Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
9
, No
.
6
,
Decem
ber
201
9
, p
p.
5235
~
5243
IS
S
N:
20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v
9
i
6
.
pp5235
-
52
43
5235
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Self
-
or
ganizin
g
map (SO
M) for s
pecies di
stributio
n modell
ing of
birds speci
es at
kenyir
lan
dscape
Sa
lw
ana
M
oham
ad
Asm
ara
1
,
Gertr
ude
D
av
id
2
,
M
ohd
Tajuddin
Abd
ull
ah
3
,
Wan Isni
Sofiah
Wan
Din
4
, Da
nakor
n
Nin
carean
a/l Eh
Pho
n
5
, Ahm
ad Fi
rd
au
s Z
ai
na
l
Ab
idi
n
6
1,
4,5,6
Facul
t
y
of C
om
pute
r
S
y
st
e
m
s a
nd
Software
Engi
n
ee
ring
,
U
nive
rsiti Mal
a
y
s
i
a
Pahang
,
Ma
lays
ia
2
,3
Instit
ute of Tr
opic
a
l
Biod
ive
rs
ity
and
Sus
ta
in
a
ble
Dev
el
opm
en
t,
Univ
ersit
i
Mal
a
y
si
a Te
r
engga
n
u,
Mal
a
y
s
ia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Ja
n 1
0
, 2
01
9
Re
vised
Ma
y
2
9
, 2
01
9
Accepte
d
J
un
27
, 201
9
Ide
nti
f
y
i
ng
whic
h
biodi
ver
sit
y
spec
i
es
are
m
ore
dom
ina
nt
tha
n
ot
her
s
in
an
y
are
a
is
a
ver
y
cha
llenging
t
a
sk.
Thi
s
is
be
ca
use
of
th
e
a
bundant
of
biodi
ver
sit
y
spe
ci
es
that
m
a
y
b
e
come
the
m
aj
ori
t
y
spe
ci
es
in
an
y
par
ti
cu
la
r
reg
ion.
Thi
s
situ
at
ion
creat
e
a
larg
e
dataset
with
a
complex
var
i
abl
es
to
b
e
ana
l
y
sed
.
Moreo
ver
,
the
r
esponds of
orga
nism
s a
n
d
envi
ronm
ental fac
tors
are
occ
urre
d
in
a
no
n
-
li
ne
ar
cor
r
el
a
tion.
The
eff
or
t
to
do
so
is
rea
l
l
y
i
m
porta
nt
in
orde
r
to
conse
r
ve
th
e
biod
ive
r
sit
y
of
n
at
ure
.
To
under
st
an
d
t
he
compl
e
x
rel
a
ti
onships
that
exi
st
bet
wee
n
spec
ie
s
distri
b
uti
on
and
the
ir
habi
t
at
,
we
ana
l
y
sed
the
intera
c
ti
ons
among
bird
dive
rsi
t
y
,
spati
al
distri
bu
tion
and
la
n
d
use
t
y
pes
a
t
Ke
n
y
ir
la
ndsc
ape
i
n
Te
r
engga
nu
,
Malay
s
ia
b
y
usi
ng
art
if
icial
neur
al
n
et
work
(
AN
N)
m
et
hod
of
self
-
orga
ni
zi
ng
m
ap
(SO
M)
ana
l
y
sis.
SO
M
per
form
s
an
uns
uper
vised
and
n
on
-
li
ne
ar
ana
l
y
s
is
on
a
complex
and
la
rge
dat
ase
t.
It
is
ca
p
abl
e
to
h
andle
t
he
non
-
l
ine
ar
co
rre
lation
b
et
we
e
n
orga
nism
and
env
ironment
al
f
ac
tors
b
ecause
SO
M
ide
ntifie
s
c
luste
rs
and
r
e
la
ti
onships
bet
wee
n
va
ria
b
l
es
without
the
f
ixe
d
assum
pti
ons
of
li
nea
rity
or
norm
al
ity
.
The
result
sug
geste
d
that
SO
M
ana
l
y
s
is
was
suite
d
for
under
standi
n
g
the
relati
onshi
ps
bet
wee
n
bird
spec
ie
s
assem
bla
ges
and
habi
tat
cha
ra
cteri
sti
cs
.
Ke
yw
or
d
s
:
ANN
Con
se
r
vation
Ken
yi
r
lan
ds
ca
pe
SO
M
Sp
eci
es
distrib
ution
Copyright
©
201
9
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Salwana
Mo
ha
m
ad
Asm
ara,
Faculty
of Com
pu
te
r
Syst
e
m
and So
ftwar
e
En
gin
eeri
ng,
Un
i
ver
sit
i M
al
ay
sia
Pah
a
ng,
26300 Gam
bang,
K
uan
ta
n,
Pa
hang, M
al
ay
sia
.
Em
a
il
: sal
wan
am
oh
a
m
ad@
um
p.
edu
.m
y
1.
INTROD
U
CTION
Descr
i
bing
the
su
it
able
env
ir
on
m
ental
con
di
ti
on
s
for
biod
iversity
sp
eci
es
is
an
app
r
oa
ch
to
asses
s
the
geog
raphi
c
distribu
ti
on
of
the
s
pecies
[1
]
.
The
r
ea
so
n
is
that
di
stribu
ti
on
of
s
pecies
is
relat
ed
to
env
i
ronm
ental
conditi
on.
E
nvir
on
m
ental
conditi
ons
that
are
su
it
able
for
the
s
pecies
m
us
t
be
descr
i
bed
a
nd
identify
the
ar
ea
to
m
at
ch
sp
eci
es
occu
r
re
nc
e
data
[2
]
.
The
en
vir
on
m
ental
data
i
s
descr
ibes
by
the
data
of
cl
i
m
at
e
and
to
pograp
hy
in
a
form
of
te
ch
ni
cal
,
sp
at
ia
l
an
d
te
m
po
ral
data
ty
pes
[
3].
T
he
re
are
tw
o
ty
pes
of
input
that
are
req
ui
red
f
or
pr
e
dicti
ng
the
spe
ci
es
distribu
ti
on
w
hich
are
th
e
locat
ion
s
of
occurre
nce
of
sp
eci
e
s
and
en
vir
onm
e
ntal
va
riables.
Both
i
nputs
m
us
t
be
fit
into
a
proces
s
to
quantify
a
nd
est
i
m
at
e
the
geog
raphic
distrib
ution
of
a
sp
eci
es
whic
h
is
known
as
sp
eci
es
dis
tribu
ti
on
m
od
el
li
ng
(SDM).
SD
M
m
et
hods
are
nu
m
erical
m
eth
ods
t
hat
will
qu
a
ntify
the
c
orrelat
ion
bet
wee
n
t
he
data
of
s
pecies
with
en
vir
onm
ent
al
factor
[4,
5]. Th
e tas
k t
o
assess the s
pecies distrib
ut
ion
ca
n
be
do
ne
b
y app
ly
in
g
a comm
un
it
y
e
colo
gy appr
oac
h
that
stud
ie
s
the
s
pe
ci
es
asse
m
blages
of
a
re
gion
[
6].
Sp
eci
es
assem
blages
are
cl
assifi
cat
ion
s
of
sp
eci
e
s
that
co
-
occ
ur
in
the
sa
m
e
place
and
at
the
sam
e
t
i
m
e
[7
]
.
The
be
nef
it
of
the
cl
assifi
cat
ion
s
is
that
the
sta
bility
of
sp
eci
es
assem
blages
m
ay
be
us
ed
t
o
de
fine
represe
ntati
ve
and
ref
e
re
nce
sit
es
fo
r
bio
l
ogic
al
assessm
e
nt
[
8].
Assem
blages
of
co
-
occ
urrin
g
inv
asi
ve
sp
eci
e
s
from
any
so
ur
ce
reg
i
ons
are
us
ed
to
ide
ntif
y
and
highli
ght
new
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
6
,
Dece
m
ber
201
9
:
52
35
-
5243
5236
sp
eci
es
th
reats
for
the
relat
ed
reg
i
on
[9
]
.
I
n
that
way,
a
la
rge
nu
m
ber
of
spe
ci
es
can
be
an
al
yse
d
co
ncurre
ntly
and ra
nked
in
a
p
a
rtic
ular regi
on b
a
sed
on sp
eci
es correlat
io
ns
.
The
data
of
sp
eci
es
with
en
vir
on
m
ental
factor
co
ns
ist
of
a
huge
nu
m
ber
of
var
ia
bles.
Th
us
,
it
is
dif
f
ic
ult
to
highli
ght
the
dat
a
pat
te
rn
i
n
the
dat
aset
[10,
11]
.
