dugagjin / tiny-csharp-neural-network

tiny-csharp-neural-network

XOR example

const double maxIterations = 1e5;
var architecture = new List<int> {2, 3, 1};
var brain = new Network(architecture);

var samples = new List<Sample> {
    new Sample {Input = new List<double> {0, 0}, Output = new List<double> {0}},
    new Sample {Input = new List<double> {0, 1}, Output = new List<double> {1}},
    new Sample {Input = new List<double> {1, 0}, Output = new List<double> {1}},
    new Sample {Input = new List<double> {1, 1}, Output = new List<double> {0}}
};

var random = new Random();
for (var i = 0; i < maxIterations; i++) {
    var index = random.Next(0, samples.Count);
    var sample = samples.ElementAt(index);
    var (input, expected) = (sample.Input, sample.Output);
    var predicted = brain.Predict(input);
    brain.Learn(predicted, expected);
}

Console.WriteLine("input\toutput\tpredicted");
samples.ForEach(sample => {
    var (input, output) = (sample.Input, sample.Output);
    var inputs = string.Join("|", input);
    var expected = string.Join("|", output);
    var predicted = string.Join("|", brain.Predict(input));
    Console.WriteLine($"{inputs}\t{expected}\t{predicted}");
});

build dll

cd src
dotnet build
dotnet publish --configuration Release  --self-contained