Synapses

A neural networks library for C#!

// run
dotnet add package SynapsesCSharp --version 7.4.1
// in the directory of your project

Neural Network

Create a neural network

Import Synapses, call NeuralNetwork.init and provide the size of each layer.

using SynapsesCSharp;
int[] layers = {4, 6, 5, 3};
NeuralNetwork neuralNetwork = NeuralNetwork.init(layers);

neuralNetwork has 4 layers. The first layer has 4 input nodes and the last layer has 3 output nodes. There are 2 hidden layers with 6 and 5 neurons respectively.

Get a prediction

double[] inputValues = {1.0, 0.5625, 0.511111, 0.47619};
double[] prediction =
    NeuralNetwork.prediction(neuralNetwork, inputValues);

prediction should be something like [ 0.8296, 0.6996, 0.4541 ].

Note that the lengths of inputValues and prediction equal to the sizes of input and output layers respectively.

Fit network

double learningRate = 0.5;
double[] expectedOutput = {0.0, 1.0, 0.0};
NeuralNetwork fitNetwork =
    NeuralNetwork.fit(
        neuralNetwork,
        learningRate,
        inputValues,
        expectedOutput
    );

fitNetwork is a new neural network trained with a single observation.

To train a neural network, you should fit with multiple datapoints

Create a customized neural network

The activation function of the neurons created with NeuralNetwork.init, is a sigmoid one. If you want to customize the activation functions and the weight distribution, call NeuralNetwork.customizedInit.

ActivationFunction activationF(int layerIndex) {
    switch (layerIndex) {
        case 0:
            return ActivationFunction.sigmoid;
        case 1:
            return ActivationFunction.identity;
        case 2:
            return ActivationFunction.leakyReLU;
        default:
            return ActivationFunction.tanh;
    }
}

double weightInitF(int _layerIndex) {
    Random rnd = new Random();
    return 1.0 - 2.0 * rnd.NextDouble();
}

NeuralNetwork customizedNetwork =
    NeuralNetwork.customizedInit(
        layers,
        activationF,
        weightInitF
    );

Visualization

Call NeuralNetwork.toSvg to take a brief look at its svg drawing.

The color of each neuron depends on its activation function while the transparency of the synapses depends on their weight.

string svg = NeuralNetwork.toSvg(customizedNetwork);

Save and load a neural network

JSON instances are compatible across platforms! We can generate, train and save a neural network in Python and then load and make predictions in Javascript!

toJson

Call NeuralNetwork.toJson on a neural network and get a string representation of it. Use it as you like. Save json in the file system or insert into a database table.

string json = NeuralNetwork.toJson(customizedNetwork);

ofJson

NeuralNetwork loadedNetwork = NeuralNetwork.ofJson(json);

As the name suggests, NeuralNetwork.ofJson turns a json string into a neural network.

Encoding and decoding

One hot encoding is a process that turns discrete attributes into a list of 0.0 and 1.0. Minmax normalization scales continuous attributes into values between 0.0 and 1.0. You can use DataPreprocessor for datapoint encoding and decoding.

The first parameter of DataPreprocessor.init is a list of tuples (attributeName, discreteOrNot).

Dictionary<string, string> setosaDatapoint =
    new Dictionary<string, string>()
        {
            {"petal_length", "1.5"},
            {"petal_length", "1.5"},
            {"petal_width", "0.1"},
            {"sepal_length", "4.9"},
            {"sepal_width", "3.1"},
            {"species", "setosa"}
        };

Dictionary<string, string> versicolorDatapoint =
    new Dictionary<string, string>()
        {
            {"petal_length", "3.8"},
            {"petal_width", "1.1"},
            {"sepal_length", "5.5"},
            {"sepal_width", "2.4"},
            {"species", "versicolor"}
        };

Dictionary<string, string> virginicaDatapoint =
    new Dictionary<string, string>()
        {
            {"petal_length", "6.0"},
            {"petal_width", "2.2"},
            {"sepal_length", "5.0"},
            {"sepal_width", "1.5"},
            {"species", "virginica"}
        };

IEnumerable<Dictionary<string, string>> dataset =
    new List<Dictionary<string, string>>()
    {
        setosaDatapoint,
        versicolorDatapoint,
        virginicaDatapoint
    };

DataPreprocessor dataPreprocessor =
    DataPreprocessor.init(
        new (string, bool)[]
            {
                ("petal_length", false),
                ("petal_width", false),
                ("sepal_length", false),
                ("sepal_width", false),
                ("species", true)
            },
        dataset
    );

IEnumerable<double[]> encodedDatapoints = dataset.Select(x =>
    DataPreprocessor
        .encodedDatapoint(dataPreprocessor, x)
);

encodedDatapoints equals to:

[ [ 0.0     , 0.0     , 0.0     , 1.0     , 0.0, 0.0, 1.0 ],
  [ 0.511111, 0.476190, 1.0     , 0.562500, 0.0, 1.0, 0.0 ],
  [ 1.0     , 1.0     , 0.166667, 0.0     , 1.0, 0.0, 0.0 ] ]

Save and load the preprocessor by calling DataPreprocessor.toJson and DataPreprocessor.ofJson.

Evaluation

To evaluate a neural network, you can call Statistics.rootMeanSquareError and provide the expected and predicted values.

IEnumerable<(double[], double[])> expectedWithOutputValues =
    new List<(double[], double[])>()
    {
        (new double[] {0.0, 0.0, 1.0}, new double[] {0.0, 0.0, 1.0}),
        (new double[] {0.0, 0.0, 1.0}, new double[] {0.0, 1.0, 1.0})
    };

double rmse = Statistics
    .rootMeanSquareError(expectedWithOutputValues);