A Highly Functional Decision Paradigm Based on Nonlinear Adaptive Genetic Algorithm.
Abstract
In the first year of this Phase 2 research program, POC refined the GA Optimizer by rewriting it in DLL format and optimizing the fuzzy logic-based GA control rules. We also evaluated existing neural network training methods using the GA, and examined areas in which our algorithm can improve their performance. POC has also initiated a new line of optimization development tools, the route optimizer. Initial development efforts demonstrate the proof of concept, and show that the algorithm can be applied to many optimization problems; these include making financial predictions, medical predictions, medical diagnoses, and market classifications, as well as modeling manufacturing processes and the anticipated resulting product quality, classifying biological organisms estimating job costs, detecting fraud, and many others. Finally, POC has begun developing the protocol of a development tool combining fuzzified genetic algorithm (GA) and a neural network (NN). This tool will find the optimal structure for the NN by training based on various combinations of the input data, and will optimize it by using NN performance as a fitness value.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 30, 1996
- Accession Number
- ADA317032
Entities
People
- Jeongdal Kim