# Estimating Hidden Markov Models (HMMs)

## Estimating Hidden Markov Models (HMMs)

### Recap of Hidden Markov Models (HMMs)

We donâ€™t see state

Each state produces a visible output. this output is drawn from a distribution for each state.

We observe a sequence of outputs, not states.

### Estimating HMMs with the Viterbi algorithm

Assume we know transition matrix. and starting probls

Given we observe sequence of outputs, what were most likely actual paths?

Virbiti returns this

### Estimating HMMs with the forward algorithm

Given we have observed outputs, what is the chance of being in a certain state at a certain time?

### Estimating HMMs with the forward-backward algorithm

We calculate state x at time t given all obs.

### Baum-Welch algorithm

### Kalman filters