
A HMM is a Hidden Markov Model. Markov models are often used to model observation sequences. The fundamental assumption in a markov model is that the probability of an observation (event) can only depend on the previous observation. A HMM can be visualised as a graph with a number of states. If states are connected they have line connecting them. The following picture shows a HMM with two states, labeled "Rainy" and "Sunny". Each state can emit three symbols (these are not visible in the graph).
For an introduction into HMM's see Rabiner (1989).
© djmw, April 10, 2013