## Neural Computation

August 1, 2002, Vol. 14, No. 8, Pages 1929-1958
(doi: 10.1162/089976602760128072)
© 2002 Massachusetts Institute of Technology
Bayesian A* Tree Search with Expected O(N) Node Expansions: Applications to Road Tracking
Article PDF (711.55 KB)
Abstract

Many perception, reasoning, and learning problems can be expressed as Bayesian inference. We point out that formulating a problem as Bayesian inference implies specifying a probability distribution on the ensemble of problem instances. This ensemble can be used for analyzing the expected complexity of algorithms and also the algorithm-independent limits of inference. We illustrate this problem by analyzing the complexity of tree search. In particular, we study the problem of road detection, as formulated by Geman and Jedynak (1996). We prove that the expected convergence is linear in the size of the road (the depth of the tree) even though the worst-case performance is exponential. We also put a bound on the constant of the convergence and place a bound on the error rates.