Code viewer for World: Character Recognition: (1-...
// Submission by StevenMacManus 20214549

// Cloned by StevenMacManus on 9 Dec 2020 from World "Character recognition 1 layer neural network" by StevenMacManus 
// Please leave this clone trail here.
 


// Cloned by StevenMacManus on 4 Dec 2020 from World "Character recognition neural network (clone by StevenMacManus)" by StevenMacManus 
// Please leave this clone trail here.
 


// Cloned by StevenMacManus on 1 Dec 2020 from World "Character recognition neural network" by "Coding Train" project 
// Please leave this clone trail here.
 

// Port of Character recognition neural network from here:
// https://github.com/CodingTrain/Toy-Neural-Network-JS/tree/master/examples/mnist
// with many modifications 


// --- defined by MNIST - do not change these ---------------------------------------

const PIXELS        = 28;                       // images in data set are tiny 
const PIXELSSQUARED = PIXELS * PIXELS;

// number of training and test exemplars in the data set:
const NOTRAIN = 60000;
const NOTEST  = 10000;

var weights = [];
var CurrentScore = 0;
var TrainedScore = 0;
var Memory = {Model: [], Score: 0};
var ModelLoad = false;
var EarlyStopCount = 0;
var PreviousScore = 0;
var BestScore = 0;
var checkpoint = [];

// User Code: ------------------------------------------------------------------
// Arrays to store Balanced Accuracy Metrics
var ClassPrecision = new Array(10);
var ClassRecall = new Array(10);
var ClassF1 = new Array(10);
var ClassMatrix = new Array(10);

for (i = 0; i < 10; i++)
{
    ClassMatrix[i] = new Array(10);
}

for (i = 0; i < 10; i++)
    for (j = 0; j < 10; j++)
    {
      ClassMatrix[i][j] = 0;  
    }


//--- can modify all these --------------------------------------------------

// no of nodes in network 
const noinput  = PIXELSSQUARED;
const nooutput = 10;
const nohidden = 4*PIXELS;

var learningrate_initial = 0.1;   // default 0.1  
var learningrate = learningrate_initial;

// User Code: ------------------------------------------------------------------
// FilterInputs sets whether or not we filter the training images
// SHIFT is the upper limit of Pixels to shift the images by
const FilterInputs = false;
const SHIFT = 5; 

// should we train every timestep or not 
let do_training = true;

// how many to train and test per timestep 
const TRAINPERSTEP = 60;
const TESTPERSTEP  = 10;

// multiply it by this to magnify for display 
const ZOOMFACTOR    = 7;                        
const ZOOMPIXELS    = ZOOMFACTOR * PIXELS; 

// 3 rows of
// large image + 50 gap + small image    
// 50 gap between rows 

const canvaswidth = ( PIXELS + ZOOMPIXELS ) + 50;
const canvasheight = ( ZOOMPIXELS * 3 ) + 100;


const DOODLE_THICK = 14;    // thickness of doodle lines 
const DOODLE_BLUR = 3;      // blur factor applied to doodles 

let mnist;      
// all data is loaded into this 
// mnist.train_images
// mnist.train_labels
// mnist.test_images
// mnist.test_labels


let nn;

let trainrun = 1;
let train_index = 0;

let testrun = 1;
let test_index = 0;
let total_tests = 0;
let total_correct = 0;

// images in LHS:
let doodle, demo;
let doodle_exists = false;
let demo_exists = false;

let mousedrag = false;      // are we in the middle of a mouse drag drawing?  


// save inputs to global var to inspect
// type these names in console 
var train_inputs, test_inputs, demo_inputs, doodle_inputs;


// Matrix.randomize() is changed to point to this. Must be defined by user of Matrix. 

function randomWeight()
{
    return ( AB.randomFloatAtoB ( -0.5, 0.5 ) );
            // Coding Train default is -1 to 1
}    

// User Code: ------------------------------------------------------------------
// Functions to save and recall the trained model from the server
function restoreData()
{
    AB.restoreData ( function ( a ) 
    { 
        LoadWeights(a);
        
        trained_nn = new NeuralNetwork(  noinput, nohidden, nooutput );
        
        for (i = 0; i < Memory.Model.bias_h.rows; i++)
            for (j = 0; j < Memory.Model.bias_h.cols; j++)
            {
                trained_nn.bias_h.data[i][j] = Memory.Model.bias_h.data[i][j];
            }
            
        for (i = 0; i < Memory.Model.bias_o.rows; i++)
            for (j = 0; j < Memory.Model.bias_o.cols; j++)
            {
                trained_nn.bias_o.data[i][j] = Memory.Model.bias_o.data[i][j];
            }
            
