let width = 400;

let height = 400;

let t = 0;

let ab = [];

//Creating a training set consisting of the first five letters of the latin alphabet

let training = [];

let input = [];

//let b = 0.1; //Bias

let n = 25; //Number of pixels

//let N = 0; //Training iterations

let η = 0.05; //Learning rate

let px = 50;

let py = 50;

let r = [0.3, 0.59, 0.11];

let c;

let N=14

function preload() {

  for (let q = 0; q < N;q++) {

    ab.push(loadImage("3a/" + q + ".png"));

  }

}

function setup() {

  createCanvas(width, height);

  for (let i = 0; i < N; i++) {

    training.push(pixelate(ab[i]));

  }

  //print(training)

  for (let i = 0; i < N; i++) {

    let temp = [];

    for (let j = 0; j < px * py; j++) {

      temp.push(1 / px / py);

    }

    brain.push(temp);

  }

  seeB = createButton("See");

  seeB.mousePressed(see);

  learnB = createButton("Learn")

  learnB.mousePressed(learn)

  studyB = createButton("Study")

  studyB.mousePressed(study)

  focusB=createButton("Focus")

  focusB.mousePressed(focus)

​

  resizeCanvas(2 * width, height + 100);

}

​

function draw() {}

//This trains the machine and adjusts the brain on a random training image

function learn() {

  //Here I create a new random training image

  //t = random([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13]);

  //print(output)

  let output = think(training[t]);

  //Here I adjust the brain according to the output and the expected output(training)

  for (let i = 0; i < brain.length; i++) {

    for (let j = 0; j < brain[i].length; j++) {

      brain[i][j] +=

        (int(t == i) - output[i]) *

        output[i] *

        (1 - output[i]) *

        training[t][j] *

        η;

    }

  }

}

function study(){

  for(let i=0;i<1000;i++){

    t = random([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13]);

    learn()

  }

}

function focus(){

  for(let i=0;i<1000;i++){

    learn()

  }

}

function see(a) {

  background(0);

   t = random([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13]);

  //This creates the drawn image

  for (let i = 0; i < px * py; i++) {

    fill(training[t][i] * 255);

    rect(

      ((i % px) * width) / px,

      (floor(i / px) * height) / py,

      width / px,

      height / py

    );

  }

  tint(255, 255);

  image(ab[t], width, 0, width, height);

  //This draws the machine's guess

    for (let i = 0; i < N; i++) {

      tint(255, think(training[t])[i]*255);

      image(ab[i], (i * 2 * width) / 14, height, (2 * width) / 14, 100);

    }

}

function think(a) {

  return matMul(brain, a);

}

//This is the scalar product to calculate each pixel of the output picture

function dot(v, w) {

  let S = 0;

  for (let i = 0; i < v.length; i++) {

    S += v[i] * w[i];

  }

  return S;

}

//This performs all the scalar products, takes the sigmoid on the result and produces the total image

function matMul(A, x) {

  let S = [];

  for (let i = 0; i < A.length; i++) {

    S.push(sigmoid(dot(A[i], x)));

  }

  return S;

}

//The sigmoid function makes sure our values are between 0 and 1

function sigmoid(x) {

  return 1 / (1 + exp(-x));

}

function pixelate(a) {

  let ret = [];

  image(a, 0, 0, width, height);

  loadPixels();

  n = pixels.length / 4;

  for (let j = 0; j < py; j++) {

    for (let i = 0; i < px; i++) {

      let tgrb = 0;

      for (let k = 0; k < width / px; k++) {

        for (let m = 0; m < height / py; m++) {

          for (let l = 0; l < 3; l++) {

            tgrb +=

              r[l] *

              pixels[

                4 * ((i * width) / px + k + width * ((height / py) * j + m)) + l

              ];

          }

        }

      }

      ret.push(((tgrb / n) * px * py) / 255);

    }

  }

  return ret;

}

//Clicking a pixel on the left flips that pixel to the opposite colour

/*function mousePressed() {

  input[floor((10 * mouseX) / width) + 5 * floor((5 * mouseY) / height)] = int(

    !input[floor((10 * mouseX) / width) + 5 * floor((5 * mouseY) / height)]

  );

  print(input[floor(width / mouseX / 10) + 5 * floor(height / mouseY / 5)]);

}*/

​