6. ベイズ推論の例
ある両親から3人連続して女の子が生まれた.
次に男の子が生まれる確率は?
6
P 0,0,0{ }| p( )= p
xk∑
1− p( )
n− xk∑ = 1− p( )
3
P p( )=
p0
1− p( )
0
B 1,1( )
=1 B α,β( )= pα−1
1− p( )
β−1
dp
0
1
∫
P p | 0,0,0{ }( )=
1− p( )
3
B 1,4( )
B 1,4( )= 1− p( )
3
dp
0
1
∫ = −
1− p( )
4
4
0
1
=
1
4
E p[ ]= p⋅ 4 1− p( )
3
dp
0
1
∫ = −p 1− p( )
4
0
1
+ 1− p( )
4
dp
0
1
∫ =
1
5
p y x( )=
p x,y( )
p x( )
=
p x y( )p y( )
p x y( )p y( )dy∫
7. ヒトもベイズ推論している
A Bayesian approach can contribute to an
understanding of the brain on multiple levels, by
giving normative predictions about how an ideal
sensory system should combine prior knowledge
and observation, by providing mechanistic
interpretation of the dynamic functioning of the
brain circuit, and by suggesting optimal ways of
deciphering experimental data. Bayesian Brain
brings together contributions from both
experimental and theoretical neuroscientists
7(Doya+ 2006)
13. 混合ガウス分布 (GMM)
13
p y θ( )= π j N y µk ,Σk( )
k=1
K
∑
N y µ,Σ( )=
1
2π( )
D
2 Σ
1
2
exp −
1
2
y − µ( )T
Σ−1
y − µ( )
14. GMMのベイズ推定
共役事前分布を導入
14
p y θ( )= π j N y µk ,Σk( )
k=1
K
∑
NormalInvWishart µk ,Σk µ0,Σ0,κ0,ν0( )
Dir π1,...,πK
α
K
,...,
α
K
(www.singularpoint.org/blog/r/inverse-wishart-graphic)