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.
8
,
No
.
3
,
J
u
n
e
201
8
,
p
p
.
1854
~
1
8
6
2
I
SS
N:
2
0
8
8
-
8708
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ec
e
.
v8
i
3
.
p
p
1
8
5
4
-
186
2
1854
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e
.
co
m/
jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
I
JE
C
E
A Survey
of Ma
c
hine Lea
rning
Te
chniques
for Sel
f
-
tuning
H
a
do
o
p P
er
for
ma
nce
M
d.
Arm
a
n
ur
Ra
h
m
a
n
1
,
J
.
H
o
s
s
en
2
,
Ven
k
a
t
a
s
es
ha
ia
h C
3
,
CK
H
o
4
,
T
a
n K
i
m
G
eo
k
5
,
Aziz
a
Su
lt
a
na
6
,
J
es
m
ee
n M
.
Z
.
H
.
7
,
F
er
do
us
H
o
s
s
a
in
8
1,
2,
3,
5
,7
,8
F
a
c
u
lt
y
o
f
En
g
in
e
e
rin
g
a
n
d
T
e
c
h
n
o
lo
g
y
,
M
u
lt
im
e
d
ia Un
iv
e
rsity
,
M
e
lak
a
,
7
5
4
5
0
,
M
a
lay
sia
4
F
a
c
u
lt
y
o
f
Co
m
p
u
ti
n
g
a
n
d
In
f
o
r
m
a
ti
c
s,
M
u
lt
im
e
d
ia Un
iv
e
rsit
y
,
Cy
b
e
rja
y
a
,
6
3
1
0
0
,
M
a
lay
sia
6
F
a
c
u
lt
y
o
f
Co
m
p
u
ti
n
g
a
n
d
En
g
i
n
e
e
rin
g
,
Dh
a
k
a
In
tern
a
ti
o
n
a
l
Un
iv
e
rsit
y
,
Dh
a
k
a
,
1
2
0
5
,
Ba
n
g
lad
e
sh
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
J
an
2
9
,
2
0
1
8
R
ev
i
s
ed
Ma
r
2
0
,
2
0
1
8
A
cc
ep
ted
Ma
r
3
0
,
2
0
1
8
T
h
e
A
p
a
c
h
e
H
a
d
o
o
p
f
ra
m
e
w
o
r
k
is
a
n
o
p
e
n
so
u
rc
e
im
p
le
m
e
n
tatio
n
o
f
M
a
p
Re
d
u
c
e
f
o
r
p
r
o
c
e
ss
in
g
a
n
d
sto
rin
g
b
ig
d
a
ta.
Ho
w
e
v
e
r,
to
g
e
t
th
e
b
e
st
p
e
rf
o
r
m
a
n
c
e
f
ro
m
th
is
is
a
b
i
g
c
h
a
ll
e
n
g
e
b
e
c
a
u
se
o
f
it
s
lar
g
e
n
u
m
b
e
r
c
o
n
f
ig
u
ra
ti
o
n
p
a
ra
m
e
t
e
rs.
In
th
i
s
p
a
p
e
r,
t
h
e
c
o
n
c
e
p
t
o
f
c
rit
ica
l
issu
e
s
o
f
H
a
d
o
o
p
sy
ste
m
,
b
ig
d
a
ta
a
n
d
m
a
c
h
in
e
lea
rn
in
g
h
a
v
e
b
e
e
n
h
ig
h
li
g
h
ted
a
n
d
a
n
a
n
a
ly
sis
o
f
so
m
e
m
a
c
h
in
e
lea
rn
in
g
tec
h
n
iq
u
e
s
a
p
p
li
e
d
so
f
a
r,
f
o
r
im
p
ro
v
in
g
th
e
H
a
d
o
o
p
p
e
rf
o
rm
a
n
c
e
is
p
re
se
n
ted
.
T
h
e
n
,
a
p
r
o
m
isin
g
m
a
c
h
in
e
lea
rn
in
g
tec
h
n
iq
u
e
u
si
n
g
d
e
e
p
lea
r
n
i
n
g
a
lg
o
rit
h
m
is
p
ro
p
o
se
d
f
o
r
H
a
d
o
o
p
sy
ste
m
p
e
rf
o
r
m
a
n
c
e
i
m
p
ro
v
e
m
e
n
t.
K
ey
w
o
r
d
:
Had
o
o
p
HDFS
Ma
ch
i
n
e
l
ea
r
n
i
n
g
Ma
p
R
ed
u
ce
P
ar
am
eter
Co
p
y
rig
h
t
©
2
0
1
8
In
s
ti
t
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts
re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Md
.
A
r
m
a
n
u
r
R
ah
m
a
n
,
Facu
lt
y
o
f
E
n
g
i
n
ee
r
i
n
g
a
n
d
T
e
ch
n
o
lo
g
y
,
Mu
lti
m
ed
ia
U
n
iv
er
s
it
y
,
Me
lak
a,
7
5
4
5
0
,
Ma
lay
s
ia
.
E
m
ail:
ar
m
an
.
b
d
m
ai
l@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
p
u
r
p
o
s
e
o
f
th
is
p
ap
er
is
th
r
ee
f
o
ld
:
T
o
p
r
o
v
id
es
a
b
r
ief
d
escr
ip
tio
n
of
th
e
co
n
ce
p
t
o
f
m
ac
h
i
n
e
le
ar
n
in
g
,
b
ig
d
ata
an
d
H
ad
o
o
p
s
y
s
te
m
.
T
o
p
r
esen
t
a
s
y
s
te
m
a
tic
an
al
y
s
is
o
f
e
x
is
t
in
g
tec
h
n
iq
u
e
s
in
ter
m
s
o
f
p
er
f
o
r
m
a
n
ce
,
p
ar
a
m
eter
s
,
d
ataset
an
d
s
y
s
te
m
co
n
f
i
g
u
r
atio
n
.
T
o
p
r
o
p
o
s
e
a
p
r
o
m
i
s
in
g
tec
h
n
iq
u
e
u
s
i
n
g
d
e
ep
lear
n
i
n
g
al
g
o
r
ith
m
f
o
r
i
m
p
r
o
v
in
g
t
h
e
H
ad
o
o
p
s
y
s
te
m
p
er
f
o
r
m
an
ce
i
n
p
r
o
ce
s
s
in
g
b
ig
d
ata.
A
r
o
ad
m
ap
o
f
t
h
i
s
p
ap
er
is
g
iv
e
n
i
n
F
ig
u
r
e
1
.
I
n
Sectio
n
1
an
d
Sectio
n
2
,
H
ad
o
o
p
s
y
s
te
m
w
it
h
Ma
p
R
ed
u
ce
(
MR),
Had
o
o
p
Dis
tr
ib
u
ted
Fil
e
S
y
s
te
m
(
H
D
FS
)
an
d
Y
A
R
N
h
a
v
e
b
ee
n
d
is
cu
s
s
ed
.
T
h
en
,
a
d
is
cu
s
s
io
n
ab
o
u
t
b
ig
d
ata
an
d
V’
s
,
clas
s
i
f
icatio
n
o
f
m
ac
h
in
e
lear
n
in
g
an
d
ex
i
s
ti
n
g
m
ac
h
i
n
e
lear
n
i
n
g
alg
o
r
ith
m
s
is
p
r
esen
ted
.
I
n
S
ec
tio
n
2
.
2
,
cr
itical
i
s
s
u
es
in
H
ad
o
o
p
s
y
s
te
m
ar
e
d
is
c
u
s
s
e
d
.
I
n
Sectio
n
3
,
a
p
r
o
m
i
s
in
g
ap
p
licatio
n
o
f
De
ep
L
ea
r
n
i
n
g
al
g
o
r
ith
m
to
i
m
p
r
o
v
e
th
e
H
ad
o
o
p
p
er
f
o
r
m
an
ce
is
d
is
c
u
s
s
ed
.
A
s
u
m
m
ar
y
i
s
p
r
esen
ted
in
Secti
o
n
4.
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
A
S
u
r
ve
y
o
f Ma
ch
in
e
Lea
r
n
in
g
Tech
n
iq
u
es fo
r
S
elf
-
tu
n
i
n
g
H
a
d
o
o
p
P
erfo
r
ma
n
ce
(
Md
.
A
r
ma
n
u
r
R
a
h
m
a
n
)
1855
Fig
u
r
e
1
.
R
o
ad
m
ap
o
f
th
i
s
s
u
r
v
e
y
1
.
1
H
a
do
o
p sy
s
t
e
m
T
h
e
A
p
ac
h
e
H
ad
o
o
p
f
r
a
m
e
wo
r
k
m
ai
n
l
y
co
n
s
is
t
s
o
f
th
r
ee
c
o
m
p
o
n
en
t:
MR,
HDFS
an
d
Y
AR
N.
T
h
e
r
o
le
o
f
MR
is
d
ata
p
r
o
ce
s
s
i
n
g
.
T
h
e
r
o
le
o
f
HDFS
is
to
m
a
n
ag
e
s
to
r
ag
e
w
h
ic
h
i
s
d
o
n
e
b
y
b
r
ea
k
in
g
a
f
ile
i
n
to
m
u
ltip
le
b
lo
ck
s
an
d
co
p
y
i
n
g
ea
ch
o
f
th
e
m
i
n
to
t
h
r
ee
d
i
f
f
er
en
t
s
er
v
er
s
[1
]
-
[
4]
.
MR
i
s
co
n
s
id
er
ed
as
a
p
r
o
g
r
am
m
i
n
g
p
r
o
ce
s
s
th
at
co
m
p
r
i
s
e
s
o
f
J
o
b
T
r
ac
k
er
(
m
an
a
g
e
th
e
tas
k
)
an
d
T
ask
T
r
ac
k
er
(
r
u
n
t
h
e
tas
k
)
.
Fig
u
r
e
2
s
h
o
w
s
t
h
e
h
i
g
h
-
lev
el
ar
ch
itect
u
r
e
o
f
H
ad
o
o
p
.
Ha
d
o
o
p
w
o
r
k
s
w
it
h
t
w
o
n
o
d
es
n
a
m
el
y
m
a
s
ter
n
o
d
e
an
d
s
lav
e
n
o
d
e.
Un
d
er
th
e
m
aster
n
o
d
e
,
th
er
e
ar
e
T
ask
T
r
ac
k
er
an
d
J
o
b
T
r
ac
k
er
in
MR
L
a
y
er
a
n
d
Na
m
eNo
d
e
an
d
Data
No
d
e
in
HDFS
la
y
er
.
Un
d
er
s
la
v
e
n
o
d
e
,
th
er
e
ar
e
T
ask
T
r
ac
k
e
r
in
MR
la
y
er
an
d
Data
No
d
e
in
HDFS
la
y
er
.
T
ask
T
r
ac
k
er
o
f
Ma
s
ter
No
d
e
co
n
tacts
w
it
h
t
h
e
J
o
b
T
r
ac
k
er
an
d
J
o
b
T
r
ac
k
er
co
n
tacts
w
it
h
t
h
e
s
la
v
e
n
o
d
e
T
ask
T
r
ac
k
er
.
Na
m
eNo
d
e
co
n
ta
cts
w
it
h
all
Data
No
d
e.
Fig
u
r
e
2
.
Hig
h
le
v
el
A
r
c
h
itect
u
r
e
o
f
Had
o
o
p
1
.
1
.
1
.
H
a
do
o
p f
ile
s
y
s
t
e
m
(
H
DF
S)
HDFS
is
d
esig
n
ed
f
o
r
r
eliab
l
e
an
d
e
f
f
icie
n
t
s
to
r
ag
e
o
f
v
er
y
lar
g
e
d
ataset
s
[
5
]
.
Had
o
o
p
en
ab
les
s
ca
lin
g
to
a
lar
g
e
n
u
m
b
er
o
f
h
o
s
ts
a
n
d
d
ata
p
ar
titi
o
n
i
n
g
i
n
m
an
y
n
o
d
es
f
o
r
p
er
f
o
r
m
i
n
g
c
o
m
p
u
tatio
n
s
.