These
data
a
re
ve
ry
com
plex
to
be
analy
sed
beca
us
e
of
a
la
rg
e
a
m
ou
nt
of
spe
ci
es
and
the
respo
nd
of
or
gan
i
sm
s
to
env
ir
on
m
ental
factor
s
is
a
non
-
li
nea
r
c
orrelat
ion
[
12]
.
To
sim
plify
the
analy
sis
of
a
com
plex
dataset
,
cl
assifi
cat
ion
te
ch
niqu
es
are
com
m
on
ly
us
ed.
Most
sta
ti
stics
m
et
ho
ds
ha
nd
le
the
data
a
s
a
li
near
data
analy
sis.
The
li
near
analy
s
is
is
no
t
ver
y
eff
ic
ie
nt
t
o
cat
er
the
non
-
li
near
co
rr
el
at
ion
be
ha
viour
betwee
n
orga
ni
s
m
s
and
env
ir
on
m
enta
l
factor
in
an
op
ti
m
u
m
w
ay
[13
]
.
2.
RELATE
D
W
ORKS
SO
M
is
a
n
unsu
pe
r
vised
le
a
rn
i
ng
a
nd
non
-
li
near
m
et
ho
d
to
a
naly
se
va
rio
us
data
set
s
,
inclu
di
ng
tho
se
w
it
h
m
issi
ng
values
[14
,
15]
.
T
he
as
sessm
ent
of
com
m
un
it
y
patter
ns
a
nd
c
orre
la
ti
on
s
m
a
y
be
m
or
e
accurate
t
han
t
he
c
omm
on
sta
ti
sti
cal
m
et
ho
ds
.
T
he
SO
M
m
et
ho
d
has
be
en
wi
dely
us
e
d
to
i
den
ti
fy
s
pecies
com
m
un
it
y pat
te
rn
s a
nd
t
heir rel
at
ion
s
hip
s
w
it
h
env
i
ronm
ental
co
nd
it
io
ns
i
n
aq
uatic
h
a
bitat
s.
Howe
ver, it
h
as
no
t
bee
n
fu
ll
y
us
e
d
f
or ter
restrial
h
a
bitat
s and
for bio
div
e
rs
it
y con
ser
vatio
n plan
nin
g
[13
]
.
The
us
e
of
art
ific
ia
l
neu
ral
ne
tworks
(AN
N
),
s
uch
a
s
sel
f
-
orga
nizin
g
m
aps
(
SO
M
),
prov
e
d
to
be
us
ef
ul
a
nd
ef
fici
ent
in
m
any
cl
i
m
atic,
env
ir
on
m
ental
,
ecol
og
ic
al
a
pp
li
cat
ion
s
a
nd
s
peci
es
distrib
utio
n
[
10
,
12
,
16,
17]
.
A
NN
can
pr
e
dict
the
ta
rg
et
values
base
d
on
t
he
le
arn
i
ng
proce
ss
[18
,
19]
.
T
he
widely
us
e
d
of
SO
M
analy
sis
in
these
do
m
ai
ns
is
because
of
t
he
SO
M
abili
ty
to
correla
te
s
the
va
riables
w
it
ho
ut
de
pe
nding
on
the
fixe
d
li
nea
rity
assum
ption
s.
A
stu
dy
ha
s
bee
n
do
ne
on
a
nt
c
omm
u
niti
es
by
cl
us
t
erin
g
an
d
a
naly
sing
ind
ic
at
or
sp
e
ci
es
us
in
g
S
O
M
[12]
.
The
resu
lt
s
assess
the
unde
rstan
di
ng
the
relat
io
ns
hi
ps
betwee
n
an
t
com
m
un
it
ie
s
a
nd
s
pecies
a
nd
ha
bitat
cha
rac
te
risti
cs.
SO
M
has
bee
n
ap
pl
ie
d
to
vis
ualiz
e
broa
d
patte
r
ns
i
n
sp
eci
es
ric
hnes
s
by
ta
xo
no
m
i
c
gro
up
[
13]
.
SO
M
hav
e
al
so
been
us
e
d
i
n
the
m
od
el
li
ng
of
water
qual
it
y
and
resou
rces.
A
stud
y
has
bee
n
done
on
t
he
a
ppli
cat
ion
of
S
OM
te
ch
nique
s
com
bin
ed
wi
th
K
-
m
eans
cl
us
te
ri
ng
to
m
od
el
wate
r
qual
it
y
data
in
the
treat
m
e
nt
of
dr
in
king
water
[14]
.
An
ef
fort
ha
ve
bee
n
un
der
t
aken
to
descr
i
be
t
he
st
at
e
of
local
w
at
er
bo
dies
a
nd
po
ll
ution
in
the
area
of
int
erest
with
the
help
of
S
OM
[20]
.
Tsai
et
al
.,
[
21]
hav
e
ex
plore
d
the
relat
io
nship
betwee
n
fi
sh
com
m
un
it
y
and
water
qu
al
it
y
by
SO
M
wh
ic
h
is
su
it
ably
ref
le
ct
s
the
sp
at
ia
l
char
act
erist
ic
s
of
fishe
ry
sam
pling
sit
es.
Self
-
org
an
iz
in
g
m
ap
(SOM)
c
om
bin
ed
with
the
K
-
m
e
ans
al
gorith
m
is
us
ed
to
dete
rm
ine
the
sp
at
iotem
po
ral
patte
rn
of
the
wa
te
r
qu
al
it
y
data
and
identify
in
g
the
so
urces
of
po
ll
ution i
n
t
he Kl
ang Ri
ver ba
sin
[22]
.
In
t
his
stu
dy,
we
descr
i
be
how
SO
M
can
be
a
pp
li
ed
f
or
the
stu
dy
of
ecolo
gical
co
m
m
un
it
ie
s
an
d
distrib
ution
of
bir
d
s
pecies.
The
pr
e
sentat
ion
of
SO
M
a
da
pted
t
o
b
io
di
ver
sit
y
data
is
descr
i
bed
by
trai
ning
SO
M
on
e
xa
m
ple
data
of
Ken
yi
r
la
ndsc
ape
i
n
Te
reng
ganu,
Ma
la
ysi
a.
Sele
ct
in
g
t
he
m
os
t
ecol
ogic
al
l
y
represe
ntati
ve
sp
eci
es
is
a
m
et
hod
t
hat
e
na
ble
to
dec
reas
e
the
siz
e
of
l
arg
e
dataset
s
ba
sed
on
t
he
prefer
re
d
la
nd
us
e
ty
pe
of
eac
h
pa
rtic
ul
ar
sp
eci
es.
A
sel
f
-
organ
iz
in
g
m
ap
(S
OM
),
wh
ic
h
is
an
unsu
pe
r
vised
a
rtific
ia
l
neural
netw
ork,
ca
n
be
us
e
d
to
gen
e
rate
va
lues
that
i
nd
i
cat
e
the
stre
ngth
o
f
ass
ociat
ion
of
a
s
pecie
s
with
a sp
eci
es a
sse
m
blage.
SO
M
is
an
art
ific
ia
l
neu
ral
netw
ork
(
ANN)
m
et
ho
d
tha
t
le
arn
to
cl
ust
er
groups
of
si
m
il
ar
inp
ut
patte
rn
s
f
or
a
hi
gh
dim
ensional
input
s
pace
i
n
a
non
-
li
near
fash
i
on
onto
a
low
dim
ensional
ou
t
pu
t
la
ye
r
[15]
.
It
cal
culat
e
the
prob
a
bili
ty
densi
ty
fu
ncti
on
of
in
put
da
ta
us
ing
a
n
un
s
uper
vise
d
le
arn
i
ng
al
go
rithm
.
The
m
et
ho
d
al
so
ena
ble
to
do
cl
us
te
ri
ng,
vi
su
al
iz
ing
a
nd
a
bs
tract
io
n
of
com
plex
data.
T
he
m
ai
n
ob
j
ect
ive
of
data
vis
ualisa
ti
on
is
t
o
gain
i
ns
ig
ht
into
i
nfor
m
at
ion
by
m
app
i
ng
data
int
o
an
un
der
sta
ndable
gr
a
phic
al
fo
rm
[23]
.
Fi
gure 1
sh
ows
the struct
ur
e o
f
a
n
S
O
M
co
ns
ist
of
t
wo
la
ye
rs:
a
n
i
nput
la
ye
r
a
nd o
ut
pu
t
la
ye
r
.
T
he
in
put
la
ye
r
con
ta
in
one
ne
uron
for
each
va
riable
in
the
data
set
.
The
outp
ut
la
ye
r
neur
on
s
a
re
connecte
d
to
ever
y
neur
on in
t
he
i
nput lay
er t
hro
ugh
a
djust
able
weig
hts
of
pa
r
a
m
et
ers.
3.
