public void CanParseTabDelimtedData()
{
var data = "1" + Environment.NewLine
+ "\"2.2\"\t\t0.3e1" + Environment.NewLine
+ "'4'\t5\t6";
var reader = new DelimitedReader <SparseMatrix>('\t')
{
CultureInfo = CultureInfo.InvariantCulture
};
var matrix = reader.ReadMatrix(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0, matrix[0, 0]);
Assert.AreEqual(0.0, matrix[0, 1]);
Assert.AreEqual(0.0, matrix[0, 2]);
Assert.AreEqual(2.2, matrix[1, 0]);
Assert.AreEqual(3.0, matrix[1, 1]);
Assert.AreEqual(0.0, matrix[1, 2]);
Assert.AreEqual(4.0, matrix[2, 0]);
Assert.AreEqual(5.0, matrix[2, 1]);
Assert.AreEqual(6.0, matrix[2, 2]);
}
public void CanParseCommaDelimitedData()
{
var data = "a,b,c" + Environment.NewLine
+ "(1,2)" + Environment.NewLine
+ "\"2.2\",0.3e1" + Environment.NewLine
+ "'(4,-5)',5,6" + Environment.NewLine;
var reader = new DelimitedReader <DenseMatrix>(',')
{
HasHeaderRow = true,
CultureInfo = CultureInfo.InvariantCulture
};
var matrix = reader.ReadMatrix(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0, matrix[0, 0].Real);
Assert.AreEqual(2.0, matrix[0, 0].Imaginary);
Assert.AreEqual(0.0, matrix[0, 1]);
Assert.AreEqual(0.0, matrix[0, 2]);
Assert.AreEqual(2.2, matrix[1, 0]);
Assert.AreEqual(3.0, matrix[1, 1]);
Assert.AreEqual(0.0, matrix[1, 2]);
Assert.AreEqual(4.0, matrix[2, 0].Real);
Assert.AreEqual(-5.0, matrix[2, 0].Imaginary);
Assert.AreEqual(5.0, matrix[2, 1]);
Assert.AreEqual(6.0, matrix[2, 2]);
}
public void CanParseWhiteSpaceDelimitedData()
{
var data = "1" + Environment.NewLine
+ "\"(2.2,3.3)\" 0.3e1" + Environment.NewLine
+ "'4' 5 6" + Environment.NewLine;
var reader = new DelimitedReader <UserDefinedMatrix>
{
CultureInfo = CultureInfo.InvariantCulture
};
var matrix = reader.ReadMatrix(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0, matrix[0, 0]);
Assert.AreEqual(0.0, matrix[0, 1]);
Assert.AreEqual(0.0, matrix[0, 2]);
Assert.AreEqual(2.2, matrix[1, 0].Real);
Assert.AreEqual(3.3, matrix[1, 0].Imaginary);
Assert.AreEqual(3.0, matrix[1, 1]);
Assert.AreEqual(0.0, matrix[1, 2]);
Assert.AreEqual(4.0, matrix[2, 0]);
Assert.AreEqual(5.0, matrix[2, 1]);
Assert.AreEqual(6.0, matrix[2, 2]);
}
public void CanParseComplex32CommaDelimitedData()
{
var data = "a,b,c" + Environment.NewLine
+ "(1,2)" + Environment.NewLine
+ "\"2.2\",0.3e1" + Environment.NewLine
+ "'(4,-5)',5,6" + Environment.NewLine;
var reader = new DelimitedReader
{
Delimiter = ",",
HasHeaderRow = true,
FormatProvider = CultureInfo.InvariantCulture
};
var matrix = reader.ReadMatrix <Complex32>(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0f, matrix[0, 0].Real);
Assert.AreEqual(2.0f, matrix[0, 0].Imaginary);
Assert.AreEqual((Complex32)0.0f, matrix[0, 1]);
Assert.AreEqual((Complex32)0.0f, matrix[0, 2]);
Assert.AreEqual((Complex32)2.2f, matrix[1, 0]);
Assert.AreEqual((Complex32)3.0f, matrix[1, 1]);
Assert.AreEqual((Complex32)0.0f, matrix[1, 2]);
Assert.AreEqual(4.0f, matrix[2, 0].Real);
Assert.AreEqual(-5.0f, matrix[2, 0].Imaginary);
Assert.AreEqual((Complex32)5.0f, matrix[2, 1]);
Assert.AreEqual((Complex32)6.0f, matrix[2, 2]);
}
/// <summary>
/// Load and return the digits dataset (classification).
/// Each datapoint is a 8x8 image of a digit.