Sto
r
ag
e
o
f
f
ile
s
y
s
te
m
m
etad
ata
an
d
ap
p
lic
atio
n
d
ata
ar
e
d
o
n
e
b
y
s
ep
ar
ately
i
n
HDFS
[
3
]
.
HDFS
s
to
r
es
m
e
tad
ata
in
a
d
ed
icate
d
s
er
v
er
ca
lled
Nam
e
n
o
d
e.
A
p
p
licatio
n
d
ata
is
s
to
r
ed
o
n
an
o
th
er
s
er
v
er
ca
lled
Dat
a
No
d
e
[
6
]
,
[
7
]
.
1
.
1
.
2
.
M
a
pReduce
MR
i
s
a
p
r
o
g
r
a
m
m
i
n
g
p
r
o
ce
s
s
w
h
ich
co
n
tai
n
s
t
w
o
f
u
n
ctio
n
s
n
a
m
el
y
m
ap
an
d
r
ed
u
ce
[
6
]
,
[
8
]
.
I
n
ter
m
s
o
f
d
esi
g
n
i
n
g
,
M
R
co
m
p
r
i
s
es
o
f
th
r
ee
m
ain
co
m
p
o
n
en
ts
:
p
r
o
g
r
a
m
m
i
n
g
,
s
to
r
a
g
e
d
esig
n
i
n
g
,
an
d
s
ch
ed
u
lin
g
.
I
n
p
r
o
g
r
a
m
m
i
n
g
s
tep
,
m
ap
f
u
n
c
tio
n
r
ec
eiv
e
s
i
n
p
u
t
as
v
al
u
e
an
d
k
e
y
f
r
o
m
u
s
er
an
d
p
ass
e
s
t
h
e
o
u
tp
u
t
to
r
ed
u
ce
f
u
n
ctio
n
f
o
r
f
u
r
th
er
p
r
o
ce
s
s
i
n
g
a
n
d
g
en
er
atin
g
t
h
e
r
esu
l
t
[
8
]
,
[
3
]
.
T
h
e
r
ed
u
ce
f
u
n
ctio
n
p
r
o
ce
s
s
es
th
e
o
u
tp
u
t
f
r
o
m
t
h
e
m
ap
f
u
n
ctio
n
t
h
r
o
u
g
h
s
h
u
f
f
li
n
g
,
s
o
r
tin
g
a
n
d
m
er
g
i
n
g
o
f
d
a
ta
[
7]
,
[
9
]
.
Fig
u
r
e
3
s
h
o
w
s
M
R
w
o
r
k
p
r
o
ce
s
s
.
M
a
ste
r
No
d
e
T
a
sk
T
r
a
c
k
e
r
Jo
b
T
r
a
c
k
e
r
N
a
me
N
o
d
e
D
a
t
a
N
o
d
e
T
a
sk
T
r
a
c
k
e
r
D
a
t
a
N
o
d
e
T
a
sk
T
r
a
c
k
e
r
S
l
a
v
e
N
o
d
e
S
l
a
v
e
N
o
d
e
Ma
p
R
ed
u
ce
L
a
y
er
HDFS
L
a
y
er
D
a
t
a
N
o
d
e
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.
8
,
No
.
3
,
J
u
n
e
2
0
1
8
:
1
8
5
4
–
186
2
1856
Fig
u
r
e
3
.
Ma
p
R
ed
u
ce
p
r
o
ce
s
s
1
.
1
.
3
.
P
a
ra
m
et
er
Had
o
o
p
c
o
n
tain
s
o
v
er
2
0
0
p
ar
a
m
eter
s
w
h
er
e
th
e
s
etti
n
g
s
o
f
th
ese
p
ar
a
m
e
ter
s
d
eter
m
i
n
e
it
ef
f
ec
tiv
e
p
er
f
o
r
m
a
n
ce
[
1
0
-
15]
.
Am
o
n
g
th
ese,
ab
o
u
t 3
0
p
ar
am
eter
s
ca
n
d
ir
ec
tl
y
i
m
p
ac
t th
e
p
er
f
o
r
m
a
n
ce
.
2.
M
ACH
I
NE
L
E
AR
NIN
G
T
E
CH
NIQU
E
S AN
D
CR
I
T
I
C
AL
I
S
SU
E
S
2
.
1
.
Cla
s
s
if
ica
t
io
n o
f
m
a
chi
ne
lea
rning
a
lg
o
rit
h
m
s
Ma
ch
i
n
e
L
ea
r
n
in
g
(
ML
)
s
i
m
p
ly
r
ef
er
s
to
th
e
in
telli
g
e
n
ce
o
f
a
m
ac
h
in
e
w
h
er
e
th
e
m
a
ch
in
e
ca
n
p
r
o
v
id
e
d
ec
is
io
n
[
1
6
]
,
[
1
7
]
.
I
t
h
as
g
r
ea
tl
y
i
m
p
ac
ted
i
n
f
o
r
m
at
io
n
s
cien
ce
in
th
e
s
ec
to
r
s
o
f
p
r
ed
ictio
n
,
class
i
f
icatio
n
,
i
m
a
g
e
r
ec
o
g
n
i
tio
n
,
co
m
p
u
ter
v
i
s
io
n
,
s
p
ee
c
h
p
r
o
ce
s
s
in
g
,
n
at
u
r
al
lan
g
u
a
g
e
u
n
d
er
s
tan
d
i
n
g
,
n
eu
r
o
s
cie
n
ce
,
h
ea
lt
h
,
a
n
d
I
o
T
(
I
n
ter
n
et
o
f
T
h
in
g
s
)
.
ML
al
g
o
r
ith
m
is
r
eq
u
ir
ed
to
p
r
o
ce
s
s
i
n
f
o
r
m
atio
n
f
r
o
m
t
h
e
v
er
it
y
o
f
d
ata
w
it
h
in
a
li
m
ited
ti
m
e
d
u
r
atio
n
.
I
t
is
ch
alle
n
g
e
d
b
y
th
e
e
m
er
g
en
ce
o
f
b
i
g
d
ata
[
1
8
]
,
[
7
]
.
Ma
ch
in
e
L
ea
r
n
i
n
g
i
s
ca
te
g
o
r
ized
in
to
s
u
p
er
v
is
ed
,
u
n
s
u
p
er
v
i
s
ed
,
s
e
m
i
-
s
u
p
er
v
is
ed
a
n
d
r
ein
f
o
r
ce
m
en
t
lear
n
in
g
[
1
9
]
.
Mo
s
t p
o
p
u
lar
m
ac
h
i
n
e
lear
n
i
n
g
alg
o
r
it
h
m
s
ar
e
s
h
o
w
n
i
n
Fi
g
u
r
e
4.
a.
Su
p
er
v
i
s
ed
L
ea
r
n
i
n
g
:
Su
p
er
v
i
s
ed
lear
n
i
n
g
i
s
s
k
i
lled
b
y
lab
elled
in
s
tan
ce
s
,
li
k
e
an
i
n
p
u
t
as
t
h
e
e
x
p
ec
te
d
r
esu
lt i
s
k
n
o
w
n
.
S
u
p
er
v
i
s
ed
lear
n
in
g
d
eliv
er
s
d
ataset
co
m
p
r
i
s
in
g
o
f
b
o
th
s
tr
u
ct
u
r
e
an
d
lab
els.
b.
Un
s
u
p
er
v
i
s
ed
L
ea
r
n
in
g
:
U
n
s
u
p
er
v
is
ed
le
ar
n
i
n
g
co
n
d
u
cted
d
ata
w
h
er
e
n
o
p
r
ev
io
u
s
lab
els
a
n
d
its
ai
m
i
s
to
d
is
co
v
er
d
ata
an
d
tr
ac
e
s
i
m
ila
r
ities
a
m
o
n
g
th
e
o
b
j
ec
ts
.
T
h
is
is
a
tech
n
iq
u
e
o
f
ex
p
lo
r
i
n
g
l
ab
els
s
in
ce
th
e
d
ata
its
elf
.
U
n
s
u
p
er
v
i
s
ed
lear
n
in
g
f
u
n
c
tio
n
s
w
ell
o
n
t
h
e
tr
an
s
ac
tio
n
al
d
ataset.
c.
Se
m
i
-
s
u
p
er
v
i
s
ed
L
ea
r
n
in
g
:
Se
m
i
-
s
u
p
er
v
i
s
ed
lear
n
in
g
a
n
d
s
u
p
er
v
i
s
ed
lear
n
in
g
ca
n
u
s
e
th
e
s
a
m
e
ap
p
licatio
n
b
u
t
s
e
m
i
-
s
u
p
er
v
is
ed
lear
n
in
g
ca
n
d
o
to
g
eth
er
w
it
h
lab
eled
to
u
n
lab
eled
d
ata
b
ec
au
s
e
o
f
lear
n
in
g
.
d.
R
ein
f
o
r
ce
m
e
n
t
L
ea
r
n
in
g
:
R
ei
n
f
o
r
ce
m
en
t
lear
n
i
n
g
is
f
r
eq
u
e
n
tl
y
u
s
ed
in
n
a
v
ig
a
tio
n
,
g
a
m
i
n
g
a
n
d
r
o
b
o
tics
.
T
h
is
lear
n
i
n
g
m
et
h
o
d
w
h
ic
h
c
o
n
n
ec
t
s
b
y
a
d
y
n
a
m
ic
s
itu
a
tio
n
i
n
w
h
er
e
it
h
a
s
to
p
er
f
o
r
m
a
p
ar
ticu
lar
ai
m
ex
ce
p
t a
tr
ain
er
ex
p
licit
l
y
s
a
y
i
n
g
it
w
h
et
h
er
h
a
s
ap
p
r
o
ac
h
ed
its
ai
m
[
2
0
]
.
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
A
S
u
r
ve
y
o
f Ma
ch
in
e
Lea
r
n
in
g
Tech
n
iq
u
es fo
r
S
elf
-
tu
n
i
n
g
H
a
d
o
o
p
P
erfo
r
ma
n
ce
(
Md
.
A
r
ma
n
u
r
R
a
h
m
a
n
)
1857
Fig
u
r
e
4
.
Ma
ch
in
e
L
ea
r
n
in
g
Alg
o
r
ith
m
s
2
.
2
.
Ana
ly
s
is
o
f
cr
it
ica
l iss
ue
s
in
h
a
do
o
p s
y
s
t
e
m
T
h
e
cr
itical
is
s
u
e
s
in
Had
o
o
p
s
y
s
te
m
f
o
r
b
ig
d
ata
p
r
o
ce
s
s
in
g
ar
e
d
ep
icted
in
Fig
u
r
e
5
.