PRO
CED
U
R
E OF S
OM
APPL
IC
ATIO
N
The
le
arn
i
ng
s
te
ps
are
desc
ribed
in
detai
l
t
o
get
the
insig
ht
of
the
S
O
M
m
e
tho
d
to
be
ap
plied
in
the case st
ud
y.
The pr
oce
dure
s
re
qu
i
red to a
pp
ly
S
OM i
nclud
i
ng m
ai
n
ste
ps
:
3.1.
D
ata g
atheri
ng
and n
or
maliz
at
ion
The
m
et
ho
do
l
og
y
of
sam
pli
ng
for
data
ga
therin
g
is
de
scribe
d
in
the
nex
t
sect
io
n
of
Sam
plin
g
Me
thods.
Af
te
r
the
da
ta
gath
erin
g,
data
nor
m
al
iz
ation
is
done
t
o
en
sure
that
al
l
var
ia
ble
s
have
the
sam
e
ra
nk
in
the
de
velo
pm
ent
of
t
he
S
OM.
By
no
rm
al
iz
ing
the
dat
a,
the
SO
M
m
ay
visu
al
iz
es
overall
perform
ance
of
the
s
pecies
distrib
ution
an
d
as
sem
blage
in
a
bette
r
way.
Da
ta
norm
al
iz
ation
ca
us
es
S
OM
ne
uro
ns
t
o
e
xpan
se
towa
rd h
i
gh
li
gh
te
d pa
rts of t
he
d
at
a
distrib
ut
ion
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Self
-
org
an
izi
ng
map
(
SOM)
for
s
pecies
distr
ibu
ti
o
n modell
ing of
bir
d
…
(
Sa
lw
ana
M
ohama
d @
Asm
ara
)
5237
Figure
1.
Str
uc
ture of
tw
o
-
di
m
ension
al
SOM
[16]
3.2. D
ata tr
aining
An
in
put
vect
or
f
r
om
the
da
ta
m
at
rix
is
introd
uced
to
the
it
era
ti
ve
trai
nin
g
proce
dure
to
form
the
S
OM.
T
he
SO
M
us
es
a
ty
pe
of
le
ar
nin
g
that
is
cal
le
d
c
om
petitive,
uns
up
e
r
vised
,
or
sel
f
-
or
gani
zi
ng
proce
dure
to
m
at
ch
each
inp
ut
vect
or
wit
h
a
neuron
in
the
SO
M
[
16
]
.
The
ne
uro
n
wi
th
the
cl
os
est
m
at
ch
to
the
pr
ese
nted
input
patte
rn
i
s
cal
le
d
winner
ne
uron
or
best
m
a
tc
hin
g
un
it
(BMU)
.
The
winner
and
it
s
neig
hbours
pr
e
def
i
ned
in
th
e
al
gorithm
up
da
te
their
weig
ht
vector
s
acc
ordin
g
to
the
S
O
M
le
arn
in
g
ru
l
es
as
fo
ll
ows:
w
ij
(
t
+
1
)
−
w
ij
+
α
(
t
)
h
jc
(
t
)
[
x
i
(
t
)
−
w
ij
(
t
)
]
(1)
Wh
e
re
w
ij
(
t
)
is
a
we
igh
t
betwee
n
a
node
i
in
the
input
la
ye
r
a
nd
a
node
j
in
th
e
ou
t
pu
t
la
ye
r
at
it
erati
on
ti
m
e
t,
α
(t)
is
a
le
arn
i
ng
rate
fa
ct
or
w
hich
is
a
decr
easi
ng
f
un
ct
io
n
of
t
he
it
erati
on
ti
m
e
t,
an
d
h
jc
(
t
)
is
a
neig
hbor
hood
f
unct
io
n
(a
sm
oo
thin
g
ke
rn
el
def
i
ned
ov
e
r
th
e
la
tt
ic
e
points)
that
def
i
nes
t
he
s
iz
e
of
neig
hbour
hood
of
the
winni
ng
no
de
c
to
be
update
d
durin
g
the
le
ar
ni
ng
proces
s.
At
the
final
sta
ge
the r
e
fer
e
nce
ve
ct
or
s
of all
th
e act
ivate
d ne
uro
ns
a
re
update
d.
3.3.
Ex
tra
c
tin
g
in
fo
rm
ati
on
f
rom
the tr
ain
ed SO
M
The
res
ult
can
be
vis
ualiz
ed
an
d
t
he
data
can
be
cl
us
te
r
ed
when
the
trai
ning
of
S
O
M
is
do
ne
.
The
trai
ne
d
S
OM
m
ap
is
a
visu
al
iz
at
ion
t
oo
l
that
will
gi
ve
the
visio
n
of
the
data
ex
pl
or
at
io
n.
T
he
fi
rst
-
le
vel
SO
M
cl
us
te
rs
input
data
int
o
a
ce
rtai
n
nu
m
ber
of
cl
us
te
rs
an
d
t
he
sec
on
d
-
le
vel
cl
us
te
r
s
the
outp
ut
ne
uro
ns
into as
m
any d
iffe
ren
t
reg
i
ons
as r
e
qu
i
red.
4.
SAMPLIN
G
METHO
DS
The
a
naly
sis
of
SO
M
has
been
done
t
o
the
dataset
of
bir
ds
in
Keny
ir
la
nd
sca
pe,
Tereng
ganu
,
Ma
la
ysi
a
fr
om
2015
to
2016
.
Ther
e
a
re
53
sp
eci
es
of
bi
r
ds
from
20
fa
m
ily
t
ypes
fou
nd
in
the
stu
dy
sit
e
.
Ther
e
are
tw
o
ty
pes
of
sam
pling
m
et
hod
th
at
have
been
use
d
t
o
c
ollec
t
the
da
ta
wh
ic
h
are
po
i
nt
co
un
t
an
d
m
ist
n
et
ti
ng
.
4
.
1.
P
oint
c
ount
Ra
lph
et
al
.
[
24
]
sta
te
d
that
point
c
ount
i
s
one
of
the
m
os
t
c
o
m
m
on
ly
us
e
d
m
et
hod
in
stu
dyin
g
the abu
nd
a
nce,
d
ist
rib
utio
n
a
nd ec
ology o
f forest
b
ir
ds. P
oint
co
unt
was
c
hose
n
bec
ause
it
en
a
ble obse
rvers t
o
locat
e
an
d
obs
erv
e
rain
f
or
est
bir
ds
t
hro
ugh
sta
ndin
g
at
a
fixe
d
locat
io
n
in
a
fixe
d
ti
m
e
w
hich
al
s
o
he
lps
in
identify
in
g
bir
ds
that
a
re
hard
to
obser
ve.
A
ll
obser
ve
d
bi
rd
s
wer
e
ide
nt
ifie
d
up
to
spe
ci
es
le
vel
an
d
extra
inf
or
m
at
ion
suc
h
as
sta
tus
of
distribu
ti
on,
protect
io
n
sta
tu
s
and
c
onser
va
ti
on
sta
tus
of
I
UCN
Re
d
List
2014
was
ref
e
rr
e
d
from
pu
blishe
d
m
at
erial
s.
Bi
rd
s
ide
ntific
at
ion
was
ai
de
d
by
Ro
bs
on
[
25]
,
Stra
nge
a
nd
Jey
araj
asi
ng
a
m
[26]
, D
a
vis
on a
nd F
ook
[
27]
and
Wong
[
28]
with
addit
io
nal in
form
at
io
n by
W
el
ls
[29
]
.
4
.2
.
Mis
t
ne
tt
i
ng
Mi
st
netti
ng
is
ano
t
her
m
et
ho
d
t
hat
is
m
ore
sp
eci
fic
in
t
erm
s
of
identi
ficat
ion
that
al
so
incl
ud
es
crypti
c
sp
eci
e
s,
sam
pling
of
ge
netic
m
ater
ia
ls,
par
asi
te
and
m
or
phol
og
ic
al
data
co
ll
ect
ion
on
ca
ptured
bir
ds
[30]
.
Mi
st
-
netti
ng
is
al
so
vital
in
or
de
r
to
reco
nfi
rm
t
he
sp
eci
es
obs
erv
e
d
duri
ng
point
-
c
ount
surv
ey
and
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
6
,
Dece
m
ber
201
9
:
52
35
-
5243
5238
it
ben
efit
s
in
c
aptu
rin
g
in
div
i
du
al
s
of
seve
ra
l
sp
eci
es
[
31
]
.
In
a
ddit
ion
,
t
his
m
e
tho
d
fo
c
us
m
or
e
on
the
r
esults
of
s
pecies
dist
rib
u
ti
on
rathe
r
than
a
bundan
ce
[32]
.
Nets
data
we
re
rec
orde
d
s
uch
as
t
heir
heig
ht
an
d
G
PS
read
i
ng.
