Interference of waves based usage of an optimization algorithm for finding rules in an agent system
Tomáš Cádrik, Marián Mach, Peter Sinčák · 2014
There are many controllers using algorithms based on neural networks, evolutionary classifier systems and others, which are saving the discovered data in the form of a model characteristic to this algorithm. If we want to use it in the agent domain, we need to have this method in the agent to interpret the data. With the extension of using cloud with a combination of robotics, it is not very effective to use this approach. So we need to build a method which can extract the data in the form of rules from the algorithms and then we can use only the extracted rules. This paper introduces a method which extracts rules from the interaction of the learned model of Zeroth Level Classifier System with the environment. The new method uses an interference of waves inspired approach based on a simple evolutionary algorithm.