Altruism in evolutionary algorithms for the reconstruction of brain activity
Jens Haueisen, Thomas R. Knösche · Max Planck Digital Library · 2006
Abstract. Evolutionary strategies are used for a variety of optimization problems such as to find an optimal design for a technical device, the design of a new drug or the localization of electric activity in the brain or the heart. They emulate natural evolution based on a Darwinian „survival of the fittest“. However, in natural evolution, altruism (unselfishness) is a commonly observed phenomenon, too. We have developed a number of evolutionary strategies realizing different kinds of altruistic behaviour. With the help of simulations on standardized mathematical optimization problems and data analysis involving source localization of magnetoencephalographic data we compared our new algorithms to each other and to equivalent algorithms without altruism. We found for altruism between individuals, that only a certain amount of altruism improved the performance, too much altruism worsens the performance. Key words : Inverse problems, Optimization, Evolutionary strategies, Magnetoencephalography