        for (i = 0; i < Memory.Model.weights_ih.rows; i++)
            for (j = 0; j < Memory.Model.weights_ih.cols; j++)
            {
                trained_nn.weights_ih.data[i][j] = Memory.Model.weights_ih.data[i][j];
            }
            
        for (i = 0; i < Memory.Model.weights_ho.rows; i++)
            for (j = 0; j < Memory.Model.weights_ho.cols; j++)
            {
                trained_nn.weights_ho.data[i][j] = Memory.Model.weights_ho.data[i][j];
            }
    });
}

function LoadWeights( a )
{
    Memory.Model = a.Model;
    Memory.Score = a.Score;
}

function saveData()
{
    Memory.Model = nn;
    Memory.Score = CurrentScore;
    AB.saveData( Memory )
    console.log("Saving Model Parameters")
}

// User Code: ------------------------------------------------------------------
// Function to shuffle the training data, must ensure that the order is 
// Maintained between images and label
function shuffleTrainingData()

{
  
  var currentIndex = mnist.train_images.length
  var temporaryValue, randomIndex;
  
  console.log("Shuffling Training Data")

  while (0 !== currentIndex) {
      
    randomIndex = Math.floor(Math.random() * currentIndex);
    currentIndex -= 1;

    temporaryData = mnist.train_images[currentIndex];
    temporaryLabel = mnist.train_labels[currentIndex];
    
    mnist.train_images[currentIndex] = mnist.train_images[randomIndex];
    mnist.train_images[randomIndex] = temporaryData;
    
    mnist.train_labels[currentIndex] = mnist.train_labels[randomIndex];
    mnist.train_labels[randomIndex] = temporaryLabel;
  }
}

// User Code: ------------------------------------------------------------------
// Function to save checkpoints, and stop learning when accuracy decreases

function EarlyStop(CurrentScore)
{

    if (testrun == 1) 
    {
        PreviousScore = CurrentScore;
        BestScore = PreviousScore;
    }
    else
    {
        if (BestScore > CurrentScore) 
        {
            EarlyStopCount++
        }
        else 
        {
            BestScore = CurrentScore;
            checkpoint = new NeuralNetwork(  nn, nohidden, nooutput );
            EarlyStopCount = 0;
        }
        
        PreviousScore = CurrentScore;
    }
    
    if (EarlyStopCount == 2)
    {
        do_training = false;
        console.log("Early Stop: Reverting Model to Checkpoint")
        trained_nn = checkpoint;
        ModelLoad = true;
    }
}


// CSS trick 
// make run header bigger 
 $("#runheaderbox").css ( { "max-height": "95vh" } );



//--- start of AB.msgs structure: ---------------------------------------------------------
// We output a serious of AB.msgs to put data at various places in the run header 
var thehtml;

  // 1 Doodle header 
  thehtml = "<hr> <h1> 1. Doodle </h1> Top row: Doodle (left) and shrunk (right). <br> " +
        " Draw your doodle in top LHS. <button onclick='wipeDoodle();' class='normbutton' >Clear doodle</button> <br> ";
   AB.msg ( thehtml, 1 );

  // 2 Doodle variable data (guess)
  
  // 3 Training header
  thehtml = "<hr> <h1> 2. Training </h1> Middle row: Training image magnified (left) and original (right). <br>  " +
        " <button onclick='do_training = false;' class='normbutton' >Stop training</button>" +
        " <button onclick='restoreData(); ModelLoad = true; do_training = false;' class='normbutton' >Load Weights</button> <br> ";
  AB.msg ( thehtml, 3 );
     
  // 4 variable training data 
  
  // 5 Testing header
  thehtml = "<h3> Classification Report </h3> " ;
  AB.msg ( thehtml, 5 );
           
  // 6 variable testing data 
  
  // 7 Demo header 
  thehtml = "<hr> <h1> 3. Demo </h1> Bottom row: Test image magnified (left) and  original (right). <br>" +
        " The network is <i>not</i> trained on any of these images. <br> " +
        " <button onclick='makeDemo();' class='normbutton' >Demo test image</button> <br> ";
   AB.msg ( thehtml, 7 );
   
  // 8 Demo variable data (random demo ID)
  // 9 Demo variable data (changing guess)
  
const greenspan = "<span style='font-weight:bold; font-size:x-large; color:darkgreen'> "  ;

