Population-Based Automatic Programming - Part 2: Performance on Some Example Problems.
J.A. Schoonees, E. Jakoet · 1997
A new automatic programming paradigm, called population-based automatic programming (PBAP) was proposed in Part 1 of this paper. The algorithm is in the style of genetic programming, but uses population-based incremental learning (PBIL) instead of the genetic algorithm. It produces a computer program by intelligent stochastic search of the space of all possible programs with a given set of functions and terminals. The new method was tested on four example problems from the original work on genetic programming by Koza: from optimal control, robotic planning, symbolic regression, and the Boolean multiplexer problem respectively. 1