C
l
u
st
e
r
i
n
g
A
l
g
o
r
i
t
h
ms
k
-
M
e
a
n
s
k
-
M
e
d
i
a
n
s
Ex
p
e
c
t
a
t
i
o
n
M
a
x
i
m
i
sa
t
i
o
n
H
i
e
r
a
r
c
h
i
c
a
l
C
l
u
st
e
r
i
n
g
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
D
e
e
p
B
o
l
t
z
man
n
M
a
c
h
i
n
e
S
t
a
c
k
e
d
A
u
t
o
-
E
n
c
o
d
e
r
s
D
e
e
p
B
e
l
i
e
f
N
e
t
w
o
r
k
s
P
a
r
t
i
a
l
L
e
a
st
S
q
u
a
r
e
s R
e
g
r
e
ssi
o
n
P
r
i
n
c
i
p
a
l
C
o
m
p
o
n
e
n
t
A
n
a
l
y
si
s
S
a
mm
o
n
M
a
p
p
i
n
g
P
r
i
n
c
i
p
a
l
C
o
m
p
o
n
e
n
t
R
e
g
r
e
ssi
o
n
M
i
x
t
u
r
e
D
i
scr
i
mi
n
a
n
t
A
n
a
l
y
si
s
M
u
l
t
i
d
i
me
n
s
i
o
n
a
l
S
c
a
l
i
n
g
P
r
o
j
e
c
t
i
o
n
P
u
r
su
i
t
L
i
n
e
a
r
D
i
scri
mi
n
a
n
t
A
n
a
l
y
si
s
F
l
e
x
i
b
l
e
D
i
scri
m
i
n
a
n
t
A
n
a
l
y
si
s
Q
u
a
d
r
a
t
i
c
D
i
scr
i
mi
n
a
n
t
A
n
a
l
y
si
s
M
a
c
h
i
n
e
L
e
a
r
n
i
n
g
A
l
g
o
r
i
t
h
ms
D
e
e
p
L
e
a
r
n
i
n
g
A
l
g
o
r
i
t
h
ms
En
se
mb
l
e
A
l
g
o
r
i
t
h
ms
A
r
t
i
f
i
c
i
a
l
N
e
u
r
a
l
N
e
t
w
o
r
k
A
l
g
o
r
i
t
h
ms
R
e
g
r
e
ssi
o
n
A
l
g
o
r
i
t
h
ms
I
n
st
a
n
c
e
-
b
a
se
d
A
l
g
o
r
i
t
h
ms
R
e
g
u
l
a
r
i
z
a
t
i
o
n
A
l
g
o
r
i
t
h
ms
D
e
c
i
si
o
n
T
r
e
e
A
l
g
o
r
i
t
h
ms
B
a
y
e
si
a
n
A
l
g
o
r
i
t
h
ms
A
sso
c
i
a
t
i
o
n
R
u
l
e
L
e
a
r
n
i
n
g
A
l
g
o
r
i
t
h
ms
D
i
me
n
si
o
n
a
l
i
t
y
R
e
d
u
c
t
i
o
n
A
l
g
o
r
i
t
h
ms
O
t
h
e
r
A
l
g
o
r
i
t
h
ms
O
r
d
i
n
a
r
y
L
e
a
st
S
q
u
a
r
e
s R
e
g
r
e
ssi
o
n
M
u
l
t
i
v
a
r
i
a
t
e
A
d
a
p
t
i
v
e
R
e
g
r
e
ssi
o
n
S
p
l
i
n
e
s
S
t
e
p
w
i
se
R
e
g
r
e
ssi
o
n
L
i
n
e
a
r
R
e
g
r
e
ssi
o
n
L
o
g
i
st
i
c
R
e
g
r
e
ssi
o
n
L
o
c
a
l
l
y
Esti
m
a
ted
S
c
a
tt
e
rp
lo
t
S
m
o
o
th
i
n
g
R
a
n
d
o
m F
o
r
e
st
S
t
a
c
k
e
d
G
e
n
e
r
a
l
i
z
a
t
i
o
n
(
b
l
e
n
d
i
n
g
)
G
r
a
d
i
e
n
t
B
o
o
st
e
d
R
e
g
r
e
ssi
o
n
T
r
e
e
s
G
r
a
d
i
e
n
t
B
o
o
st
i
n
g
M
a
c
h
i
n
e
s (G
B
M
)
B
o
o
st
i
n
g
B
o
o
t
st
r
a
p
p
e
d
A
g
g
r
e
g
a
t
i
o
n
(
B
a
g
g
i
n
g
)
A
d
a
B
o
o
st
L
e
a
st
-
A
n
g
l
e
R
e
g
r
e
ssi
o
n
(
L
A
R
S
)
R
i
d
g
e
R
e
g
r
e
ssi
o
n
L
e
a
st
A
b
so
l
u
t
e
S
h
r
i
n
k
a
g
e
a
n
d
S
e
l
e
c
t
i
o
n
O
p
e
r
a
t
o
r
El
a
st
i
c
N
e
t
P
e
r
c
e
p
t
r
o
n
B
a
c
k
-
P
r
o
p
a
g
a
t
i
o
n
H
o
p
f
i
e
l
d
N
e
t
w
o
r
k
R
a
d
i
a
l
B
a
s
i
s F
u
n
c
t
i
o
n
N
e
t
w
o
r
k
A
p
r
i
o
r
i
a
l
g
o
r
i
t
h
m
Ec
l
a
t
a
l
g
o
r
i
t
h
m
k
-
N
e
a
r
e
st
N
e
i
g
h
b
o
u
r
(
k
N
N
)
L
e
a
r
n
i
n
g
V
e
c
t
o
r
Q
u
a
n
t
i
z
a
t
i
o
n
S
e
l
f
-
O
r
g
a
n
i
z
i
n
g
M
a
p
(
S
O
M
)
L
o
c
a
l
l
y
W
e
i
g
h
t
e
d
L
e
a
r
n
i
n
g
N
a
t
u
r
a
l
L
a
n
g
u
a
g
e
P
r
o
c
e
ssi
n
g
R
e
i
n
f
o
r
c
e
me
n
t
L
e
a
r
n
i
n
g
C
o
mp
u
t
a
t
i
o
n
a
l
i
n
t
e
l
l
i
g
e
n
c
e
(
e
v
o
l
u
t
i
o
n
a
r
y
a
l
g
o
r
i
t
h
ms,
e
t
c
.
)
G
r
a
p
h
i
c
a
l
M
o
d
e
l
s
C
o
mp
u
t
e
r
V
i
s
i
o
n
(
C
V
)
R
e
c
o
mm
e
n
d
e
r
S
y
st
e
ms
B
a
y
e
si
a
n
N
e
t
w
o
r
k
N
a
i
v
e
B
a
y
e
s
M
u
l
t
i
n
o
m
i
a
l
N
a
i
v
e
B
a
y
e
s
G
a
u
ssi
a
n
N
a
i
v
e
B
a
y
e
s
B
a
y
e
si
a
n
B
e
l
i
e
f
N
e
t
w
o
r
k
A
v
e
r
a
g
e
d
O
n
e
-
D
e
p
e
n
d
e
n
c
e
Es
t
i
m
a
t
o
r
s
M5
C
l
a
ssi
f
i
c
a
t
i
o
n
a
n
d
R
e
g
r
e
ssi
o
n
T
r
e
e
C
4
.
5
a
n
d
C
5
.
0
I
t
e
r
a
t
i
v
e
D
i
c
h
o
t
o
mi
se
r
3
D
e
c
i
si
o
n
S
t
u
mp
C
h
i
-
sq
u
a
r
e
d
A
u
t
o
m
a
t
i
c
I
n
t
e
r
a
c
t
i
o
n
D
e
t
e
c
t
i
o
n
(
C
H
A
I
D
)
C
o
n
d
i
t
i
o
n
a
l
D
e
c
i
si
o
n
T
r
e
e
s
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.
8
,
No
.
3
,
J
u
n
e
2
0
1
8
:
1
8
5
4
–
186
2
1858
Fig
u
r
e
5
.
I
s
s
u
es o
f
Had
o
o
p
s
y
s
te
m
f
o
r
B
ig
d
ata
p
r
o
ce
s
s
in
g
2
.
2
.
1
.
P
er
f
o
r
m
a
nce
A
n
u
m
b
er
o
f
m
o
d
el
s
h
a
v
e
b
ee
n
p
u
b
l
is
h
ed
in
th
e
last
f
e
w
y
ea
r
s
o
n
H
ad
o
o
p
an
d
MR
s
y
s
te
m
to
o
p
tim
ize
th
e
p
er
f
o
r
m
an
ce
b
y
d
if
f
er
e
n
t
t
y
p
e
s
o
f
s
el
f
-
t
u
n
in
g
tech
n
iq
u
es.
C
o
m
p
ar
i
s
o
n
w
i
th
d
ef
a
u
lt
H
ad
o
o
p
s
y
s
te
m
s
is
m
ad
e
to
f
i
n
d
o
u
t
t
h
e
p
ar
a
m
eter
s
in
v
o
lv
ed
i
n
m
a
n
ag
i
n
g
s
y
s
te
m
p
er
f
o
r
m
a
n
ce
s
,
u
n
d
er
s
ta
n
d
t
h
e
w
a
y
th
o
s
e
p
ar
a
m
eter
s
w
o
r
k
an
d
lo
o
k
f
o
r
th
e
li
m
itat
io
n
s
o
f
d
ef
a
u
lt
H
ad
o
o
p
s
y
s
te
m
s
.
I
n
o
r
d
er
to
im
p
r
o
v
e
H
ad
o
o
p
s
y
s
te
m
p
er
f
o
r
m
a
n
ce
f
o
r
p
r
o
ce
s
s
i
n
g
b
ig
d
ata,
a
n
e
f
f
icien
t
al
g
o
r
ith
m
is
cr
u
cial
f
o
r
o
p
ti
m
izi
n
g
t
h
e
p
er
f
o
r
m
a
n
ce
.
Ma
n
y
r
esear
ch
er
s
h
a
v
e
d
e
v
elo
p
ed
a
d
if
f
er
e
n
t a
l
g
o
r
ith
m
f
o
r
i
m
p
r
o
v
in
g
H
ad
o
o
p
s
y
s
te
m
p
er
f
o
r
m
a
n
ce
co
m
p
ar
ed
to
th
e
d
ef
au
lt c
o
n
f
ig
u
r
atio
n
.
Dif
f
er
en
t
t
y
p
e
o
f
ML
al
g
o
r
ith
m
h
a
v
e
b
ee
n
ap
p
lied
in
H
ad
o
o
p
s
y
s
te
m
f
o
r
p
er
f
o
r
m
an
ce
i
m
p
r
o
v
e
m
en
t.
A
l
g
o
r
ith
m
s
ar
e:
a.
R
an
d
o
m
-
Fo
r
est
On
e
o
f
th
e
p
o
p
u
lar
alg
o
r
ith
m
s
o
f
M
L
is
R
a
n
d
o
m
Fo
r
est
.
A
m
o
n
g
m
an
y
o
th
er
ap
p
r
o
a
ch
es
f
o
r
H
ad
o
o
p
p
er
f
o
r
m
a
n
ce
i
m
p
r
o
v
e
m
en
t,
R
an
d
o
m
-
Fo
r
est
A
p
p
r
o
ac
h
to
Au
to
-
T
u
n
i
n
g
Had
o
o
p
’
s
C
o
n
f
i
g
u
r
atio
n
(
R
FHO
C
)
,
is
a
m
o
d
el
to
tu
n
e
th
e
co
n
f
ig
u
r
atio
n
p
ar
a
m
e
ter
s
au
to
m
atica
ll
y
i
n
o
p
ti
m
iz
in
g
t
h
e
p
er
f
o
r
m
a
n
ce
f
o
r
a
s
p
ec
i
f
ic
ap
p
licatio
n
t
h
at
r
u
n
s
o
n
a
p
ar
ticu
lar
clu
s
ter
.
T
h
e
R
FHO
C
e
s
tab
lis
h
es
t
w
o
m
o
d
els
b
ased
o
n
th
e
r
an
d
o
m
-
f
o
r
est
ap
p
r
o
ac
h
th
at
w
o
r
k
w
it
h
t
h
e
m
ap
an
d
r
ed
u
ce
s
tag
es
i
n
a
s
i
m
ilar
l
y
.
Fiv
e
H
ad
o
o
p
p
r
o
g
r
am
s
n
a
m
e
l
y
w
o
r
d
co
u
n
t,
ter
a
s
o
r
t,
s
o
r
t,
A
d
j
lis
t
a
n
d
I
n
v
er
ted
-
I
n
d
ex
ar
e
u
s
ed
in
th
e
ev
al
u
atio
n
o
f
R
FHO
C
.
T
h
e
e
v
al
u
atio
n
s
h
o
w
s
t
h
at
th
e
p
er
f
o
r
m
a
n
ce
h
as
b
ee
n
s
p
ee
d
ed
u
p
b
y
an
a
v
er
ag
e
f
ac
to
r
o
f
2
.