Bi
rds
wer
e
i
den
ti
fie
d
un
ti
l
sp
e
ci
es
le
vel
[
33
]
a
nd
t
he
e
xtern
al
m
or
ph
ologica
l
dat
a
of
ca
pt
ur
e
d
bir
ds
wer
e
recor
ded
includi
ng
t
heir
bro
od
i
ng
patc
h,
m
ou
lt
ing
sta
ge
an
d
th
e
nets
that
they
wer
e
cap
tu
red.
T
he
bir
ds
wer
e
m
ark
ed
be
fore
r
el
ease
d
at
the
captur
es
sit
es.
Ca
ptu
re
d
bir
ds
we
re
re
le
ased
nearby
to
the
spot
they
were
captu
red to a
void
distu
rb
a
nce
to
thei
r dail
y ac
ti
viti
es
[34]
.
5.
STUDY
SITE
S
The
var
i
ous
s
tud
y
sit
es
we
re
c
ho
se
n
due
to
it
s
diff
e
re
nt
f
or
e
st
ty
pe
s
in
orde
r
to
determ
ine
the
relat
io
nship
of
t
he
div
e
rs
it
y
of
birds
acc
ordin
g
t
o
s
pecific
ha
bitat
ty
pes.
T
he
stu
dy
w
ere
car
ried
out
at
11
sit
es
in
Ken
yi
r
la
nd
sca
pe
w
hi
ch
are
Tan
jung
Me
nton
g,
B
uw
e
h
Ri
ve
r,
Be
l
uk
a
r
Bu
kit
1,
Sekayu
Wate
rf
al
l,
Ken
yi
r
Re
sea
r
ch
Stat
io
n,
Be
l
uk
a
r
Bu
kit
2,
Sekayu
A
gr
ic
ul
ture
Pa
rk,
P
usa
t
Lat
ihan
Re
la
W
il
ay
ah
Ti
m
ur
1,
Pu
sat
Lat
iha
n
Re
la
W
il
ay
ah
Ti
m
ur
2,
Saok
W
at
erf
al
l
a
nd
Be
luk
ar
B
ukit
3.
Ta
ble
1
re
pr
ese
nts
the
c
od
e
of
the
stu
dy
sit
es
th
at
will
be
a
ppli
ed
in
t
he
S
OM.
T
he
stu
dy
sit
es
wer
e
cl
assifi
ed
int
o
si
x
la
nd
us
e
ty
pe
s
w
hich
are
f
orest
ve
ge
ta
ti
on
,
lo
w
l
and
di
pter
ocarp
f
orest
,
m
elalueca
f
or
e
st,
or
c
ha
rd,
pea
t
swam
p
fore
st
and
plantat
io
n.
Eac
h
la
nd
us
e
ty
pe
presents
b
ir
ds
with a
va
riet
y o
f
d
if
fer
e
nt
ha
bitat
s.
Table
1
.
C
odes
of study si
te
Co
d
e
Site
A
Tanju
n
g
M
en
to
n
g
B
Bu
weh
Riv
er
C
Belu
k
ar
Bu
k
it 1
D
Sek
ay
u
W
aterfall
E
Ken
y
ir
Research
S
tatio
n
F
Belu
k
ar
Bu
k
it 2
G
Sek
ay
u
Agricu
ltu
r
e Par
k
H
Pu
sat Latihan
Rela
W
ila
y
ah
Ti
m
u
r
1
I
Pu
sat
Latihan
Rela
W
ila
y
ah
Ti
m
u
r
2
J
Sao
k
W
ate
rf
all
K
Belu
k
ar
Bu
k
it 3
6.
DA
T
A ANAL
YS
IS
We
us
e
d
the
SO
M
al
gorith
m
to
char
act
er
ise
the
distrib
ution
patt
erns
of
birds.
T
he
structu
re
of
a
SO
M
co
ns
is
ts
of
a
n
input
data
la
ye
r
an
d
ou
t
pu
t
data
la
ye
r.
The
i
nput
la
ye
r
is
com
p
os
e
d
of
53
ne
uro
ns
(one
per
bir
d
sp
eci
es)
co
nnect
ed
t
o
the
4
sam
pling
da
ta
se
ts.
Traini
ng
of
t
he
S
O
M
and
t
he
cl
us
te
rin
g
proce
dures wer
e p
e
rfor
m
ed
usi
ng
Mat
la
b.
7.
STRUCT
UR
I
NG
I
N
DE
X
The
S
I
was
or
igi
nally
de
velo
ped
t
o
def
i
ne
sp
eci
e
s
sho
wi
ng
t
he
strongest
i
nf
l
uen
ce
on
the
orga
nizat
ion
of
the
S
OM
m
ap.
The
SI
is
the
value
ind
i
cat
ing
the
relat
ive
i
m
po
rtance
of
each
s
pecies
i
n
determ
ining
th
e
distrib
ution
patte
rn
s
of
t
he
sa
m
ples
in
the
SO
M.
T
her
e
fore,
the
set
of
sp
eci
es
sho
wing
hi
gh
SI
ca
n be c
onsidere
d
as
the i
ndic
at
or sp
eci
es
.
The
SI
is
cal
cu
la
te
d
fr
om
the
su
m
of
the
rati
os
of
the
distanc
e
betwee
n
th
e
weigh
ts
of
al
l
sp
eci
es
in
the
S
OM
an
d
the
to
polo
gical
distance
bet
ween
tw
o
S
O
M
un
it
s.
This
res
ults
in
r
ep
resen
ti
ng
distr
ibu
ti
on
gr
a
dients
f
or
each
sp
eci
es
in
the
trai
ned
SO
M.
A
str
uc
turing
ind
e
x
of
s
pecies
i
,
SIi
,
is
express
ed
in
the equat
io
n
as
foll
ow
s:
=
∑
∑
|
−
|
|
|
−
|
|
−
1
=
1
=
1
(2)
Wh
e
re
w
ij
and
w
i
k
are
the
connecti
on
weig
ht
s
of
s
pecies
i
(
in
the
in
pu
t
la
ye
r)
in
S
OM
unit
s
an
d
k,
||
r
j
−
r
k
||
is
the
topolo
gica
l
distance
bet
ween
un
it
s
j
and
k
,
a
nd
S
is
the
total
nu
m
ber
of
SO
M
outp
ut
un
it
s
.
SI
co
ns
ide
rs
th
e
distribu
ti
on
gr
a
dients
of
each
sp
eci
es
in
the
SO
M
m
ap.
Sp
eci
es
sho
wing
a
stron
g
gr
adient
disp
la
y
a
high
SI
val
ue,
w
herea
s
sp
eci
es
showi
ng
a
wea
k
gr
a
dient
prese
nt
a
low
SI
va
lue.
Th
us
,
the
highe
r
t
he
val
ue of
SI,
the m
or
e rele
va
nt the va
riabl
e is to t
he
str
uc
ture of
the m
ap
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Self
-
org
an
izi
ng
map
(
SOM)
for
s
pecies
distr
ibu
ti
o
n modell
ing of
bir
d
…
(
Sa
lw
ana
M
ohama
d @
Asm
ara
)
5239
8.
SOM M
AP
G
RID
The
S
OM
m
ap
sp
eci
fy
the
nu
m
ber
s
of
r
ow
s
and
c
olu
m
ns
in
the
gri
d.
We
determ
ine
the
nu
m
ber
of
rows
an
d
col
um
ns
is
set
to
10
.
Fig
ure
2
sho
ws
the
S
OM
m
ap
gri
d
f
or
ou
t
pu
t
ne
uro
n.
T
he
def
a
ult
top
ol
og
y
of
the
S
OM
is
he
xago
nal.
T
his
fig
ur
e
sho
ws
t
he
neur
on
loca
ti
on
s
i
n
the
to
po
l
og
y,
an
d
in
dicat
es
ho
w
m
any
of
the
trai
ning
da
ta
are
associat
ed
with
eac
h
of
the
ne
uro
ns
(
cl
us
te
r
cente
rs
).
T
he
to
po
l
ogy
is
a
10
-
by
-
10
gri
d,
so
t
her
e a
re
10
0 neur
on
s
.
Figure
2
.
S
OM m
ap
gr
id
for o
utput ne
uro
n
9.
CLUS
TE
RI
N
G LA
ND TYP
E USE
Ther
e
a
re
53
el
e
m
ents
in
eac
h
input
vecto
r
wh
ic
h
is
on
e
pe
r
bir
d
sp
eci
es.
The
weig
ht
ve
ct
or
s
w
hi
c
h
is
cl
us
te
r
ce
nt
ers
fall
within
the
ou
t
pu
t
ve
ct
or
s
pace
of
t
he
100
ne
uro
ns.