/// ================= ==============
/// Classes 10
/// Samples per class ~180
/// Samples total 1797
/// Dimensionality 64
/// Features integers 0-16
/// ================= ==============
/// </summary>
/// <returns>Instance of <see cref="DigitsDataset"/>.</returns>
public static DigitsDataset Load()
{
var reader = new DelimitedReader {
Sparse = false, Delimiter = " "
};
var assembly = Assembly.GetAssembly(typeof(DigitsDataset));
string descr;
using (var sr = new StreamReader(assembly.GetManifestResourceStream("Sharpkit.Learn.Datasets.Data.digits.rst")))
{
descr = sr.ReadToEnd();
}
using (var datastream = assembly.GetManifestResourceStream("Sharpkit.Learn.Datasets.Data.digits.csv.gz"))
{
var data = reader.ReadMatrix <double>(new GZipStream(datastream, CompressionMode.Decompress));
return(new DigitsDataset
{
Data = data.SubMatrix(0, data.RowCount, 0, data.ColumnCount - 1),
Target = data.Column(data.ColumnCount - 1).ToArray().Cast <int>().ToArray(),
Description = descr
});
}
}
public void TestBackPropagationWithoutRegularisaion()
{
NeuralNetwork.CostFunctionWithThetaParameter backProp =
t =>
{
var backPropagationResult = NeuralNetwork.BackPropagation(Xm, ym, t, new List <int>()
{
4
}, 4, 0);
return(backPropagationResult);
};
var bothThetas = NeuralNetwork.PackThetas(new List <Matrix <double> >()
{
theta1NN, theta2NN
});
var resultNumericalGradient = NeuralNetwork.ComputeNumericalGradient(backProp, bothThetas);
var pathNumericalGradients = Path.Combine(Utils.GetAssemblyPath(), "..\\..\\TestData\\NumericalGradientForBackPropagationWithoutRegularisation.txt");
var matrixNumericalGradients = DelimitedReader.Read <double>(pathNumericalGradients);
Assert.IsTrue(MatricesEqual(resultNumericalGradient, matrixNumericalGradients));
var resultBackPropagation = NeuralNetwork.BackPropagation(Xm, ym, bothThetas, new List <int>()
{
4
}, 4, 0);
var pathGradientForBackPropagation = Path.Combine(Utils.GetAssemblyPath(), "..\\..\\TestData\\GradientForBackPropagationWithoutRegularisation.txt");
var matrixGradientForBackPropagation = DelimitedReader.Read <double>(pathGradientForBackPropagation);
Assert.IsTrue(Equalities.DoubleEquals(resultBackPropagation.Item1, 3.08744915815864));
Assert.IsTrue(MatricesEqual(resultBackPropagation.Item2, matrixGradientForBackPropagation));
Assert.IsTrue(MatricesEqual(resultBackPropagation.Item2, resultNumericalGradient, 0.000000001));
}
public void CanParseCommaDelimitedData()
{
var data = "a,b,c" + Environment.NewLine
+ "1" + Environment.NewLine
+ "\"2.2\",0.3e1" + Environment.NewLine
+ "'4',5,6" + Environment.NewLine;
var reader = new DelimitedReader<DenseMatrix>(',')
{
HasHeaderRow = true,
CultureInfo = CultureInfo.InvariantCulture
};
var matrix = reader.ReadMatrix(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0f, matrix[0, 0]);
Assert.AreEqual(0.0f, matrix[0, 1]);
Assert.AreEqual(0.0f, matrix[0, 2]);
Assert.AreEqual(2.2f, matrix[1, 0]);
Assert.AreEqual(3.0f, matrix[1, 1]);
Assert.AreEqual(0.0f, matrix[1, 2]);
Assert.AreEqual(4.0f, matrix[2, 0]);
Assert.AreEqual(5.0f, matrix[2, 1]);
Assert.AreEqual(6.0f, matrix[2, 2]);
}
public void TestForwardPropagation()
{
var resultFeedForward = NeuralNetwork.ForwardPropagation(Xm, new List <Matrix <double> >()
{
theta1NN, theta2NN
});
var h = resultFeedForward.OutputActivation;
var a1 = resultFeedForward.NodesActivations.ElementAt(0);
var a2 = resultFeedForward.NodesActivations.ElementAt(1);
var z2 = resultFeedForward.NodesOutputs.ElementAt(0);
var z3 = resultFeedForward.NodesOutputs.ElementAt(1);
var pathH = Path.Combine(Utils.GetAssemblyPath(), "..\\..\\TestData\\FeedForward_h.txt");
var patha1 = Path.Combine(Utils.GetAssemblyPath(), "..\\..\\TestData\\FeedForward_a1.txt");
var patha2 = Path.Combine(Utils.GetAssemblyPath(), "..\\..\\TestData\\FeedForward_a2.txt");
var pathz2 = Path.Combine(Utils.GetAssemblyPath(), "..\\..\\TestData\\FeedForward_z2.txt");
var pathz3 = Path.Combine(Utils.GetAssemblyPath(), "..\\..\\TestData\\FeedForward_z3.txt");
var wantedH = DelimitedReader.Read <double>(pathH);
var wanteda1 = DelimitedReader.Read <double>(patha1);
var wanteda2 = DelimitedReader.Read <double>(patha2);
var wantedz2 = DelimitedReader.Read <double>(pathz2);
var wantedz3 = DelimitedReader.Read <double>(pathz3);
Assert.IsTrue(MatricesEqual(h, wantedH));
Assert.IsTrue(MatricesEqual(a1, wanteda1));
