Proposal of a function discovery system model using non-linear optimization method

Abstract

In this paper, we propose a new function discovery system model (S-Poly model) for the artificial life type discovery system (S-System), by incorporating non-linear optimization technique in it. This model uses Polytope method for optimizing the values of the constants of the function under discovery process while evaluating the fitness of it with the observation data. The new model significantly reduces the number of generation required for the discovery. Also. we propose two other models combining Polytope with Genetic Programming (GP-Poly model) and mutation (Mut-Poly model). All these models are found effective in the function discovery, while S-Poly model is found to be the fastest among the three.