SCOT modeling, training and Homogeneity testing

Joint work with M. Malyutov, P. Grosu and T. Zhang.
We call Stochastic COntext Trees a n-Markov Chains with every state of a string independent symbols in its more remote past than the context of length determined by the preceding symbols of this state. Its stationary distribution over contexts is iteratively evaluated explicitly in several examples. We analyze several models viewed as simplified approaches to financial modeling. Our publication is on http://www.jip.ru/2014/275-283-2014.pdf