public double TrainNeuralNetwork()
{
var chromosome = new KasandraChromosome(_fitness.TotalNumberOfWeights);
var population = new Population(minPopulation, maxPopulation, chromosome);
_geneticAlgorithm = new GeneticAlgorithm(population, _fitness, new EliteSelection(), new UniformCrossover(), new UniformMutation(true));
_geneticAlgorithm.MutationProbability = mutationRate;
_geneticAlgorithm.CrossoverProbability = crossoverProbabilty;
_geneticAlgorithm.Termination = new GenerationNumberTermination(maxIterations);
_geneticAlgorithm.TerminationReached += _geneticAlgorithm_TerminationReached;
_geneticAlgorithm.GenerationRan += _geneticAlgorithm_GenerationRan;
_geneticAlgorithm.Start();
return 0;
}
private void _geneticAlgorithm_GenerationRan(object sender, EventArgs e)
{
var ga = (GeneticAlgorithm)sender;
//if(ga.GenerationsNumber%50 == 0 || ga.GenerationsNumber == 1 || ga.GenerationsNumber < 100)
Console.WriteLine("" + ga.GenerationsNumber + "\t" + Math.Abs(ga.BestChromosome.Fitness.Value));
winner = (KasandraChromosome) ga.BestChromosome;
if (winner.Fitness.Value > bestEverFitness)
{
bestEver = winner;
bestEverFitness = winner.Fitness.Value;
}
}
private void _geneticAlgorithm_TerminationReached(object sender, EventArgs e)
{
var ga = (GeneticAlgorithm) sender;
winner = (KasandraChromosome) ga.BestChromosome;
if (winner.Fitness.Value < bestEver.Fitness.Value)
winner = bestEver;
}
