EvoDebates: Fit for the future - Understanding the algorithm
Select a name or photograph to read the full position statement and to post your own comments.
"These shortcomings will lead to an EC winter, much like the AI-winter when the big claims and promises of AI were not fulfilled and there was a lack of killer applications."
"We need a general theory of adaptation that includes physical grounding."
"From a theoretical point of view, multi-objective optimisation forces many concepts to be generalised, which is not always trivial, and raises new issues or stresses old ones. As such, it is still very much an open area for research."
"The decision to use an EC approach is generally motivated by the vast number of choices for alternative implementations of its genetic operators. A GA, for example, is an entity open to design and alteration. In this context, theoretical foundations of EC matter."
"A solid theoretical background should not be neglected in order to help give confidence in our domain and attract industrialists."
"EC is a very small subspace within the space of all possible search algorithms, and therefore might some day become less popular…. It is important therefore to concentrate our scientific efforts on the things that will have longer lasting"
"Efforts in the theoretical direction should be enforced, particularly by convincing high-level researchers in probability theory that evolutionary algorithms are a rich-enough domain."
"We need to reinvent evolution step by step, so that understanding the phenomena and improving the evolutionary algorithms go hand-in-hand."
"We need an integrated understanding of algorithms and problems."