1
1
tim
es
w
h
er
e
t
h
e
m
a
x
i
m
u
m
s
p
ee
d
ca
n
r
u
n
u
p
to
7
.
4
ti
m
es
co
m
p
ar
e
d
to
co
s
t
-
b
ased
o
p
ti
m
izatio
n
(
C
B
O
)
ap
p
r
o
ac
h
[
1
]
.
b.
Su
p
p
o
r
t V
ec
to
r
R
eg
r
ess
io
n
Su
p
p
o
r
t
Vec
to
r
R
eg
r
ess
io
n
(
SVR
)
is
also
o
n
e
o
f
th
e
m
o
s
t
p
o
p
u
lar
alg
o
r
ith
m
s
o
f
ML
.
SVR
m
o
d
el
i
s
co
n
s
id
er
ed
as
o
n
e
o
f
t
h
e
b
es
t
a
m
o
n
g
M
L
ap
p
r
o
ac
h
es
in
ter
m
s
o
f
ac
c
u
r
ac
y
an
d
co
m
p
u
tati
o
n
al
e
f
f
ic
ien
c
y
.
SVR
a
u
to
-
tu
n
i
n
g
m
ec
h
an
is
m
h
as
i
n
te
g
r
ated
m
ac
h
i
n
e
-
lear
n
i
n
g
p
er
f
o
r
m
an
ce
m
o
d
el
an
d
i
n
tellig
e
n
t
s
ea
r
ch
alg
o
r
ith
m
f
o
r
an
e
f
f
ec
ti
v
e
e
x
p
lo
r
atio
n
o
f
p
ar
a
m
eter
s
p
ac
e
a
n
d
ef
f
icie
n
t
tr
ai
n
i
n
g
m
o
d
els.
T
h
e
SVR
m
o
d
el
p
er
f
o
r
m
a
n
ce
w
as
m
ea
s
u
r
ed
i
n
t
w
o
p
r
o
g
r
a
m
s
:
s
o
r
t
a
n
d
w
o
r
d
co
u
n
t
a
n
d
co
m
p
ar
ed
Star
f
is
h
m
o
d
el.
I
t
w
a
s
s
h
o
w
n
th
a
t
th
e
SVR
m
o
d
el
p
er
f
o
r
m
an
ce
i
n
cr
e
m
en
t
w
as
3
9
%
w
h
i
le
Star
f
is
h
m
o
d
el
p
er
f
o
r
m
an
ce
in
c
r
e
m
en
t
w
a
s
1
3
%
w
h
e
n
it
is
an
al
y
z
ed
in
s
o
r
t
p
r
o
g
r
am
s
.
B
u
t
in
w
o
r
d
co
u
n
t
p
r
o
g
r
a
m
,
th
e
Star
f
is
h
m
o
d
el
p
er
f
o
r
m
a
n
ce
i
n
cr
e
m
e
n
t
w
a
s
ei
th
er
s
i
m
ilar
at
4
0
% o
r
s
lig
h
tl
y
b
etter
w
it
h
a
r
ate
o
f
5
%
[
2
1
]
.
c.
Su
p
p
o
r
t V
ec
to
r
Ma
ch
in
e
I
n
an
o
t
h
er
m
o
d
el
k
n
o
w
n
a
s
A
u
to
m
ated
R
e
s
o
u
r
ce
A
ll
o
ca
tio
n
an
d
C
o
n
f
ig
u
r
atio
n
o
f
Ma
p
R
ed
u
c
e
E
n
v
ir
o
n
m
e
n
t
i
n
th
e
C
lo
u
d
(
AR
OM
A
)
t
h
e
allo
ca
tio
n
o
f
r
eso
u
r
ce
s
an
d
co
n
f
i
g
u
r
atio
n
o
f
p
ar
a
m
eter
s
ar
e
au
to
m
ated
t
h
r
o
u
g
h
t
w
o
-
s
ta
g
e
o
p
tim
iza
tio
n
f
r
a
m
e
w
o
r
k
a
n
d
ML
.
T
h
is
m
o
d
el
f
o
cu
s
ed
o
n
th
e
w
a
y
to
r
ed
u
cin
g
th
e
co
s
t
o
f
b
i
g
d
at
a
p
r
o
ce
s
s
in
g
in
H
ad
o
o
p
s
y
s
te
m
.
T
h
e
r
esu
lt
d
ep
icts
t
h
at
t
h
e
co
s
t
o
f
p
r
o
ce
s
s
i
n
g
1
0
GB
d
ata
w
it
h
A
R
OM
A
au
t
o
-
tu
n
i
n
g
i
s
3
6
ce
n
ts
u
s
i
n
g
5
m
ed
iu
m
VM
s
(
Vir
tu
a
l
Ma
ch
i
n
e
)
w
h
er
e
t
h
e
o
n
e
w
it
h
o
u
t
AR
OM
A
a
u
to
-
t
u
n
in
g
is
5
1
ce
n
t
s
u
s
i
n
g
6
m
ed
i
u
m
VM
s
.
I
t
also
s
h
o
w
s
th
at
r
es
o
u
r
ce
allo
ca
tio
n
u
n
d
er
A
R
O
M
A
m
ec
h
a
n
is
m
ca
n
co
s
t
less
co
m
p
ar
ed
to
th
e
d
ef
au
lt
o
n
e.
On
a
v
er
ag
e
AR
OM
A’
s
co
s
t
ef
f
icien
c
y
i
s
2
5
%
[
2
2
]
.
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
A
S
u
r
ve
y
o
f Ma
ch
in
e
Lea
r
n
in
g
Tech
n
iq
u
es fo
r
S
elf
-
tu
n
i
n
g
H
a
d
o
o
p
P
erfo
r
ma
n
ce
(
Md
.
A
r
ma
n
u
r
R
a
h
m
a
n
)
1859
d.
K
-
m
ea
n
s
++
Un
li
k
e
t
h
e
AR
OM
A
m
ec
h
an
i
s
m
,
m
a
n
y
o
th
er
m
ec
h
a
n
i
s
m
s
j
u
s
t
f
o
cu
s
o
n
i
m
p
r
o
v
i
n
g
t
h
e
p
er
f
o
r
m
an
ce
s
b
y
r
ed
u
cin
g
t
h
e
t
i
m
e
o
f
d
ata
p
r
o
ce
s
s
i
n
g
.
P
r
o
f
ilin
g
a
n
d
P
er
f
o
r
m
an
ce
An
al
y
s
i
s
-
b
ased
S
y
s
te
m
(
P
P
A
B
S)
u
s
e
s
t
w
o
-
p
h
a
s
e
(
A
n
a
l
y
ze
r
a
n
d
R
ec
o
g
n
izer
)
f
r
a
m
e
w
o
r
k
t
h
at
o
p
er
ates
o
n
K
-
m
ea
n
s
++
clu
s
te
r
i
n
g
f
o
r
an
al
y
z
i
n
g
an
d
class
i
f
icatio
n
ap
p
r
o
ac
h
f
o
r
r
ec
o
g
n
izi
n
g
t
h
e
j
o
b
s
.
T
h
e
ex
p
er
i
m
e
n
tal
r
esu
lts
s
h
o
w
t
h
a
t
th
e
p
r
o
ce
s
s
in
g
ti
m
e
f
o
r
B
ig
Da
ta
h
a
s
b
ee
n
r
e
d
u
ce
d
in
T
er
aSo
r
t
an
d
W
o
r
d
C
o
u
n
t
m
et
h
o
d
s
.
I
n
a
n
ex
p
er
i
m
en
t
w
it
h
1
0
GB
in
p
u
t
d
ata
s
et,
t
h
e
u
s
u
al
ac
c
o
m
p
li
s
h
m
e
n
t
ti
m
e
f
o
r
T
er
aS
o
r
t
an
d
W
o
r
d
C
o
u
n
t
d
ec
r
ea
s
e
s
b
y
3
8
.
4
%
an
d
1
8
.
7
% r
esp
ec
tiv
el
y
.
T
h
e
r
ea
s
o
n
f
o
r
th
i
s
m
is
m
atc
h
in
p
er
f
o
r
m
an
ce
is
c
h
ar
ac
ter
is
tics
o
f
v
ar
io
u
s
j
o
b
s
[
2
3
]
.
e.
T
r
ee
-
B
ased
R
eg
r
ess
io
n
T
h
is
ap
p
r
o
ac
h
h
as
t
w
o
p
h
ase
s
f
ir
s
t
o
n
e
is
p
r
ed
ictio
n
p
h
ase
an
d
th
e
s
e
c
o
n
d
o
n
e
i
s
o
p
ti
m
iz
atio
n
p
h
ase.
I
n
th
e
p
r
ed
ictio
n
p
h
ase,
th
e
p
er
f
o
r
m
a
n
ce
o
f
Ma
p
R
ed
u
ce
j
o
b
is
esti
m
ated
an
d
in
t
h
e
o
p
ti
m
i
za
tio
n
p
h
ase
,
a
s
ea
r
ch
f
o
r
o
n
an
av
er
ag
e
o
p
tim
u
m
co
n
f
i
g
u
r
atio
n
p
ar
a
m
ete
r
is
m
ad
e
by
in
v
o
k
i
n
g
th
e
p
r
ed
icto
r
r
ep
ea
te
d
ly
.
I
t
is
r
ep
o
r
ted
th
at
th
i
s
ap
p
r
o
a
ch
ca
n
h
elp
t
h
e
u
s
er
to
in
cr
e
m
en
t
th
e
p
er
f
o
r
m
a
n
ce
r
eg
ar
d
in
g
2
to
8
ti
m
e
s
b
etter
th
an
p
r
ed
ictio
n
p
h
ase
[
2
4
]
.
2
.
2
.
2
.
P
a
ra
m
e
t
er
s
A
b
o
v
e
al
g
o
r
ith
m
s
h
a
v
e
b
ee
n
u
s
ed
p
ar
am
eter
s
i
n
T
ab
le
1
.
P
ar
am
eter
s
ar
e
th
e
m
ai
n
f
ac
t
o
r
s
th
at
p
la
y
an
i
m
p
o
r
tan
t
r
o
le
in
H
ad
o
o
p
s
y
s
te
m
f
o
r
p
er
f
o
r
m
an
ce
i
m
p
r
o
v
e
m
en
t.
T
h
e
li
m
itatio
n
o
f
d
ef
a
u
lt
H
ad
o
o
p
s
y
s
te
m
is
th
at
th
e
p
ar
a
m
eter
s
ar
e
f
ix
ed
at
d
ef
a
u
lt
v
al
u
e
s
.
Am
o
n
g
th
e
3
0
ef
f
ec
ti
v
e
p
ar
am
eter
s
d
if
f
er
e
n
t
m
o
d
els
u
s
ed
d
if
f
er
e
n
t
p
ar
a
m
eter
co
n
f
ig
u
r
at
io
n
s
.
T
h
e
T
a
b
le
1
.
s
h
o
w
s
s
o
m
e
m
o
s
t e
f
f
ec
tiv
e
p
a
r
a
m
eter
s
,
w
h
ich
w
er
e
u
s
ed
in
o
p
tim
izin
g
H
ad
o
o
p
p
er
f
o
r
m
a
n
ce
[
2
5
]
-
[
2
8
]
.
2
.
2
.
3
.
Da
t
a
s
et
A
p
p
lied
ab
o
v
e
al
g
o
r
it
h
m
s
h
a
v
e
b
ee
n
u
s
ed
t
h
ese
d
ataset
s
.
I
n
d
e
v
elo
p
in
g
a
n
e
f
f
ec
tiv
e
m
o
d
el
f
o
r
i
m
p
r
o
v
i
n
g
H
ad
o
o
p
p
er
f
o
r
m
a
n
ce
,
m
o
s
t
o
f
t
h
e
M
L
alg
o
r
it
h
m
h
av
e
u
s
ed
m
u
l
tip
le
d
atasets
i
n
o
r
d
er
t
o
m
a
k
e
s
u
r
e
th
at
t
h
e
p
er
f
o
r
m
an
ce
d
o
es
n
o
t
d
ec
r
ea
s
e
ag
ain
s
t
th
e
d
ata
s
i
ze
.