Assert.IsTrue(MatricesEqual(a2, wanteda2));
Assert.IsTrue(MatricesEqual(z2, wantedz2));
Assert.IsTrue(MatricesEqual(z3, wantedz3));
}
void Start()
{
trainedModelValues = DelimitedReader.Read <double>(Application.dataPath + "/trained.csv", false, ",", false, System.Globalization.CultureInfo.InvariantCulture.NumberFormat).Column(0);
ship = Instantiate(spaceShipPrefab, Vector3.zero, Quaternion.identity);
ship.GetComponent <SpaceShip>().InitializeBrain(trainedModelValues);
ship.GetComponent <SpaceShip>().StartEngine(s => Restart());
}
void Start()
{
Control.UseNativeMKL();
G = DelimitedReader.Read <float>("Gmatrix" + objname + ".csv", false, ",", false);
D = DelimitedReader.Read <float>("Dmatrix" + objname + ".csv", false, ",", false);
//print("Gmatrix" + objname + ".csv");
}
public void CanParseCommaDelimitedData()
{
var data = "a,b,c" + Environment.NewLine
+ "1" + Environment.NewLine
+ "\"2.2\",0.3e1" + Environment.NewLine
+ "'4',5,6" + Environment.NewLine;
var reader = new DelimitedReader
{
Delimiter = ",",
HasHeaderRow = true,
FormatProvider = CultureInfo.InvariantCulture
};
var matrix = reader.ReadMatrix <double>(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0, matrix[0, 0]);
Assert.AreEqual(0.0, matrix[0, 1]);
Assert.AreEqual(0.0, matrix[0, 2]);
Assert.AreEqual(2.2, matrix[1, 0]);
Assert.AreEqual(3.0, matrix[1, 1]);
Assert.AreEqual(0.0, matrix[1, 2]);
Assert.AreEqual(4.0, matrix[2, 0]);
Assert.AreEqual(5.0, matrix[2, 1]);
Assert.AreEqual(6.0, matrix[2, 2]);
}
public void CanParseCommaDelimitedData()
{
const string data =
@"a,b,c
1
""2.2"",0.3e1
'4',5,6
";
var reader = new DelimitedReader<DenseMatrix>(',')
{
HasHeaderRow = true
};
var matrix = reader.ReadMatrix(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0f, matrix[0, 0]);
Assert.AreEqual(0.0f, matrix[0, 1]);
Assert.AreEqual(0.0f, matrix[0, 2]);
Assert.AreEqual(2.2f, matrix[1, 0]);
Assert.AreEqual(3.0f, matrix[1, 1]);
Assert.AreEqual(0.0f, matrix[1, 2]);
Assert.AreEqual(4.0f, matrix[2, 0]);
Assert.AreEqual(5.0f, matrix[2, 1]);
Assert.AreEqual(6.0f, matrix[2, 2]);
}
public void CanParsePeriodDelimitedData()
{
const string data =
@"a.b.c
1
""2,2"".0,3e1
'4,0'.5,0.6,0
";
var reader = new DelimitedReader<DenseMatrix>('.')
{
HasHeaderRow = true,
CultureInfo = new CultureInfo("tr-TR")
};
var matrix = reader.ReadMatrix(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0f, matrix[0, 0]);
Assert.AreEqual(0.0f, matrix[0, 1]);
Assert.AreEqual(0.0f, matrix[0, 2]);
Assert.AreEqual(2.2f, matrix[1, 0]);
Assert.AreEqual(3.0f, matrix[1, 1]);
Assert.AreEqual(0.0f, matrix[1, 2]);
Assert.AreEqual(4.0f, matrix[2, 0]);
Assert.AreEqual(5.0f, matrix[2, 1]);
Assert.AreEqual(6.0f, matrix[2, 2]);
}
public void CanParsePeriodDelimitedData()
{
var data = "a.b.c" + Environment.NewLine
+ "1" + Environment.NewLine
+ "\"2,2\".0,3e1" + Environment.NewLine
+ "'4,0'.5,0.6,0" + Environment.NewLine;
var reader = new DelimitedReader
{
Delimiter = ".",
HasHeaderRow = true,
FormatProvider = new CultureInfo("tr-TR")
};
var matrix = reader.ReadMatrix <double>(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0, matrix[0, 0]);
Assert.AreEqual(0.0, matrix[0, 1]);
Assert.AreEqual(0.0, matrix[0, 2]);
Assert.AreEqual(2.2, matrix[1, 0]);
Assert.AreEqual(3.0, matrix[1, 1]);
Assert.AreEqual(0.0, matrix[1, 2]);
Assert.AreEqual(4.0, matrix[2, 0]);
Assert.AreEqual(5.0, matrix[2, 1]);
Assert.AreEqual(6.0, matrix[2, 2]);
}
public void CanParseWhiteSpaceDelimitedData()
{
var data = "1" + Environment.NewLine
+ "\"2.2\" 0.3e1" + Environment.NewLine
+ "'4' 5 6" + Environment.NewLine;
var reader = new DelimitedReader
{
FormatProvider = CultureInfo.InvariantCulture
};
var matrix = reader.ReadMatrix <float>(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0f, matrix[0, 0]);
Assert.AreEqual(0.0f, matrix[0, 1]);
Assert.AreEqual(0.0f, matrix[0, 2]);
Assert.AreEqual(2.2f, matrix[1, 0]);
Assert.AreEqual(3.0f, matrix[1, 1]);
Assert.AreEqual(0.0f, matrix[1, 2]);
Assert.AreEqual(4.0f, matrix[2, 0]);
Assert.AreEqual(5.0f, matrix[2, 1]);
Assert.AreEqual(6.0f, matrix[2, 2]);
}
public void CanParsePeriodDelimitedData()
{
var data = "a.b.c" + Environment.NewLine
+ "1" + Environment.NewLine
+ "\"2,2\".0,3e1+0,2e1i" + Environment.NewLine
+ "'4,0'.5,0.6,0" + Environment.NewLine;
var reader = new DelimitedReader <DenseMatrix>('.')
