Genetic programming using adaptable stochastic flow of control
Artur Rataj · Theoretical and Applied Informatics · 2010
This paper discusses a replacement of a typical, deterministic flow of control in the evolving program with a probabilistic one, which is then adapted using a supervised learning, until it hopefully converges back to a deterministic flow of control. The probabilistic flow of control allows for a continuous modification of the evolved program, which internally can consist of a set of competing paths of execution.