R
an
d
o
m
Fo
r
est
m
o
d
el
h
a
s
u
s
ed
a
d
ataset
o
f
2
-
5
GB
f
o
r
tr
ain
in
g
ti
m
e
a
n
d
5
0
GB
to
1
T
B
w
h
ile
ca
r
r
y
i
n
g
o
u
t
t
h
e
ev
a
lu
at
io
n
.
T
h
e
b
en
ch
m
ar
k
h
a
s
b
ee
n
co
llected
f
r
o
m
P
u
m
a
an
d
HiB
e
n
ch
s
u
ite
s
[
1
]
.
SVR
b
ased
a
u
to
-
t
u
n
i
n
g
ap
p
r
o
ac
h
h
as
u
s
ed
a
d
ataset
o
f
8
0
GB
an
d
2
4
0
GB
in
SNB
cl
u
s
ter
an
d
ap
p
lied
th
e
s
o
r
t
a
n
d
wo
r
d
co
u
n
t
f
r
o
m
H
iB
en
ch
b
en
ch
m
ar
k
s
u
ite
[
2
1
]
.
AR
OM
A
m
o
d
el
h
a
s
u
s
ed
a
d
ataset
o
f
5
GB
,
1
0
GB
an
d
2
0
GB
an
d
u
s
ed
R
a
n
d
o
m
T
ex
tW
r
iter
an
d
R
an
d
o
m
W
r
iter
to
o
ls
i
n
H
ad
o
o
p
p
ac
k
ag
e
to
p
r
o
d
u
ce
d
ata
o
f
d
if
f
er
en
t
s
izes
b
ec
au
s
e
o
f
th
e
So
r
t,
Gr
ep
an
d
W
o
r
d
C
o
u
n
t
p
r
o
g
r
a
m
s
[
2
2
]
.
T
h
e
P
P
A
B
S
m
o
d
el
u
s
ed
1
GB
,
5
GB
an
d
1
0
GB
d
atasets
.
A
t
r
ain
in
g
s
et
o
f
M
R
ap
p
licatio
n
s
co
n
s
i
s
ti
n
g
o
f
3
w
ell
-
k
n
o
w
n
s
ets
n
a
m
el
y
:
Had
o
o
p
E
x
a
m
p
les
s
et,
HiB
en
ch
a
n
d
Had
o
o
p
B
en
ch
m
ar
k
s
s
et
(
I
n
tel
i
m
p
le
m
en
ted
b
en
ch
m
ar
k
s
et)
[
2
3
]
.
T
ab
le
1
.
C
o
n
f
ig
u
r
atio
n
P
ar
a
m
eter
s
in
Had
o
o
p
s
y
s
te
m
P
a
r
a
me
t
e
r
P
a
p
e
r
/
R
e
f
e
r
e
n
c
e
i
o
.
so
r
t
.
f
a
c
t
o
r
[
1
]
,
[
2
1
-
2
4
]
map
r
e
d
.
j
o
b
.
sh
u
f
f
l
e
.
me
r
g
e
.
p
e
r
c
e
n
t
[
1
]
,
[
2
2
]
,
[
2
4
]
map
r
e
d
.
o
u
t
p
u
t
.
c
o
mp
r
e
ss
[
1
]
,
[
2
4
]
map
r
e
d
.
i
n
me
m.me
r
g
e
.
t
h
r
e
sh
o
l
d
[
1
]
,
[
2
4
]
map
r
e
d
.
r
e
d
u
c
e
.
t
a
s
k
s
[
1
]
,
[
2
1
]
,
[
2
2
]
i
o
.
so
r
t
.
s
p
i
l
l
.
p
e
r
c
e
n
t
[
1
]
,
[
2
2
-
2
4
]
map
r
e
d
.
j
o
b
.
sh
u
f
f
l
e
.
i
n
p
u
t
.
b
u
f
f
e
r
.
p
e
r
c
e
n
t
[
1
]
,
[
2
2
]
,
[
2
4
]
i
o
.
so
r
t
.
r
e
c
o
r
d
.
p
e
r
c
e
n
t
[
1
]
,
[
2
2
]
,
[
2
4
]
i
o
.
so
r
t
.
m
b
[
1
]
,
[
2
2
-
2
4
]
map
r
e
d
.
c
o
mp
r
e
ss.ma
p
.
o
u
t
p
u
t
[
1
]
map
r
e
d
.
t
a
s
k
t
r
a
c
k
e
r
.
ma
p
.
t
a
sk
s.ma
x
i
m
u
m
[
2
1
]
,
[
2
3
]
map
r
e
d
.
t
a
s
k
t
r
a
c
k
e
r
.
r
e
d
u
c
e
.
t
a
s
k
s.m
a
x
i
mu
m
[
2
1
]
,
[
2
3
]
,
[
2
4
]
map
r
e
d
.
c
h
i
l
d
.
j
a
v
a
.
o
p
t
s
[
2
3
]
map
r
e
d
.
r
e
d
u
c
e
.
p
a
r
a
l
l
e
l
.
c
o
p
i
e
s
[
2
2
]
,
[
2
3
]
d
f
s.
b
l
o
c
k
.
s
i
z
e
[
2
3
]
map
r
e
d
.
ma
p
.
o
u
t
p
u
t
.
c
o
m
p
r
e
ss
[
2
3
]
,
[
2
4
]
map
r
e
d
.
j
o
b
.
r
e
d
u
c
e
.
i
n
p
u
t
.
b
u
f
f
e
r
.
p
e
r
c
e
n
t
[
2
2
]
,
[
2
4
]
M
a
p
H
e
a
p
S
i
z
e
[
2
4
]
M
a
p
T
a
sk
s
M
a
x
[
2
4
]
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.
8
,
No
.
3
,
J
u
n
e
2
0
1
8
:
1
8
5
4
–
186
2
1860
P
a
r
a
me
t
e
r
P
a
p
e
r
/
R
e
f
e
r
e
n
c
e
S
p
l
i
t
S
i
z
e
[
2
4
]
H
t
t
p
T
h
r
e
a
d
[
2
4
]
R
e
d
u
c
e
H
e
a
p
S
i
z
e
[
2
4
]
R
e
d
u
c
e
T
a
sk
sN
u
m
[
2
4
]
R
e
d
u
c
e
C
o
p
y
N
u
m
[
2
4
]
R
e
d
u
c
e
S
l
o
w
st
a
r
t
[
2
4
]
JV
M
R
e
u
se
[
2
4
]
2
.
2
.
4
.
Sy
s
t
e
m
co
nfig
ura
t
io
n
T
h
e
s
y
s
te
m
co
n
f
ig
u
r
atio
n
h
a
s
b
ee
n
u
s
ed
i
n
ab
o
v
e
ML
a
lg
o
r
ith
m
s
f
o
r
i
m
p
r
o
v
i
n
g
.
T
h
e
d
if
f
er
en
t
m
o
d
el
s
h
a
v
e
co
m
p
ar
ed
t
h
e
p
er
f
o
r
m
an
ce
w
it
h
d
if
f
er
en
t
s
y
s
te
m
.
Fo
r
ex
a
m
p
le,
R
an
d
o
m
-
Fo
r
est
m
o
d
el
h
a
s
co
m
p
ar
ed
its
o
u
tco
m
e
ag
ai
n
s
t
C
B
O
b
ased
ap
p
r
o
ac
h
an
d
th
e
co
n
f
i
g
u
r
atio
n
h
a
s
b
ee
n
d
o
n
e
in
a
s
i
m
ilar
m
a
n
n
er
also
.
I
t
h
as
u
s
ed
1
0
Su
g
a
n
s
er
v
er
s
p
r
ep
ar
ed
w
it
h
I
n
tel
-
Xeo
n
C
P
U
-
E5
-
2
4
0
7
2
.
2
0
GHz
an
d
q
u
ad
-
co
r
e
p
r
o
ce
s
s
o
r
an
d
3
2
GB
P
C
3
m
e
m
o
r
y
co
n
n
ec
ted
th
r
o
u
g
h
g
ig
a
b
it
E
th
er
n
et
[
1
]
.
SVR
m
o
d
el
h
as
u
s
ed
HiB
en
ch
b
en
ch
m
ar
k
f
o
r
W
o
r
d
C
o
u
n
t
a
n
d
So
r
t
b
e
n
ch
m
ar
k
w
h
ile
it
h
as
u
s
ed
t
w
o
clu
s
ter
s
n
a
m
el
y
San
d
y
B
r
id
g
e
(
SNB
)
an
d
Z
T
clu
s
ter
.
S
VR
p
er
f
o
r
m
ed
th
e
ex
p
er
i
m
e
n
t
o
n
a
s
er
v
er
w
h
ic
h
is
a
d
u
a
l
-
co
r
e
I
n
tel
R
C
o
r
eT
M
i5
-
2
5
4
0
M
p
r
o
ce
s
s
o
r
r
u
n
n
i
n
g
at
2
.
6
0
GHz
an
d
4
GB
m
ai
n
m
e
m
o
r
y
[
2
1
]
.
On
th
e
o
th
er
,
h
a
n
d
SVM
m
o
d
el
w
a
s
i
m
p
le
m
en
ted
o
n
s
etti
n
g
s
o
f
7
HP
P
r
o
-
L
ia
n
t
B
L
4
6
0
C
G6
b
l
ad
e
s
er
v
er
to
g
eth
e
r
w
i
th
a
H
P
E
VA
s
to
r
ag
e
ar
ea
n
et
w
o
r
k
th
at
co
m
p
r
is
ed
o
f
1
0
Gb
p
s
E
th
er
n
e
t
an
d
8
Gb
p
s
Fi
b
r
e/iSC
SI
d
u
a
l
ch
a
n
n
els.
T
h
e
s
m
all
a
n
d
m
ed
iu
m
VM
(
Vir
tu
al
Ma
ch
i
n
e)
u
s
ed
w
er
e
co
n
tai
n
ed
w
ith
1
v
C
P
U,
2
GB
R
A
M
an
d
5
0
GB
h
ar
d
d
is
k
s
p
ac
e
a
n
d
2
v
C
P
Us,
4
GB
R
A
M
a
n
d
8
0
GB
h
ar
d
d
is
k
s
p
ac
e
r
esp
ec
tiv
ely
[
2
2
]
.
I
n
ad
d
itio
n
,
K
-
m
ea
n
s
+
+
m
o
d
el
h
a
s
u
s
ed
a
clu
s
ter
t
h
at
co
n
tain
s
f
i
v
e
Dat
aNo
d
es
an
d
o
n
e
Nam
eNo
d
e.
T
h
e
Na
m
eNo
d
e
an
d
Data
No
d
e
r
u
n
o
n
C
P
U
o
f
2
E
C
2
C
o
m
p
u
te,
a
n
d
1
E
C
2
C
o
m
p
u
te
Un
i
t,
t
h
e
m
e
m
o
r
y
o
f
3
0
0
GB
an
d
2
0
0
GB
r
esp
ec
tiv
el
y
[
2
3
]
.
B
e
s
id
es,
T
r
ee
-
B
ased
R
eg
r
ess
io
n
m
ac
h
i
n
e
lear
n
in
g
alg
o
r
it
h
m
h
as
ev
al
u
ated
it
s
p
er
f
o
r
m
a
n
ce
o
n
8
n
o
d
es P
d
ef
au
lt
s
ett
in
g
w
h
er
e
ea
ch
n
o
d
e
co
n
tain
s
ei
g
h
t I
n
tel
i7
-
4
7
7
0
co
r
es,
3
2
GB
R
A
M
a
n
d
2
T
B
d
is
k
s
p
ac
e
[
2
4
]
.
3.
