Global optimization via neural network approximation of inverse coordinate mappings

Authors:
V. D. Koshur;K. V. Pushkaryov
Affiliations:
Institute of Space and Information Technology of Siberian Federal University, Krasnoyarsk, Russia 660074;Institute of Space and Information Technology of Siberian Federal University, Krasnoyarsk, Russia 660074
Venue:
Optical Memory and Neural Networks
Year:
2011

Citing 2
Cited 0

Solution of Large-Scale Problems of Global Optimization on the Basis of Parallel Algorithms and Cluster Implementation of Computing Processes

PaCT '09 Proceedings of the 10th International Conference on Parallel Computing Technologies
Artificial Intelligence: A Modern Approach

Artificial Intelligence: A Modern Approach

Quantified Score

Hi-index	0.00

Visualization

Abstract

The novel heuristic method of global optimization via neural network approximation of inverse coordinate mappings is suggested. The method deals with continuous objective functions of multiple variables with multiple extremes. The search space must be a multidimensional box. No derivatives of a function are required to exist. The method is built on Generalized Regression Neural Networks (GRNNs). An example of application of the method to a smooth function with multiple extremes is provided. The method has been compared with the genetic algorithm and particle swarm optimization through an experiment. The description and results of the experiment are included.