{
HasHeaderRow = true,
CultureInfo = new CultureInfo("tr-TR")
};
var matrix = reader.ReadMatrix(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0, matrix[0, 0]);
Assert.AreEqual(0.0, matrix[0, 1]);
Assert.AreEqual(0.0, matrix[0, 2]);
Assert.AreEqual(2.2, matrix[1, 0]);
Assert.AreEqual(3.0, matrix[1, 1].Real);
Assert.AreEqual(2.0, matrix[1, 1].Imaginary);
Assert.AreEqual(0.0, matrix[1, 2]);
Assert.AreEqual(4.0, matrix[2, 0]);
Assert.AreEqual(5.0, matrix[2, 1]);
Assert.AreEqual(6.0, matrix[2, 2]);
}
private ResultModel item_predict(string pathFile)
{
ResultModel result = new ResultModel();
// load data
Matrix <double> data = DelimitedReader.Read <double>(pathFile, false, "\t", false, System.Globalization.CultureInfo.InvariantCulture.NumberFormat);
result.raw_data = data;
int n_users = data.Column(0).Distinct().Count();
int n_items = data.Column(1).Distinct().Count();
double test_split = 0.1;
double train_split = 1 - test_split;
int train_row = (int)(data.RowCount * train_split);
int train_col = data.ColumnCount;
int test_row = data.RowCount - train_row;
int test_col = train_col;
Matrix <double> train_data = DenseMatrix.Create(train_row, train_col, 0);
Matrix <double> test_data = DenseMatrix.Create(test_row, test_col, 0);
train_data = data.SubMatrix(0, train_row, 0, train_col);
test_data = data.SubMatrix(train_row, test_row, 0, test_col);
// Create two user-item matrices, one for training and another for testing
Matrix <double> train_data_matrix = DenseMatrix.Create(n_users, n_items, 0);
for (int line = 0; line < train_data.RowCount; line++)
{
int userId = (int)(train_data[line, 0]);
int itemId = (int)(train_data[line, 1] - 1);
train_data_matrix[userId, itemId] = train_data[line, 2];
}
Matrix <double> test_data_matrix = DenseMatrix.Create(n_users, n_items, 0);
for (int line = 0; line < test_data.RowCount; line++)
{
int userId = (int)(test_data[line, 0]);
int itemId = (int)(test_data[line, 1] - 1);
test_data_matrix[userId, itemId] = test_data[line, 2];
}
result.train_data = train_data_matrix;
result.test_data = test_data_matrix;
// consine similarity calculate
Matrix <double> item_similarity = Recommender_System.cosin(train_data_matrix.Transpose());
// predict
Matrix <double> item_prediction = Recommender_System.predict(train_data_matrix, item_similarity, "item");
result.predic_data = item_prediction;
// rmse error
double item_rmse = RMSE.calRMSE(item_prediction, test_data_matrix);
result.rmse = item_rmse;
return(result);
}
private void button1_Click(object sender, EventArgs e)
{
//takes a matrix in a text file (1,1),(3,4) and the same underneath.. and reads them in
//args for method = file path, booleal true if sparse matrix, string delimiter, boolean true if it has headers
Matrix <Complex32> fileMatrix = DelimitedReader.Read <Complex32>(@"D:\My Documents\Paul Docs\Education, training, employment\Job\1 Research\C#\2 Tutorials Vids\3 Odds n Sods\3 Math.Net Vector math\csv.txt", false, ",", false);
textBox1.AppendText(fileMatrix.ToString("F2"));
}
float Alpha = 0.000020f; //rayleigh damping, metal
void Start()
{
forceamp = 1000f;
forceoffset = 0.1f;
G = DelimitedReader.Read <float>("Gmatrix.csv", false, ",", false);
D = DelimitedReader.Read <float>("Dmatrix.csv", false, ",", false);
M = DelimitedReader.Read <float>("Mmatrix.csv", false, ",", false);
}
public void TestSigmoid()
{
var sigmoid = NeuralNetwork.Sigmoid(X);
var path = Path.Combine(Utils.GetAssemblyPath(), "..\\..\\TestData\\Sigmoid.txt");
var matrixToCompare = (DenseMatrix)DelimitedReader.Read <double>(path);
Assert.IsTrue(MatricesEqual(sigmoid, matrixToCompare));
}
/// <summary>
/// Initialize current model.
