let net = NeuralNetwork::Train(&data);

// Uses
use cogent::{
    NeuralNetwork,
    EvaluationData,MeasuredCondition
};
use ndarray::{Array2,Axis};

// Setup
// ----------
// 784-ReLU->800-Softmax->10
let mut net = NeuralNetwork::new(784,&[
    Layer::Dense(800,Activation::ReLU),
    Layer::Dense(10,Activation::Softmax)
]);

// Setting training and testing data.
// `get_mnist_dataset(bool)` simply gets MNIST dataset.
// The boolean specifies if it is the MNIST testing data (`true`) or training data (`false`).

// Sets training and testing data.
let (mut train_data,mut train_labels):(Array2<f32>,Array2<usize>) = get_mnist_dataset(false);
let (test_data,test_labels):(Array2<f32>,Array2<usize>) = get_mnist_dataset(true);

// Execution
// ----------
// Trains until no notable accuracy improvements are being made over a number of iterations.
// By default this would end training if 0.5% accuracy improvement was not seen over 12 iterations/epochs.

net.train(&mut train_data,&mut train_labels)
    .evaluation_data(EvaluationData::Actual(&test_data,&test_labels)) // Sets evaluation data
    .l2(0.1) // Implements L2 regularisation with a 0.1 lambda vlaue
    .tracking() // Prints backpropgation progress within each iteration
    .log_interval(MeasuredCondition::Iteration(1)) // Prints evaluation after each iteration
    .go();

// If evaluation data is not set manually it will simply shuffle and split off a random group from training data to be evaluation data.
// In the case of MNIST where training and evaluation datasets are given seperately, it makes sense to set it as such.

// Evaluation
// ----------
let (cost,correctly_classified):(f32,u32) = net.evaluate(&test_data,&test_labels,None); // (cost,examples correctly classified)
println!("Cost: {:.2}",cost);
println!(
    "Accuracy: {}/{} ({:.2}%)",
    correctly_classified,
    test_data.len_of(Axis(1)),
    correctly_classified as f32 / test_data.len_of(Axis(1)) as f32
);