//[InlineData(new int[] { 6, 3 })]
public void ImageRecognitionTest()
{
int size = 64;
int numberOfOutputs = 1;
int numberOfInputs = size * size;
var context = getImgRecognitionDescriptor(numberOfInputs);
LearningApi api = new LearningApi(context);
api.UseActionModule <object, double[][]>((notUsed, ctx) =>
{
return(getImageData(size, $"{Directory.GetCurrentDirectory()}\\MLPerceptron\\TrainingImages"));
//return getSomOtherData($"{Directory.GetCurrentDirectory()}\\MLPerceptron\\TestFiles\\TrainingData.csv");
});
// High number of hidden neurons in first layer brings network to constant result for everything.
// var hiddenLayerNeurons = new int[] { size*size, 3 };
var hiddenLayerNeurons = new int[] { 6000, 9, 3 };
api.UseMLPerceptron(0.01, 6, 1, 1, hiddenLayerNeurons);
Stopwatch sw = new Stopwatch();
sw.Start();
IScore score = api.Run() as IScore;
sw.Stop();
Trace.WriteLine($"Duration:{(double)(sw.ElapsedMilliseconds / 1000 / 60)} min");
var testImageData = getImageData(size, $"{Directory.GetCurrentDirectory()}\\MLPerceptron\\TestingImages");
//var testImageData= getSomOtherData($"{Directory.GetCurrentDirectory()}\\MLPerceptron\\TestFiles\\TestData.csv");
MLPerceptronResult result = api.Algorithm.Predict(testImageData, api.Context) as MLPerceptronResult;
// float accuracy = MLPHelpers.GetAccuracy(testImageData, result.results, numberOfOutputs);
}
public void UnitTestSdr(int iterations, double learningrate, int batchSize, int[] hiddenLayerNeurons, int iterationnumber)
{
LearningApi api = new LearningApi();
api.UseActionModule <object, double[][]>((notUsed, ctx) =>
{
List <double[]> rows = new List <double[]>();
ctx.DataDescriptor = new DataDescriptor();
var trainingFiles = Directory.GetFiles($"{Directory.GetCurrentDirectory()}\\MLPerceptron\\TestFiles\\Sdr");
int rowCnt = 0;
foreach (var file in trainingFiles)
{
using (var reader = new StreamReader(file))
{
string line;
while ((line = reader.ReadLine()) != null)
{
var tokens = line.Split(",");
List <string> newTokens = new List <string>();
foreach (var token in tokens)
{
if (token != " ")
{
newTokens.Add(token);
}
}
tokens = newTokens.ToArray();
if (rowCnt == 0)
{
ctx.DataDescriptor.Features = new LearningFoundation.DataMappers.Column[tokens.Length - 1];
for (int i = 1; i < tokens.Length; i++)
{
ctx.DataDescriptor.Features[i - 1] = new LearningFoundation.DataMappers.Column
{
Id = i,
Index = i,
Type = LearningFoundation.DataMappers.ColumnType.BINARY
};
}
ctx.DataDescriptor.LabelIndex = -1;
}
// We have 65 features and digit number in file. to encode digits 0-9.
// Digits can be represented as 9 bits.
double[] row = new double[tokens.Length - 1 + 10];
for (int i = 0; i < tokens.Length; i++)
{
row[i] = double.Parse(tokens[i], CultureInfo.InvariantCulture);
}
//
// This code encodes 9 digit classes as last 9 bits of training vector.
