public static LSTMLayer readLSTM(string path)
{
FileStream fs = new FileStream(path, FileMode.Open);
BinaryFormatter bf = new BinaryFormatter();
LSTMLayer ps = bf.Deserialize(fs) as LSTMLayer;
fs.Close();
return(ps);
}
public static void Run()
{
System.DateTime currentTime = System.DateTime.Now;
readword();
Global.inputDim = 100;
Global.hiddenDim = 100;
readfourword();
Global.randn = new Normal();
DataSet X = new DataSet();
if (Global.isRead == 1)
{
string path = "model//deepNetwork";
Global.upLSTMLayer = LSTMLayer.readLSTM(path + "//lstmmodel.txt");
Global.upLSTMLayerr = LSTMLayer.readLSTM(path + "//lstmmodelr.txt");
Global.GRNNLayer1 = GRNNLayer.readGRNN(path + "//grnnmodel1.txt");
Global.GRNNLayer2 = GRNNLayer.readGRNN(path + "//grnnmodel2.txt");
Global.GRNNLayer3 = GRNNLayer.readGRNN(path + "//grnnmodel3.txt");
Global.GRNNLayer4 = GRNNLayer.readGRNN(path + "//grnnmodel4.txt");
Global.feedForwardLayer = FeedForwardLayer.readFF(path + "//feedforwardmodel.txt");
Global.wordEmbedding = LSTMLayer.getSerializeWordembedding(path + "//embedding.txt", Global.wordEmbedding);
}
else
{
Global.GRNNLayer1 = new GRNNLayer();
Global.GRNNLayer2 = new GRNNLayer();
Global.GRNNLayer3 = new GRNNLayer();
Global.GRNNLayer4 = new GRNNLayer();
Global.upLSTMLayer = new LSTMLayer();
Global.upLSTMLayerr = new LSTMLayer();
Global.feedForwardLayer = new FeedForwardLayer();
}
Trainer.train(X);
Global.swLog.Close();
System.DateTime currentTime_1 = System.DateTime.Now;
Console.WriteLine(currentTime_1 - currentTime);
Console.Read();
}
public static void train(DataSet X)
{
if (Global.mode == "test")
{
Console.WriteLine("begain testing......");
testAccuracy1 = runtestIteration(X.Testing, false);
Console.WriteLine("Epoch test f-score: {0}", (testAccuracy1 * 100).ToString("f3"));
//Console.WriteLine("Epoch test best f-score: {0}", (testAccuracy * 100).ToString("f3"));
Global.swLog.WriteLine("Epoch test fscore: {0}", (testAccuracy1 * 100).ToString("f3"));
//Global.swLog.WriteLine("Epoch test best fscore: {0}", (testAccuracy * 100).ToString("f3"));
Postprocessing.transfer("data/temp/" + "test_raw.txt");
Console.WriteLine("Finished");
}
//else if (Global.mode == "test")
//{
// Console.WriteLine("begain testing......");
// testAccuracy1 = runtestIteration(X.Testing, false);
// Postprocessing.transfer(Global.readFile);
// Console.WriteLine("predict fininshed");
// Global.swLog.WriteLine("predict fininshed");
//}
else if (Global.mode == "train")
{
for (int iter = 0; iter < Global.trainIter; iter++)
{
DateTime begin = DateTime.Now;
Console.WriteLine("\niter: {0}", iter + 1);
Global.swLog.WriteLine("\niter: {0}", iter + 1);
double trainAccuracy = runtrainIteration(X.Training, X.Testing, true, iter);
if (double.IsNaN(trainAccuracy) || double.IsInfinity(trainAccuracy))
{
Console.WriteLine("WARNING: invalid value for training loss. Try lowering learning rate.");
}
//test
testAccuracy1 = runtestIteration(X.Testing, false);
if (testAccuracy <= testAccuracy1)
{
LSTMLayer.SerializeWordembedding("model//deepNetwork//embedding");
//LSTMLayer.SerializeBigramWordembedding();
Global.upLSTMLayer.saveLSTM("model//deepNetwork//lstmmodel.txt");
Global.upLSTMLayerr.saveLSTM("model//deepNetwork//lstmmodelr.txt");
Global.GRNNLayer1.saveGRNN("model//deepNetwork//grnnmodel1.txt");
Global.GRNNLayer2.saveGRNN("model//deepNetwork//grnnmodel1.txt");
Global.GRNNLayer3.saveGRNN("model//deepNetwork//grnnmodel1.txt");
Global.GRNNLayer4.saveGRNN("model//deepNetwork//grnnmodel1.txt");
Global.feedForwardLayer.saveFFmodel("model//deepNetwork//feedforwardmodel.txt");
testAccuracy = testAccuracy1;
}
DateTime end = DateTime.Now;
// Console.WriteLine("train f-score: {0}", (trainAccuracy*100).ToString("f3"));
Console.WriteLine("test f-score: {0}", (testAccuracy * 100).ToString("f3"));
Console.WriteLine("test f-score: {0}", (testAccuracy1 * 100).ToString("f3"));
Console.WriteLine("time used: {0}", end - begin);
