Genetic algorithms and trees: part 1: recognition trees (the fixed width case)

All available information about the physical world, is a result of measurements. So, every object or process can be described by an array of attributes. It is therefore important to be able to classify and recognize these defining arrays as efficiently as possible. In order to realize this, we introduce a formal context for this problem and a genetic algorithm on decision trees. The formal model includes the relative cost of different measurements and the frequency of occurrence of the objects to be classified. It enables us to define schemata for the genetic algorithm and to prove an equivalent of the classical Schema Theorem. Also included are some theoretical consequences of applying the algorithm to recognition of bitmaps