/// </summary>
public override void Initialize()
{
try
{
// Try to initialize model state from files.
layers = new List <Layer>();
int count = 0;
int inputCount = Config.Inputs;
foreach (var layer in Config.Layers)
{
var state = DelimitedReader.Read <double>(Path(String.Format("layers/{0}", count)));
// If either number of neurons or number of inputs
// (omitting bias) is wrong, format is incorrect.
if (state.RowCount != layer.NeuronCount || state.ColumnCount - 1 != inputCount)
{
throw new Exception("Incorrect format");
}
var neurons = new List <Neuron>();
foreach (var neuronState in state.ToRowArrays())
{
var weights = Vector <double> .Build.Dense(neuronState).SubVector(1, neuronState.Length - 1);
neurons.Add(new Neuron(inputCount, weights, neuronState[0], layer.Function.ToString()));
}
layers.Add(new Layer(neurons));
// Set current layer neuron count as next layer input count.
inputCount = layer.NeuronCount;
count++;
}
Loaded = true;
}
catch
{
// If fails, generate new one from scratch.
layers = new List <Layer>();
foreach (var layer in Config.Layers)
{
int inputCount;
if (layers.Count == 0)
{
// First layer input count is network input count.
inputCount = Config.Inputs;
}
else
{
// All subsequent layers input count is previous layer size.
inputCount = layers.Last().Size;
}
layers.Add(Layer.Generate(layer.NeuronCount, inputCount, layer.Function.ToString()));
}
}
}
public static RawMatrix Read(StreamReader str)
{
var mtx = new RawMatrix()
{
Scale = Double.Parse(str.ReadLine()),
m = DelimitedReader.Read <double>(str),
BlockSize = ulong.Parse(str.ReadLine())
};
return(mtx);
}
public void Reader_ReturnVal_Split()
{
IEnumerable <String> strings = null;
using (var reader = new DelimitedReader(TestInputFileBlank))
{
strings = reader.Read();
}
Assert.That(strings, Is.EquivalentTo(new[] { "Shelby Macias", "3027 Lorem St.|Kokomo|Hertfordshire|L9T 3D5|England", "1 66 890 3865-9584", "*****@*****.**" }));
}
private DenseMatrix readMatrixChild(XmlNode node, string childName)
{
XmlElement element = node [childName];
if (element == null)
{
return(null);
}
return(DenseMatrix.OfMatrix(DelimitedReader.Read <double> (new StringReader(element.InnerText))));
}
static public RawVector Read(StreamReader str)
{
RawVector vct = new RawVector(0)
{
BlockSize = ulong.Parse(str.ReadLine()),
Scale = Double.Parse(str.ReadLine()),
v = DelimitedReader.Read <double>(str).Column(0),
};
return(vct);
}
/// <summary>
/// Load and return the diabetes dataset (regression).
/// ============== ==================
/// Samples total 442
/// Dimensionality 10
/// Features real, -.2 < x < .2
/// Targets integer 25 - 346
/// ============== ==================
/// </summary>
/// <returns>
/// Instance of <see cref="DiabetesDataset"/> class.
