An Incremental Error Correcting Evaluation Algorithm For Recursion Networks without Circuits
A recursion network consists of a directed graph together with an evaluation function v on the set of nodes of the graph, such that the value of every nonterminal node recursively results from the values of its successors. An example is a game graph with a grundy-function. Consider the problem to determine the value of a specified node ‘root’, when instead of v only an erroneous estimation function \( \hat v\) on the set of all nodes is given. (Typically v and \( \hat v\) will coincide in many, but not in all nodes.)
This paper presents a simple algorithm which yields the correct value v (root), if on every longpath with starting node ‘root’ the estimates \( \hat v\) provide “more correct than wrong information” about the v-values. A longpath is a path that cannot be prolonged.
KeywordsDirected Graph Rooted Tree Terminal Node Game Graph Wrong Information
Unable to display preview. Download preview PDF.
- 1.I. Althöfer, An incremental negamax algorithm, 1987.Google Scholar
- 2.R. Gellman, S. Dreyfus, “Applied Dynamic Programming”, Princeton University Press, Princeton (New Jersey), 1962.Google Scholar
- 5.J. Pearl, “Heuristics-Intelligent Search Strategies for Computer Problem Solving”, Addison-Wesley, Reading (Massachusetts ), 1985.Google Scholar
- 7.W.W. Peterson, E.J. Weldon Jr., “Error Correcting Codes”, 2. ed., MIT Press, Cambridge (Massachusetts ), 1972.Google Scholar