//--- end of AB.msgs structure: ---------------------------------------------------------




function setup() 
{
    
  createCanvas ( canvaswidth, canvasheight );

  doodle = createGraphics ( ZOOMPIXELS, ZOOMPIXELS );       // doodle on larger canvas 
  doodle.pixelDensity(1);
  
// JS load other JS 
// maybe have a loading screen while loading the JS and the data set 

      AB.loadingScreen();
 
 $.getScript ( "/uploads/themacmac/matrix.js", function()
 {
   $.getScript ( "/uploads/themacmac/1_layer_nn.js", function()
   {
        $.getScript ( "/uploads/codingtrain/mnist.js", function()
        {
            console.log ("All JS loaded");
            nn = new NeuralNetwork(  noinput, nohidden, nooutput );
            nn.setLearningRate ( learningrate );
            nn.setActivationFunction ( sigmoid );
            loadData();
        });
   });
 });
 
 restoreData();
}



// load data set from local file (on this server)

function loadData()    
{
  loadMNIST ( function(data)    
  {
    mnist = data;
    console.log ("All data loaded into mnist object:")
    console.log(mnist);
    shuffleTrainingData();
    AB.removeLoading();     // if no loading screen exists, this does nothing 
  });
}



function getImage ( img )      // make a P5 image object from a raw data array   
{
    let theimage  = createImage (PIXELS, PIXELS);    // make blank image, then populate it 
    theimage.loadPixels();        
    
    for (let i = 0; i < PIXELSSQUARED ; i++) 
    {
        let bright = img[i];
        let index = i * 4;

            theimage.pixels[index + 0] = bright;
            theimage.pixels[index + 1] = bright;
            theimage.pixels[index + 2] = bright;
            theimage.pixels[index + 3] = 255;
    }
    
    theimage.updatePixels();
    return theimage;
}

// User Code: ------------------------------------------------------------------
// Function to filter the MNIST images
function Filter (img)
{
    aug_img = getImage ( img );
    aug_img.filter (THRESHOLD, 0.4);
    aug_img.filter (BLUR, 1);
    aug_img.resize ( PIXELS, PIXELS );     
    aug_img.loadPixels();

  // set up inputs   
  let aug_inputs = [];
  
  for (let i = 0; i < PIXELSSQUARED ; i++) 
  {
     aug_inputs[i] = aug_img.pixels[i * 4];
  }
  
  return aug_inputs
}


// User Code: ------------------------------------------------------------------
// 4 functions to shift the image by one pixel in each cardinal direction
function ShiftImageLeft( arr )
{
    
    let pic = [];
    let shifted_pic = [];
    var temp;

    for (let i = 0; i < PIXELS; i++)
        for (let j = 0; j < PIXELS; j++)
        {
            pic.push(arr[j + i*PIXELS]);
        }
    for (let i = 0; i < PIXELS; i++)
        for (let j = 0; j < PIXELS; j++)
        {
            // Note temp, can set third line to temp instead of 0 if wrapping
            // is desired
            temp = pic[i*PIXELS];
            shifted_pic[i*PIXELS + j] = pic[i*PIXELS + j + 1];
            shifted_pic[PIXELS * (1 + i) - 1] = 0;
        }

return shifted_pic
}

function ShiftImageRight( arr )
{
    let pic = [];
    let shifted_pic = [];
    var temp;

    for (let i = 0; i < PIXELS; i++)
        for (let j = 0; j < PIXELS; j++)
        {
            pic.push(arr[j + i*PIXELS]);
        }
    for (let i = 0; i < PIXELS; i++)
        for (let j = 0; j < PIXELS; j++)
        {
            // Note temp, can set third line to temp instead of 0 if wrapping
            // is desired
            temp = pic[PIXELS * (1 + i) - 1];
            shifted_pic[i*PIXELS + j] = pic[i*PIXELS + j - 1];
            shifted_pic[i*PIXELS] = 0;
        }   
        
return shifted_pic
}

function ShiftImageUp( arr )
{
    let pic = [];
    let shifted_pic = [];
    var temp;

    for (let i = 0; i < PIXELS; i++)
        for (let j = 0; j < PIXELS; j++)
        {
            pic.push(arr[j + i*PIXELS]);
        }
 
    for (let i = 0; i < PIXELS; i++)
        for (let j = 0; j < PIXELS; j++)
        {
            // Note temp, can set third line to temp instead of 0 if wrapping
            // is desired
            temp = pic[j];
            shifted_pic[i*PIXELS + j] = pic[i*PIXELS + j+PIXELS];
            shifted_pic[(PIXELS * (PIXELS - 1)) + j] = 0;
        }
    
