Evolution of an Effective Brain-Computer Interface Mouse via Genetic Programming with Adaptive Tarpeian Bloat Control

The Tarpeian method for bloat control has been shown to be a robust technique to control bloat. The covariant Tarpeian method introduced last year, solves the problem of optimally setting the parameters of the method so as to achieve full control over the dynamics of mean program size. However, the theory supporting such a technique is applicable only in the case of fitness proportional selection and for a generational system with crossover only. In this paper, we propose an adaptive variant of the Tarpeian method, which does not suffer fromthis limitation. Themethod automatically adjusts the rate of application of Tarpeian bloat control so as to achieve a desired program size dynamics. We test the method in a variety of standard benchmark problems as well as in a real-world application in the field of Brain Computer Interfaces, obtaining excellent results.