Variable-order Markov model: Difference between revisions

Revision as of 09:15, 21 April 2007

Revision as of 09:17, 21 April 2007

Line 35:

In practical settings there is seldom sufficient data to accurately estimate the [[exponentexponential growth|exponential]] growing]] number of [[conditional probability]] components as the order of the [[Markov chain]] increases.

The Variable Order Markov model assumes that in realistic settings, there are certain realizations of states (represented by contexts) in which some past states are independent from the future states, accordingly, ''a great reduction in the number of model parameters can be achieved''.