GPの木構造表現のためのノード突然変異

概要

論文の詳細を見る
In Genetic Programming(GP) programs or functions are evolved by means of a Genetic Algorithm(GA) in order to solve a given problem. The most frequently used representation of programs in GP is the tree representation whose syntax is similar to a Lisp s-expression : one non-terminal node represents a function, and branches from that node point to the arguments of the function. A leaf, or terminal node, represents a constant or a variable. The two genetic operators, crossover and mutation, used to recombine programs are different from the genetic operators of GA. Crossover swaps two sub-trees of two programs, and mutation delete one sub-tree of a program and replace it by an enterely new sub-tree. On the contrary in GA, mutation replaces only one gene by another. Thus there is a need to define a new mutation for GP that would replace one node's label by another : a node-mutation.
社団法人電子情報通信学会の論文
1996-03-11