The use of sophisticated machine intelligence approaches for solving complex problems in science and engineering has increased steadily over the last several decades. Within this growing trend, which relies heavily on state-of-the-art optimization and design strategies, the methodology known as memetic algorithms (MAs) is one of those in the forefront of successful computational intelligence techniques in solving complex problems. This paper represents an important milestone in the development of memetic algorithm towards designing new breeds of self-configuring solvers that acclimatize to suit the problem or environment at hand.

  Does it describe a new discovery, methodology, or synthesis of knowledge?

“Memetic Algorithms (MAs) represent one of the recent most successful stories of computational intelligence that capture the power of both biological and cultural selection.”

A unique feature of MAs is the choice of memes, otherwise known as cultures or individual learning procedures, and recent studies have shown that this choice significantly affects the performances of the problem search. To handle this issue, which has impeded the potential success of memetic computing in optimization, this paper introduces "Meta-Lamarckian learning" to establish the concept of collaborations between diverse problem-specific memes that favor neighborhood structures containing high-quality solutions, while presenting a methodology along with several schemes for adaptive choice of multiple memes that leads to better solution quality and time efficiency while the memetic search progresses.

  Would you summarize the significance of your paper in layman’s terms?

A strong capability to engineer reliable and high-quality products is necessary in all businesses in order to remain competitive within an increasingly global economy, which is constantly being exposed to high commercial pressures. Many real-life scenarios of problem-solving involve searching for an optimum solution. Hence, optimization is now a part and parcel of problem-solving in many areas of the arts, business & finance, design, science, and engineering, including those which are directly applicable to our daily life.

Given the restricted theoretical knowledge available for choosing an optimization solver that best suits a given ad-hoc black box problem, our paper introduced the concept of "Meta-Lamarckian learning" and presented several strategies for assisting designers in the automatic adaptation of memetic solvers that yield robust and higher quality designs on complex optimization problems efficiency, thus leading to lower time-to-market and better quality at lower cost.

  How did you become involved in this research and were any particular problems encountered along the way?

Memetic Algorithms represent a unique class of population-based meta-heuristic search methods that are inspired by Darwinian principles of natural selection and, on the other hand, by Richard Dawkins’ notion of a meme, thus rendering a methodology that balances generality and problem-specificity.

Nevertheless, with the plethora of memes in the form of heuristics and/or numerical approaches available for use, it is a perplexing task for novice users and almost impossible, even for experts, to know in advance how to choose meme(s) that are the most suitable for solving a problem in hand—an outcome that is often referred to as the "No Free Lunch" theorem for search, which proves that no algorithm is better than another over all possible problems.

Our driving belief is that technology and research, as a whole, should be used in the service of industry, society, and humanity. With this in mind, our research goal is to design new breeds of self-configuring strategies that acclimatize to suit the specific problem or environment at hand.

  Where do you see your research leading in the future?

MAs represent one of the recent most successful stories of computational intelligence that capture the power of both biological and cultural selection. In recent years, there has been increasing research interest in the field of Memetic Computing. There is now a large body of evidence revealing that memetic algorithms not only converge to high quality solutions, but also search vast, and sometimes noisy, solution spaces more efficiently than their conventional counterparts. Thus, memetic algorithms are the preferred methodology for solving many real-world applications.

Internationally, there have been several ongoing research activities relating to Memetic Algorithms and/or Memetic Computing.

A Task Force on Memetic Algorithms in the IEEE Computational Intelligence Society (CIS), Emergent Technology Technical Committee (ETTC), was founded in 2006. The role of the ETTC within the Computational Intelligence Society of the IEEE is to identify new directions in research and application within the scope of the CIS. In addition, special issues dedicated to work on Memetic Algorithms have been edited in several top-quality journals recently. In 2007/08, the first issue of a new journal dedicated to memetic computing is also expected to be published. In international conferences, there are now also many special sessions dedicated to the topic of MAs research.

The work in the paper represents only an initial step towards acquiring a clear understanding and demonstration of natural computing, particularly, the new science of memetics in the real world. It serves as an important initial milestone of memetic computing towards designing new breeds of self-configuring solvers that acclimatize to suit the problem or environment at hand.

At the moment, there remains a lack of adequate research efforts placed on the theoretical aspect of memetic computing. Hence, one challenging task that we aspire to work on is the design of theoretic and formal memetic models and frameworks that balance evolution and individual learning while the search progresses for maximum effectiveness and efficiency.

In the short term, our intent is to further improve the quality, speed, and generality with which memetics can be made applicable to real world problems. However, since the value of complex memetic algorithms also stands with its applications, rather than only the theory itself, it makes good sense to establish collaborations with researchers across a variety of new fields.