Adaptable swarm intelligence framework
Mihai Cuibus, Rodica Potolea · 2012
Modern software systems must be continuously adapted to current performance and usability requirements. Indicators like overhead, computational complexity, parameter tuning, or ease of design and implementation are getting increasingly harder to accomplish due to constant increase in system dimensions like code size, API (Application Programming Interface), deployment size, component communication, network lag etc. Furthermore, many entities rely on classic, highly deterministic algorithms that are little or not capable of changing strategies on the fly. Lately, bio-inspired algorithms have successfully tackled this problem with significant, positive results. We propose a framework that may prove useful in obtaining better performance by automatically selecting and combining the best swarm intelligence algorithms with the best parameter selection.