P
RO
SPEC
T
O
F
AP
P
L
I
C
AT
I
O
N
O
F
DE
E
P
L
E
A
RNIN
G
A
L
G
O
RI
T
H
M
T
O
H
ADO
O
P
P
E
RF
O
RM
ANCE IM
P
RO
VE
M
E
NT
Had
o
o
p
is
an
in
te
g
r
al
p
ar
t
o
f
p
r
o
ce
s
s
in
g
b
i
g
d
ata.
Had
o
o
p
p
er
f
o
r
m
a
n
ce
i
s
an
i
m
p
ed
i
m
e
n
t
in
g
etti
n
g
ef
f
icien
t
s
er
v
ice
as
t
h
e
p
ar
a
m
eter
s
ar
e
n
o
t
s
el
f
-
t
u
n
ed
.
Di
f
f
e
r
en
t
M
L
al
g
o
r
ith
m
h
as
b
ee
n
p
r
o
p
o
s
ed
to
im
p
r
o
v
e
th
e
p
er
f
o
r
m
an
ce
s
b
y
allo
w
i
n
g
au
to
-
t
u
n
i
n
g
o
f
th
e
m
o
s
t e
f
f
ec
ti
v
e
p
ar
a
m
eter
s
.
T
h
e
an
a
l
y
s
is
o
f
p
er
f
o
r
m
an
ce
w
it
h
d
if
f
er
e
n
t
M
L
alg
o
r
it
h
m
s
s
h
o
ws
th
a
t
t
h
e
s
el
f
-
t
u
n
i
n
g
h
a
s
i
m
p
r
o
v
ed
H
ad
o
o
p
s
y
s
te
m
p
er
f
o
r
m
an
ce
in
co
m
p
ar
i
s
o
n
w
it
h
d
ef
a
u
lt
p
ar
a
m
eter
co
n
f
i
g
u
r
atio
n
.
Ho
w
e
v
er
,
t
h
er
e
i
s
a
n
ee
d
f
o
r
a
n
e
w
m
o
d
el
to
f
u
r
th
e
r
i
m
p
r
o
v
e
H
ad
o
o
p
s
y
s
te
m
p
er
f
o
r
m
an
ce
w
i
th
r
esp
ec
t
to
s
p
ee
d
an
d
ac
cu
r
ac
y
.
De
ep
lear
n
in
g
h
as
b
ee
n
ad
o
p
ted
w
it
h
m
o
s
t
p
o
p
u
lar
T
h
ea
n
o
[
2
9
]
,
T
en
s
o
r
f
lo
w
[
3
0
]
,
C
af
ee
[
3
1
]
lib
r
ar
y
in
m
a
n
y
s
ec
to
r
s
o
f
b
ig
d
ata
p
r
o
ce
s
s
in
g
an
d
it
w
a
s
f
o
u
n
d
to
re
s
u
lt
in
i
m
p
r
o
v
ed
p
er
f
o
r
m
an
c
e.
Dee
p
L
ea
r
n
in
g
a
lg
o
r
it
h
m
s
a
r
e
u
s
ed
i
n
p
r
o
ce
s
s
i
n
g
b
i
g
d
ata
in
m
a
n
y
g
ia
n
t
tec
h
co
m
p
a
n
ies
i
n
cl
u
d
in
g
Go
o
g
l
e,
Face
b
o
o
k
,
Am
az
o
n
an
d
s
o
o
n
.
T
h
e
au
th
o
r
s
f
ee
l
th
er
e
i
s
a
s
co
p
e
f
o
r
ap
p
ly
i
n
g
d
ee
p
lear
n
in
g
al
g
o
r
ith
m
s
f
o
r
s
elf
-
tu
n
i
n
g
an
d
i
m
p
r
o
v
in
g
H
ad
o
o
p
s
y
s
te
m
p
er
f
o
r
m
an
ce
[
7
]
.
4.
CO
NCLU
SI
O
N
I
n
th
is
p
ap
er
a
b
r
ief
r
e
v
ie
w
o
f
th
e
co
n
ce
p
ts
o
f
b
ig
d
ata,
H
a
d
o
o
p
s
y
s
te
m
is
p
r
ese
n
ted
,
s
el
f
-
t
u
n
i
n
g
o
f
H
ad
o
o
p
p
ar
am
eter
s
a
n
d
M
L
a
lg
o
r
ith
m
s
.
Sel
f
-
t
u
n
i
n
g
o
f
H
ad
o
o
p
s
y
s
te
m
p
ar
a
m
eter
s
u
s
in
g
ML
alg
o
r
it
h
m
s
h
a
s
b
ee
n
f
o
u
n
d
to
i
m
p
r
o
v
e
p
er
f
o
r
m
a
n
ce
co
m
p
ar
ed
to
d
ef
a
u
l
t
p
ar
a
m
eter
co
n
f
i
g
u
r
atio
n
.
T
h
e
p
r
o
s
p
ec
t
o
f
t
h
e
ap
p
licatio
n
o
f
d
ee
p
lear
n
in
g
alg
o
r
ith
m
s
f
o
r
s
elf
-
t
u
n
in
g
in
H
ad
o
o
p
s
y
s
te
m
to
i
m
p
r
o
v
e
s
p
ee
d
an
d
ac
cu
r
ac
y
o
f
p
er
f
o
r
m
a
n
ce
is
p
r
o
p
o
s
ed
b
y
t
h
e
au
th
o
r
s
.
RE
F
E
R
E
NC
E
S
[1
]
Be
i
Z,
Yu
Z,
Zh
a
n
g
H,
X
i
o
n
g
W
,
X
u
C,
Eec
k
h
o
u
t
L
a
n
d
F
e
n
g
S
,
2
0
1
6
,
"
RF
HO
C:
A
Ra
n
d
o
m
-
F
o
re
st
A
p
p
ro
a
c
h
to
A
u
to
-
T
u
n
in
g
Ha
d
o
o
p
’s C
o
n
f
ig
u
ra
ti
o
n
",
IEE
E
T
ra
n
s.
P
a
ra
ll
e
l
Dist
rib
.
S
y
st.
,
27
,
p
p
.
1
4
7
0
-
14
83
.
[2
]
T
e
c
h
n
o
lo
g
y
I
2
0
1
6
S
u
rv
e
y
o
n
P
e
r
f
o
r
m
a
n
c
e
o
f
H
a
d
o
o
p
M
a
p
re
d
u
c
e
Op
ti
m
iza
ti
o
n
M
e
th
o
d
s
,
2
,
p
p
.
1
1
4
-
21
.
[3
]
Uz
u
n
k
a
y
a
C,
En
sa
ri
T
a
n
d
Ka
v
u
ru
c
u
Y
,
2
0
1
5
,
"
Ha
d
o
o
p
Eco
sy
ste
m
a
n
d
Its
A
n
a
l
y
sis
o
n
T
w
e
e
ts
",
Pro
c
e
d
ia
-
S
o
c
.
Beh
a
v
.
S
c
i.
,
1
9
5
,
p
p
.
1
8
9
0
-
1
8
9
7
.
[4
]
Bo
n
if
a
c
io
A
S
,
M
e
n
o
ll
i
A
a
n
d
S
il
v
a
F
,
2
0
1
4
,
"
Ha
d
o
o
p
M
a
p
Re
d
u
c
e
Co
n
f
ig
u
ra
ti
o
n
P
a
ra
m
e
te
rs
a
n
d
S
y
ste
m
P
e
rf
o
rm
a
n
c
e
:
a
S
y
ste
m
a
ti
c
Re
v
ie
w
"
,
2
0
1
4
In
t.
C
o
n
f
.
Pa
r
a
ll
e
l
Distr
ib
.
Pr
o
c
e
ss
.
T
e
c
h
.
A
p
p
l.
,
p
p
.
1
-
7
.
[5
]
Do
n
g
y
u
F
e
n
g
,
L
ig
u
Zh
u
a
n
d
L
e
i
Zh
a
n
g
,
2
0
1
6
,
"
Re
v
ie
w
o
f
h
a
d
o
o
p
p
e
rf
o
rm
a
n
c
e
o
p
ti
m
iza
ti
o
n
"
,
2
0
1
6
2
n
d
IE
EE
I
n
t.
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
A
S
u
r
ve
y
o
f Ma
ch
in
e
Lea
r
n
in
g
Tech
n
iq
u
es fo
r
S
elf
-
tu
n
i
n
g
H
a
d
o
o
p
P
erfo
r
ma
n
ce
(
Md
.
A
r
ma
n
u
r
R
a
h
m
a
n
)
1861
Co
n
f.
Co
mp
u
t.
Co
mm
u
n
.
,
p
p
.
65
-
6
8
.
[6
]
Re
h
a
n
M
a
n
d
G
a
n
g
o
d
k
a
r
D
,
2015
,
"
Ha
d
o
o
p
,
M
a
p
Re
d
u
c
e
a
n
d
HD
F
S
:
A
De
v
e
lo
p
e
rs
P
e
rsp
e
c
ti
v
e
",
Pro
c
e
d
ia
-
Pro
c
e
d
ia
C
o
mp
u
t.
S
c
i.
,
48
,
p
p
.
45
-
50
.
[7
]
Ou
ss
o
u
s
A
,
Be
n
jello
u
n
F
,
A
it
A
a
n
d
Be
lfk
ih
S
,
2
0
1
7
,
"
Big
Da
ta
tec
h
n
o
lo
g
ies
:
A
su
rv
e
y
",
J
.
Kin
g
S
a
u
d
Un
iv.
-
Co
mp
u
t
.
In
f.
S
c
i.
[8
]
Jia
n
g
D,
Oo
i
B
C,
S
h
i
L
a
n
d
W
u
S
,
2
0
1
0
,
"
T
h
e
p
e
rf
o
rm
a
n
c
e
o
f
M
a
p
Re
d
u
c
e
"
,
Pro
c
.
V
L
DB
E
n
d
o
w.
,
3
,
p
p
.
4
7
2
-
4
83
.
[9
]
Zh
a
n
g
B,
Křik
a
v
a
F
,
Ro
u
v
o
y
R
a
n
d
S
e
in
t
u
rier
L
,
2015
,
"
S
e
lf
-
c
o
n
fig
u
ra
ti
o
n
o
f
th
e
n
u
m
b
e
r
o
f
c
o
n
c
u
rre
n
tl
y
ru
n
n
in
g
M
a
p
Re
d
u
c
e
j
o
b
s i
n
a
h
a
d
o
o
p
c
l
u
s
ter
",
Pro
c
.
-
IEE
E
In
t.
Co
n
f.
Au
t
o
n
.
Co
m
p
u
t
,
.
ICAC
2
0
1
5
,
p
p
.
1
4
9
-
1
50
.
[1
0
]
L
e
e
G
J
a
n
d
F
o
rtes
J A
B
,
2
0
1
6
,
"
Ha
d
o
o
p
p
e
rf
o
rm
a
n
c
e
se
l
f
-
tu
n
in
g
u
sin
g
a
f
u
z
z
y
-
p
re
d
ictio
n
a
p
p
ro
a
c
h
",
Pro
c
.
-
2
0
1
6
IEE
E
In
t.
Co
n
f.
Au
t
o
n
.
Co
m
p
u
t
.
I
CAC
2
0
1
6
,
p
p
.
55
-
64
.
[1
1
]
G
e
n
k
in
M
,
De
h
n
e
F
,
P
o
sp
e
lo
v
a
M
,
Ch
e
n
Y
a
n
d
Na
v
a
rro
P
,
2
0
1
6
,
"
A
u
to
m
a
ti
c
,
o
n
-
li
n
e
tu
n
i
n
g
o
f
YA
RN
c
o
n
tain
e
r
m
e
m
o
r
y
a
n
d
CP
U
p
a
ra
m
e
ters
".
[1
2
]
L
i
C,
Zh
u
a
n
g
H,
L
u
K,
S
u
n
M
,
Z
h
o
u
J,
Da
i
D an
d
Z
h
o
u
X
,
2
0
1
4
,
"
A
n
a
d
a
p
ti
v
e
a
u
t
o
-
c
o
n
f
ig
u
ra
ti
o
n
to
o
l
f
o
r
h
a
d
o
o
p
"
,
Pro
c
.
