next up previous
Next: Functions of random variables Up: Appendix Previous: Entropy

Mutual information

Mutual information is defined as

$\displaystyle I(X,Y)$ $\textstyle =$ $\displaystyle H(X)+H(Y)-H(X,Y)$  
  $\textstyle =$ $\displaystyle E\{ -log\;\;p(X)\}+E\{ -log\;\;p(Y)\}+E\{ -log\;\;p(X,Y)\}$  
  $\textstyle =$ $\displaystyle E\{ log\;\frac{p(X,Y)}{p(X)\;p(Y)} \}$  

Mutual information measures the amount of information shared between the two random variables $X$ and $Y$. Since

\begin{displaymath}p(X,Y)=p(X\vert Y) \; p(Y)=p(Y\vert X) \; p(X) \end{displaymath}

we have
$\displaystyle I(X,Y)$ $\textstyle =$ $\displaystyle E\{ log\;\frac{p(X,Y)}{p(X)\;p(Y)} \}$  
  $\textstyle =$ $\displaystyle E\{ log\;\frac{p(X\vert Y)}{p(X)} \}=H(X)-H(X\vert Y)$  
  $\textstyle =$ $\displaystyle E\{ log\;\frac{p(Y\vert X)}{p(Y)} \}=H(Y)-H(Y\vert X)$  



Ruye Wang 2018-03-26