Department of Electrical Engineering and Automation, Aalto University School of Electrical Engineering, Finland
Received date: January 22, 2014; Accepted date: January 25, 2014; Published date: February 13, 2014
Citation: XZ Gao (2014) Hybrid Nature-Inspired Computing (NIC) Methods: Motivation and Prospection. Swarm Intell and Evol Comput 3:e102. doi:
Copyright: © 2014 Gao XZ. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Visit for more related articles at International Journal of Swarm Intelligence and Evolutionary Computation
XZ Gao (2014) Hybrid Nature-Inspired Computing (NIC) Methods: Motivation and Prospection. Swarm Intell and Evol Comput 3:e102.
We can observe that the aforementioned NIC methodologies share a lot of similarities, e.g., adaptation, learning, and evolution. On the other hand, there are some distinct differences among themselves, and all the NIC methods have advantages and disadvantages. However, compared with their hybridizations, the individual NIC methods are not always effective enough. Additionally, practicing engineers often face the difficulty of choosing the best suited NIC methods to meet particular engineering requirements. The inspiration and fusion of these NIC techniques are shown in Figure 1. How to combine the existing NIC methods together so that they can benefit from each other and achieve superior performances has been a popular research subject. The most important topics focus on the different fusion types of merging the NIC methods, such as application of efficient local search strategies in the NIC to build up novel hybrid approaches, solution diversity maintenance techniques for finding suitable balance between exploration and exploitation of the hybrid NIC methods, and comparison of performances of effectiveness, convergence, and computational complexities of the hybrid NIC algorithms using statistics techniques. In addition, the robustness of the hybrid NIC methods against the parameter variation and â€˜noisyâ€™ fitness landscapes and their adaptation capability to the time-varying environments should be examined as well. Therefore, the following issues are usually under investigation in the current research work of the fusion of these NIC methods.
(1) Exploration of different fusion types of merging the NIC methods together, i.e., how to integrate certain NIC methods into the others so that their short comings are overcome and their advantages can be enhanced.
(2) Examining the effects of different meta-heuristic operations, e.g., GA crossover & mutation and BF chemo taxis and swarming, on the hybrid NIC methods.
(3) How to apply efficient local search strategies in the MC to build up novel hybrid NIC approaches.
(4) Study of the appropriate selections of the coefficients in the hybrid NIC methods.
(5) Investigation of the maintenance techniques for finding suitable balance between exploration and exploitation of the hybrid NIC methods.
(6) Comparison of the performances including effectiveness, convergence, and computational complexities of the hybrid NIC algorithms using statistics approaches.
(7) Test of the robustness of the hybrid NIC methods against the parameter variation and â€˜noisyâ€™ fitness landscapes.
(8) Study of the adaptation capability of the hybrid NIC methods to the time-varying environments.
Although various hybrid NIC methods have been shown to offer better optimization performances than that of the standalone versions, the â€˜No Free Lunch (NFL)â€™ theorem is a fundamental barrier to the exaggerated claims of the power and efficiency of any specific optimization algorithms . In other words, there is no single optimization method in practice that can be the best for all kinds of engineering problems. Therefore, one possible way to handle the negative implication of the NFL theorem is to develop hybrid solutions based on the fusion of existing ones and restrict the applications of a given algorithm to only a particular type of optimization tasks.