1.最大似然函数定义
Y={y1,y2,…yn}
p(y1,y2,…yn)=p(y1)p(y2)…p(yn)
即y1,y2,…yn为独立同分布
似然函数:
likelihood= ∏ j = 1 N p ( y i ) \prod_{j=1}^{N}p(y_i) ∏j=1Np(yi)
L δ = ∏ j = 1 N p δ ( y i ) L_\delta=\prod_{j=1}^{N}p_\delta(y_i) Lδ=∏j=1Npδ(yi)
L δ = ∏ p δ ( y j ) = ∏ j = 1 N ∑ z p δ ( y j ∣ z ) p δ ( z ) L_\delta=\prod p_{\delta}(y_j)=\prod_{j=1}^{N}\sum_zp_{\delta}(y_j|z)p_{\delta}(z)