IEE
E
In
t.
Co
n
f.
En
g
.
Co
m
p
lex
C
o
mp
u
t
.
S
y
st.
IC
ECCS
,
p
p
.
69
-
72
.
[1
3
]
Ke
rk
C
W
a
n
d
Zah
id
M
S
M
,
2
0
1
5
, "
A
u
to
-
T
u
n
e
d
Ha
d
o
o
p
M
a
p
Re
d
u
c
e
f
o
r
ECG
A
n
a
l
y
sis
",
p
p
.
3
2
9
-
3
34
.
[1
4
]
P
o
sp
e
lo
v
a
M
a
n
d
A
ff
a
irs
P
,
2
0
1
5
,
"
Re
a
l
T
i
m
e
A
u
to
tu
n
i
n
g
f
o
r
M
a
p
Re
d
u
c
e
o
n
Ha
d
o
o
p
/
YA
R
N
b
y
Re
a
l
T
i
m
e
A
u
to
tu
n
i
n
g
f
o
r
M
a
p
Re
d
u
c
e
o
n
H
a
d
o
o
p
/
YA
R
N
.
[1
5
]
W
il
lk
e
T
L
,
2016
,
"
G
u
n
th
e
r :
S
e
a
rc
h
-
Ba
se
d
A
u
to
-
T
u
n
in
g
o
f
M
a
p
Re
d
u
c
e
G
u
n
th
e
r :
S
e
a
rc
h
-
Ba
se
d
A
u
to
-
tu
n
i
n
g
o
f
M
a
p
Re
d
u
c
e
".
[1
6
]
Jo
rd
a
n
M
I
a
n
d
M
it
c
h
e
ll
T
M
,
2
0
1
5
,
"
M
a
c
h
in
e
lea
rn
i
n
g
:
T
re
n
d
s,
p
e
rsp
e
c
ti
v
e
s,
a
n
d
p
r
o
sp
e
c
ts
"
,
S
c
ien
c
e
(
8
0
-
.
)
,
3
4
9
,
pp.
2
5
5
-
2
60
.
[1
7
]
Qiu
J,
W
u
Q,
Din
g
G
,
X
u
Y
a
n
d
F
e
n
g
S
,
2
0
1
6
,
"
A
su
rv
e
y
o
f
m
a
c
h
in
e
lea
rn
in
g
f
o
r
b
ig
d
a
ta
p
r
o
c
e
ss
in
g
",
EURA
S
IP
J
.
Ad
v
.
S
ig
n
a
l
Pro
c
e
ss
.
,
2
0
1
6
,
67
.
[1
8
]
T
sa
i
C
-
W
,
L
a
i
C
-
F
,
Ch
a
o
H
-
C
a
n
d
V
a
silak
o
s A
V
,
2
0
1
5
,
"
Big
d
a
ta an
a
ly
ti
c
s: a su
rv
e
y
",
J
.
Bi
g
Da
ta
2
,
21
.
[1
9
]
S
o
lan
k
i
K an
d
Dh
a
n
k
a
r
A
,
2017
,
"
A
re
v
ie
w
o
n
M
a
c
h
in
e
L
e
a
rn
in
g
T
e
c
h
n
iq
u
e
s
",
8
,
p
p
.
7
7
8
-
7
82
.
[2
0
]
S
in
g
h
Y,
B
h
a
ti
a
P
K
a
n
d
S
a
n
g
w
a
n
O
,
2
0
0
7
,
"
A
re
v
ie
w
o
f
stu
d
ies
o
n
m
a
c
h
in
e
lea
rn
in
g
T
e
c
h
n
iq
u
e
s
",
In
t.
J
.
Co
m
p
u
t
.
S
c
i.
S
e
c
u
r.
,
1
,
p
p
.
3
9
5
-
39
9
[2
1
]
Yig
it
b
a
si
N,
W
il
lk
e
T
L
,
L
ia
o
G
a
n
d
Ep
e
m
a
D
,
2013
,
"
T
o
w
a
rd
s
m
a
c
h
in
e
lea
rn
in
g
-
b
a
se
d
a
u
to
-
t
u
n
i
n
g
o
f
M
a
p
Re
d
u
c
e
",
Pro
c
.
-
IEE
E
C
o
mp
u
t.
S
o
c
.
A
n
n
u
.
In
t.
S
y
mp
.
M
o
d
e
l.
A
n
a
l
.
S
im
u
l.
Co
m
p
u
t
.
T
e
l
e
c
o
mm
u
n
.
S
y
st.
M
AS
COTS
,
p
p
.
11
-
20
.
[2
2
]
L
a
m
a
P
a
n
d
Zh
o
u
X
,
2
0
1
2
,
"
A
ROM
A
:
A
u
to
m
a
ted
Re
so
u
rc
e
A
ll
o
c
a
ti
o
n
a
n
d
Co
n
f
ig
u
ra
ti
o
n
o
f
M
a
p
Re
d
u
c
e
En
v
iro
n
m
e
n
t
in
th
e
Cl
o
u
d
",
Pro
c
.
9
th
In
t
.
Co
n
f.
Au
t
o
n
.
Co
m
p
u
t.
-
I
CAC
’1
2
,
63
.
[2
3
]
W
u
D
,
2
0
1
3
,
"
A
P
ro
f
il
in
g
a
n
d
P
e
rf
o
rm
a
n
c
e
A
n
a
l
y
sis
Ba
s
e
d
S
e
lf
-
tu
n
in
g
S
y
ste
m
f
o
r
Op
ti
m
iza
t
io
n
o
f
Ha
d
o
o
p
M
a
p
Re
d
u
c
e
Clu
ste
r
C
o
n
f
ig
u
ra
ti
o
n
".
[2
4
]
Ch
e
n
C
O,
Z
h
u
o
Y
Q,
Ye
h
C
C,
L
in
C
M
a
n
d
L
iao
S
W
,
2
0
1
5
,
"
M
a
c
h
in
e
L
e
a
rn
in
g
-
Ba
se
d
Co
n
f
ig
u
ra
ti
o
n
P
a
ra
m
e
ter
T
u
n
in
g
o
n
Ha
d
o
o
p
S
y
ste
m
",
Pro
c
.
-
2
0
1
5
IEE
E
In
t.
Co
n
g
r
.
Bi
g
Da
t
a
,
Bi
g
Da
t
a
C
o
n
g
r.
,
2
0
1
5
,
p
p
.
3
8
6
-
3
92
.
[2
5
]
He
g
e
r
D
,
2013
,
"
Ha
d
o
o
p
P
e
rf
o
rm
a
n
c
e
T
u
n
in
g
-
A
P
ra
g
m
a
ti
c
&
Itera
ti
v
e
A
p
p
ro
a
c
h
"
,
C.
J
.
,
p
p
.
1
-
16
.
[2
6
]
De
v
ice
s A
M
,
2
0
1
2
,
"
Ha
d
o
o
p
P
e
r
f
o
r
m
a
n
c
e
T
u
n
in
g
G
u
id
e
"
,
p
p
.
1
-
22
.
[2
7
]
S
h
rin
iv
a
s
J
,
2
0
1
1
,
"
In
tro
d
u
c
ti
o
n
Ha
d
o
o
p
C
o
n
f
ig
u
ra
ti
o
n
T
u
n
in
g
Be
st
P
ra
c
ti
c
e
s
JV
M
Co
n
f
ig
u
ra
ti
o
n
T
u
n
i
n
g
OS
Co
n
f
ig
u
ra
ti
o
n
T
u
n
in
g
Co
n
c
lu
si
o
n
&
F
u
t
u
re
Dire
c
ti
o
n
",
T
e
c
h
n
o
lo
g
y
,
9
,
p
p
.
2
0
1
1
-
2
0
1
1
.
[2
8
]
M
in
-
Z
h
e
n
g
J
,
2
0
1
5
,
"
Re
se
a
rc
h
o
n
t
h
e
p
e
rf
o
rm
a
n
c
e
o
p
ti
m
iza
ti
o
n
o
f
h
a
d
o
o
p
i
n
b
ig
d
a
ta
e
n
v
iro
n
m
e
n
t
",
In
t.
J
.
Da
ta
b
a
se
T
h
e
o
ry
Ap
p
l.
,
8
,
p
p
.
2
9
3
-
3
0
4
.
[2
9
]
T
h
e
T
h
e
a
n
o
De
v
e
lo
p
m
e
n
t
T
e
a
m
,
"
T
h
e
a
n
o
:
A
P
y
th
o
n
f
ra
m
e
w
o
rk
f
o
r
f
a
st
c
o
m
p
u
tatio
n
o
f
m
a
th
e
m
a
ti
c
a
l
e
x
p
re
ss
io
n
s.
Av
a
il
a
b
le
",
a
rX
iv
:1
6
0
5
.
0
2
6
8
8
.
[3
0
]
A
b
a
d
i
M
,
A
g
a
r
w
a
l
A
,
Ba
rh
a
m
P
,
Bre
v
d
o
E
,
Ch
e
n
Z
,
Cit
r
o
C,
e
t
a
l.
,
"
T
e
n
so
rF
lo
w
:
L
a
rg
e
-
S
c
a
le
M
a
c
h
in
e
L
e
a
rn
in
g
o
n
He
tero
g
e
n
e
o
u
s Distrib
u
ted
S
y
ste
m
s,"
Co
RR,
2
0
1
6
.
[3
1
]
Jia
Y,
S
h
e
lh
a
m
e
r
E,
Do
n
a
h
u
e
J
,
Ka
ra
y
e
v
S
,
L
o
n
g
J,
G
irsh
ick
R,
G
u
a
d
a
rra
m
a
S
a
n
d
Da
rre
ll
T
,
2
0
1
4
,
"
Ca
ff
e
:
Co
n
v
o
l
u
ti
o
n
a
l
A
rc
h
it
e
c
tu
re
f
o
r
F
a
st F
e
a
tu
re
Em
b
e
d
d
in
g
".
B
I
O
G
RAP
H
I
E
S
O
F
AUTH
O
RS
M
d
.
Ar
m
a
n
u
r
R
a
h
m
a
n
re
c
e
i
v
e
d
th
e
B.
S
c
.
d
e
g
re
e
in
c
o
m
p
u
te
r
sc
ien
c
e
a
n
d
e
n
g
in
e
e
rin
g
f
ro
m
A
sia
n
Un
iv
e
rsit
y
o
f
B
a
n
g
lad
e
sh
(
A
UB)
in
2
0
1
0
.
He
is
c
u
rre
n
tl
y
w
o
rk
in
g
to
w
a
rd
th
e
M
E
n
g
S
c
d
e
g
re
e
a
t
th
e
M
u
lt
ime
d
ia
Un
iv
e
rs
it
y
(M
M
U),
M
a
la
y
sia
.
His
r
e
se
a
r
c
h
in
tere
st
in
c
lu
d
e
p
e
rf
o
rm
a
n
c
e
o
p
ti
m
iza
ti
o
n
o
f
b
ig
d
a
ta sy
ste
m
,
d
a
ta
m
in
in
g
,
m
a
c
h
in
e
lea
rn
in
g
a
n
d
im
a
g
e
p
ro
c
e
ss
in
g
.
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.
8
,
No
.
3
,
J
u
n
e
2
0
1
8
:
1
8
5
4
–
186
2
1862
Dr
.
