public RandomizedTrainer(SimpleNeuralNetworkFactory factory, double[][][] trainingData, ILogger <RandomizedTrainer> logger)
{
// todo: consider going back to double[,,,]
TrainingData = trainingData;
NumberOfDataSets = trainingData.Length;
NumberOfInputs = trainingData[0].Length;
NumberOfOutputs = trainingData[0][0].Length;
NumberOfHiddenNodes = 4;
Factory = factory;
Logger = logger;
}
public static void Main(string[] args)
{
//#error version
var data = new double[3][][];
for (var x = 0; x < data.Length; x++)
{
data[x] = new double[][] { new double[] { x }, new double[] { x *2 } };
}
var loggerFactory = LoggerFactory.Create(x =>
{
x.AddConsole();
x.SetMinimumLevel(LogLevel.Debug);
});
var factory = new SimpleNeuralNetworkFactory(1234);
var trainer = new RandomizedTrainer(factory, data, loggerFactory.CreateLogger <RandomizedTrainer>());
for (var x = 0; x < 5; x++)
{
trainer.Train();
Console.WriteLine($"epoch {x}: {trainer.BestError}");
}
// {5,5},
//{-6,-6}
//};
//var data = new double[,]
//{
//{1,1},
//{5,5},
//{-6,-6}
//};
//for (var x = 0; x < data.GetLength(0); x++)
//{
//var output = nn.run(data[x, 0]);
//var delta = data[x, 1] - output[0];
//Console.WriteLine($"[ {data[x, 0]} -> {output[0]}, {delta} ]");
////foreach (var o in output)
////{
////Console.Write(o + " ");
////}
//// Console.WriteLine("]");
//}
}