for (int k = 0; k < 10; k++)
{
if (double.Parse(tokens[0], CultureInfo.InvariantCulture) == k)
{
row[tokens.Length - 1 + k] = 1;
}
else
{
row[tokens.Length - 1 + k] = 0;
}
}
rows.Add(row);
}
}
rowCnt++;
}
return(rows.ToArray());
});
//int[] hiddenLayerNeurons = { 6 };
// Invoke the MLPerecptronAlgorithm with a specific learning rate, number of iterations
api.UseMLPerceptron(learningrate, iterations, batchSize, iterationnumber, hiddenLayerNeurons);
MLPerceptronAlgorithmScore score = api.Run() as MLPerceptronAlgorithmScore;
api.Save("SdrMnistModel");
}
public void UnitTestCreditApproval()
{
// Read the csv file which contains the training data
using (var readerTrainData = new StreamReader($"{Directory.GetCurrentDirectory()}\\MLPerceptron\\TestFiles\\TrainingDataCreditApproval.csv"))
{
int lineCountTrainData = 0;
int indexTrainData = 0;
int numberOfOutputs = 1;
while (readerTrainData.ReadLine() != null)
{
lineCountTrainData++;
}
double[][] data = new double[lineCountTrainData - 1][];
List <double>[] listTrainData = new List <double> [lineCountTrainData - 1];
LearningApi api = new LearningApi();
api.UseActionModule <object, double[][]>((notUsed, ctx) =>
{
ctx.DataDescriptor = new DataDescriptor();
readerTrainData.DiscardBufferedData();
readerTrainData.BaseStream.Seek(0, SeekOrigin.Begin);
var firstLine = readerTrainData.ReadLine();
var firstLineValues = firstLine.Split(',');
double[] maxValue = new double[15];
double[] minValue = new double[15];
double[] meanValue = new double[15];
List <double>[] listDataSpecifics = new List <double> [15];
ctx.DataDescriptor.Features = new LearningFoundation.DataMappers.Column[firstLineValues.Length - 1];
for (int i = 0; i < (firstLineValues.Length - 1); i++)
{
ctx.DataDescriptor.Features[i] = new LearningFoundation.DataMappers.Column
{
Id = i,
Name = firstLineValues[i],
Type = LearningFoundation.DataMappers.ColumnType.NUMERIC,
Index = i,
};
}
ctx.DataDescriptor.LabelIndex = firstLineValues.Length - 1;
for (int i = 0; i < 15; i++)
{
listDataSpecifics[i] = new List <double>();
}
while (!readerTrainData.EndOfStream)
{
var line = readerTrainData.ReadLine();
var values = line.Split(',');
listTrainData[indexTrainData] = new List <double>();
for (int i = 0; i < 15; i++)
{
listDataSpecifics[i].Add(Convert.ToDouble(values[i], CultureInfo.InvariantCulture));
}
for (int i = 0; i < values.Length; i++)
{
listTrainData[indexTrainData].Add(Convert.ToDouble(values[i], CultureInfo.InvariantCulture));
}
data[indexTrainData] = listTrainData[indexTrainData].ToArray();
indexTrainData++;
}
for (int i = 0; i < 15; i++)
{
maxValue[i] = listDataSpecifics[i].Max();
minValue[i] = listDataSpecifics[i].Min();
meanValue[i] = (listDataSpecifics[i].Sum()) / (lineCountTrainData - 1);
}
for (int i = 0; i < lineCountTrainData - 1; i++)
{
for (int j = 0; j < 15; j++)
{
data[i][j] = (data[i][j] - meanValue[j]) / (maxValue[j] - minValue[j]);
}
}
return(data);
});
int[] hiddenLayerNeurons = { 9 };
// Invoke the MLPerecptronAlgorithm with a specific learning rate, number of iterations, number of hidden layers
api.UseMLPerceptron(0.1, 20, 1, 1, hiddenLayerNeurons);
IScore score = api.Run() as IScore;
// Read the csv file which contains the training data
using (var readerTestData = new StreamReader($"{Directory.GetCurrentDirectory()}\\MLPerceptron\\TestFiles\\TestDataCreditApproval.csv"))