A
n
i
nput
ve
ct
or
is
a
ve
ct
or
wit
h
featur
e
s
of
vari
ables
an
d
at
tr
ibu
te
s
wh
e
reas
the
at
trib
utes
represe
nted
by
11
sit
es
of
sam
pling
.
The
input
vecto
r
is
c
orre
la
te
d
with
eac
h
un
it
.
T
he
S
OM
m
ap
gri
d
sh
ows
al
l
poss
ible
analy
sis
usi
ng
a
def
i
ned
set
of
ou
t
pu
t
vecto
rs.
Me
anwhil
e,
th
e
input
vect
or
s
will
be
s
or
te
d
on
t
he
gr
id
so
that
si
m
il
ar
inp
ut
vect
or
s
a
re
cl
os
e
to
each
oth
e
r
and
dissim
il
ar
inp
ut
vect
or
s
are
far
from
each
oth
e
r.
A
visu
al
iz
at
ion
too
l
f
or
the
S
OM
is
the
wei
gh
t
dist
ance m
at
rix
as
in Figu
re
3
.
In
Fig
ur
e
3,
th
e
blu
e
he
xa
gons
represe
nt
th
e
ne
urons.
T
he
red
li
nes
c
onne
ct
neig
hbouri
ng
ne
uro
ns.
The
col
or
s
in
the
reg
i
on
s
c
onta
inin
g
the
re
d
li
nes
sh
ow
the
distances
be
tween
ne
uro
ns.
The
da
rk
e
r
colo
rs
repr
ese
nt
la
rg
e
r
distance
s,
a
nd
the
li
ghte
r
c
olors
re
pr
e
sent
sm
a
ll
er
distances.
The
SO
M
netw
ork
a
pp
e
ars
t
o
hav
e
cluster
ed
the lan
d use ty
pe
int
o four dis
ti
nct g
r
oups as
in Figu
re
4.
Figure
3. S
OM
weig
ht
distanc
e m
a
trix
of
nei
ghbour
distances
Figure
4. Cl
us
t
ers of
sp
eci
es
di
stribu
ti
on
on
SO
M
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
6
,
Dece
m
ber
201
9
:
52
35
-
5243
5240
These
cl
us
te
rs
ref
le
ct
ed
t
he
la
nd
us
e
ty
pe
at
each
of
sam
pling
sit
e
wh
ic
h
a
re
cl
us
te
r
1
c
or
respo
nd
s
to
forest
veg
et
at
i
on.
Cl
us
te
r
2
c
orres
ponds
to
peat
swam
p
forest.
Cl
us
te
r
3
corres
ponds
to
low
la
nd
di
ptero
ca
rp
forest
an
d
m
elalueca
f
or
est
.
Cl
us
te
r
4
c
orr
e
sponds
to
or
chard
a
nd
pla
ntati
on
.
The
t
r
ai
ned
m
ap
of
SO
M
consi
sti
ng
of
node
s
relat
e
d
to
their
outp
ut
ve
ct
or
s
is
prese
nted
by
t
wo
-
di
m
ension
al
(
2D)
c
om
po
ne
nt
pl
anes
.
The
wei
gh
t
ve
ct
or
relat
ed
t
o
each
ne
uro
n
will
transf
e
rs
to
bec
om
e
the
cl
us
te
r
cente
r
of
a
n
in
pu
t
ve
ct
or.
The
va
riable
di
stribu
ti
on
val
ues
are
repres
ented
on
th
e
trai
ne
d
m
ap
by
di
vid
e
d
sect
ion
s
.
Eac
h
sect
ion
of
the
va
riable
di
stribu
ti
on
is
di
ff
e
ren
ti
at
ed
by
dif
fer
e
nt
c
olours.
T
hese
c
om
po
nen
t
plane
s
co
uld
be
c
ompare
d
i
n
order
to
ide
ntif
y
si
m
i
l
arit
ie
s
between
var
ia
bl
es.
As
su
m
ption
ca
n
be
m
ade
that
the
in
pu
ts
are
hi
gh
ly
c
orr
el
at
ed
if
the
relat
ion
patte
rn
s
of
tw
o
in
puts
we
re
ver
y
sim
il
ar.
The
c
om
po
ne
nt
planes
re
pr
es
enting
pr
ob
a
bi
li
ty
of
occurre
nce
as s
how
n
in
Fig
ure
5
.
Figure
5. S
OM for
al
l
pro
bab
i
li
ty
o
f
bir
d spe
ci
es o
cc
urren
c
e
So
m
e
patte
rn
s
of
sim
il
arity
cou
ld
be
disti
ng
ui
sh
ed
s
uc
h
as
s
i
m
i
la
r
distrib
ution
for
the
s
pe
ci
es
of
Ixos
ma
l
accensi
s
,
Sta
chy
ris
eryt
hr
opte
r
a,
Pel
lorn
eu
m
cap
ist
ra
t
um,
Prionochilu
s
percussus
,
Dicaeum
tri
gin
os
ti
gm
a,
Ire
na
puel
la, O
tus bakk
amoe
na
and
Cy
mb
ir
hy
nchus
macr
orhync
hos
.
Th
e
s
e
sp
eci
es
was s
how
n
to
prefe
r
cl
us
t
er
3
a
nd
4
w
hi
ch
co
rr
es
pond
s
to
lo
w
la
nd
dip
te
r
ocarp
for
est
,
m
e
la
lueca
forest,
or
c
hard
an
d
plantat
io
n
la
nd
us
e
ty
pe
.
O
ther
s
pecies
with
sim
il
ar
distribu
ti
on
are
Pycnon
otu
s
go
i
avier,
Tric
ha
st
om
a
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Self
-
org
an
izi
ng
map
(
SOM)
for
s
pecies
distr
ibu
ti
o
n modell
ing of
bir
d
…
(
Sa
lw
ana
M
ohama
d @
Asm
ara
)
5241
ma
l
accense
,
St
ach
yri
s
poli
oc
ephala
,
P
rio
no
c
hilus
pe
rc
us
s
us
,
Aegit
hin
a
ti
phia
has
bee
n
re
ported
to
be
par
ti
cula
rly
com
m
on
in
cl
us
te
r
2
w
hich
co
r
respo
nd
s
to
pe
at
swa
m
p
fo
re
st
and
cl
us
te
r
3
an
d
4.
S
pecies
of
Geopeli
a
stri
a
ta,
P
hilento
ma
pyr
hopter
a
a
nd
Eryt
hrur
a
pr
asi
na
wa
s
s
how
n
to
be
i
n
cl
us
te
r
1
wh
i
ch
is
corres
ponds
to
fo
re
st
veg
et
at
i
on.
The
a
vaila
bili
ty
of
hab
it
a
t
su
ch
as
the
c
har
act
erist
ic
of
the
veg
et
at
io
n
and
the
f
ood
s
our
ces
has
a
la
r
ge
im
pact
on
bir
d
distrib
ut
ion
an
d
bir
ds
de
pe
nd
so
le
l
y
on
their
ha
bitat
for
su
r
viv
al
[35
,
36]
.
The
refo
re,
the
analy
sis
re
su
lt
shows
t
he
certai
n
ha
bitat
that
hav
e
the
food
s
ource
f
ro
m
the
ve
getat
io
n
w
hich
is
a
c
r
ucial
in
dicat
or
to
c
onser
ve
t
he
bir
d
s
pecie
s
that
hav
e
th
e
sig
nificant
r
ole
in
ecolo
gy.
Also,
t
he
pr
ese
nce
of
bir
ds
in
a
pa
rtic
ular
a
rea
i
s
m
ai
nly
influ
enced
by
othe
r
cru
ci
al
c
har
act
erist
ic
s
su
c
h
as
su
it
abl
e
hab
it
at
,
sh
el
te
r,
le
ss
com
petit
ion
and
pre
da
ti
on
[
33
]
.
Ra
m
li
et
al
.
[33]
al
so
sta
te
d
that
bird
popula
ti
on
siz
e
in
a
sp
eci
fic
area
is
m
ai
nly
aff
ect
ed
by
f
ood
sup
ply
,
sea
so
na
l
breedi
ng,
m
igrator
y
act
ivit
ie
s
and
ha
bitat
cha
ng
e
s.
Ha
bitat
sel
ect
ion
of
dif
f
eren
t
bir
ds
va
ries
f
ro
m
sp
eci
e
s
to
sp
eci
es.
Bi
rd
s
m
ay
resp
ond
t
o
diff
e
re
nt v
e
get
at
ion
al
str
uctu
r
e
[
37
,
38]
.
10.
CONCL
US
I
O
N
The
SO
M
wa
s
fo
un
d
to
be
an
eff
ect
ive
analy
ti
cal
too
l
fo
r
un
der
sta
ndin
g
the
respo
ns
es
of
bi
r
d
sp
eci
es
to
dif
f
eren
t
ha
bitat
a
nd
assess
the
relat
ion
s
hip
of
each
sp
eci
es
assem
blages
i
n
each
la
nd
use
ty
pe.