/// </returns>
public static DiabetesDataset Load()
{
var reader = new DelimitedReader { Sparse = false, Delimiter = " " };
var assembly = Assembly.GetAssembly(typeof(DiabetesDataset));
using (var datastream = assembly.GetManifestResourceStream("Sharpkit.Learn.Datasets.Data.diabetes_data.csv.gz"))
using (var targetstream = assembly.GetManifestResourceStream("Sharpkit.Learn.Datasets.Data.diabetes_target.csv.gz"))
{
var data = reader.ReadMatrix<double>(new GZipStream(datastream, CompressionMode.Decompress));
var target = reader.ReadMatrix<double>(new GZipStream(targetstream, CompressionMode.Decompress));
return new DiabetesDataset { Data = data, Target = target.Column(0) };
}
}
public void ShouldParseWhenLastStringHasNoTerminator()
{
var bytes = new List <byte>();
bytes.AddRange(Encoding.ASCII.GetBytes("abc"));
bytes.Add(0);
bytes.AddRange(Encoding.ASCII.GetBytes("def"));
var reader = new DelimitedReader(new StreamReader(new MemoryStream(bytes.ToArray())));
Assert.Equal("abc", reader.Read());
Assert.Equal("def", reader.Read());
Assert.Null(reader.Read());
}
public static RawMatrix Read(StreamReader str)
{
var mtx = new RawMatrix()
{
Scale = Double.Parse(str.ReadLine()),
m = DelimitedReader.Read <double>(str),
BlockSize = ulong.Parse(str.ReadLine())
};
Max = Math.Max(Max, mtx.m.Enumerate().Max(x => (x < 0) ? -x : x));
return(mtx);
}
public void Reader_WithOut_Split()
{
IEnumerable <String> strings = null;
bool success = false;
using (var reader = new DelimitedReader(TestInputFileBlank))
{
success = reader.TryRead(out strings);
}
Assert.That(strings, Is.EquivalentTo(new[] { "Shelby Macias", "3027 Lorem St.|Kokomo|Hertfordshire|L9T 3D5|England", "1 66 890 3865-9584", "*****@*****.**" }));
Assert.That(success, Is.True);
}
public void TestCostFunction()
{
var tupleJAndGrad = NeuralNetwork.CostFunction(theta, X, y);
var pathJ = Path.Combine(Utils.GetAssemblyPath(), "..\\..\\TestData\\J.txt");
var matrixJ = (DenseMatrix)DelimitedReader.Read <double>(pathJ);
var pathGrad = Path.Combine(Utils.GetAssemblyPath(), "..\\..\\TestData\\Grad.txt");
var matrixGrad = (DenseMatrix)DelimitedReader.Read <double>(pathGrad);
Assert.IsTrue(Equalities.DoubleEquals(tupleJAndGrad.Item1, matrixJ[0, 0]));
Assert.IsTrue(MatricesEqual(tupleJAndGrad.Item2, matrixGrad));
}
public bool Load(string folderPath)
{
bool result = false;
string layerPath = folderPath;
if (Directory.Exists(folderPath) == true)
{
var activationsFilePath = String.Format("{0}\\Activations.csv", layerPath);
if (File.Exists(activationsFilePath))
{
Matrix <double> a = DelimitedReader.Read <double>(activationsFilePath, sparse: false, delimiter: ",");
var arrays = a.ToRowArrays();
Activations = DenseVector.Build.DenseOfArray(arrays[0]);
}
else
{
throw new System.IO.DirectoryNotFoundException(activationsFilePath);
}
if (Name != "Input")
{
var biasesFilePath = String.Format("{0}\\Biases.csv", layerPath);
if (File.Exists(biasesFilePath))
{
Matrix <double> a = DelimitedReader.Read <double>(biasesFilePath, sparse: false, delimiter: ",");
var arrays = a.ToRowArrays();
Biases = DenseVector.Build.DenseOfArray(arrays[0]);
}
else
{
throw new System.IO.DirectoryNotFoundException(biasesFilePath);
}
var weightsFilePath = String.Format("{0}\\Weights.csv", layerPath);
if (File.Exists(weightsFilePath))
{
Weights = DelimitedReader.Read <double>(weightsFilePath, sparse: false, delimiter: ",");
}
else
{
throw new System.IO.DirectoryNotFoundException(weightsFilePath);
}
}
result = true;
}
else
{
throw new System.IO.DirectoryNotFoundException(folderPath);
}
return(result);
}
static void Main(string[] args)
{
Utils.SetNumberDecimalSeparatorToDotInCultureSettings();
var x_data = DelimitedReader.Read <double>(Path.Combine(Utils.GetAssemblyPath(), "..\\..\\TestData\\x_data.txt"));
var y_data = DelimitedReader.Read <double>(Path.Combine(Utils.GetAssemblyPath(), "..\\..\\TestData\\y_data.txt"));
NeuralNetwork.CostFunctionWithThetaParameter backProp =
tt =>
{
var backPropagationResult = NeuralNetwork.BackPropagation(x_data, y_data, tt, new List <int>()
{
25
}, 10, 0);
return(backPropagationResult);
};
var thetas = NeuralNetwork.RandomInitialiseWeights(400, 10, new List <int>()
{
25
});
var theta = NeuralNetwork.PackThetas(thetas);
var resultGradientDescent = NeuralNetwork.GradientDescent(backProp, theta, 1, 300);
DelimitedWriter.Write("..\\..\\TestData\\ThetaGradient.txt", resultGradientDescent.Item1, "thetaGradient");
DelimitedWriter.Write("..\\..\\TestData\\JHistory.txt", resultGradientDescent.Item2, "JHistory");
var resultTheta = resultGradientDescent.Item1;
//var resultComputeNumericalGradient = NeuralNetwork.ComputeNumericalGradient(backProp, resultTheta);
//resultComputeNumericalGradient.Save("D:/test/numericalGradients.txt");
var result5 = NeuralNetwork.BackPropagation(x_data, y_data, resultTheta, new List <int>(25), 10, 0);
var JJ = result5.Item1;
DelimitedWriter.Write("..\\..\\TestData\\GradientForBackPropagation.txt", result5.Item2, "gradBackPropagation");
var resultThetaLoadedFromFile = (DenseMatrix)DelimitedReader.Read <double>("..\\..\\TestData\\thetaGradient.txt");
var thetasList = NeuralNetwork.UnpackThetas(resultThetaLoadedFromFile, 400, new List <int>()
{
25
}, 10);
var result = NeuralNetwork.GetPredictions(x_data, y_data, thetasList);
Console.WriteLine(result.Item1);
DelimitedWriter.Write("..\\..\\TestData\\Predictions.txt", result.Item2, "");
}
private void OnEnable()
{
if (readflag == 0)
{
Control.UseNativeMKL();
G = DelimitedReader.Read <float>("Gmatrix" + objname + ".csv", false, ",", false);
D = DelimitedReader.Read <float>("Dmatrix" + objname + ".csv", false, ",", false);
displacedVertices = this.GetComponentInParent <MeshFilter>().sharedMesh.vertices;
Sound = this.GetComponent <Sound>();
soundsamples = new float[44100];
D_data = D.ToRowMajorArray();
readflag = 0;
}
}
/// <summary>
/// Initialize current model.
