Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I want to do inference in a Hidden Markov Modell (Gaussian Mixture), given observed continoues variables $Y$ and latent discrete variables $X$. For this I need to compute the probability of an observation message $\mu_{Y \rightarrow X}(x_t) =: \varrho_t)x_t = P(y_t|x_t)$.

But how can I obtain the probability $P(y_t|x_t)$ for some given time $t$? I only have the values of $Y$, the (actually hidden) corresponding values of $X$, and a transition probability table $P(x'|x)$ but no probabilities what so ever for $y_t$.

How can I approach this?

share|cite|improve this question
up vote 1 down vote accepted

Since X is discrete and your Y is a Gaussian mixture, the most simple thing to assume is that $$y_t | x_t = a \sim N(\mu_a, \sigma_a^2).$$

Now, given a sequence of observations and hidden states $(x_t, y_t)_{t=1}^T$, you can estimate $P(y_t|x_t = a)$ using the maximum likelihood principle. That is, take all times t such that $x_t = a$, ${\cal T}_a = \{t : x_t = a\}$, and estimate $(\mu_a, \sigma_a)$ from corresponding $\{y_t\}_{{\cal T}_a}$. Perform this for all possible values that X can take on.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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