?
EBc dSüDZE
? 11.1 ?y
? 11.2 ??0"¥s ?ZE
? 11.3 ?Ys ?¥W¤ZE
? 11.4 s)
?ZE
? 11.5 
e
?ZEeo
? 11.6
??¥ù5
11.1 ?y
11.1 ?y
? μSü"'
?YX?T
s ?
[ p?
é?s ?
? dSü"'
?Y??
n51 p
"'"é?s ? (
? )
? dSüDZEvás1
v ?
? ?à
q
áf
9¥ZE
? ?"'WM
?
¥ZE
11.1 ?y
Hierarchical
Clustering
sa
?
Hierarchical
Clustering
sa
?
Partitional
Clustering
s v
?
Partitional
Clustering
s v
?
K-Series
K-Series
PCA/SVD based methods
PCA/SVD based methods
PDF based methods
PDF based methods
Graph Theory based methods
Graph Theory based methods
Top-down Division
Top-down Division
Bottom-up Agglomerative
Bottom-up Agglomerative
Other methods
Other methods
11.2 ??0"¥s ?ZE
11.2 ??0"¥s ?ZE
?
±Xü+? bWs1 ??? u×
? u×
à
q
áf
^??¥
??? u×?B? ?
? ?/me?
U è x=0 y=0?
??
ü
V[ü (x,y)
ü
s?
1??? u×
11.2 ??0"¥s ?ZE
11.2 ??0"¥s ?ZE
? g?ZE
? ? bW?°¤s??? u×1? ?
4
7B? bW?51?e?b
? ùsB?US"d?"d/
¥

q
áf
V[¨H?à
q
áV
Ub
? ?T
H?à
q
áf
?C??
T5
NUSà

B?Tséb
11.2 ??0"¥s ?ZE
? g?ZE
?
? B? bW?¥??s ?
}|{ SyyuS
uS
T
ii
i
∈=

¥g?1"' "
 S
i
?¨°ZmZE9à
q
áf
s?à
q
áf
¥?[#?-W¥!?
!?)S<°? u
i
¥?
ü

sé
11.2 ??0"¥s ?ZE
11.2 ??0"¥s ?ZE
? g?ZE
?
? ??
a¥US"d u
i
? ?
T¥ZE
Pg? {u
i
T
y }¥ZμKvZμ
v ?-Ws ?¥?9 V
v
?
@?"1 p¥ u
i
^"'xZμ ?¥Kv+?
′?¥+?_

? iù5?"ê4¥ u
i

Hi?

3?
¥H?
áf
11.2 ??0"¥s ?ZE
? g?ZE
?
?
E??
1.9
"'xZμ?¥Kv+?′?¥+?_

u
i
ü"'
g?? u
i

2.¨°ZmE pH?à
q
áf
3.s?H?à
q
áf
¥ò?!??t!?
T<°? u
i
¥?
ü

s?+?0"
4. ?T
àμ!?5¨/B?Kv¥+?′}9
5.
¤?¥ò?0"é?]"¥V?°à
?
0"?
^??1?
11.2 ??0"¥s ?ZE
? ??0"s ?¥Y}
E
? I
n
SμB?s
)|()|(
||,||
,
1
i
i
i
i
iiii
i
c
i
yp
N
N
yf
SNNN
S
Γ=Γ
===ΓΓ
Γ=

=
1F ?¥ ?Hqà
q
á
o?M? O ?
U
11.2 ??0"¥s ?ZE
? ??0"s ?¥Y}
E
?
?
?0"
?-W¥,  ?,
[]
[]
Kv
"S
P

∑∑

==
Γ?Γ=
Γ?Γ
dyypyfyfJ
dyypyfyf
c
i
c
j
ji
ji
)()|()|(
)()|()|(
11
2
2
11.2 ??0"¥s ?ZE
? ??0"s ?¥Y}
E
?
? I
nü y
k
V Γ
i
?M? Γ
j
?/? J ¥?M
^9 ? )(,),,(
1
)()]|()|([2
)(]2[
2
KyyK
N
ff
dyypfyfyfc
dyypfcJ
kji
iji
i
==?
Γ?Γ+
=?


