- Hidden Markov Model(HMM)
- a simple representation of a stochastic process
- the hidden state of the process is represented by a single
state variable at each point in time
- the observation is represented by an observation variable
- the probability over the state sequence can be decomposed as
follows
- Bayesian Belief Network
- a general way of representing joint probability distribution
with chain rule and conditional independence assumptions
- it allows for an arbitrary set of hidden variables, with
arbitrary conditional independence assumptions
- the probability distribution over a set of random variables
is given by
- different representations