Timeline for Why maximizing the expected value of log likelihood under the posterior distribution of latent variables maximize the observed data log-likelihood?
Current License: CC BY-SA 4.0
3 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Sep 27, 2020 at 22:50 | vote | accept | Dibya Prakash Das | ||
| Sep 27, 2020 at 22:49 | comment | added | Dibya Prakash Das | I also realize now that the last equation in the question $∑p(Z|X,θ^{old})\ln p(Z|X,θ)= −(KL(q||p) + const)$ can be rearranged and written as $∑p(Z|X,θ^{old})\ln p(Z|X,θ) + KL(q||p) = - const$ and so we see that maximizing the first term is equivalent to minimizing the second term and thus after getting the new improved estimate for $q(Z)$, $KL(q||p)$ value goes down and makes $\ln p(X|\theta)$ increase as we originally intended. | |
| Sep 27, 2020 at 22:37 | history | answered | Dibya Prakash Das | CC BY-SA 4.0 |