AIC / BIC

定義

最尤推定量\(\hat{\theta}_{(x_1, \dots, x_n)}\)について

\[\begin{split} \begin{align} AIC &:= \sum^n_{i=1} - \log p(x_i | \hat{\theta}_{(x_1, \dots, x_n)} ) + d\\ BIC &:= \sum^n_{i=1} - \log p(x_i | \hat{\theta}_{(x_1, \dots, x_n)} ) + \frac{d}{2} \log n\\ \end{align} \end{split}\]