/// <summary>
/// Create an annealing trainer.
/// </summary>
/// <param name="method">The method to use.</param>
/// <param name="training">The training data to use.</param>
/// <param name="argsStr">The arguments to use.</param>
/// <returns>The newly created trainer.</returns>
public IMLTrain Create(IMLMethod method,
IMLDataSet training, String argsStr)
{
if (!(method is BasicNetwork))
{
throw new TrainingError(
"Invalid method type, requires BasicNetwork");
}
ICalculateScore score = new TrainingSetScore(training);
IDictionary <String, String> args = ArchitectureParse.ParseParams(argsStr);
var holder = new ParamsHolder(args);
int populationSize = holder.GetInt(
MLTrainFactory.PropertyPopulationSize, false, 5000);
IMLTrain train = new MLMethodGeneticAlgorithm(() =>
{
IMLMethod result = (IMLMethod)ObjectCloner.DeepCopy(method);
((IMLResettable)result).Reset();
return(result);
}, score, populationSize);
return(train);
}
public void Train(int mode = 0, int epochs = 1, int maxTemp = 10, int minTemp = 2, int cycles = 10, int timerTimeout = 5000)
{
var encogScore = new EncogScore(maze, timerTimeout);
encogScore.EncogCycleComplete += EncogScore_EncogCycleComplete;
encogScore.MazeCycleComplete += EncogScore_MazeCycleComplete;
IMLTrain train;
if (mode == 0)
{
//Simulated Annealing
train = new NeuralSimulatedAnnealing(network, encogScore, maxTemp, minTemp, cycles);
}
else
{
//Genetic Algorithm
train = new MLMethodGeneticAlgorithm(() => { return(network); }, encogScore, cycles);
}
for (int epoch = 1; epoch <= epochs; epoch++)
{
train.Iteration();
TrainingIterationComplete?.Invoke(epoch, train.Error);
}
//SaveTrainingData();
}
public void StartTraining()
{
var Scorer = new GameScore();
var Trainer = new MLMethodGeneticAlgorithm(GetNetwork, Scorer, 200)
{
ThreadCount = 8
};
//Trainer.ThreadCount = 1;
var s = new BestStrategy();
s.Init(Trainer);
Trainer.Strategies.Add(s);
for (int i = 0; i < 1000; i++)
{
Trainer.Iteration();
Console.WriteLine(string.Format("Epoch {0} finished : {1} Hash : {2}", i, Trainer.Genetic.BestGenome.Score, Trainer.Genetic.BestGenome.GetHashCode()));
if ((i % 10) == 0)
{
SaveTrainer(Trainer);
}
}
SaveTrainer(Trainer);
System.IO.File.WriteAllText(@".\Result.txt", Trainer.Genetic.BestGenome.ToString());
}
public static BasicNetwork train(Render render)
{
BasicNetwork network = CreateNetwork(6, 2, 8);
IMLTrain train;
train = new MLMethodGeneticAlgorithm(() =>
{
BasicNetwork result = CreateNetwork(6, 2, 8);
((IMLResettable)result).Reset();
return(result);
}, new Tester(), 2000);
int epoch = 1;
for (int i = 0; i < 80; i++)
{
train.Iteration();
Console.WriteLine(@"Epoch #" + epoch + @" Score:" + train.Error);
epoch++;
if (train.Error >= 37)
{
break;
}
}
return((BasicNetwork)train.Method);
}
public static void LanderTrain()
{
Console.WriteLine("\n\nEncog network structure: \nNumber of Input neurons 3 \nNumber of output neurons 1 ");
int hiddenLayers = Util.GetInput("Number of hidden layers [default 1]:", 1);
int hiddenLayerNeurons = Util.GetInput("Number of hidden layer neurons [default 100]: ", 100);
int numOfEpoch = Util.GetInput("Number of Epochs [default 10]:", 10);
int type = Util.GetInput("\nSelect a training method [Annealing - 0][Genetic - 1 default]:", 1);
BasicNetwork network = CreateNetwork(hiddenLayers, hiddenLayerNeurons);
var pilotScore = new PilotScore();
IMLTrain train;
if (type == 0)
{
int startTemp = Util.GetInput("Start Temperature [default 10]:", 10);
int endTemp = Util.GetInput("Stop Temperature [default 2]:", 2);
int cycles = Util.GetInput("Number of Cycles [default 10]:", 10);
train = new NeuralSimulatedAnnealing(
network, pilotScore, startTemp, endTemp, cycles);
}
else
{
int PopulationSize = Util.GetInput("Population Size [default 10]:", 10);
train = new MLMethodGeneticAlgorithm(() => {
BasicNetwork result = CreateNetwork(hiddenLayers, hiddenLayerNeurons);
((IMLResettable)result).Reset();
return(result);
}, pilotScore, PopulationSize); // population size
}
Console.WriteLine("\n\nTraining: \n");
System.Diagnostics.Stopwatch stopwatch = new Stopwatch();
stopwatch.Start();
for (int i = 1; i <= numOfEpoch; i++) // num of epochs
{
train.Iteration();
Console.WriteLine($"Epoch#: {i} \t Score: {train.Error}");
}
stopwatch.Stop();
Console.WriteLine("\nThe total number of times it tried the Lunar Lander for training: " + NeuralPilot.CycleCount);
Console.WriteLine($"Elapsed: {stopwatch.Elapsed}\n");
int showPredictedOutput = Util.GetInput("Show landing simulation for AI prediction? [No - 0, Yes - 1 default]: ", 1);
if (showPredictedOutput == 1)
{
network = (BasicNetwork)train.Method;
var pilot = new NeuralPilot(network, true);
pilot.ScorePilot();
Console.WriteLine("hit enter to continue...");
Console.ReadLine();
}
}
/// <summary>
/// This is where encog network settings are configured and training is being processed.
