public void Play()
{
var iris = AyxCsvReader.Instance.ReadCsvFileDataTable("Data/iris.csv");
var labels = DataFormat.DataTableToList(iris, "Species").ToList();
var codec = new OneHotCodec(labels);
var y = codec.Encode(labels);
var X = DataFormat.DataTableToTensor(iris, 1, 2, 3, 4);
var model = new NeuralNetwork()
.AddFullLayer(5)
.AddReLU()
.AddFullLayer(5)
.AddReLU()
.AddFullLayer(codec.Length)
.AddSoftmax()
.UseCrossEntropyLoss()
.UseGradientDescent(0.001);
var trainer = new Trainer(model, 16, 200, true);
trainer.StartTrain(X, y, null, null);
var machine = new ClassificationMachine(model);
machine.LabelCodec = codec;
while (true)
{
Console.Write("input Sepal.Length:");
var a1 = double.Parse(Console.ReadLine());
Console.Write("input Sepal.Width:");
var a2 = double.Parse(Console.ReadLine());
Console.Write("input Petal.Length:");
var a3 = double.Parse(Console.ReadLine());
Console.Write("input Petal.Width:");
var a4 = double.Parse(Console.ReadLine());
var input = new TensorOld(new double[] { a1, a2, a3, a4 }, 1, 4);
Console.WriteLine($"you input is {machine.Predict(input).First()}, Codec:{machine.LastCodecResult}, Raw:{machine.LastRawResult}");
}
}
public void OneHotTest()
{
var oh = new OneHotCodec(data);
Assert.Equal(6, oh.Length);
var test = new List <string> {
"a", "c", "d", "a"
};
var expected = new TensorOld(new double[, ]
{
{ 1, 0, 0, 0, 0, 0 },
{ 0, 0, 1, 0, 0, 0 },
{ 0, 0, 0, 1, 0, 0 },
{ 1, 0, 0, 0, 0, 0 },
});
var encode = oh.Encode(test);
Assert.Equal(expected, encode);
var decode = oh.Decode(encode);
Assert.Equal(decode, test);
}