    return shifted_pic;
}

function ShiftImageDown( arr )
{
    let pic = [];
    let shifted_pic = [];
    var temp;

    for (let i = 0; i < PIXELS; i++)
        for (let j = 0; j < PIXELS; j++)
        {
            pic.push(arr[j + i*PIXELS]);
        }
 
    for (let i = 0; i < PIXELS; i++)
        for (let j = 0; j < PIXELS; j++)
        {
            // Note temp, can set third line to temp instead of 0 if wrapping
            // is desired
            temp = pic[(PIXELS * (PIXELS - 1)) + j];
            shifted_pic[i*PIXELS + j] = pic[i*PIXELS + j-PIXELS];
            shifted_pic[j] = 0;
        }
    return shifted_pic;
}

// User Code: ------------------------------------------------------------------
// Function that calls the shift functions to shift the image by up to a user 
// defined limit of pixels in each cardinal and ordinal direction

function ShiftInputsByN(img)
{
    let rand1 = Math.random()
    let rand2 = Math.random()
    
    for (i = 0; i < Math.ceil(rand2*SHIFT); i++)
    {
        if (rand1 < 0.1) img = img;
        else if (0.1 < rand1 < 0.2) img = ShiftImageUp( img );
        else if (0.2 < rand1 < 0.3) img = ShiftImageLeft( ShiftImageUp( img ) ) ;
        else if (0.3 < rand1 < 0.4) img = ShiftImageLeft(img); 
        else if (0.4 < rand1 < 0.5) img = ShiftImageDown( ShiftImageLeft( img ) ); 
        else if (0.5 < rand1 < 0.6) img = ShiftImageDown(img); 
        else if (0.6 < rand1 < 0.7) img = ShiftImageRight( ShiftImageDown( img ) ); 
        else if (0.7 < rand1 < 0.8) img = ShiftImageRight( img ); 
        else img = ShiftImageUp(ShiftImageRight( img ) );
    }
    
    return img;
}

function getInputs ( img )      // convert img array into normalised input array 
{
    let inputs = [];
    for (let i = 0; i < PIXELSSQUARED ; i++)          
    {
        let bright = img[i];
        inputs[i] = bright / 255;       // normalise to 0 to 1
    } 
    return ( inputs );
}

 

function trainit (show)        // train the network with a single exemplar, from global var "train_index", show visual on or off 
{
  // User Code: ----------------------------------------------------------------
  // Now shifting the inputs
  let img   = ShiftInputsByN( mnist.train_images[train_index]);
  let label = mnist.train_labels[train_index];
  
  if (FilterInputs) img = Filter(img)
  
  // optional - show visual of the image 
  if (show)                
  {
    var theimage = getImage ( img );    // get image from data array 
    image ( theimage,   0,                ZOOMPIXELS+50,    ZOOMPIXELS,     ZOOMPIXELS  );      // magnified 
    image ( theimage,   ZOOMPIXELS+50,    ZOOMPIXELS+50,    PIXELS,         PIXELS      );      // original
  }

  // set up the inputs
  let inputs = getInputs ( img );       // get inputs from data array 

  // set up the outputs
  let targets = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0];
  targets[label] = 1;       // change one output location to 1, the rest stay at 0 

  train_inputs = inputs;        // can inspect in console 
  nn.train ( inputs, targets );

  thehtml = "<br> Learning Rate: " + nn.learning_rate.toFixed(5) + "<br> Epoch: " + trainrun + "<br><br> Training Sample: " + train_index ;
  
  AB.msg ( thehtml, 4 );

  train_index++;
  if ( train_index == NOTRAIN ) 
  {
    train_index = 0;
    trainrun++;
  }
}

// User Code: ------------------------------------------------------------------
// 2 Functions for calculating Balanced Accuracy Metrics
function SumRow(matrix, index)
{
    let Sum = 0;
    
    for (i = 0; i < (matrix.length); i++)
    {
    Sum = Sum + matrix[index][i];
    }
    return Sum
}

function SumColumn(matrix, index)
{
    let Sum = 0;

    for (i = 0; i < (matrix.length); i++)
    {
    Sum = Sum + matrix[i][index];
    }
    return Sum
}


function testit()    // test the network with a single exemplar, from global var "test_index"
{ 
  // User Code: ----------------------------------------------------------------
  // Now shifting the inputs
  let img   = ShiftInputsByN(mnist.test_images[test_index]);
  let label = mnist.test_labels[test_index];
  
  // set up the inputs
  let inputs = getInputs ( img ); 
  
  test_inputs = inputs;        // can inspect in console 
  let prediction = nn.predict(inputs); 
  let guess         = findMax(prediction);      // the top output 
  
// User Code: ------------------------------------------------------------------
// Calculate Precision, Recall, and F1 for each digit
// Can console.table(ClassMatrix) for a truth table.

