Using Genetic Programming in Industrial Statistical Model Building

Abstract

The chapter summarizes the practical experience of integrating genetic programming and statistical modeling at The Dow Chemical Company. A unique methodology for using Genetic Programming in statistical modeling of designed and undesigned data is described and illustrated with successful industrial applications. As a result of the synergistic efforts, the building technique has been improved and the model development cost and time can be significantly reduced. In case of designed data Genetic Programming reduced costs by suggesting transformations as an alternative to doing additional experimentation. In case of undesigned data Genetic Programming was instrumental in reducing the model building costs by providing alternative models for consideration.

Future Trends in Soft Computing Industrial Applications

“…An example of a fruitful collaboration between fundamental model developers and soft computing developers is given in section IIC and described in [16]. Recently, a joint effort between statisticians and soft computing developers demonstrated significant improvement in using genetic programming in industrial statistical model building [20].…”

An Evolution of a Complete Program using XML-based Grammar Definition

“…More recently, GP has been used to solve problems in various fields, such as: medical (Guo andNandi 2006, Hong andCho 2004), Railway platform allocation (Clarke, Hinde, Withall, Jackson, Phillips, Brown and Watson 2010), robotics (Konig and Schmeck 2009), and symbolic regressions (Castillo, Kordon, Sweeney and Zirk. 2005).…”

Challenges in Open-Ended Problem Solving with Genetic Programming

“…There is the matter of perception because many humans rightfully distrust solutions that are either not understandable or not conventional. There is also the matter of hidden consequences, because GP L R u n G P Figure 17-2 depicts the protocol from (Castillo, Kordon et al 2004), which describes a means of using GP to identify transforms that eliminate lack-of-fit in a Box-Behnkin design. In Castillo et Ö/.…”

“…Anecdotal evidence of this has often come up during the U-M Center for the Study of Complex Systems Workshop on Genetic Programming Theory and Practice. For example, see (Caplan and Beker 2004;Castillo, Kordon, et al 2004;MacLean and Wollesen, et al 2004).…”

“…For the most part, GP solutions in the real world do seem to require additional vetting and handling before they are adopted into practice e.g., (Caplan and Beker 2004;Castillo, Kordon et al 2004;MacLean, Wollesen et al 2004). The challenges of implementing a GP solution do increase if the form of a solution differs greatly from the representational Implementing would include the actual building of that circuit.…”