{
int lineCountTestData = 0;
int indexTestData = 0;
while (readerTestData.ReadLine() != null)
{
lineCountTestData++;
}
double[][] testData = new double[lineCountTestData - 1][];
List <double>[] listTestData = new List <double> [lineCountTestData - 1];
readerTestData.DiscardBufferedData();
readerTestData.BaseStream.Seek(0, SeekOrigin.Begin);
var firstLine = readerTestData.ReadLine();
var firstLineValues = firstLine.Split(',');
double[] maxValue = new double[15];
double[] minValue = new double[15];
double[] meanValue = new double[15];
List <double>[] listDataSpecifics = new List <double> [15];
for (int i = 0; i < 15; i++)
{
listDataSpecifics[i] = new List <double>();
}
while (!readerTestData.EndOfStream)
{
var line = readerTestData.ReadLine();
var values = line.Split(',');
listTestData[indexTestData] = new List <double>();
for (int i = 0; i < 15; i++)
{
listDataSpecifics[i].Add(Convert.ToDouble(values[i], CultureInfo.InvariantCulture));
}
for (int i = 0; i < values.Length; i++)
{
listTestData[indexTestData].Add(Convert.ToDouble(values[i], CultureInfo.InvariantCulture));
}
testData[indexTestData] = listTestData[indexTestData].ToArray();
indexTestData++;
}
for (int i = 0; i < 15; i++)
{
maxValue[i] = listDataSpecifics[i].Max();
minValue[i] = listDataSpecifics[i].Min();
meanValue[i] = (listDataSpecifics[i].Sum()) / (lineCountTestData - 1);
}
for (int i = 0; i < lineCountTestData - 1; i++)
{
for (int j = 0; j < 15; j++)
{
testData[i][j] = (testData[i][j] - meanValue[j]) / (maxValue[j] - minValue[j]);
}
}
// Invoke the Predict method to predict the results on the test data
var result = ((MLPerceptronResult)api.Algorithm.Predict(testData, api.Context)).results;
int creditApprovedCorrectClassification = 0, creditDeclinedCorrectClassification = 0;
// Check how many samples of the test data have been correctly classified by the neural network
for (int i = 0; i < lineCountTestData - 1; i++)
{
for (int j = 0; j < numberOfOutputs; j++)
{
if (testData[i][(testData[i].Length - numberOfOutputs) + j] == (result[i * numberOfOutputs + j] >= 0.5 ? 1 : 0))
{
if (testData[i][(testData[i].Length - numberOfOutputs) + j] == 0)
{
creditDeclinedCorrectClassification++;
}
else
{
creditApprovedCorrectClassification++;
}
}
}
}
}
}
}
public void UnitTestMNIST(int iterations, double learningrate, int batchSize, int[] hiddenLayerNeurons, int testCaseNumber)
{
MNISTFileRead mnistObj = new MNISTFileRead();
MNISTFileRead.ReadMNISTTrainingData();
MNISTFileRead.ReadMNISTTestData();
// TODO
// test by using same test cases in dependence on number of iterations.
// test by dependence on number of layers
// test by dependence on number of neurons in layers.
// [2 layers]: [x,x] [1,2] [1,3] [1,4]
// Read the csv file which contains the training data
int numberOfOutputs = 10;
LearningApi api = new LearningApi();
api.UseActionModule <object, double[][]>((notUsed, ctx) =>
{
ctx.DataDescriptor = new DataDescriptor();
ctx.DataDescriptor.Features = new LearningFoundation.DataMappers.Column[MNISTFileRead.firstLineValues.Length - 1];
for (int i = 0; i < (MNISTFileRead.firstLineValues.Length - 1); i++)
{
ctx.DataDescriptor.Features[i] = new LearningFoundation.DataMappers.Column
{
Id = i,
Name = MNISTFileRead.firstLineValues[i],