11.2 ??0"¥s ?ZE
? ??0"s ?¥Y}
E
?
? ?B[v? 0
? ?=[?1 | %?
μv?Jv
? y
k
??V Γ
j
?M?
P Kv¥ Γ
i
?
)|()|(
ji
yfyf Γ?Γ
)|(
i
yf Γ
11.2 ??0"¥s ?ZE
? ??0"s ?¥Y}
E
?
?
E??
1. Sê?B?
Ss
2. S?
B?? y9
iü y×?
s
¥?
P Kv¥ ??b
3. ?Tμ ??é?
 ?Y¥M
*
1×ˉ
B
??b
)|(
i
yf Γ
)|(
i
yf Γ
11.3 ?Ys ?¥W¤ZE
11.3 ?Ys ?¥W¤ZE
?
?1?
? "'WM
?¥

? 5f

?é
¥
YS
?
"S
? ?
=í
íM
?ú
? ?Wí
íM
??
11.3 ?Ys ?¥W¤ZE
? C- (′
E
K-Means,K- (′ )
? Klμ
üZ5 m
i
^ ?
= (′
? L
!Xüμ

Ssü Γ
k
?¥B?í
í
yM? Γ
j
?
H J¥?M

∑∑
=Γ∈
=
c
iy
i
i
myJ
1
2
||||
22
||||
1
||||
1
j
j
j
k
k
k
my
N
N
my
N
N
J?
+
+?
=?
11.3 ?Ys ?¥W¤ZE
? C- (′
E
?
?
E??
1.ê4B?
Ssi9
ò ? (′
2.ê4B?"' y
!ê? Γ
k
?bsY9
ü y
M? 
eò ??/?¥?J
3. ?
μ?J ?v? 05?M? yb?5M? y ?

3Kl?J ¥ ?
4.÷?M1 ?¥ (′[# J ′b
5. ? ??Y} N Q J?M5T?b?52b
11.3 ?Ys ?¥W¤ZE
? C- (′
E
?
? 
Ss¥ ??
? 
S}V?
¥ ??
?
:??¥s

? C- (′
Ei?
M?
? 
Ss¥ZE
? ÷? (′¥
H
?
?
"¥?
%?
? ??
11.3 ?Ys ?¥W¤ZE
? C- (′
E
?
?
HWˉ1 O(N)
? e?^
LC
?
a¨?, o?,s?¥

? B?
Ss
ù?
11.3 ?Ys ?¥W¤ZE
? ?"'M
?
¥
?
E
? ?¨B?,, ?}VB? ?
? K
j
V[
^B?f
B??"??
? "'-WM
?¥

? 5f
Kl¥
"S
),(
j
Ky?
∑∑
=Γ∈
=
c
jy
j
j
KyJ
1
),(
11.3 ?Ys ?¥W¤ZE
? "'M
?
¥
?
E
?
?
E?? ?
? C- (′
1.ê4
Ssi9

S
2.×?s
¥ò?"'
3.??i×ˉ 2-3°à
l ?
? C- (′
E =[ ? (′1
x
f  ?T1"'
-W¥M
?

),(min),(
k
k
jj
KyKyy?=?Γ∈ ?T
11.3 ?Ys ?¥W¤ZE
? "'M
?
¥
?
E
?
?
E
l ?¥ sHq5f
J
@
? ??- -¥s ??¥
? ??-a¥s ??¥
)
~
,()
~
,
~
(),,()
~
,( ΚΓ≤ΚΓΚΓ≤ΚΓ JJJJ
*
1 ?T
K,Γ
K
~
,
~
Γ
11.3 ?Ys ?¥W¤ZE
? "'M
?
¥
?
E
?
? ?
f

a¨?ò ?1?
s?
|
|log
2
1
)(
)(
2
1
),
)
,(
)}(
)(
2
1
exp{
|
|)2(
1
),(
1
1
2/12/
jij
T
ij
jjj
ij
T
i
j
d
jj
mymyKy
mV
mymyVyK
Σ+?Σ?=?
Σ=
Σ
Σ
=

M
?

