In this chapter, we introduced algorithms for doing inference over our HMM models. We looked at the forward-backward algorithm to do predictions for our hidden states given the observations. We also discussed the Viterbi algorithm, which is used to compute the most probable states in our model.
In all these algorithms, we assumed that we knew the transition and the emission probabilities of the model. But in real-world problems, we need to compute these values from the data. In the next chapter, we will introduce algorithms for computing transition and emission probabilities using the maximum-likelihood approach.