Type = LearningFoundation.DataMappers.ColumnType.NUMERIC,
Index = i,
};
}
ctx.DataDescriptor.LabelIndex = MNISTFileRead.firstLineValues.Length - 1;
return(MNISTFileRead.trainingData);
});
//int[] hiddenLayerNeurons = { 6 };
// Invoke the MLPerecptronAlgorithm with a specific learning rate, number of iterations
api.UseMLPerceptron(learningrate, iterations, batchSize, testCaseNumber, hiddenLayerNeurons);
IScore score = api.Run() as IScore;
// Invoke the Predict method to predict the results on the test data
var result = ((MLPerceptronResult)api.Algorithm.Predict(MNISTFileRead.testData, api.Context)).results;
int accurateResults = 0;
// Check if the test data has been correctly classified by the neural network
for (int i = 0; i < MNISTFileRead.testData.Length; i++)
{
accurateResults++;
for (int j = 0; j < numberOfOutputs; j++)
{
//Assert.True(testData[i][(testData[i].Length - numberOfOutputs) + j] == (result[i * numberOfOutputs + j] >= 0.5 ? 1 : 0));
if (MNISTFileRead.testData[i][(MNISTFileRead.testData[i].Length - numberOfOutputs) + j] != (result[i * numberOfOutputs + j] >= 0.5 ? 1 : 0))
{
accurateResults--;
break;
}
}
}
/*
* float loss = expectedResults / (numberOfOutputs) / testData.Length;
* Debug.WriteLine($"{hiddleLayerToString(hiddenLayerNeurons)} - Loss: {loss}");
*/
double accuracy = ((double)accurateResults * numberOfOutputs) / result.Length;
Debug.WriteLine($"Accuracy: {accuracy}");
}
public void UnitTestOne()
{
// Read the csv file which contains the training data
using (var readerTrainData = new StreamReader($"{Directory.GetCurrentDirectory()}\\MLPerceptron\\TestFiles\\TrainingData.csv"))
{
int lineCountTrainData = 0;
int indexTrainData = 0;
int numberOfOutputs = 0;
while (readerTrainData.ReadLine() != null)
{
lineCountTrainData++;
}
double[][] data = new double[lineCountTrainData - 1][];
List <double>[] listTrainData = new List <double> [lineCountTrainData - 1];
LearningApi api = new LearningApi();
api.UseActionModule <object, double[][]>((notUsed, ctx) =>
{
ctx.DataDescriptor = new DataDescriptor();
readerTrainData.DiscardBufferedData();
readerTrainData.BaseStream.Seek(0, SeekOrigin.Begin);
var firstLine = readerTrainData.ReadLine();
var firstLineValues = firstLine.Split(',');
foreach (var value in firstLineValues)
{
if (value.Contains("Output"))
{
numberOfOutputs++;
}
}
ctx.DataDescriptor.Features = new LearningFoundation.DataMappers.Column[firstLineValues.Length - numberOfOutputs];
for (int i = 0; i < (firstLineValues.Length - numberOfOutputs); i++)
{
ctx.DataDescriptor.Features[i] = new LearningFoundation.DataMappers.Column
{
Id = i,
Name = firstLineValues[i],
Type = LearningFoundation.DataMappers.ColumnType.NUMERIC,
Index = i,
};
}
ctx.DataDescriptor.LabelIndex = firstLineValues.Length - numberOfOutputs;
while (!readerTrainData.EndOfStream)
{
var line = readerTrainData.ReadLine();
var values = line.Split(',');
listTrainData[indexTrainData] = new List <double>();
foreach (var value in values)
{
listTrainData[indexTrainData].Add(Convert.ToDouble(value, CultureInfo.InvariantCulture));
}
data[indexTrainData] = listTrainData[indexTrainData].ToArray();
indexTrainData++;
}
return(data);
});
int[] hiddenLayerNeurons = { 6 };
// Invoke the MLPerecptronAlgorithm with a specific learning rate, number of iterations, and number of hidden layer neurons