//Global.swLog.WriteLine("train f-score: {0}", (trainAccuracy * 100).ToString("f3"));
Global.swLog.WriteLine("test f-score: {0}", (testAccuracy * 100).ToString("f3"));
Global.swLog.WriteLine("test f-score: {0}", (testAccuracy1 * 100).ToString("f3"));
Global.swLog.WriteLine("time used: {0}", end - begin);
}
}
}
public static double runtrainIteration(List <DataSeq> X, List <DataSeq> Xtest, bool train, int iter)
{
List <DataSeq> x = new List <DataSeq>();
if (train)
{
x = shuffle(X);//shuffle every window (point)
}
TrainThread runThread = new TrainThread(train);
List <ManualResetEvent> manualEvents = new List <ManualResetEvent>();
List <DataStep> temp = new List <DataStep>();
int i = 0, j = 0;
int length = x.Count();
while (i < length)
{
if (i != 0 && i % 16 == 0)
{
testAccuracy1 = runtestIteration(Xtest, false);
if (testAccuracy <= testAccuracy1)
{
LSTMLayer.SerializeWordembedding("model//deepNetwork//embedding");
//LSTMLayer.SerializeBigramWordembedding();
Global.upLSTMLayer.saveLSTM("model//deepNetwork//lstmmodel.txt");
Global.upLSTMLayerr.saveLSTM("model//deepNetwork//lstmmodelr.txt");
Global.GRNNLayer1.saveGRNN("model//deepNetwork//grnnmodel1.txt");
Global.GRNNLayer2.saveGRNN("model//deepNetwork//grnnmodel1.txt");
Global.GRNNLayer3.saveGRNN("model//deepNetwork//grnnmodel1.txt");
Global.GRNNLayer4.saveGRNN("model//deepNetwork//grnnmodel1.txt");
Global.feedForwardLayer.saveFFmodel("model//deepNetwork//feedforwardmodel.txt");
testAccuracy = testAccuracy1;
}
Console.WriteLine("test f-score: {0}", (testAccuracy * 100).ToString("f3"));
Console.WriteLine("test1 f-score: {0}", (testAccuracy1 * 100).ToString("f3"));
Global.swLog.WriteLine("test f-score: {0}", (testAccuracy * 100).ToString("f3"));
Global.swLog.WriteLine("test f-score: {0}", (testAccuracy1 * 100).ToString("f3"));
}
for (int k = 0; k < Global.nThread && i < length; k++, i++)
{
ManualResetEvent mre = new ManualResetEvent(false);
param pa = new param(x[i]);
pa.mre = mre;
pa.datastep = x[i].datasteps;
manualEvents.Add(mre);
for (int m = 0; m < x[i].datasteps.Count; m++)
{
temp.Add(x[i].datasteps[m]);
}
ThreadPool.QueueUserWorkItem(new WaitCallback(runThread.run), pa);
}
WaitHandle.WaitAll(manualEvents.ToArray());
if (train)
{
UpdateWeight_rmProp(temp);
}
manualEvents.Clear();
temp.Clear();
}
return(runThread.accword / runThread.totalword);
}
static void Main(string[] args)
{
Console.WriteLine("Choose running mode: 1. training, 2. testing");
string mode = Console.ReadLine();
if (mode == "1")
{
Global.mode = "train";
}
else if (mode == "2")
{
Global.mode = "test";
}
//Console.WriteLine("Choose reading mode: 1. read saved model, 2. read model trained on MSR dataset.");
//string read = Console.ReadLine();
//string modes = Console.ReadLine();
//if (modes == "1")
//{
// Global.isRead = true;
//}
//else if (modes == "2")
//{
// Global.isRead = false;
//}
Console.WriteLine("Choose the bigram feature mode: 1. read bigram features, 2. create bigram features");
string bigramfeature = Console.ReadLine();
if (bigramfeature == "1")
{
Global.isReadBigramfeature = true;
}
else if (bigramfeature == "2")
{
Global.isReadBigramfeature = false;
}
readword();
readbigramword();
readidoimword();
Global.randn = new Normal();
Global._UpLSTMLayer = LSTMLayer.readLSTM("model\\lstmmodel.txt");
Global._UpLSTMLayerr = LSTMLayer.readLSTM("model\\lstmmodelr.txt");
// Global._LSTMLayerr = LSTMLayer.readLSTM("model\\lstmmodelr.txt");
Global._GRNNLayer1 = GRNNLayer.readGRNN("model\\grnnmodel1.txt");
Global._GRNNLayer2 = GRNNLayer.readGRNN("model\\grnnmodel2.txt");
Global._GRNNLayer3 = GRNNLayer.readGRNN("model\\grnnmodel3.txt");
Global._GRNNLayer4 = GRNNLayer.readGRNN("model\\grnnmodel4.txt");
Global._feedForwardLayer = FeedForwardLayer.readFF("model\\feedforwardmodel.txt");
LSTMLayer.getSerializeWordembedding();
LSTMLayer.getSerializeBigramWordembedding();
DataSet X = new DataSet();
Trainer.train(X);
Global.swLog.Close();
}