Monte Carlo method for problems of hierarchical inference
Los Alamos Scientific Lab., N. Mex. Indiana Univ., Bloomington. Dept. of Physics (US), Christopher N Everett · 1975
A simple Monte Carlo method is presented for estimating the probability p(Sx.•••,Sr|h) of the state vector (Si,...,Sr) of the K terminal (data) nodes in a general inference tree (with an arbi trary number of nodes, each having a prescribed number of possible states) assuming the "hypothesis" node to be in state h.From this information, the probability p(h|Si,...,Sr) of hypothesis h, assuming an observed state vector (Si,...,Sr) at the terminal nodes, is then computed from Bayes' formula.The whole routine is easily coded, and a flow chart is included.I.