A Little About the GARAGe
The MSU Genetic Algorithms Research and Applications Group (the
GARAGe) is a multi-disciplinary unit interested in the application of genetic
algorithms, genetic programming and other forms of evolutionary computation to real-world problems, as well as fundamental
research on GA and GP. We are doing or have recently done a number of interesting
projects, both in terms of GA/GP fundamental research and in GA/GP applications. Projects involving Prof. Goodman include:
- Goodman's new NSF Cyber-enabled Discovery and Innovation grant (Sept., 2009) with Profs. Johannes Bauer, a telecomm policy expert, and Kurt DeMaagd, an expert in internet growth and modeling, will study "Improving Governance of Next-Generation ICT Infrastructure." This three-year grant will put three graduate research assistants (from Telecommunications, Information Studies and Media and from Electrical and Computer Engineering, working as a team developing agent-based models of enterprises or sectors, classes of customers, and non-profit and government agencies, geographically distributed, and using evolutionary computation both to parameterize them and to evolve governance structures that drive their behaviors in desired directions.
- research on sustainable and scalable evolutionary methods, resulting
in development of the
Hierarchical Fair Competition principle.
- automated design, including research on composite material design
and multi-objective design of automotive components for crashworthiness,
weight savings, and other characteristics (see spinoff company, Red
- automated design
of mechatronic systems using bond graphs and genetic programming
(NSF). Current funding (2007-2010) is from the Danish Research Council to Technical University of Denmark, where Assoc. Prof. Zhun Fan leads the project. New Ph.D. student Jean-Francois Dupuis worked on the problem in the GARAGe during 2007-2008, with Goodman.
- multi-objective "compatible" control algorithms, "MOCC," in which evolutionary methods are used to evolve controllers for such systems as commercial greenhouses, minimizing energy usage while keeping environmental parameters within bounds around the optimum. This research is joint with visiting professor Lihong Xu, of Tongji University, Shanghai, and a series of his Ph.D. students (2006-2010).
- evolutionary improvement of medical diagnosis through pattern classication methods applied to high-dimensionality sets of laboratory test data, with Visiting Prof. Ping Wu, East China Normal University, 2008-2009.
- recognition of breast tumors in microwave imaging data, using evolutionary methods for solving the inverse problem from a novel microwave imaging system being developed by Visiting Prof. Meng Yao, East China Normal University, 2009-2010.
- agent-based modeling, with parameter optimization via evolutionary methods, of an emergency urban evacuation scenario, with undergrad Professorial Assistant Matt Durak, for use at the Academy of Critical Incident Analysis, John Jay College, City University of New York.
- GA operator modeling, with Visiting Prof. Zhenhua Li, China University of Geosciences (Wuhan) (2006-2007)
- mathematical modeling, by Bulent Buyukbozkirli, of genetic algorithm
behavior, aimed at gaining insight useful for configuring genetic algorithms
to solve particular problems
- parallelization of GAs/GPs including use of hierarchical decomposition
of problem domains and heterogeneous design spaces using concepts such
as iiga, the injection island GA
- nesting of irregular shapes using feature matching and GAs
- multiple population topologies and interchange methodologies
- mobile communications infrastructure optimization
- scheduling applications, including job-shop scheduling
- configuration applications, particularly physics applications of optimal
molecule configurations for particular systems like C60 (buckyballs)
- protein folding and protein/ligand docking
- plant floor layout
- and many others.
We have three main GA/GP faculty (Punch, Goodman, and Ofria) and 6 other
faculty members collaborating on GA/GP applications. Charles Ofria, Associate Professor in Computer Science and Engineering, studies evolution
using the AVIDA software for artificial life, which he developed with
Chris Adami, now collaborating closely with Richard Lenski (a microbiologist
and evolutionary biologist). A large number of graduate students and post-docs work with Ofria on "digital evolution" projects in the GARAGe's sister laboratory, the Devo-Lab.
The GARAGe currently has 6 graduate
students, 3 visiting professors and and 1 visiting scholar (a Ph.D. student at Tongji University) working on various
projects, as well as other established collaborations with researchers in the
former Soviet Union, China, and Denmark.
In 1999, Erik Goodman and colleague Ron Averill founded a company, Red
Cedar Technology, building on their design automation experiences
in the GARAGe and creating the HEEDS software, which combines evolutionary
methods with many others. HEEDS is now used by many companies around the world, for design of products as diverse as heart stents and aircraft structures.