Patt
ern
s
an
d
relat
ion
s
hip
in
ecolog
ic
al
c
omm
un
it
ie
s
c
an
be
re
pr
e
se
nted
vis
ually
and
the
unpre
dicte
d
structu
res
t
hat
m
igh
t
be
r
evea
le
d
by
the
a
nal
ysi
s
al
so
can
be
disco
ve
red.
Howe
ver,
S
O
M
has
a
li
m
it
a
ti
on
to
cl
assify
data
be
cause
e
ver
y
S
OM
is
dissim
i
l
ar
[
7]
.
It
will
fi
nd
a
sm
all
different
sim
il
ari
ties
am
on
g
the
s
a
m
ple
vecto
rs
each
ti
m
e
the
pr
el
i
m
i
nar
y
co
ndit
ions
are
cha
ng
e
d.
In
orde
r
to
ha
nd
le
this
lim
itati
on
,
the
c
hange
i
n
rank
of
each
s
pecies
m
us
t
be
ob
se
rved
a
nd
identifie
d
t
o
ge
ner
at
e
a
valid
SO
M
ra
nk
s
e
ach
tim
e
the
data
are
resam
pled
an
d
a
ne
w
S
OM
m
od
el
create
d.
Des
pite
the
li
m
it
a
ti
on
,
S
OM
and
m
achine
le
arn
i
ng
m
et
hods
m
a
y
be
ef
fici
ently
app
li
ed
for
a
co
m
plex
and
no
nl
inear
data.
It
m
ay
pr
ov
i
de
m
or
e
acc
ur
at
e
as
sessm
ents
of
sp
eci
es
patte
rn
a
nd
ec
ology
proce
ss
es
than
t
rad
it
ion
al
sta
ti
sti
cal
m
et
ho
ds.
T
h
e
app
li
cat
io
n
of
SO
M
m
et
ho
ds
in
the
correla
ti
on
b
et
wee
n
sp
eci
e
s
distrib
ution
a
nd
ec
ology
assessm
ent
cou
ld
be
an
ef
fort
to
dev
el
op
strat
e
gy
for
conser
ving
biodive
rsity
en
vi
r
on
m
ents.
ACKN
OWLE
DGE
MENTS
Appreci
at
ion
c
onveyed
to
U
niv
e
rsiti
Ma
la
ysi
a
Paha
ng
f
or
pro
j
ect
fina
ncin
g
unde
r
R
DU1
80363
.
Ma
ny
tha
nk
s
to
our
pa
rtne
r
s
from
Un
ive
r
sit
i
Ma
la
ysi
a
Tereng
ganu
f
or
their
co
ntri
buti
on
in
fiel
d
data
colle
ct
ion
a
nd
ideas
propose
d i
n
this
p
a
per.
REFERE
NCE
S
[1]
L
.
Ca
y
u
ela
e
t
al
.
,
“
Speci
es
distr
i
buti
on
m
odel
ing
in
the
tropics:
p
roble
m
s,
pote
n
tiali
t
ie
s,
and
th
e
r
ole
of
b
iol
ogi
ca
l
dat
a
for effe
ct
iv
e
spec
ie
s
conse
r
vat
ion
,
”
Tr
op.
C
onserv.
S
ci
.
,
vo
l. 2, no. 3, pp. 319
–
352,
2009
.
[2]
R.
G.
Pear
son,
“
Speci
es’
distri
buti
on
m
odel
ing
for
conse
rva
ti
o
n
educ
at
ors
and
pra
ctiti
on
ers,”
Less
ons Conserv
.
,
vol. 3, pp. 54
–
89,
2010
.
[3]
K.
Voigt
,
“
Envi
r
onm
ent
al
informati
on
databa
ses,
”
Encyc
l
.
Comput
.
Chem
.
,
1998.
[4]
J.
Mill
er,
“
Species di
stribu
ti
on
m
odel
ing
,
”
Geogr.
Compas
s
,
vol. 4
,
no
.
6
,
pp
.
490
–
509,
2010
.
[5]
J.
El
it
h
,
M.
A.
Burgm
an,
and
H.
M.
Rega
n,
“
Mapping
epi
stemic
u
nce
rt
ai
nt
ie
s
and
vague
concept
s
i
n
pre
dic
t
ions
of
spec
ie
s d
istri
but
i
on,
”
Ec
o
l. Modell.
,
vol
.
15
7,
no.
2,
pp
.
313
–
329
,
2002.
[6]
D.
R.
Paini,
S.
P.
W
orne
r,
D.
C.
Cook,
P.
J.
De
Barr
o,
and
M.
B
.
Thomas,
“
Us
ing
a
self
-
orga
n
iz
i
ng
m
ap
to
pre
di
ct
inva
sive
spe
cies:
sensiti
vity
to
d
a
ta
err
o
rs
and
a
c
om
par
ison
with
expe
rt
op
ini
on
,
”
J.
App
l.
Ec
o
l.
,
v
ol.
47
,
no.
2
,
pp
.
290
–
298,
2010
.
[7]
S.
P
.
W
orne
r
and
M.
Gevre
y
,
“
Modell
ing
glob
a
l
insec
t
p
est
spec
ie
s
assem
bla
ge
s
to
det
ermine
r
isk
of
inva
sion,
”
J.
Appl. Ecol.
,
v
ol.
43
,
no
.
5
,
pp
.
858
–
867,
2006
.
[8]
R.
Cér
éghi
no
,
J.
L.
Gira
ud
el
,
an
d
A.
Com
pin,
“
Spati
al
an
aly
sis
of
strea
m
inv
erteb
rates
distri
b
u
tion
in
the
Adou
r
Garonne
dra
ina
g
e
basin
(Franc
e)
,
using
Kohonen
self
orga
nizing
m
aps,
”
Ec
ol
.
Mode
ll.
,
vol
.
146,
no.
1
–
3,
pp.
167
–
180,
De
c. 2001.
[9]
M.
A.
McGeoc
h,
P.
Genove
si,
P.
J.
Bel
li
ngham,
M.
J.
Costel
lo,
C.
McGranna
ch
an,
and
A.
Shep
par
d,
“
Priorit
i
zing
spec
ie
s,
pa
thwa
ys,
and
si
te
s
to
a
c
hie
ve
conse
rv
at
i
on
ta
rg
et
s
for
b
i
ologi
c
al
inva
sio
n,
”
Bi
ol
.
In
vasio
ns
,
vol.
18,
no.
2
,
pp.
299
–
314
,
20
16.
[10]
T.
-
S.
Chon
and
Y.
-
S.
Park,
“
Self
-
Organi
zi
ng
M
ap
A2
-
Jørgen
sen,
Sven
Eri
k,
”
B.
D.
B.
T.
-
E
.
of
E.
Fath,
Ed
.
Oxford:
Aca
d
emic
Press
,
2008,
p
p.
3203
–
3210
.
[11]
J.
L.
Gi
rau
de
l
a
nd
S.
Le
k
,
“
A
compari
son
of
self
-
orga
n
iz
ing
m
ap
al
gorit
hm
an
d
som
e
conventiona
l
stat
isti
c
a
l
m
et
hods for
e
cological
comm
unity
ordination,”
E
col
.
Mode
ll
.
,
vo
l
.
146
,
no
.
1
–
3
,
p
p.
329
–
339
,
De
c
.
2001
.
[12]
S.
-
H.
Park,
S.
H
osoishi,
K.
Ogat
a,
and
Y
.
Kuboki,
“
Cluste
ring
o
f
ant
comm
unit
i
es
and
indicat
or
spec
ie
s
an
aly
si
s
using se
lf
-
orga
n
i
zi
ng
m
aps,
”
C
.
R
.
B
iol.
,
vol. 337,
no.
9
,
pp
.
545
–
5
52,
Sep
.
2014
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
6
,
Dece
m
ber
201
9
:
52
35
-
5243
5242
[13]
M.
E.
Hopton
and
A.
L.
Ma
yer
,
“
Us
ing
self
-
orga
nizing
m
aps
to
expl
ore
pa
tt
ern
s
in
spec
i
e
s
ric
hness
and
prote
c
ti
on,
”
Bi
o
div
ers.
Conser
v.
,
vol. 15, no. 14,
pp.
4477
–
4494
,
2006.
[14]
P.
Juntunen,
M.
Li
ukkonen
,
M. Leht
ol
a,
and
Y. H
il
tune
n
,
“
Cluste
r
anal
y
s
is b
y
se
lf
-
orga
ni
zi
ng
m
a
ps: An
appl
ication
to
the
m
ode
ll
i
ng
of
wate
r
qual
ity
in
a
tre
a
tment
proc
ess,”
App
l.
