3
1
]
X
.
W
a
n
g
,
et
a
l.
,
“
Co
m
m
u
n
i
t
y
p
re
se
rv
in
g
n
e
t
w
o
rk
e
m
b
e
d
d
in
g
,”
i
n
T
h
e
3
1
st
AAAI
Co
n
fer
e
n
c
e
o
n
Arti
fi
c
ia
l
In
telli
g
e
n
c
e
,
p
p.
2
0
3
-
2
0
9
,
2
0
1
7
.
[3
2
]
S
.
A
v
i
y
e
n
te,
“
Co
m
p
re
ss
e
d
S
e
n
sin
g
F
ra
m
e
w
o
rk
f
o
r
EE
G
Co
m
p
re
ss
io
n
,
”
IEE
E
W
o
rk
sh
o
p
o
n
S
t
a
ti
stica
l
S
i
g
n
a
l
Pro
c
e
ss
in
g
,
p
p
.
1
8
1
-
1
8
4
,
2
0
0
7
.
[3
3
]
C
.
W
a
n
g
,
e
t
a
l.
,
“
M
GA
E
:
M
a
rg
in
a
li
z
e
d
g
ra
p
h
a
u
t
o
e
n
c
o
d
e
r
f
o
r
g
ra
p
h
c
lu
ste
rin
g
,”
i
n
CIKM
’1
7
,
p
p.
889
-
8
9
8
,
2
0
1
7
.
[3
4
]
J.
Ch
ian
g
a
n
d
R.
K.
W
a
rd
,
“
En
e
rg
y
-
E
ff
icie
n
t
Da
ta
R
e
d
u
c
ti
o
n
T
e
c
h
n
iq
u
e
s
f
o
r
W
irele
ss
S
e
i
z
u
re
De
te
c
ti
o
n
S
y
ste
m
s,
”
S
e
n
so
rs
,
v
o
l
.
1
4
,
n
o
.
2
,
p
p
.
2
0
3
6
-
2
0
5
1
,
2
0
1
4
.
[3
5
]
W
.
Din
g
,
Y.
L
u
,
F
.
Y
a
n
g
,
W
.
Da
i,
P
.
L
i,
S
.
L
iu
,
a
n
d
J.
S
o
n
g
,
“
S
p
e
c
trall
y
e
ff
icie
n
t
CS
I
a
c
q
u
isit
io
n
f
o
r
p
o
w
e
r
li
n
e
c
o
m
m
u
n
ica
ti
o
n
s:
A
Ba
y
e
sia
n
c
o
m
p
re
ss
i
v
e
se
n
sin
g
p
e
rsp
e
c
ti
v
e
,
”
IEE
E
J
.
S
e
l.
Are
a
s
C
o
mm
u
n
.
,
v
o
l.
3
4
,
n
o
.
7
,
p
p
.
2
0
2
2
-
2
0
3
2
,
j
u
l
.
2
0
1
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.