static void Main(string[] args)
{
Console.WriteLine("Sine Wave Test");
int seed;
using (var rng = new RNGCryptoServiceProvider())
{
var buffer = new byte[sizeof(int)];
rng.GetBytes(buffer);
seed = BitConverter.ToInt32(buffer, 0);
}
RandomProvider.SetSeed(seed);
var filename = "SineWave.xml";
var serializer = new DataContractSerializer(typeof(IEnumerable <Layer>), new Type[] { typeof(LSTM), typeof(HyperbolicTangent), typeof(Sigmoid), typeof(FullyConnected), typeof(Activation), typeof(Identity) });
var trainingDataList = new List <ValueTuple <double[], double[]> >();
var dataSize = 100;
var maxLength = 200;
Model model;
for (int i = 0, j = 0; i < dataSize; i++, j += maxLength)
{
var x = new double[maxLength];
var y = new double[maxLength];
for (var k = 0; k < maxLength; k++)
{
x[k] = Math.Sin((j + k) * 0.01 * Math.PI);
y[k] = Math.Sin((j + k + maxLength) * 0.01 * Math.PI);
}
trainingDataList.Add(ValueTuple.Create <double[], double[]>(x, y));
}
if (File.Exists(filename))
{
using (XmlReader xmlReader = XmlReader.Create(filename))
{
model = new Model((IEnumerable <Layer>)serializer.ReadObject(xmlReader));
}
}
else
{
int epochs = 100;
int iterations = 1;
model = new Model(
new LSTM(1, 128, maxLength, true, false, (fanIn, fanOut) => Initializers.LeCunNormal(fanIn),
new FullyConnected(128, maxLength, (fanIn, fanOut) => Initializers.LeCunNormal(fanIn),
new Activation(maxLength, new Identity()))));
model.Stepped += (sender, args) =>
{
if (iterations % 10 == 0)
{
Console.WriteLine("Epoch {0}/{1}", iterations, epochs);
}
iterations++;
};
Console.WriteLine("Training...");
var stopwatch = Stopwatch.StartNew();
model.Fit(trainingDataList, epochs, 10, new SGD(), new MeanSquaredError());
stopwatch.Stop();
Console.WriteLine("Done ({0}).", stopwatch.Elapsed.ToString());
}
var dataList = new List <IEnumerable <double> >();
var data = trainingDataList[0];
var vector = model.Predict(data.Item1);
for (var i = 0; i < data.Item1.Length; i++)
{
dataList.Add(new double[] { data.Item1[i], data.Item2[i], vector[i] });
}
using (var fs = new FileStream("SineWaveTest.csv", FileMode.Create, FileAccess.Write, FileShare.Read))
using (var s = new StreamWriter(fs, System.Text.Encoding.UTF8))
{
dataList.ForEach(x =>
{
s.Write(String.Join(",", x));
s.Write(Environment.NewLine);
});
fs.Flush();
}
XmlWriterSettings settings = new XmlWriterSettings();
settings.Indent = true;
settings.Encoding = new System.Text.UTF8Encoding(false);
using (XmlWriter xmlWriter = XmlWriter.Create(filename, settings))
{
serializer.WriteObject(xmlWriter, model.Layers);
xmlWriter.Flush();
}
}
public Decoder(int sequenceLength, int vocabularySize, int wordVectorSize, int hiddenSize) : base(sequenceLength, sequenceLength * vocabularySize)
{
this.embedding = new Embedding(sequenceLength, vocabularySize, wordVectorSize, (fanIn, fanOut) => 0.01 * Initializers.LeCunNormal(fanIn));
this.recurrent = new LSTM(wordVectorSize, hiddenSize, sequenceLength, true, false, (fanIn, fanOut) => Initializers.LeCunNormal(fanIn));
this.attention = new Attention(hiddenSize, sequenceLength);
this.fullyConnected = new FullyConnected(hiddenSize * 2, sequenceLength, sequenceLength * vocabularySize, (fanIn, fanOut) => Initializers.LeCunNormal(fanIn));
this.weights = new double[this.embedding.Weights.Length + this.recurrent.Weights.Length + this.fullyConnected.Weights.Length];
for (int i = 0; i < this.embedding.Weights.Length; i++)
{
this.weights[i] = this.embedding.Weights[i];
}
for (int i = 0, j = this.embedding.Weights.Length; i < this.recurrent.Weights.Length; i++, j++)
{
this.weights[j] = this.recurrent.Weights[i];
}
for (int i = 0, j = this.embedding.Weights.Length + this.recurrent.Weights.Length; i < this.fullyConnected.Weights.Length; i++, j++)
{
this.weights[j] = this.fullyConnected.Weights[i];
}
}