api.UseMLPerceptron(0.1, 10000, 1, 1, hiddenLayerNeurons);
IScore score = api.Run() as IScore;
// Read the csv file which contains the test data
using (var readerTestData = new StreamReader($"{Directory.GetCurrentDirectory()}\\MLPerceptron\\TestFiles\\testData.csv"))
{
int lineCountTestData = 0;
int indexTestData = 0;
while (readerTestData.ReadLine() != null)
{
lineCountTestData++;
}
double[][] testData = new double[lineCountTestData - 1][];
List <double>[] listTestData = new List <double> [lineCountTestData - 1];
readerTestData.DiscardBufferedData();
readerTestData.BaseStream.Seek(0, SeekOrigin.Begin);
readerTestData.ReadLine();
while (!readerTestData.EndOfStream)
{
var line = readerTestData.ReadLine();
var values = line.Split(',');
listTestData[indexTestData] = new List <double>();
foreach (var value in values)
{
listTestData[indexTestData].Add(Convert.ToDouble(value, CultureInfo.InvariantCulture));
}
testData[indexTestData] = listTestData[indexTestData].ToArray();
indexTestData++;
}
// Invoke the Predict method to predict the results on the test data
var result = ((MLPerceptronResult)api.Algorithm.Predict(testData, api.Context)).results;
int expectedResults = 0;
Debug.WriteLine("\r\nTraining completed.");
Debug.WriteLine("-------------------------------------------");
Debug.WriteLine($"Testing {lineCountTestData - 1} samples.");
// Check if the test data has been correctly classified by the neural network
for (int i = 0; i < lineCountTestData - 1; i++)
{
Debug.WriteLine("");
for (int j = 0; j < numberOfOutputs; j++)
{
//Assert.True(testData[i][(testData[i].Length - numberOfOutputs) + j] == (result[i * numberOfOutputs + j] >= 0.5 ? 1 : 0));
if (testData[i][(testData[i].Length - numberOfOutputs) + j] == (result[i * numberOfOutputs + j] >= 0.5 ? 1 : 0))
{
expectedResults++;
}
Debug.Write($"{i.ToString("D2")} - ");
for (int k = 0; k < api.Context.DataDescriptor.Features.Length; k++)
{
Debug.Write($"[{testData[i][k]}] ");
}
Debug.Write("\t\t");
Debug.Write($"Expected: {testData[i][(testData[i].Length - numberOfOutputs) + j]} - Predicted: {testData[i][(testData[i].Length - numberOfOutputs) + j]} - Result: {testData[i][(testData[i].Length - numberOfOutputs) + j] == (result[i * numberOfOutputs + j] >= 0.5 ? 1 : 0)} --\t\t");
}
}
float accuracy = (float)expectedResults / (float)numberOfOutputs / (float)testData.Length;
Debug.WriteLine("-------------------------------------------");
Debug.WriteLine($"Accuracy: {accuracy * 100}%");
}
}
}
//[InlineData(new int[] { 7, 2 })]
//[InlineData(new int[] { 8, 2 })]
//[InlineData(new int[] { 9, 2 })]
//[InlineData(new int[] { 10, 2 })]
//[InlineData(new int[] { 20, 2 })]
//[InlineData(new int[] { 30, 2 })]
//[InlineData(new int[] { 30, 20, 10, 5 })]
//[InlineData(new int[] { 15, 10, 5, 2 })]
//[InlineData(new int[] { 50, 30, 20, 10 })]
//[InlineData(new int[] { 10, 7, 5, 3 })]
//[InlineData(new int[] { 125, 77, 45, 34 , 19, 12, 9, 3})]
//[InlineData(new int[] { 50, 30, 20 })]
public void UnitTestIris(int iterations, double learningrate, int batchSize, int[] hiddenLayerNeurons, int iterationnumber)
{
// TODO
// test by using same test cases in dependence on number of iterations.
// test by dependence on number of layers
// test by dependence on number of neurons in layers.
