let testingData;

loadData()

async function loadData() {

​

  let test = [

    { image: [10,8,51,4], label: 1},

    { image: [2,3,21,34], label: 2},

    { image: [20,13,31,44], label: 3},

    { image: [13,17,61,64], label: 4},

    { image: [15,19,1,24], label: 5},

  ].map(val => {

      const temp = val;

      const label = temp.label;

      temp.label = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0];

      temp.label[label] = 1;

      temp.labelNum = label;

      return temp;

    })

  testingData = test;

​

  let train = [

    { image: [2,3,41,4], label: 2},

  ].map(val => {

      const temp = val;

      const label = temp.label;

      temp.label = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0];

      temp.label[label] = 1;

      return temp;

    })

  trainingData = train;

  addData();

}

​

const options = {

  inputs: 784,

  outputs: 1,

  task: 'classification',

  debug: true

}

const nn = ml5.neuralNetwork(options);

​

/*

The data is an array of objects:

[{..},{..},{..},...{..},{..}]

​

Where each object looks like this:

{image: Array(784), label: Array(10)}

​

Where "image" is an array of greyscale pixel values

and label is an array representing the label, for example

0 => [1,0,0,0,0,0,0,0,0,0]

1 => [0,1,0,0,0,0,0,0,0,0]

​

*/

function addData() {

  for(let datapt of trainingData) {

    nn.addData(datapt.image, datapt.label);

  }

  console.log('DATA ADDED!');

  train();

}

​

function train() {

  nn.normalizeData();

  const options = {

    epochs: 20

  }

  nn.train(options, finishedTraining)

}

​

async function finishedTraining() {

  console.log('FINISHED TRAINING!');

  console.log('Use test(index) to test NN')

}

​

function test(index) {

  console.log('Label:' +testingData[index].labelNum);

  console.log(testingData[index].image);

  nn.classify(testingData[index].image, gotResults);

}

​

function gotResults(error, results) {

  if(error) console.error(error);

  else console.log(results);

}