So
ft
Comput.
,
v
ol.
13
,
no.
7
,
pp.
3191
–
3196
,
Jul.
2013.
[15]
T.
Kohonen
,
“
The
self
-
org
anizin
g
m
ap,
”
Neuroc
omputing
,
vo
l. 2
1,
no
.
1
–
3
,
pp
.
1
–
6,
Nov.
1998.
[16]
A.
M.
Kalt
eh,
P.
Hjorth,
and
R.
Berndt
ss
on,
“
Revi
ew
of
the
self
-
orga
nizing
m
ap
(S
OM
)
a
pproa
ch
in
wat
er
resourc
es:
An
alys
is,
m
odel
li
ng
an
d
application
,
”
E
nvi
ro
n.
Mod
el.
S
oft
w.
,
vol
.
23
,
no
.
7
,
pp
.
835
–
845
,
Jul.
2008
.
[17]
D.
-
H.
Kim
,
W
.
-
S.
Cho,
and
T.
-
S
.
Chon,
“
Self
-
or
gani
z
ing
m
ap
an
d
spec
ie
s
abund
a
nce
distr
ibut
ion
of
strea
m
bent
hi
c
m
ac
roinve
rt
ebr
a
te
s
in
rev
ea
l
i
ng
comm
unit
y
pat
te
rns
in
diffe
ren
t
sea
sons
,
”
Ec
ol
.
I
nf
orm
.
,
vol.
17,
pp.
14
–
29
,
Sep
.
2013.
[18]
Kam
al
Moham
m
ed
Alhenda
wi
,
“
Predic
ti
ng
th
e
eff
e
ct
iv
ene
ss
of
web
informat
io
n
s
y
stems
using
neur
a
l
n
et
works
m
odel
ing:
fr
ame
work &
empir
ical te
sting
,
”
Int
.
J
.
Sof
tw. E
ng
.
Co
mput.
Syst
.
,
vol
.
4,
no
.
1
,
pp
.
61
–
74,
201
8.
[19]
H.
Q.
and
H.
I.
-
K.
A.
R.
Ajibo
y
e,
R.
Abdulla
h
-
Ars
hah,
“
Eva
luating
the
eff
ec
t
o
f
d
at
ase
t
siz
e
on
pre
dic
t
ive
m
odel
using supe
rvised
learni
ng
techniq
ue,
”
Int
.
J. Sof
tw
.
Eng
.
Comput
.
S
ci
.
,
vo
l. 1, pp. 75
–
84,
2015
.
[20]
E.
Olkows
ka,
B
.
Kudłak
,
S.
Ts
akov
ski,
M.
Ru
m
an,
V.
Sim
eonov,
and
Z
.
Polk
ows
ka,
“
As
ses
s
m
ent
of
the
wa
t
er
qual
ity
of
Kłodn
ic
a
River
ca
t
ch
m
ent
using
self
-
orga
nizing
m
aps,
”
Sci.
Total
Environ.
,
vol.
476
–
477,
pp.
477
–
484,
Apr.
2014.
[21]
W.
-
P.
Tsai
,
S
.
-
P.
Huang,
S.
-
T.
Cheng,
K.
-
T.
Sh
ao,
and
F
.
-
J.
Ch
ang,
“
A
dat
a
-
m
i
ning
fra
m
ework
for
expl
oring
th
e
m
ult
i
-
relati
on
bet
we
en
fish
spec
i
es
and
wa
ter
qualit
y
t
hrough
self
-
o
rga
nizing
m
a
p,
”
Sci
.
Total Env
iron.
,
vol. 579, pp.
474
–
483,
Feb
.
2
017.
[22]
S.
M.
Sharif,
F.
M.
Kus
in,
Z.
H.
As
ha’
ari
,
and
A.
Z.
Ari
s,
“
Chara
c
te
ri
zation
of
W
at
er
Quali
t
y
Condit
ions
in
th
e
Klang
Rive
r
Ba
sin,
Malay
sia
U
sing
Self
Organi
zi
ng
Map
and
K
-
m
ea
ns
Algorit
hm
,
”
Proce
dia
E
nvi
ron.
Sci.
,
v
o
l
.
30,
pp
.
73
–
78
,
2
015.
[23]
S.
Moham
ad,
“Ge
ner
ating
visual
isation
for
crim
e
sce
ne
inve
st
iga
ti
on
b
ase
d
on
proba
bil
i
t
y
r
e
sult
of
knowledge
-
base
d
s
y
s
te
m
,
”
I
nt.
J. Comput. Appl.
Te
chnol.
,
vo
l.
56
,
no
.
1
,
pp
.
1
8
–
26,
2017
.
[24]
C.
J.
R
al
ph,
S.
Droege
,
and
J.
R.
Sauer
,
“
Ma
nagi
ng
and
m
onit
oring
b
irds
u
sing
point
coun
ts:
standa
rds
an
d
appl
i
ca
t
ions,
p
.
161
–
16
8.
,
”
CJ
R
alph,
JR
Sauer,
S.
Dr
oege
[
te
ch
.
eds.
]
,
Monit.
bi
rd
Popul
.
by
point
count
s.
USD
A
For.
Serv
.
G
en.
Tech.
Re
p
.
PSW
-
GTR
-
149
,
1995.
[25]
C.
Robson,
Fiel
d
guide t
o
the bi
rds
of
South
-
Eas
t
Asia
.
Uni
te
d
Ki
ngdom
:
New Holl
and
Publishers,
2002.
[26]
M.
Strange a
nd
A.
Je
y
araja
sing
a
m
,
Bi
rds
:
a
phot
ographic
guid
e
t
o
the
b
irds of
P
e
ninsular Mal
aysi
a
and
Singapore
.
VV
B
La
ufe
rsw
e
il
er
,
1993
.
[27]
G.
Davison,
A
p
hotographic
gu
i
de
to
birds of
Peninsular Mal
aysi
a
and
Singapore
.
New Hol
la
nd
,
2003.
[28]
T.
S.
W
on
g,
A
Naturali
st’s
Guide
to
the
Bi
rds
o
f
Borneo:
Sabah
,
Sarawak,
Brun
ei
and
Ka
li
mant
an
.
John
Be
auf
o
y
Pub.,
2012
.
[29]
D.
R.
W
e
ll
s,
“
T
he
birds
of
the
T
hai
-
Mal
a
y
Penin
sula.
Vol
.
2
.
Th
e
Pass
eri
ne
B
irds
of
Southern
Tha
i
la
nd,
Peninsular
Malay
s
ia a
nd
Si
ngapor
e.”
Prin
ceton:
Prin
ce
ton
Univer
sit
y
Pr
ess, 2007.
[30]
K.
Chm
el
,
J.
Ri
ege
rt
,
L.
Pau
l,
a
nd
V.
Novotný
,
“
Vert
ical
str
at
if
ic
a
ti
on
of
an
av
ia
n
comm
unity
i
n
New
Guine
an
tropi
c
al
r
ai
nfor
e
st,”
Popul.
E
col.
,
vol
.
58
,
no
.
4
,
p
p.
535
–
547
,
201
6.
[31]
N.
M.
Azm
an,
N
.
S.
A.
La
ti
p
,
S.
A.
M.
Sah,
M.
A
.
M.
M.
Akil,
N.
J.
Shafie
,
and
N.
L.
Khair
uddin
,
“
Avian
dive
rs
i
t
y
and
fe
edi
ng
gu
ilds
in
a
se
conda
r
y
for
est,
an
o
il
pal
m
pla
n
ta
t
ion
and
a
padd
y
f
ie
l
d
in
rip
arian
ar
e
as
of
th
e
Keri
a
n
Rive
r
B
asin, Per
ak,
Ma
lay
si
a,”
T
rop.
life
Sci. R
es
.
,
vol. 22, no. 2,
p.
45
,
2011
.
[32]
J.
V
Rem
sen
Jr
and
D.
A.
Goo
d,
“
Misus
e
of
d
at
a
from
m
ist
-
net
ca
p
ture
s
to
a
ss
ess
rel
at
ive
ab
undanc
e
in
bird
popula
ti
ons
,
”
Au
k
,
pp
.
381
–
398
,
1996.
[33]
R.
Ramli
,
Z
.
Ya
’co
b,
and
R.
Ha
shim
,
“
Diver
sit
y
of
Birds
In
Ke
naboi
Forest
R
e
serve
,
Je
le
bu
,
N
ege
ri
S
embila
n,
Malay
s
ia,”
Mala
ysian
J
.
S
ci.
,
vol. 28, no. 4, pp. 46
5
–
480,
2009
.
[34]
E.
Nor
H
ashim
and
R.
