Although Evolutionary Algorithms are very good at mimicking adaptation within a species to optimize solutions for difficult problems, creating algorithms that can mimic the development of two or more species from a common ancestor has been a challenge. There are versions of Evolutionary Algorithms that have some characteristics of speciation, but none that match natural processes. Such algorithms would be a good step in the development of a general purpose Evolutionary Algorithm and would help in understanding the principles of evolution. In regards to this research, we consider a population to be distinct (and a separate species) if it is made up of individuals that are unable to produce viable offspring with individuals from the other population or if offspring are produced, they are sterile. The short term goal, which is reasonable for this special session, is to have individuals of differing species choose not to mate and if they do produce offspring, the offspring do not continue to reproduce. In this way, the gene pools for each of the species will be isolated.
The purpose of this special session is to bring together people working on Evolutionary Algorithms that tend toward or have the potential for speciation. Some possible topics of interest include:Evolutionary algorithms mimicking allopatric or sympatric speciation
Environments for research in natural speciation
Biologically-inspired models of interactive agents
Use of topology in populations
Formation of sub-populations
Selection criteria in evolutionary algorithms
Multimodal function optimization
Paper SubmissionPapers should be submitted through the IEEE CEC 2019 paper submission website. Please specify that your paper is submitted to the Special Session on Speciation. All papers accepted and presented at CEC2019 will be included in the conference proceedings.
The submission deadline, date of notification, and the final paper submission dealine are the same as for regular conference papers -- these dates can be found at cec2019.org
OrganizersGary ParkerProfessor Gary Parker is in the Department of Computer Science at Connecticut College where he is the Director of the Autonomous Agent Learning Lab. He received his BA in Zoology at the University of Washington, MS in Computer Science at the Naval Postgraduate School, and PhD in Computer Science and Cognitive Science at Indiana University. His research focuses on evolutionary computation and methodologies for learning in autonomous robots. He developed the cyclic genetic algorithm (CGA), which is a variant of the standard GA, to provide a means for learning cycles of behavior, and punctuated anytime learning (PAL), which he is using to incorporate CGAs into a real-time learning system that uses periodic tests on the robot to continually improve the quality of controller learning. Additional research has resulted in a method using PAL to co-evolve cooperative individuals in a team of robots. He has over 80 peer-reviewed publications, including conference papers (two of which received best paper awards), journal articles, and book chapters. He was the chair of the 7th and 8th International Symposium on Intelligent Automation and Control (ISIAC2008 and ISIAC2010); has organized three IEEE WCCI special sessions; has been on the program committee / reviewer for over 50 conferences; served as a reviewer for 17 journals; has been on the board of associate editors for two journals; is a member of three IEEE Computational Intelligence Society task forces; and was presented with the World Automation Congress / AutoSoft Lifetime Achievement Award in October 2008.
Department of Computer Science, Connecticut College, New London, Connecticut, USA
Department of Information Science, University of Otago, Dunedin, Otago, New Zealand.
Associate Professor Peter Whigham is based at the Department of Information Science, University of Otago, where he teaches data science undergraduate and postgraduate classes. An early developer of grammar-based Genetic Programming, he now works in theoretical aspects of GP (e.g., topological population structure, the phenomenon of bloat, and fine-grained evaluation of GP fitness) as well as applying GP methods in evolutionary machine learning. He has also examined the effect of spatial structure in theoretical population biology, and develops computational models for understanding ecological systems. He has published in IEEE Transactions on Evolutionary Computation, Genetic Programming and Evolvable Machines, Theoretical Population Biology, Physica A, along with the conferences GECCO and CEC.