// [2 layers]: [x,x] [1,2] [1,3] [1,4]
// Read the csv file which contains the training data
using (var readerTrainData = new StreamReader($"{Directory.GetCurrentDirectory()}\\MLPerceptron\\TestFiles\\TrainingDataIris.csv"))
{
int lineCountTrainData = 0;
int indexTrainData = 0;
int numberOfOutputs = 2;
while (readerTrainData.ReadLine() != null)
{
lineCountTrainData++;
}
double[][] data = new double[lineCountTrainData - 1][];
List <double>[] listTrainData = new List <double> [lineCountTrainData - 1];
LearningApi api = new LearningApi();
api.UseActionModule <object, double[][]>((notUsed, ctx) =>
{
ctx.DataDescriptor = new DataDescriptor();
readerTrainData.DiscardBufferedData();
readerTrainData.BaseStream.Seek(0, SeekOrigin.Begin);
var firstLine = readerTrainData.ReadLine();
var firstLineValues = firstLine.Split(',');
double[] maxValue = new double[firstLineValues.Length - 1];
double[] minValue = new double[firstLineValues.Length - 1];
double[] meanValue = new double[firstLineValues.Length - 1];
List <double>[] listDataSpecifics = new List <double> [firstLineValues.Length - 1];
ctx.DataDescriptor.Features = new LearningFoundation.DataMappers.Column[firstLineValues.Length - 1];
for (int i = 0; i < (firstLineValues.Length - 1); i++)
{
ctx.DataDescriptor.Features[i] = new LearningFoundation.DataMappers.Column
{
Id = i,
Name = firstLineValues[i],
Type = LearningFoundation.DataMappers.ColumnType.NUMERIC,
Index = i,
};
}
ctx.DataDescriptor.LabelIndex = firstLineValues.Length - 1;
for (int i = 0; i < firstLineValues.Length - 1; i++)
{
listDataSpecifics[i] = new List <double>();
}
while (!readerTrainData.EndOfStream)
{
var line = readerTrainData.ReadLine();
var values = line.Split(',');
listTrainData[indexTrainData] = new List <double>();
for (int i = 0; i < firstLineValues.Length - 1; i++)
{
listDataSpecifics[i].Add(Convert.ToDouble(values[i], CultureInfo.InvariantCulture));
}
encodeIrisValues(indexTrainData, listTrainData, values);
data[indexTrainData] = listTrainData[indexTrainData].ToArray();
indexTrainData++;
}
for (int i = 0; i < firstLineValues.Length - 1; i++)
{
maxValue[i] = listDataSpecifics[i].Max();
minValue[i] = listDataSpecifics[i].Min();
meanValue[i] = (listDataSpecifics[i].Sum()) / (lineCountTrainData - 1);
}
for (int i = 0; i < lineCountTrainData - 1; i++)
{
for (int j = 0; j < firstLineValues.Length - 1; j++)
{
data[i][j] = (data[i][j] - meanValue[j]) / (maxValue[j] - minValue[j]);
}
}
return(data);
});
//int[] hiddenLayerNeurons = { 6 };
// Invoke the MLPerecptronAlgorithm with a specific learning rate, number of iterations
api.UseMLPerceptron(learningrate, iterations, batchSize, iterationnumber, hiddenLayerNeurons);
IScore score = api.Run() as IScore;
// Read the csv file which contains the test data
using (var readerTestData = new StreamReader($"{Directory.GetCurrentDirectory()}\\MLPerceptron\\TestFiles\\TestDataIris.csv"))
{
int lineCountTestData = 0;
int indexTestData = 0;
while (readerTestData.ReadLine() != null)
{
lineCountTestData++;
}
double[][] testData = new double[lineCountTestData - 1][];
List <double>[] listTestData = new List <double> [lineCountTestData - 1];
readerTestData.DiscardBufferedData();
readerTestData.BaseStream.Seek(0, SeekOrigin.Begin);
var firstLine = readerTestData.ReadLine();
var firstLineValues = firstLine.Split(',');
double[] maxValue = new double[firstLineValues.Length - 1];
double[] minValue = new double[firstLineValues.Length - 1];
double[] meanValue = new double[firstLineValues.Length - 1];
List <double>[] listDataSpecifics = new List <double> [firstLineValues.Length - 1];
for (int i = 0; i < firstLineValues.Length - 1; i++)
{
listDataSpecifics[i] = new List <double>();
}
while (!readerTestData.EndOfStream)
{
var line = readerTestData.ReadLine();
var values = line.Split(',');
listTestData[indexTestData] = new List <double>();
