HR: 1340h
AN: SM13A-0311    [Abstracts]
TI: Understanding space weather via analysis, prediction, modeling, and classification supported by Genetic Programming.
AU: * Jorgensen, A M
EM: ajorg@lanl.gov
AF: Los Alamos National Laboratory, ISR-4, MS D448, Los Alamos, NM 87545
AU: Karimabadi, H
SM13A-0311 AF: SciberQuest, Pacific Executive Plaza 777 South Highway 101, Suite 108, Solana Beach, CA 92075-2623
AB: We will demonstrate the application of genetic programming to several different space physics problems. Genetic programming is an evolutionary optimization or search technique. In contrast to most common search techniques, genetic programming can search spaces of algorithms or mathematical functions. A population of candidate solutions is allowed to evolve in accordance with the Darwinian principles of natural selection and survival of the fittest. Candidate solutions are evaluated against a truth data set. For example, if the task is to reproduce a training data set based on some inputs, the fitness will measure how well the candidate solution performs in this task. Candidate solutions which score higher are preferentially selected to producing new candidate solutions through analogs to the biological world processes of reproduction and mutation. This process can be used to derive predictive expressions, models, and algorithms for feature classification. We will illustrate the use of genetic programming for deriving high-level human-readable mathematical expressions for prediction, modeling, and feature classification.
DE: 2722 Forecasting (7924, 7964)
DE: 2753 Numerical modeling
DE: 2794 Instruments and techniques
DE: 3200 MATHEMATICAL GEOPHYSICS (0500, 4400, 7833)
DE: 9820 Techniques applicable in three or more fields
SC: SPA-Magnetospheric Physics [SM]
MN: Fall Meeting 2005