Vò ?"'?9?
"
π
11.3 ?Ys ?¥W¤ZE
? "'M
?
¥
?
E
?
? ?àf

a¨?ò ?"'"?ò1¥
?àZ_
¥0 bW ú¥ f ?
 ?
^"'??à0 bW¥
?_
"d?Kv+?′?¥+ -
?"'xZμ?¥
^?
)]()[(
)]()[(),(
),...,,(
),(
21
j
T
jjj
T
j
T
jjjj
j
dj
T
jj
myUUmy
myUUmyKy
d
juuuU
yUVyK
j

=?
=
=
11.3 ?Ys ?¥W¤ZE
? "'M
?
¥
?
E
?
? V[
Eò??¥s?
? 31?5??
¥s??
11.3 ?Ys ?¥W¤ZE
? í
#f
5
E
? í
#f
"'WM
?¥

?T y
i
^ y
j
¥? I?í
# y
j
^ y
i
¥? K?í
#
? í
#f
P¤
áMí¥? ?^
?B ?
jiKIa
ij
≠?+=,2
11.3 ?Ys ?¥W¤ZE
? í
#f
5
E
?
? Be?v
x
f  ?5 ??B1?B ?
? ?ví
#f
′ ??B1?= ?b?
^
1? ?¥
?B ?
?= ?
11.3 ?Ys ?¥W¤ZE
? í
#f
5
E
?
? ]B ??¥?-Wi, ?¤,b ?¤
>ü
?l1
?-W¥í
#f
b
? B?? 1
&¥ ?¤
>?l1 α
ii
=2N
[ò@oμB??¥
?
? ?] ?¥??i ?¤ ?¤
>1 α
ij
=0
? 9 ?
=
>
∑∑
==
=
N
i
N
j
ijwithin
L
11
α
ij
a
11.3 ?Ys ?¥W¤ZE
? í
#f
5
E
?
? ? i ?? j ?-W¥Klí
#f
′?l1
? ? i ?
=Kv ?¤
>:1 α
imax
? ? i ?D? j ?-W¥ ?¤
>?l1 β
ij
)(min
,
kl
yy
ij
a
jlik
Γ∈Γ∈

11.3 ?Ys ?¥W¤ZE
? í
#f
5
E
?
? β
ij
¥
!9
"S
^ ?T
 ?W¥Klí
#′l? ?
BZ¥ ?
=¥Kv ?¤
>
H
>}Nü
^?
¥V7?? I
nü?
 ?i
≤≤++
≤>+
>≤+
>>?+
=
maxmaxmaxmax
maxmaxmax
maxmaxmax
maxmaxmaxmax
,,
,,
,,
,)],()[(
jijiijjiij
jijiijjij
jijiijiij
jijiijjijiij
ij
if
if
if
if
αγαγααγ
αγαγαγ
αγαγαγ
αγαγαγαγ
β
11.3 ?Ys ?¥W¤ZE
? í
#f
5
E
?
? 9 ?W
>
? 5f


=
ji
ijbetween
L β
betweenwithin
LLJ +=
11.3 ?Ys ?¥W¤ZE
? í
#f
5
E
?
?
E??
1.9
 ? ?
2.¨  ? ?9

