public void Generate()
{
IRouletteWheel rouletteWheel = breedingFactory.CreateRouletteWheel(generation);
breeder.SetRouletteWheel(rouletteWheel);
breeder.CalcFitness(generation);
Solution[] newGeneration = new Solution[population];
for (int i = 0; i < population; i++)
{
newGeneration[i] = breeder.Breed();
}
logger.logBestFitSolution(writer, FindBestFit());
generation = newGeneration;
}
private void createNextGeneration(ITrainingSession bestPerformer)
{
Stopwatch watch = new Stopwatch();
watch.Start();
int numberOfTopPerformersToChoose = (int)(_generationConfig.GenerationPopulation * 0.50);
int numToBreed = (int)(_generationConfig.GenerationPopulation * 0.35);
int numToGen = (int)(_generationConfig.GenerationPopulation * 0.15);
if (numberOfTopPerformersToChoose + numToBreed + numToGen < _generationConfig.GenerationPopulation)
{
numToGen += _generationConfig.GenerationPopulation -
(numberOfTopPerformersToChoose + numToBreed + numToGen);
}
var sessions = _generation.GetBestPerformers(numberOfTopPerformersToChoose);
if (bestPerformer != null)
{
LoggerFactory.GetLogger().Log(LogLevel.Info, "Best performer found for creating generation");
if (sessions.All(s => s.NeuralNet.GetGenes() != bestPerformer.NeuralNet.GetGenes()))
{
LoggerFactory.GetLogger()
.Log(LogLevel.Info,
$"Best performer adding to sessions with eval {bestPerformer.GetSessionEvaluation()}");
sessions[sessions.Count - 1] = bestPerformer;
sessions = sessions.OrderByDescending(s => s.GetSessionEvaluation()).ToList();
LoggerFactory.GetLogger()
.Log(LogLevel.Info, $"session 0 eval: {sessions[0].GetSessionEvaluation()}");
}
else
{
LoggerFactory.GetLogger()
.Log(LogLevel.Info,
$"Best performer already in generation: not adding.");
}
}
_history.AddEval(sessions[0].GetSessionEvaluation());
var mutateChance = DetermineMutateChance();
IList <INeuralNetwork> children = _breeder.Breed(sessions, numToBreed);
var newSessions = new List <ITrainingSession>();
//Allow the very top numToLiveOn sessions to be added to next generation untouched
int numToLiveOn = sessions.Count / 10;
var sessionsToLiveOn = sessions.Take(numToLiveOn).ToList();
newSessions.AddRange(sessionsToLiveOn);
//try to mutate session that will live on 1 by 1
var mutatedTop = GetMutatedNetworks(sessionsToLiveOn.Select(s => s.NeuralNet), mutateChance);
//or each session that lived on that was mutated, remove last top performer, then mutate remaining top performers in batch and add
sessions = sessions.Skip(numToLiveOn).ToList();
var sessionSubset = sessions.Take(sessions.Count - mutatedTop.Count);
IList <INeuralNetwork> toKeepButPossiblyMutate = sessionSubset.Select(session => session.NeuralNet).ToList();
IList <INeuralNetwork> newNetworks = getNewNetworks(numToGen, mutateChance);
List <INeuralNetwork> toTryMutate = new List <INeuralNetwork>();
//try to mutate both new networks as well as all the top performers we wanted to keep
toTryMutate.AddRange(toKeepButPossiblyMutate);
toTryMutate.AddRange(newNetworks);
bool didMutate;
IList <INeuralNetwork> maybeMutated = _mutator.Mutate(toTryMutate, mutateChance, out didMutate);
List <INeuralNetwork> allToAdd = new List <INeuralNetwork>();
allToAdd.AddRange(mutatedTop);
allToAdd.AddRange(children);
allToAdd.AddRange(maybeMutated);
newSessions.AddRange(allToAdd.Select((net, sessionNumber) => new TrainingSession(net, _evaluatableFactory.Create(net), sessionNumber)));
_generation = new Generation(newSessions, _generationConfig);
watch.Stop();
LoggerFactory.GetLogger().Log(LogLevel.Debug, $"create generation runtime (sec): {watch.Elapsed.TotalSeconds}");
watch.Reset();
}
public void Solve(BionicModel model, BionicSolverSettings settings)
{
if (model == null)
{
throw new ArgumentNullException("model");
}
if (settings == null)
{
throw new ArgumentNullException("settings");
}
OnSolverStarting(new EventArgs());
IInitialPopulationGenerator initializer = InitialPopulationGeneratorFactory.CreateInitialPopulationGenerator();
ISelector selector = SelectorFactory.CreateSelector();
IBreeder breeder = BreederFactory.CreateBreeder();
// 1. Zeroing generation number
model.CurrentGeneration = 0;
OnSolverStarted(new EventArgs());
// 2. Creating initial population
CopyIndividuals(initializer.GenerateInitialPopulation(settings.InitialPopulationSize, model.Attributes), model.CurrentPopulation);
CalculateFitness(model, model.CurrentPopulation, settings);
model.ApplyFunctionalConstraints();
OnPopulationUpdated(new EventArgs());
while (model.CurrentGeneration < settings.MaxGenerations)
{
// 3. Selecting most fit individuals
IEnumerable <Individual> fitIndividuals = selector.Select(model.CurrentPopulation, model.FitnessCriterion, settings.SelectionCap);
if (fitIndividuals.Count() == 0)
{
OnSolverFinished(new SolverFinishedEventArgs("All individuals are inactive"));
return;
}
// Selected subpopulation will contain only selected active individuals
Population selectedSubPopulation = new Population(fitIndividuals.Count());
foreach (Individual fitIndividual in fitIndividuals)
{
selectedSubPopulation.Add(fitIndividual);
}
// 4. Breeding the selected ones
IEnumerable <Individual> descendants = breeder.Breed(selectedSubPopulation, model.Attributes,
settings.DescendantsRangePercent, settings.DescendantsNum, model.CurrentGeneration + 1);
if (descendants.Count() == 0)
{
OnSolverFinished(new SolverFinishedEventArgs("No descendants was created"));
return;
}
// Descendants subpopulation will contain only new individuals
Population descendantsSubPopulation = new Population(descendants.Count());
foreach (Individual descendant in descendants)
{
descendantsSubPopulation.Add(descendant);
}
// 5. Calculating fitness of the descendants
CalculateFitness(model, descendantsSubPopulation, settings);
// At this point, all descendants are still active, although
// there may be individuals that don't fit f. c.
// 6. Forming new population
if (settings.ApplyElitism)
{
// First unite descendants with their parents (elitism)
// We can do this without fear of ID collision because this
// was taken into account on the breeding step
CopyIndividuals(descendantsSubPopulation, selectedSubPopulation);
// Now this united population becomes current, we just
// throw away old individuals
model.CurrentPopulation.Clear();
CopyIndividuals(selectedSubPopulation, model.CurrentPopulation);
}
else
{
// We throw away parents and old individuals,
// descendants are taking over the place
model.CurrentPopulation.Clear();
CopyIndividuals(descendantsSubPopulation, model.CurrentPopulation);
}
model.ApplyFunctionalConstraints();
// 7. Updating current generation number
model.CurrentGeneration = model.CurrentPopulation.Max(ind => ind.Value.GenerationNumber);
OnPopulationUpdated(new EventArgs());
}
OnSolverFinished(new SolverFinishedEventArgs());
}
