NC-LAB EvoFree Group

Natural Computation Laboratory - Department of Computer Science - Missouri University of Science and Technology

The Evolutionary Freedom (EvoFree) Group

The goal of this research group is to develop novel types of self-adaptive parameterless evolutionary algorithms (EAs) which are both easier to use and more powerful than the current state-of-the-art. This group is currently working on two projects, namely Autonomous EAs and Parameterless EAs with the eventual goal of creating an EA which is fully autonomous as well as parameterless.

Autonomous EAs

The concept of the Autonomous EA is a fundamentally new concept which seeks to replace the central control of traditional EAs, such as parent selection, with autonomous control structures, such as individuals finding their own mate, and unnatural strategy parameters of traditional EAs, such as population size, are no longer explicitly controlled but instead become emergent behaviors of population dynamics. While aspects of autonomous EAs have been previously studied, the overall concept was first introduced to the EA community by our paper Toward Automating EA Configuration: the Parent Selection Stage which we presented at the 2007 IEEE Congress on Evolutionary Computation.

Parameterless EAs

The concept of Parameterless EAs is not new, but past attempts have fallen far short of achieving a truly parameterless EA. The motivation for eliminating all EA strategy parameters is two-fold: first of all, for EAs to become widespread in use they must become much easier to use by non-EA experts who are not equipped to properly set EA parameters, and, secondly, EA parameter tuning is typically very time consuming even for EA experts so eliminating parameters has the potential to benefit the EA community as well. There are many parameters to eliminate and we tackled one, namely population size, in our paper Greedy Population Sizing for Evolutionary Algorithms which we presented at the 2007 IEEE Congress on Evolutionary Computation. Another approach to eliminating parameters is a side effect of using Autonomous EAs which eskew unnatural parameters.

Group Members

	Group Photo	Alternate Group Photo
Current Members Dr. Daniel Tauritz - Department of Computer Science André Nwamba - Department of Computer Science Jason Cook - Department of Computer Science Lisa Guntly - Department of Computer Science Former Members Ekaterina Holdener née Smorodkina - Department of Computer Science Joshua Eads - Department of Computer Science	From left to right: Daniel, Jason, André, Kate	From left to right: Daniel, Jason, André, Kate

Group Photo

Alternate Group Photo

Current Members

Dr. Daniel Tauritz - Department of Computer Science

André Nwamba - Department of Computer Science

Jason Cook - Department of Computer Science

Lisa Guntly - Department of Computer Science

Former Members

Ekaterina Holdener née Smorodkina - Department of Computer Science

Joshua Eads - Department of Computer Science

From left to right: Daniel, Jason, André, Kate

Publications

Ekaterina A. Holdener née Smorodkina and Daniel R. Tauritz. Learning Offspring Optimizing Mate Selection. In Proceedings of GECCO 2008 - the Genetic and Evolutionary Computation Conference, pages 1109-1110, Atlanta, Georgia, U.S.A., July 12-16, 2008 (two-page poster abstract).
Ekaterina A. Holdener. The Art of Parameterless Evolutionary Algorithms. Ph.D. dissertation, Missouri University of Science and Technology, 2008.
Ekaterina A. Smorodkina and Daniel R. Tauritz. Greedy Population Sizing for Evolutionary Algorithms. In Proceedings of CEC 2007 - IEEE Congress on Evolutionary Computation, pages 2181-2187, Singapore, September 25-28, 2007.
Ekaterina A. Smorodkina and Daniel R. Tauritz. Toward Automating EA Configuration: the Parent Selection Stage. In Proceedings of CEC 2007 - IEEE Congress on Evolutionary Computation, pages 63-70, Singapore, September 25-28, 2007.