# ? M
ij
 M
ij
V
U y
j
^ y
i
¥
?+?í
#
3.9

#f
? L
ij
=M
ij
+ M
ji
-2I,L
ii
=2N
4. L ?
??D Kí
# ?¤??
S¥
s
5.
? ?9
γ
ij
α
imax
 α
jmax
o1 γ
ij
l
? α
imaxa
α
jmax
?¥ ?B?üi
 ?
y ?
?¤b×ˉà
àμ?¥ ?¤?
31?
),(
jiij
yy?=?
11.4 s)
?ZE
11.4 s)
?ZE
?
2¥
?
y
1
,y
2
,y
3
,y
4
,y
5
y
1
,y
2
,y
3
y
4
,y
5
y
1
,y
2
y
1
y
2
y
3
y
4
y
5
11.4 s)
?ZE
? 1?_/??sé
? 1?_
??i
? S
>f
b
B?sé?i
?1
P
>Kl
11.4 s)
?ZE
? 1?_
??i
E
1.
?"'?ò1??B ?
2.VXμ¥ ??G
?i?B ?b$
i¥
 ?
^
P
>f
Kl¥
? ?
3.×ˉ 2°à??c
μ"'¥ ?b
?
>f
?lü
? ?i¥
>
11.4 s)
?ZE
? ?]ZE¥
>f
Method Cost Function
Single-Link
Average-Link
Complete-
Link
),(min
,
ji
SxSx
xxd
jjii
∈∈
∑∑
∈∈
iijj
SxSx
ji
ji
xxd
ss
),(
1
),(max
,
ji
SxSx
xxd
jjii
∈∈
11.5 
e
?ZEeo
11.5 
e
?ZEeo
? Mixture Models
? L
!

^V k???s? E
1
,E
2
,...,E
k
3?¥b E
i
¥

q
áV
U1
? ¨V
Uy
r
^? E
i
3?¥à
q5
E¥
"¥
^K
v
?f
? B???
?f
íE°¤ pKv′b
? ¨ EM
E (Expectation-Maximum)3 %? ?ù5
)|( θyp
i
i
r
τ
∏∑
= =
=
n
r
k
i
ri
i
r
ypL
1 1
)|(),( θττθ
11.5 
e
?ZEeo
? ?m
¥
?ZE
?
?
?m
B?b
? ?
-W¥M
?·?B??′ V[
y ?m
¥H
? ?¨m
?¥Bt ?
?mé?s
? èKl
3?
Es?m¥Kl
3?

?aü
Ké¥H ????
? ?b

?0m? ??KéH …GQé?/ ?
11.5 
e
?ZEeo
? ?m
¥
?ZE
?
? èú ?Y0m
E
? Hé"B?m G¥Hé" (edge cut set)?l1H
¥"V G? ???"?¥
μH-a¤
?¥m? ?YbcK
H¥Hé"?1 G¥K
lHé"
?S MinCutb
? H ?Ym G¥H ?Y¨ k(G)V
U?l1
KlHé"¥vlb ?T k(G)=s5 G?1 s-H
?Ymb
11.5 
e
?ZEeo
? ?m
¥
?ZE
?
? ?lB? μ n???¥m G
^,ú ?Y,¥ ?
T k(G)>n/2
?
E??
1. ?"m?
?¥?b
2.s?m¥KlHé"
μ?
E?Ns?
?
0mb
3. ?TKlHé"¥H
v?m??
¥B?5Nm
^ú ?Y¥bNmü
^B? ?b?5??¥

?0m×ˉ 2-3b
4.üe5 ?"?¥??B?5s
¥?ò ?? ?
11.5 
e
?ZEeo
? ???ss¥ZE
? ? s′s3¥ZE
? ? ?¥sé×
¥ZE
? ? |y
*ü?
¥ZE
? ?L.
E¥ZE
? ……
11.6
??¥ù5
11.6
??¥ù5
?
?
^B?
_ 8
 8?¨¥
?
? ??]¥
?]¥
"Sê4?]
¥
?
E
?
E
(scale)
ù?¥ù5
? ? |
I
1"¥M
?
$
¥ò?
+?
^ V1?¥
$
? "
?5·?8
1 ??s ?¥
"$
? IóD
? [1] A,K,Jain,M,N,Murty,P,J,Flynn,Data clustering,A Review,
ACM Computing Surveys,September 1999,Volume 31 Issue 3,
? [2] Erez Hartuv,Armin Schmitt,Jorg Lang,et al,An algorithm for
clustering cDNAs for gene expression analysis,In Proceedings of
the Third Annual International Conference on Computational
Molecular Biology(RECOMB 99)
? [3] Daniel Fasulo,An Analysis of Recent Work on Clustering
Algorithms,Technical Report 01-03-02,University of Washington,
April 1999
? [4] Jeffrey D,Banfield and Adrian E,Raftery,Model-based gaussian
and non-gaussian clustering,Biometrics,49:803-821,September 1993.