Getting Started

This guide will walk you through the basic features of Brain4J, from setting up your first model to making predictions.

Creating Your First Model

Let's create a simple model to simulate the XOR operator.

Set Up Your Project

If you haven't checked it yet, follow the Installation Guide.

Initialize a Model

Create a basic model using Brain4J:

Model model = new Sequential(
        new DenseLayer(2, Activations.LINEAR), // 2 Input neurons
        new DenseLayer(32, Activations.MISH), // 32 Hidden neurons
        new DenseLayer(32, Activations.MISH), // 32 Hidden neurons
        new DenseLayer(1, Activations.SIGMOID) // 1 Output neuron for classification
);

model.compile(Loss.BINARY_CROSS_ENTROPY, new AdamW(0.1));

Train the Model

To train the model we need some data, for the XOR we can generate it:

List<Sample> samples = new ArrayList<>();

for (int x = 0; x < 2; x++) {
    for (int y = 0; y < 2; y++) {
        Tensor input = Tensors.vector(x, y);
        Tensor output = Tensors.vector(x ^ y);

        samples.add(new Sample(input, output));
    }
}

// Samples, no shuffle, 4 of batch size
ListDataSource dataSource = new ListDataSource(samples, false, 4);

// Fit the model for 50 epoches, evaluate every 10
model.fit(dataSource, 50, 10);

Make Predictions and Evaluation

Once trained, you can use the model to make predictions.

// You can evaluate the model like this
EvaluationResult evaluation = model.evaluate(dataSource);
System.out.println(evaluation.confusionMatrix());

Additionally, if you have test data, you can have Brain4J evaluate the model on it:

ListDataSource trainSource = ...;
ListDataSource testSource = ...;

// Trains with trainSource, evaluates with testSource
model.fit(trainSource, testSource, 50, 10);

Next Steps

Check out Advanced Usage