/// </summary>
public override void Initialize()
{
try
{
var state = DelimitedReader.Read <double>(Path(Config.State)).Row(0);
var bias = state[0];
var weights = state.SubVector(1, state.Count - 1);
perceptron = new Neuron(Config.Inputs, weights, bias, Config.Function.ToString());
}
catch
{
perceptron = Neuron.Generate(Config.Inputs, Config.Function.ToString());
}
}
/// <summary>
/// Load and return the digits dataset (classification).
/// Each datapoint is a 8x8 image of a digit.
/// ================= ==============
/// Classes 10
/// Samples per class ~180
/// Samples total 1797
/// Dimensionality 64
/// Features integers 0-16
/// ================= ==============
/// </summary>
/// <returns>Instance of <see cref="DigitsDataset"/>.</returns>
public static DigitsDataset Load()
{
var reader = new DelimitedReader { Sparse = false, Delimiter = " " };
var assembly = Assembly.GetAssembly(typeof(DigitsDataset));
string descr;
using (var sr = new StreamReader(assembly.GetManifestResourceStream("Sharpkit.Learn.Datasets.Data.digits.rst")))
{
descr = sr.ReadToEnd();
}
using (var datastream = assembly.GetManifestResourceStream("Sharpkit.Learn.Datasets.Data.digits.csv.gz"))
{
var data = reader.ReadMatrix<double>(new GZipStream(datastream, CompressionMode.Decompress));
return new DigitsDataset
{
Data = data.SubMatrix(0, data.RowCount, 0, data.ColumnCount - 1),
Target = data.Column(data.ColumnCount - 1).ToArray().Cast<int>().ToArray(),
Description = descr
};
}
}
public void CanParseTabDelimtedData()
{
var data = "1"
+ Environment.NewLine
+ "\"2.2\"\t\t0.3e1"
+ Environment.NewLine
+ "'4'\t5\t6";
var reader = new DelimitedReader<SparseMatrix>('\t');
var matrix = reader.ReadMatrix(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0f, matrix[0, 0]);
Assert.AreEqual(0.0f, matrix[0, 1]);
Assert.AreEqual(0.0f, matrix[0, 2]);
Assert.AreEqual(2.2f, matrix[1, 0]);
Assert.AreEqual(3.0f, matrix[1, 1]);
Assert.AreEqual(0.0f, matrix[1, 2]);
Assert.AreEqual(4.0f, matrix[2, 0]);
Assert.AreEqual(5.0f, matrix[2, 1]);
Assert.AreEqual(6.0f, matrix[2, 2]);
}
public void CanParsePeriodDelimitedData()
{
var data = "a.b.c" + Environment.NewLine
+ "1" + Environment.NewLine
+ "\"2,2\".0,3e1" + Environment.NewLine
+ "'4,0'.5,0.6,0" + Environment.NewLine;
var reader = new DelimitedReader
{
Delimiter = ".",
HasHeaderRow = true,
FormatProvider = new CultureInfo("tr-TR")
};
var matrix = reader.ReadMatrix<double>(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0, matrix[0, 0]);
Assert.AreEqual(0.0, matrix[0, 1]);
Assert.AreEqual(0.0, matrix[0, 2]);
Assert.AreEqual(2.2, matrix[1, 0]);
Assert.AreEqual(3.0, matrix[1, 1]);
Assert.AreEqual(0.0, matrix[1, 2]);
Assert.AreEqual(4.0, matrix[2, 0]);
Assert.AreEqual(5.0, matrix[2, 1]);
Assert.AreEqual(6.0, matrix[2, 2]);
}
public void CanParseComplex32CommaDelimitedData()
{
var data = "a,b,c" + Environment.NewLine
+ "(1,2)" + Environment.NewLine
+ "\"2.2\",0.3e1" + Environment.NewLine
+ "'(4,-5)',5,6" + Environment.NewLine;
var reader = new DelimitedReader
{
Delimiter = ",",
HasHeaderRow = true,
FormatProvider = CultureInfo.InvariantCulture
};
var matrix = reader.ReadMatrix<Complex32>(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0f, matrix[0, 0].Real);
Assert.AreEqual(2.0f, matrix[0, 0].Imaginary);
Assert.AreEqual((Complex32)0.0f, matrix[0, 1]);
Assert.AreEqual((Complex32)0.0f, matrix[0, 2]);
Assert.AreEqual((Complex32)2.2f, matrix[1, 0]);
Assert.AreEqual((Complex32)3.0f, matrix[1, 1]);
Assert.AreEqual((Complex32)0.0f, matrix[1, 2]);
Assert.AreEqual(4.0f, matrix[2, 0].Real);
Assert.AreEqual(-5.0f, matrix[2, 0].Imaginary);
Assert.AreEqual((Complex32)5.0f, matrix[2, 1]);
Assert.AreEqual((Complex32)6.0f, matrix[2, 2]);
}
public void CanParsePeriodDelimitedData()
{
var data = "a.b.c" + Environment.NewLine
+ "1" + Environment.NewLine
+ "\"2,2\".0,3e1+0,2e1i" + Environment.NewLine
+ "'4,0'.5,0.6,0" + Environment.NewLine;
var reader = new DelimitedReader<DenseMatrix>('.')
