How to do this step?

56 Views Asked by At

This is a slide from a Markov model lecture. I am studying probability in the context of Computer Science, namely, Markov models for traditional AI learning.

I got stuck following the conditional probability equations in the linked slide, when trying to work out the simplifications myself.

I tried to condition the inner 3 joint events using $(x_t | x_{t-1}, e_{1:t})$ but I couldn’t get it to the form shown.

$x_t$ is the Hidden Markov model state and $e_t$ is the observed effect.

Q: How do they do this conditional expectation simplification step? (see red lines I marked in slide image).

Thanks!

see image

P.S. my rep not high enough to post image directly, sorry

1

There are 1 best solutions below

9
On BEST ANSWER

Working with the expression under the sum and using Markov property and rule of conditional probability: $$ P(x_{t-1}, x_t, e_{1:t}) = P(x_{t-1}, x_t, e_1,e_2,...,e_t)= P(x_{t-1}, e_{1:t-1}) P(x_t, e_t | x_{t-1}, e_{1:t-1}) = P(x_{t-1}, e_{1:t-1}) P(x_t|x_{t-1}, e_{1:t})P(e_t|x_t, x_{t-1},e_{1:t-1}) = P(x_{t-1}, e_{1:t-1})P(x_t|x_{t-1})P(e_t|x_t) $$