The research sets the given goals:
• Analyzing the distributed genetic programming algorithm and the
experimental results of other researchers that attempt to set the optimal
parameters of the algorithm.
• Analyzing the swarm intelligence application techniques and possible
solutions to adapt them for governing the migration parameters of the
distributed GP.
• Choosing the most suitable swarm intelligence algorithm and blending
both distributed GP and swarm intelligence algorithms together to
control migration at GP algorithm runtime.
• Creating an implementation of the proposed method and using it to
generate programs automatically for simple tasks with known optimal
solutions.
• Examining the influence the swarm management rules make to the
genetic programming algorithm and comparing it with the dynamics of
the particle swarm that is led by similar rules. Also comparing the
processes in the improved genetic programming algorithm and the
ones of the GP algorithm with optimal manually chosen parameters of
distribution.
• Collecting the empirical results about the performance of the
distributed GP algorithm with adaptive migration control andcomparing them with the performance of standard genetic
programming and distributed GP algorithms. Conclusions about the
applicability and perspectives of the new migration control method
will be drawn based on the results of the assembled results. |