for (int i = 0; i < firstLineValues.Length - 1; i++)
{
listDataSpecifics[i].Add(Convert.ToDouble(values[i], CultureInfo.InvariantCulture));
}
encodeIrisValues(indexTestData, listTestData, values);
testData[indexTestData] = listTestData[indexTestData].ToArray();
indexTestData++;
}
for (int i = 0; i < firstLineValues.Length - 1; i++)
{
maxValue[i] = listDataSpecifics[i].Max();
minValue[i] = listDataSpecifics[i].Min();
meanValue[i] = (listDataSpecifics[i].Sum()) / (lineCountTestData - 1);
}
for (int i = 0; i < lineCountTestData - 1; i++)
{
for (int j = 0; j < firstLineValues.Length - 1; j++)
{
testData[i][j] = (testData[i][j] - meanValue[j]) / (maxValue[j] - minValue[j]);
}
}
// Invoke the Predict method to predict the results on the test data
var result = ((MLPerceptronResult)api.Algorithm.Predict(testData, api.Context)).results;
float expectedResults = 0;
// Check if the test data has been correctly classified by the neural network
for (int i = 0; i < lineCountTestData - 1; i++)
{
for (int j = 0; j < numberOfOutputs; j++)
{
//Assert.True(testData[i][(testData[i].Length - numberOfOutputs) + j] == (result[i * numberOfOutputs + j] >= 0.5 ? 1 : 0));
if (testData[i][(testData[i].Length - numberOfOutputs) + j] == (result[i * numberOfOutputs + j] >= 0.5 ? 1 : 0))
{
expectedResults++;
}
}
}
float loss = expectedResults / (numberOfOutputs) / testData.Length;
Debug.WriteLine($"{hiddleLayerToString(hiddenLayerNeurons)} - Loss: {loss}");
}
}
}
public void UnitTestMNISTTestRunner(int iterations, double learningrate, int batchSize, int[] hiddenLayerNeurons, int testCaseNumber)
{
int numberOfOutputs = 10;
LearningApi api = new LearningApi();
api.UseActionModule <object, double[][]>((notUsed, ctx) =>
{
ctx.DataDescriptor = new DataDescriptor();
ctx.DataDescriptor.Features = new LearningFoundation.DataMappers.Column[firstLineValues.Length - 1];
for (int i = 0; i < (firstLineValues.Length - 1); i++)
{
ctx.DataDescriptor.Features[i] = new LearningFoundation.DataMappers.Column
{
Id = i,
Name = firstLineValues[i],
Type = LearningFoundation.DataMappers.ColumnType.NUMERIC,
Index = i,
};
}
ctx.DataDescriptor.LabelIndex = firstLineValues.Length - 1;
return(trainingData);
});
//int[] hiddenLayerNeurons = { 6 };
// Invoke the MLPerecptronAlgorithm with a specific learning rate, number of iterations
api.UseMLPerceptron(learningrate, iterations, batchSize, testCaseNumber, hiddenLayerNeurons);
IScore score = api.Run() as IScore;
// Invoke the Predict method to predict the results on the test data
var result = ((MLPerceptronResult)api.Algorithm.Predict(testData, api.Context)).results;
StreamWriter resultfile = new StreamWriter($"{Directory.GetCurrentDirectory()}\\MLPerceptron\\TestFiles\\mnist_result_" + testCaseNumber.ToString() + ".csv");
double[] temparray = new double[numberOfOutputs];
int index = 0;
while (index < result.Length)
{
for (int i = index; i < index + numberOfOutputs; i++)
{
temparray[i - index] = result[i];
}
double max = temparray.Max();
resultfile.WriteLine(Array.IndexOf(temparray, max));
index = index + numberOfOutputs;
}
int accurateResults = 0;
// Check if the test data has been correctly classified by the neural network
for (int i = 0; i < testData.Length; i++)
{
accurateResults++;
for (int j = 0; j < numberOfOutputs; j++)
{
//Assert.True(testData[i][(testData[i].Length - numberOfOutputs) + j] == (result[i * numberOfOutputs + j] >= 0.5 ? 1 : 0));
if (testData[i][(testData[i].Length - numberOfOutputs) + j] != (result[i * numberOfOutputs + j] >= 0.5 ? 1 : 0))
{
accurateResults--;
break;
}
}
}
/*
* float loss = expectedResults / (numberOfOutputs) / testData.Length;
* Debug.WriteLine($"{hiddleLayerToString(hiddenLayerNeurons)} - Loss: {loss}");
*/
double accuracy = ((double)accurateResults * numberOfOutputs) / result.Length;
resultfile.WriteLine("Accuracy = {0}", accuracy.ToString());
resultfile.Close();
}