Ramli
,
“
Com
par
at
ive
stu
d
y
of
under
stor
e
y
birds
d
ive
rsit
y
inha
bi
ti
ng
lo
wland
ra
infore
s
t
virgi
n
jungl
e
res
erv
e and
r
ege
n
er
at
ed
fore
st
,
”
Sci.
World
J
.
,
vol
.
2
013,
2013
.
[35]
R.
Gill,
Monit
or
ing
the
status
of
European
and
North
Ame
rican
ce
rvids:
[
fi
na
l
report
for
UNEP
GEMS
projec
t
S
T
4101
-
84
-
02
(
PP
2551)
]
.
UN
EP
,
1990
.
[36]
K.
C.
Ta
na
lgo,
J.
A.
F.
Pineda
,
M.
E.
Agrava
n
te
,
and
Z.
M.
A
m
ero
l,
“
Bird
Diver
sit
y
and
Stru
ct
ure
in
Diff
erent
La
nd
-
use
T
y
p
es
in
Lowla
nd
South
-
Ce
ntra
l
Mindan
a
o,
Phili
ppin
es,
”
Tr
op.
Life
Sci
.
Re
s.
,
vol.
26
,
no
.
2
,
pp
.
85
–
103
,
De
c. 2
015.
[37]
C.
H.
Se
ker
c
iog
lu,
“
Eff
ec
ts
o
f
f
ore
str
y
pra
ctice
s
on
veg
etati
on
struct
ur
e
and
bir
d
comm
unity
of
Kibale
Na
ti
ona
l
Park,
Ugand
a,”
Bi
ol
.
Conserv
.
,
v
ol.
107
,
no
.
2
,
pp
.
229
–
240
,
2002
.
[38]
N.
S.
Sodhi,
L.
P
.
Koh,
D.
M.
Prawira
dilaga,
I
.
Tinulele,
D.
D.
Put
ra,
and
T
.
H.
T
.
Ta
n,
“
La
nd
use
and
conse
rva
ti
on
val
ue
for
for
est birds i
n
C
ent
r
al
Sulawesi
(Indon
esia
),”
Bi
ol
.
Con
serv.
,
vol. 122, n
o.
4
,
pp
.
547
–
55
8,
2005
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Self
-
org
an
izi
ng
map
(
SOM)
for
s
pecies
distr
ibu
ti
o
n modell
ing of
bir
d
…
(
Sa
lw
ana
M
ohama
d @
Asm
ara
)
5243
BIOGR
AP
H
I
ES
OF
A
UTH
ORS
Sal
w
ana
is
a
PhD
holde
r
in
Com
pute
r
Scie
n
ce
.
H
er
rese
arch
int
er
ests
are
dec
ision
suppor
t
s
y
stems
,
knowl
edge
eng
ineeri
n
g,
databa
se
,
si
m
ula
ti
on
and
b
iodi
ver
sit
y
info
rm
at
ic
s.
She
is
cur
ren
t
l
y
a
seni
or
lectur
er
a
t
Fa
cul
t
y
of
Com
put
er
S
y
st
ems
and
Software
Eng
ineeri
ng,
Univer
si
ti
Malay
s
ia
Pah
an
g.
Form
erly
,
sh
e
was
a
postdo
ct
ora
l
res
e
arc
h
e
r
at
Ken
y
i
r
R
e
sea
rch
Inst
it
ut
e,
Univer
siti
Malays
ia
T
ere
ngg
anu
.
Ge
rt
rude
is
a
p
ostgradua
t
e
student
at
Univer
sit
i
Malay
si
a
Te
r
en
gganu.
She
is
cur
ren
tly
pursuing
Master
degr
e
e
a
nd
her
rese
arc
h
i
s
about
birds
spec
ie
s
at
Peninsul
a
r
Malay
s
ia.
Her
rese
arc
h
int
er
est
are
a
ar
e
orn
it
hol
og
y
,
av
ia
n
e
col
o
g
y
,
bird
biod
ive
r
sit
y
and
bird
bio
dive
rsit
y
.
Mohd
Tajuddin
was
el
ec
t
ed
a
nd
induc
t
ed
as
Fell
ow
in
th
e
Aca
dem
y
of
Sci
e
nce
Mal
a
y
s
ia
o
n
the
8th
June
2013.
Resea
r
ch
disc
ipl
ine
-
Zoo
log
y,
Molec
ul
ar
E
co
l
og
y
,
Biodi
ver
si
t
y
and
Prote
ct
e
d
Area
Mana
geme
nt.
In
1970s
-
8
0s
on
the
Su
m
at
ran
rhinos
conse
rva
t
ion.
1990s
-
on
dive
rsit
y
,
distri
buti
on
,
pop
ula
ti
on
,
ec
o
log
y
,
and
spe
cies
con
serva
ti
on
and
th
ei
r
n
at
ura
l
h
abi
t
at
s
in
Peninsula
r
and
Borneo.
20
00s
-
on
m
ole
cu
la
r
e
col
og
y
a
nd
bioge
ogra
ph
y
of
m
amm
al
s
in
Borneo
and
Peninsula
r
Ma
l
a
y
si
a.
2011
-
2
014
on
zoono
ses,
genome
a
nd
m
et
age
nom
i
c
of
pr
imate
s.
2014
-
2018:
res
ea
rch
on
the
Ora
ng
As
li
sus
ta
in
a
bil
ity
;
ca
nop
y
b
i
olog
y
;
Ken
y
i
r
l
a
ndsca
pes;
isla
nds
and
wet
la
nds
in Te
ren
gg
anu, Ma
lay
s
ia.
Wan
Isni
Sofia
h
is
a
senior
le
cturer
at
th
e
Facult
y
of
Com
pute
r
S
y
stems
&
Software
Engi
n
ee
rin
g,
Univer
siti
Mal
a
ysia
Pahang.
She
holds
a
PhD
in
Inform
at
ion
Tech
nolog
y
&
Quantitat
iv
e
Sci
enc
es
,
Master
in
Com
pute
r
Sci
enc
e
a
nd
Networki
ng
and
her
first
d
e
gre
e
in
Da
ta
C
om
m
unic
at
ion
&
Networki
ng
fro
m
Univer
siti
Teknologi
Mara
Sh
ah
Alam,
Mal
a
ysia.
Her
f
ie
lds
of
rese
arch
intere
s
t
are
D
at
a
Com
muni
cation
&
Ne
t
working,
Netwo
rk
Desig
n
&
Alg
orit
hm
,
W
ireless
Sensor
Network
,
Cluste
ring
and
F
uzzy
Logi
c
.
Dan
akorn
is
cur
ren
t
l
y
a
Seni
or
Le
c
ture
r
at
the
Depa
rtme
nt
of
Graphi
c
and
Multi
m
ed
ia
Te
chno
log
y
,
Fa
c
ulty
of
Com
puter
S
y
stems
&
Sof
twar
e
Engi
ne
ering,
Univer
si
ti
M
al
a
y
si
a
Pahang
.
He
ea
rne
d
his
PhD
from
Univer
siti
Te
knolog
i
Malay
s
ia
(UTM
)
in
Educ
at
ion
al
Te
chnol
og
y
an
d
rec
e
ive
d
his
M
aste
r
in
Educ
a
tional
T
ec
hnolog
y
,
Bac
h
el
or
De
gre
e
in
Com
put
er
Scie
n
ce
and
Diploma
in
Comput
er
Scie
n
ce
(
Multi
m
edi
a)
fro
m
UTM.
His
res
ea
rch
cur
iosit
y
f
al
l
s
withi
n
but
n
ot
restr
icted
to
aug
m
ent
ed
rea
l
ity
,
virt
ual
re
alit
y
,
g
rap
hic
and
m
ultim
edi
a,
edu
ca
t
io
nal
te
chno
log
y
,
ani
m
at
ion
,
computer
-
base
d
edu
cation and
instru
cti
onal
d
esign.
Ah
mad
Fi
rdau
s
disti
nct
iv
ely
r
ec
e
ive
d
his
Ph
D
from
Univer
s
ity
of
Mal
a
y
a
,
Malay
s
ia.
He
a
l
so
obta
in
ed
his
Master
s
of
Com
p
ute
r
Scie
n
ce
(
Networki
ng)
from
Univer
sit
y
Te
knologi
Mar
a
,
Malay
s
ia.
He
is
cur
ren
tly
a
sen
ior
le
c
ture
r
at
t
he
Facul
t
y
of
Com
pute
r
Sy
st
ems
and
Software
Engi
ne
eri
ng
at
Univer
siti
Ma
lay
sia
Pahang,
Malay
si
a.
His
a
r
e
a
of
rese
arc
h
i
ncl
udes
Mobil
e
Secur
ity
,
Blo
ckchai
n
and
In
trusio
n
Detect
ion
S
y
st
em.
Evaluation Warning : The document was created with Spire.PDF for Python.