1
$\begingroup$

From Kullback-Liebler divergence of matrix factorization; \begin{equation*} \mathrm{X}\approx\mathbf{WH} \tag{1} \end{equation*}

\begin{equation*} d_{\mathrm{KL}} (\mathbf{X}\ \vert\vert \mathbf{WH})=\sum_{f,t}\left(X_{f,t}\log\frac{X_{f,t}}{WH_{f,t}}-X_{f,t}+WH_{f,t}\right) \tag{2} \end{equation*}

\begin{equation*} \stackrel{\hbox{cst.}}{\hbox{=}} \sum_{f,t}-X_{f,t}\log\sum_{k}{WH_{f,t}} + \sum_{f,t}\sum_{k}{WH_{f,t}} \tag{3} \end{equation*}

How is constant in $(3)$ derived from $(2)$? Why does $X_{f,t}\log{X_{f,t}}-X_{f,t}$ disappear?

Reference (page 77)

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.