{
HasHeaderRow = true,
CultureInfo = new CultureInfo("tr-TR")
};
var matrix = reader.ReadMatrix(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual((Complex32)1.0f, matrix[0, 0]);
Assert.AreEqual((Complex32)0.0f, matrix[0, 1]);
Assert.AreEqual((Complex32)0.0f, matrix[0, 2]);
Assert.AreEqual((Complex32)2.2f, matrix[1, 0]);
Assert.AreEqual(3.0f, matrix[1, 1].Real);
Assert.AreEqual(2.0f, matrix[1, 1].Imaginary);
Assert.AreEqual((Complex32)0.0f, matrix[1, 2]);
Assert.AreEqual((Complex32)4.0f, matrix[2, 0]);
Assert.AreEqual((Complex32)5.0f, matrix[2, 1]);
Assert.AreEqual((Complex32)6.0f, matrix[2, 2]);
}
public void CanParseWhiteSpaceDelimitedData()
{
const string data =
@"1
""2.2"" 0.3e1
'4' 5 6
";
var reader = new DelimitedReader<UserDefinedMatrix>();
var matrix = reader.ReadMatrix(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0f, matrix[0, 0]);
Assert.AreEqual(0.0f, matrix[0, 1]);
Assert.AreEqual(0.0f, matrix[0, 2]);
Assert.AreEqual(2.2f, matrix[1, 0]);
Assert.AreEqual(3.0f, matrix[1, 1]);
Assert.AreEqual(0.0f, matrix[1, 2]);
Assert.AreEqual(4.0f, matrix[2, 0]);
Assert.AreEqual(5.0f, matrix[2, 1]);
Assert.AreEqual(6.0f, matrix[2, 2]);
}
public void CanParseTabDelimitedData()
{
var data = "1" + Environment.NewLine
+ "\"2.2\"\t\t0.3e1" + Environment.NewLine
+ "'4'\t5\t6";
var reader = new DelimitedReader<SparseMatrix>('\t')
{
CultureInfo = CultureInfo.InvariantCulture
};
var matrix = reader.ReadMatrix(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual((Complex32)1.0f, matrix[0, 0]);
Assert.AreEqual((Complex32)0.0f, matrix[0, 1]);
Assert.AreEqual((Complex32)0.0f, matrix[0, 2]);
Assert.AreEqual((Complex32)2.2f, matrix[1, 0]);
Assert.AreEqual((Complex32)3.0f, matrix[1, 1]);
Assert.AreEqual((Complex32)0.0f, matrix[1, 2]);
Assert.AreEqual((Complex32)4.0f, matrix[2, 0]);
Assert.AreEqual((Complex32)5.0f, matrix[2, 1]);
Assert.AreEqual((Complex32)6.0f, matrix[2, 2]);
}
public void CanParseWhiteSpaceDelimitedData()
{
var data = "1" + Environment.NewLine
+ "\"2.2\" 0.3e1" + Environment.NewLine
+ "'4' 5 6" + Environment.NewLine;
var reader = new DelimitedReader<UserDefinedMatrix>
{
CultureInfo = CultureInfo.InvariantCulture
};
var matrix = reader.ReadMatrix(new MemoryStream(Encoding.UTF8.GetBytes(data)));
Assert.AreEqual(3, matrix.RowCount);
Assert.AreEqual(3, matrix.ColumnCount);
Assert.AreEqual(1.0f, matrix[0, 0]);
Assert.AreEqual(0.0f, matrix[0, 1]);
Assert.AreEqual(0.0f, matrix[0, 2]);
Assert.AreEqual(2.2f, matrix[1, 0]);
Assert.AreEqual(3.0f, matrix[1, 1]);
Assert.AreEqual(0.0f, matrix[1, 2]);
Assert.AreEqual(4.0f, matrix[2, 0]);
Assert.AreEqual(5.0f, matrix[2, 1]);
Assert.AreEqual(6.0f, matrix[2, 2]);
}
