let data = [

  { x: 0.99, y: 0.02, label: "right" },

  { x: 0.76, y: -0.1, label: "right" },

  { x: -1.0, y: 0.12, label: "left" },

  { x: -0.9, y: -0.1, label: "left" },

  { x: 0.02, y: 0.98, label: "down" },

  { x: -0.2, y: 0.75, label: "down" },

  { x: 0.01, y: -0.9, label: "up" },

  { x: -0.1, y: -0.8, label: "up" },

];

let classifer;

let status = "training";

​

let start, end;

​

function setup() {

  createCanvas(640, 240);

  ml5.setBackend("cpu");

  // Step 2: set your neural network options

  let options = {

    task: "classification",

    debug: true,

  };

​

  // Step 3: initialize your neural network

  classifier = ml5.neuralNetwork(options);

​

  // Step 4: add data to the neural network

  for (let item of data) {

    let inputs = [item.x, item.y];

    let outputs = [item.label];

    classifier.addData(inputs, outputs);

  }

​

  // Step 5: normalize your data;

  classifier.normalizeData();

​

  // Step 6: train your neural network

  classifier.train({ epochs: 200 }, finishedTraining);

}

// Step 7: use the trained model

function finishedTraining() {

  status = "ready";

}

​

// Step 8: make a classification

​

function draw() {

  background(255);

  textAlign(CENTER, CENTER);

  textSize(64);

  text(status, width / 2, height / 2);

  if (start && end) {

    strokeWeight(8);

    line(start.x, start.y, end.x, end.y);

  }

}

​

function mousePressed() {

  start = createVector(mouseX, mouseY);

}

​

function mouseDragged() {

  end = createVector(mouseX, mouseY);

}

​

function mouseReleased() {

  let dir = p5.Vector.sub(end, start);

  dir.normalize();

  let inputs = [dir.x, dir.y];

  classifier.classify(inputs, gotResults);

}

​

// Step 9: define a function to handle the results of your classification

function gotResults(results) {

  status = results[0].label;

}

​