/// </summary>
public static void LogisticTrain()
{
Console.WriteLine("\n\nEncog network structure: \nNumber of Input neurons 1 \nNumber of output neurons 9 ");
int hiddenLayers = Util.GetInput("Number of hidden layers [default 1]: ", 1);
int hiddenLayerNeurons = Util.GetInput("Hidden layer neurons [default 100]: ", 100);
int type = Util.GetInput("\nSelect a training method [Annealing - 0][Genetic - 1 default]:", 1);
int numOfEpoch = Util.GetInput("Number of Epochs [default 10]:", 10);
BasicNetwork network = CreateNetwork(hiddenLayers, hiddenLayerNeurons);
var pilotScore = new EncogLogisticScore();
IMLTrain train;
if (type == 0)
{
int startTemp = Util.GetInput("Start Temperature [default 10]:", 10);
int endTemp = Util.GetInput("End Temperature [default 2]:", 2);
int cycles = Util.GetInput("Cycles [default 10]:", 10);
train = new NeuralSimulatedAnnealing(network, pilotScore, endTemp, startTemp, cycles);
}
else
{
int populationSize = Util.GetInput("Population Size [default 10]:", 10);
train = new MLMethodGeneticAlgorithm(() => {
BasicNetwork result = CreateNetwork(hiddenLayers, hiddenLayerNeurons);
((IMLResettable)result).Reset();
return(result);
}, pilotScore, populationSize); // population size
}
Stopwatch watch = new Stopwatch();
watch.Start();
Console.WriteLine("\n\nTraining: \n");
for (int i = 0; i < numOfEpoch; i++) // num of epochs
{
train.Iteration();
double totalCosts = train.Error;
string currencyScore = totalCosts.ToString("$#,##0");
Console.WriteLine($"Epoch # {i} \t Score: {currencyScore,10}");
}
watch.Stop();
Console.WriteLine("\nPredicted outputs:");
network = (BasicNetwork)train.Method;
var pilot = new EncogLogisticSimulator(network, true);
pilot.CalculateScore(LogisticSimulator.GenerateCustomerOrders(), true);
Console.WriteLine($"\nElapsed: {watch.Elapsed}");
Console.WriteLine("\nThe total number of times it tried the Logistic Simulation for training: " + pilotScore.SessionCnt);
Console.ReadLine();
}
public void TestGenetic()
{
IMLDataSet trainingData = new BasicMLDataSet(XOR.XORInput, XOR.XORIdeal);
ICalculateScore score = new TrainingSetScore(trainingData);
MLMethodGeneticAlgorithm genetic = new MLMethodGeneticAlgorithm(() =>
{
return(NetworkUtil.CreateXORNetworkUntrained());
}, score, 500);
NetworkUtil.TestTraining(genetic, 0.00001);
}
public BasicNetwork SharkTrain()
{
IMLTrain train = new MLMethodGeneticAlgorithm(() =>
{
BasicNetwork result = CreateNetwork(2, 2);
((IMLResettable)result).Reset();
return(result);
}, new SharkTrainer(), 1000);
for (int i = 0; i < 6; i++)
{
train.Iteration();
Console.WriteLine("Shark Inital Train Score: " + Math.Round(train.Error, 2));
}
return((BasicNetwork)train.Method);
}
public void Init(IMLTrain train)
{
Method = (MLMethodGeneticAlgorithm)train;
Adversaries = new List <Adversaries>();
var r = new RangeRandomizer(-1, 1);
for (int i = 0; i < 6; i++)
{
Adversaries.Add(new Adversaries()
{
PlayerStart = r.NextDouble() >= 0.5 ? 1.0 : -1.0, Player = new PlayerVertical(i)
});
Adversaries.Add(new Adversaries()
{
PlayerStart = r.NextDouble() >= 0.5 ? 1.0 : -1.0, Player = new PlayerHorizontal(i)
});
}
BaseAdversaries = Adversaries.ToList();
}
/// <summary>
/// Program entry point.
/// </summary>
/// <param name="app">Holds arguments and other info.</param>
public void Execute(IExampleInterface app)
{
BasicNetwork network = CreateNetwork();
IMLTrain train;
if (app.Args.Length > 0 && String.Compare(app.Args[0], "anneal", true) == 0)
{
train = new NeuralSimulatedAnnealing(
network, new PilotScore(), 10, 2, 100);
}
else
{
train = new MLMethodGeneticAlgorithm(() =>
{
BasicNetwork result = CreateNetwork();
((IMLResettable)result).Reset();
return(result);
}, new PilotScore(), 500);
}
int epoch = 1;
for (int i = 0; i < 50; i++)
{
train.Iteration();
Console.WriteLine(@"Epoch #" + epoch + @" Score:" + train.Error);
epoch++;
}
Console.WriteLine(@"\nHow the winning network landed:");
network = (BasicNetwork)train.Method;
var pilot = new NeuralPilot(network, true);
Console.WriteLine(pilot.ScorePilot());
EncogFramework.Instance.Shutdown();
}
void SaveTrainer(MLMethodGeneticAlgorithm Trainer)
{
var Basic = ((Trainer.Genetic.BestGenome as MLMethodGenome).Phenotype as BasicNetwork);
Encog.Persist.EncogDirectoryPersistence.SaveObject(new System.IO.FileInfo("./Network.net"), Basic);
}
