Autonomous Hybridization of Agent-Based Computing.

Using agent-based systems for computing purposes, where agent becomes not only driver for realizing computing task, but a part of the computing itself is an interesting paradigm allowing for easy yet robust design of metaheuristics, making possible easy parallelization and developing new efficient computing methods. Such methods as Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO) or Evolutionary Multi Agent-System (EMAS) are examples of such algorithms. In the paper novel approach to hybridization of such computing systems is presented. A number of agents doing their computing task can agree to run other algorithm (similarly to high level hybrid proposed by Talbi). The paper focuses on presenting the background and the idea of such algorithm along with firm experimental results.