3
$\begingroup$

I have a question about one task: I have two classes w1 and w2 and we are working with 2D space. The density of each of the two classes are the folowing: \begin{equation} p(x = {(x_{1}, x_{2})}'| w_{1}) = \frac{1}{4}\cdot e ^{-\frac{x_{1}+ x_{2}}{2}} \end{equation} \begin{equation} p(x = {(x_{1}, x_{2})}'| w_{2}) = \frac{1}{16}\cdot x_{1}x_{2} \cdot e ^{-\frac{x_{1}+ x_{2}}{2}} \end{equation} How to find the equation of the boundary between the two classes.

$\endgroup$
2
  • $\begingroup$ For finding a Bayes decision boundary, you need the priors of these classes. Are they equally likely? $\endgroup$ – gunes Apr 28 '20 at 15:23
  • $\begingroup$ Yes, I think they are equally likely. $\endgroup$ – Stats8901 Apr 28 '20 at 15:46
2
$\begingroup$

If they're equally likely, you'll just solve $p(x|w_1)=p(x|w_2)$ to find the decision boundary, which yields the curve $x_1x_2=4$.

$\endgroup$
6
  • $\begingroup$ Thanks, but if they are not equal we have p(x|w1)p(w1) = p(x|w2)p(w2) which I can't solved? $\endgroup$ – Stats8901 Apr 28 '20 at 16:21
  • $\begingroup$ you can still solve it when you substitute the values. $\endgroup$ – gunes Apr 28 '20 at 16:22
  • $\begingroup$ How? Could you give some hint? $\endgroup$ – Stats8901 Apr 28 '20 at 16:26
  • 1
    $\begingroup$ $$x_1x_2=4p(w_1)/p(w_2)$$ $\endgroup$ – gunes Apr 28 '20 at 16:27
  • $\begingroup$ And this is not the equation, I guess? We must find a way to find the p(w1) and p(w2) ? $\endgroup$ – Stats8901 Apr 28 '20 at 16:31

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.