Digital Evolution and Evolutionary Algorithms | Jade Eli Technologies
When I was in my 20s I worked on a project involving Evolutionary Algorithms - I used them to try to find better mathematical solutions to a given problem than by using other approximation or problem solving strategies. The idea is to have a population where each individual represents a solution to the equation, no matter how bad it is, a set of operators which manipulate the population or cross genes from one individual to the other in a new generation and an evaluation function which is the target problem to solve. Every generation the operators perform mutation and crossover of the given individuals of the generation to create a number of offspring. Those individuals get evaluated for their quality of solving the problem and a selection function mixes a number of new individuals and some of the existing ones into the population on the new generation. And then the mechanism starts over until an abortion criteria is met, e.g. no changes in quality of the current evaluation function