Author:
Nonel Thirer
Affiliation:
HIT – Holon Institute of Technology, Israel
Keyword(s):
Genetic Algorithm, Pipeline, Resources, Subpopulations, Flexible Implementation.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Bio-inspired Hardware and Networks
;
Computational Intelligence
;
Evolutionary Computing
;
Genetic Algorithms
;
Informatics in Control, Automation and Robotics
;
Intelligent Control Systems and Optimization
;
Soft Computing
Abstract:
Many problems common to the electrical and electronics field can be solved by finding a target function and its minimum or maximum. For such problems, usually an analytical solution is not implementable, and therefore iterative algorithms are used. One such efficient algorithm is the Genetic Algorithm (GA). The GA imitates the biological evolution process, finding the solution by implementing the “natural selection” principle, which asserts that the strong has higher chances to survive. The GA is an iterative procedure which operates on a population of individuals called "chromosomes" or "possible solutions" (usually represented by a binary code) and performs several processes on the population individuals, in order to produce a new population - the same as in the biological evolution. Using the algorithm on large populations requires substantial hardware resources. Also, naturally, the amount of time necessary to reach a solution increases, due to the greater number of iterations ne
eded. In this paper, we present an FPGA pipelined based method designed to implement a GA, which provides a high-speed solution for large populations, with a minimum of resources. This outcome is obtained by a procedure which operates sequentially with parts of the population. In addition, an immigration unit is defined to provide an efficient communication between these parts in different iterations. Moreover, some possible solutions to improve our method are analyzed.
(More)