J
a
k
ir
H
o
s
se
n
is
g
ra
d
u
a
ted
in
M
e
c
h
a
n
ica
l
En
g
i
n
e
e
rin
g
fro
m
th
e
Dh
a
k
a
Un
iv
e
rsit
y
o
f
En
g
in
e
e
rin
g
a
n
d
T
e
c
h
n
o
lo
g
y
(1
9
9
7
),
M
a
ste
rs
in
C
o
m
m
u
n
ica
ti
o
n
a
n
d
Ne
tw
o
rk
En
g
in
e
e
rin
g
f
ro
m
Un
iv
e
rsiti
P
u
tra
M
a
la
y
sia
(2
0
0
3
)
a
n
d
P
h
D
in
S
m
a
rt
T
e
c
h
n
o
l
o
g
y
a
n
d
Ro
b
o
ti
c
En
g
in
e
e
rin
g
f
ro
m
Un
iv
e
rsiti
P
u
tra
M
a
lay
si
a
(2
0
1
2
)
.
He
is
c
u
rre
n
tl
y
a
S
e
n
i
o
r
L
e
c
tu
re
r
a
t
th
e
F
a
c
u
lt
y
o
f
En
g
in
e
e
rin
g
a
n
d
T
e
c
h
n
o
lo
g
y
,
M
u
lt
im
e
d
ia
U
n
iv
e
rsity
,
M
a
la
y
si
a
.
His
re
se
a
rc
h
in
tere
sts
a
re
in
th
e
a
re
a
o
f
A
rti
f
icia
l
In
telli
g
e
n
c
e
(F
u
z
z
y
L
o
g
ic,
Ne
u
ra
l
Ne
t
w
o
rk
),
In
f
e
re
n
c
e
S
y
st
e
m
s,
P
a
tt
e
rn
Clas
sif
ica
ti
o
n
,
M
o
b
i
le Ro
b
o
t
Na
v
ig
a
ti
o
n
a
n
d
I
n
t
e
ll
ig
e
n
t
C
o
n
tr
o
l.
Dr
.
Ch
i
n
th
a
k
u
n
ta
V
e
n
k
a
ta
S
e
s
h
a
i
a
h
r
e
c
e
i
v
e
d
h
is
Ba
c
h
e
lo
r
o
f
En
g
in
e
e
rin
g
(B.
E.
)
De
g
re
e
in
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
f
ro
m
S
.
V
.
Un
iv
e
rsit
y
,
A
n
d
h
ra
P
ra
d
e
sh
,
I
n
d
i
a
,
in
th
e
y
e
a
r
1
9
6
4
.
He
re
c
e
iv
e
d
M
a
ste
r
o
f
En
g
in
e
e
rin
g
(M
.
E)
d
e
g
re
e
in
Hig
h
V
o
lt
a
g
e
En
g
in
e
e
rin
g
f
ro
m
In
d
ian
In
stit
u
te o
f
S
c
ien
c
e
,
Ba
n
g
a
lo
re
in
1
9
6
6
.
He
re
c
e
iv
e
d
h
is
P
h
.
D.
d
e
g
re
e
in
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
(in
th
e
a
re
a
o
f
P
o
w
e
r
S
y
st
e
m
s)
in
1
9
7
6
f
ro
m
I.
I.
T
.
M
a
d
ra
s.
L
a
ter,
h
e
w
o
rk
e
d
in
th
e
sa
m
e
in
stit
u
te
ti
ll
2
0
0
5
.
He
w
a
s
a
p
p
o
i
n
ted
a
s
P
ro
f
e
ss
o
r
o
f
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
in
Ja
n
.
1
9
9
3
.
In
2
0
0
6
,
h
e
jo
in
e
d
th
e
F
a
c
u
lt
y
o
f
En
g
in
e
e
rin
g
a
n
d
T
e
c
h
n
o
l
o
g
y
,
M
u
lt
im
e
d
ia
Un
iv
e
rsit
y
(M
e
la
k
a
)
M
a
la
y
sia
a
n
d
is
w
it
h
th
e
m
p
re
se
n
tl
y
a
s
A
ss
o
c
iate
P
ro
f
e
ss
o
r.
His
re
se
a
r
c
h
in
tere
sts
a
re
in
th
e
a
re
a
s
o
f
El
e
c
tri
c
a
l
P
o
w
e
r
S
y
st
e
m
s,
Hig
h
V
o
lt
a
g
e
En
g
i
n
e
e
rin
g
a
n
d
In
stru
m
e
n
tatio
n
,
P
o
w
e
r
El
e
c
tro
n
ics
a
n
d
it
s
a
p
p
li
c
a
ti
o
n
t
o
g
re
e
n
tec
h
n
o
lo
g
y
so
lu
ti
o
n
s,E
lec
ti
c
P
o
w
e
r
q
u
a
li
ty
i
m
p
ro
v
e
m
e
n
t
a
n
d
El
e
c
tri
c
a
l
e
n
e
rg
y
c
o
n
se
rv
a
ti
o
n
,
P
o
w
e
r
e
ff
icie
n
t
d
e
v
ice
s an
d
Big
d
a
ta an
a
l
y
ti
c
s
.
D
r
.
H
o
C
h
in
K
u
a
n
o
b
tain
e
d
th
e
B.
S
c
.
(Ho
n
s)
in
C
o
m
p
u
ter
S
c
ien
c
e
w
it
h
El
e
c
tro
n
ics
E
n
g
in
e
e
rin
g
f
ro
m
Un
iv
e
rsit
y
Co
ll
e
g
e
L
o
n
d
o
n
,
UK
.
S
u
b
se
q
u
e
n
tl
y
,
h
e
c
o
m
p
let
e
d
h
is
M
.
S
c
.
(IT
)
a
n
d
P
h
.
D.
i
n
In
f
o
rm
a
ti
o
n
T
e
c
h
n
o
l
o
g
y
f
ro
m
M
u
lt
im
e
d
ia
Un
iv
e
rsit
y
,
M
a
la
y
sia
.
A
t
p
re
se
n
t,
h
e
is
a
P
ro
f
e
ss
o
r
a
n
d
De
a
n
a
t
th
e
F
a
c
u
lt
y
o
f
Co
m
p
u
ti
n
g
a
n
d
In
f
o
rm
a
ti
c
s,
M
u
lt
im
e
d
ia
Un
iv
e
rsit
y
,
M
a
la
y
sia
.
His
m
a
in
re
se
a
rc
h
in
tere
sts a
re
Na
tu
ra
l
Co
m
p
u
ti
n
g
,
Co
m
b
in
a
to
rial
O
p
ti
m
iza
ti
o
n
a
n
d
Da
ta M
in
i
n
g
.
Ta
n
K
i
m
G
e
o
k
re
c
e
iv
e
d
th
e
B.
E.
,
M
.
E
.
,
a
n
d
P
h
D.
d
e
g
re
e
s
a
ll
i
n
e
lec
-
tri
c
a
l
e
n
g
in
e
e
rin
g
f
ro
m
Un
iv
e
rsit
y
o
f
T
e
c
h
n
o
lo
g
y
M
a
la
y
sia
,
in
1
9
9
5
,
1
9
9
7
,
a
n
d
2
0
0
0
,
re
sp
e
c
ti
v
e
l
y
.
He
h
a
s
b
e
e
n
S
e
n
i
o
r
R&
D
e
n
g
in
e
e
r
in
E
P
COS
S
in
g
a
p
o
re
i
n
2
0
0
0
.
I
n
2
0
0
1
–
2
0
0
3
,
h
e
jo
in
e
d
Do
C
o
M
o
Eu
r
o
-
L
a
b
s
in
M
u
n
ic
h
,
G
e
rm
a
n
y
.
He
is
c
u
rre
n
tl
y
a
c
a
d
e
m
ic
sta
ff
in
M
u
lt
im
e
d
ia
Un
iv
e
rsit
y
.
His
re
se
a
rc
h
in
tere
sts
in
c
lu
d
e
ra
d
io
p
ro
p
a
g
a
ti
o
n
f
o
r
o
u
t
d
o
o
r
a
n
d
in
d
o
o
r
,
RF
ID,
m
u
lt
i
-
u
se
r
d
e
tec
ti
o
n
tec
h
n
iq
u
e
f
o
r
m
u
lt
i
-
c
a
rrier t
e
c
h
-
n
o
l
o
g
ies
,
a
n
d
A
-
G
P
S
.
Az
iza
S
u
lta
n
a
r
e
c
e
i
v
e
d
th
e
B.
S
c
.
d
e
g
re
e
in
c
o
m
p
u
ter
sc
ie
n
c
e
a
n
d
e
n
g
in
e
e
rin
g
f
ro
m
Dh
a
k
a
In
tern
a
ti
o
n
a
l
Un
iv
e
rsit
y
(DIU
)
in
2
0
1
6
.
Sh
e
is
c
u
rre
n
t
ly
se
a
rc
h
in
g
a
n
o
o
p
e
rt
u
n
it
y
to
c
o
n
ti
n
u
e
h
e
r
h
ig
h
e
r
stu
d
y
.
He
r
re
se
a
r
c
h
in
tere
st
in
c
lu
d
e
p
e
rf
o
rm
a
n
c
e
o
p
ti
m
iz
a
ti
o
n
o
f
b
ig
d
a
ta
s
y
ste
m
,
d
a
ta
m
in
in
g
,
m
a
c
h
in
e
lea
rn
in
g
a
n
d
ima
g
e
p
ro
c
e
ss
in
g
.
J
e
s
m
e
e
n
M
.
Z
.
H
.
c
u
rre
n
tl
y
a
p
o
stg
ra
d
u
a
te
st
u
d
e
n
t
i
n
E
n
g
in
e
e
rin
g
sp
e
c
ializin
g
in
A
rti
f
ic
a
l
In
telleg
e
n
c
e
f
ro
m
M
u
lt
im
e
d
ia
Un
iv
e
rsit
y
(M
M
U).
S
h
e
c
o
m
p
lete
d
a
b
a
c
h
e
lo
r’s d
e
g
re
e
in
Co
m
p
u
ter
S
c
ien
c
e
a
n
d
E
n
g
in
e
e
rin
g
f
ro
m
In
tern
a
ti
o
n
a
l
Isla
m
ic Un
i
v
e
rsit
y
Ch
it
tag
o
n
g
,
Ba
n
g
lad
e
sh
.
Fer
d
o
u
s
H
o
ss
a
i
n
r
e
c
e
iv
e
d
B.
S
c
.
d
e
g
re
e
in
c
o
m
p
u
ter
sc
ien
c
e
a
n
d
e
n
g
in
e
e
rin
g
in
2
0
1
2
a
n
d
M
.
S
c
.
d
e
g
re
e
in
In
f
o
rm
a
ti
o
n
a
n
d
C
o
m
m
u
n
ica
ti
o
n
T
e
c
h
n
o
l
o
g
y
in
2
0
1
5
f
ro
m
M
a
w
l
a
n
a
Bh
a
sh
a
n
i
S
c
ien
c
e
a
n
d
T
e
c
h
n
o
lo
g
y
Un
iv
e
rsit
y
,
Ba
n
g
lad
e
s
h
.
Cu
rre
n
tl
y
,
h
e
is
p
u
rsu
in
g
h
i
s
P
h
.
D.
d
e
g
re
e
in
F
a
c
u
lt
y
o
f
En
g
in
e
e
rin
g
a
n
d
T
e
c
h
n
o
lo
g
y
a
t
M
u
lt
im
e
d
ia Un
iv
e
rsity
,
M
a
la
y
sia
.
His
re
se
a
rc
h
a
re
a
in
c
lu
d
e
s
ra
d
io
f
re
q
u
e
n
c
y
id
e
n
ti
f
ica
ti
o
n
(RF
ID)
s
y
ste
m
,
ra
d
io
p
ro
p
a
g
a
ti
o
n
f
o
r
o
u
t
d
o
o
r
a
n
d
in
d
o
o
r,
im
a
g
e
p
ro
c
e
ss
in
g
,
a
n
d
b
ig
d
a
ta
a
n
a
ly
sis.
He
is
a
lso
a
n
a
ss
o
c
iate
m
e
m
b
e
r
o
f
Ba
n
g
lad
e
sh
Co
m
p
u
ter S
o
c
iety
,
Ba
n
g
lad
e
sh
.
Evaluation Warning : The document was created with Spire.PDF for Python.