We start by considering a simple methodology to test and validate the
feasibility of combining genetic algorithms and case-based reasoning on four
sample problem sets. In our experiments, a genetic algorithm finds and saves
solutions to a problem 
, the problem is changed slightly to 
,
and appropriate solutions to are injected into the initial
population of the genetic algorithm that is trying to solve the new problem,
. If the cases from contain good building blocks or partial
solutions, the genetic algorithm can use these building blocks or schemas and
quickly approach the solution to . The results show that compared to
a genetic algorithm that starts from scratch (from a randomly initialized
population), the genetic algorithm with injected solutions quickly finds
good solutions to and that the quality of solutions after convergence
is usually better.