Implicit fitness and heterogeneous preferences in the genetic algorithm

Authors:
Justin T.H. Smith
Affiliations:
University of New Mexico, Albuquerque, NM, USA
Venue:
Proceedings of the 12th annual conference companion on Genetic and evolutionary computation
Year:
2010

Citing 5
Cited 0

Adaptation in natural and artificial systems

Adaptation in natural and artificial systems
Echoing emergence: objectives, rough definitions, and speculations for ECHO-class models

Complexity
Robust Evolutionary Algorithm Design for Socio-economic Simulation

Computational Economics
Revisiting Individual Evolutionary Learning in the Cobweb Model --- An Illustration of the Virtual Spite-Effect

Computational Economics
Classifier fitness based on accuracy

Evolutionary Computation

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper takes an economic approach to derive an evolutionary learning model based entirely on the endogenous use of genetic operators in the employment of self-interested autonomous agents. Reproductive decisions depend on subjective tradeoffs between the quality and quantity of offspring. This avoids the imposition of an external fitness function typically used by genetic algorithms in favor of evolving, heterogeneous preferences over risky reproductive outcomes, expressed via optimal reaction functions. An application to learning in a repeated Cournot oligopoly game is developed, with analytical predictions tested against a computational simulation. With evolutionary coordination via intergenerational wealth transfers, risk-averse firms learn to cooperate to Cournot-Nash market equilibria, raising the market price above costs to fund risk-spreading reproductive strategies.