  ClassMatrix[guess][label]++
  
    for ( c = 0; c < 10; c++)
    {
        let PreDen = SumRow(ClassMatrix, c);
        let RecDen = SumColumn(ClassMatrix, c);
        ClassPrecision[c] = (ClassMatrix[c][c] / PreDen);
        ClassRecall[c] = (ClassMatrix[c][c] / RecDen);
        ClassF1[c] = ((2 * ClassPrecision[c] * ClassRecall[c])/(ClassPrecision[c] + ClassRecall[c])).toFixed(2);
    }
  
  total_tests++;
  if (guess == label)  total_correct++;

  let percent = (total_correct / total_tests) * 100 ;
  CurrentScore = percent;
  
// User Code: ------------------------------------------------------------------
// Show the Classification Report

thehtml = "<table style=width:400px> <tr> <th> </th> <th>precision</th> <th>recall</th> <th>F1</th> </tr>" 
        + "<tr> <th style=text-align:right>0</th> <th>" + ClassPrecision[0].toFixed(2) + "</th> <th>" + ClassRecall[0].toFixed(2) + "</th> <th>" + ClassF1[0] + "</th> </tr>" 
        + "<tr> <th style=text-align:right>1</th> <th>" + ClassPrecision[1].toFixed(2) + "</th> <th>" + ClassRecall[1].toFixed(2) + "</th> <th>" + ClassF1[1] + "</th> </tr>" 
        + "<tr> <th style=text-align:right>2</th> <th>" + ClassPrecision[2].toFixed(2) + "</th> <th>" + ClassRecall[2].toFixed(2) + "</th> <th>" + ClassF1[2] + "</th> </tr>" 
        + "<tr> <th style=text-align:right>3</th> <th>" + ClassPrecision[3].toFixed(2) + "</th> <th>" + ClassRecall[3].toFixed(2) + "</th> <th>" + ClassF1[3] + "</th> </tr>" 
        + "<tr> <th style=text-align:right>4</th> <th>" + ClassPrecision[4].toFixed(2) + "</th> <th>" + ClassRecall[4].toFixed(2) + "</th> <th>" + ClassF1[4] + "</th> </tr>" 
        + "<tr> <th style=text-align:right>5</th> <th>" + ClassPrecision[5].toFixed(2) + "</th> <th>" + ClassRecall[5].toFixed(2) + "</th> <th>" + ClassF1[5] + "</th> </tr>" 
        + "<tr> <th style=text-align:right>6</th> <th>" + ClassPrecision[6].toFixed(2) + "</th> <th>" + ClassRecall[6].toFixed(2) + "</th> <th>" + ClassF1[6] + "</th> </tr>" 
        + "<tr> <th style=text-align:right>7</th> <th>" + ClassPrecision[7].toFixed(2) + "</th> <th>" + ClassRecall[7].toFixed(2) + "</th> <th>" + ClassF1[7] + "</th> </tr>" 
        + "<tr> <th style=text-align:right>8</th> <th>" + ClassPrecision[8].toFixed(2) + "</th> <th>" + ClassRecall[8].toFixed(2) + "</th> <th>" + ClassF1[8] + "</th> </tr>" 
        + "<tr> <th style=text-align:right>9</th> <th>" + ClassPrecision[9].toFixed(2) + "</th> <th>" + ClassRecall[9].toFixed(2) + "</th> <th>" + ClassF1[9] + "</th> </tr>"
        + "<tr> <th></th> <th></th> <th></th> <th></th> </tr>" 
        + "<tr> <th>Accuracy: </th> <th>" + percent.toFixed(2) + "</th> <th></th> <th></th> </tr> </table>" 
  AB.msg ( thehtml, 6 );

  test_index++;
  if ( test_index == NOTEST ) 
  {
    console.log( "Completed epoch: " + testrun + ", Accuracy: " + CurrentScore.toFixed(2) + ", Stored Model Accuracy: " + Memory.Score.toFixed(2));
    
    // User Code: --------------------------------------------------------------
    // Train Until Performance Decreases, Save the model if it bests the stored model's accuracy   

    if( CurrentScore.toFixed(2) > Memory.Score.toFixed(2)) saveData();
    
    restoreData();
    
    EarlyStop(CurrentScore)
    
    // User Code: --------------------------------------------------------------
    // Implement Exponential Learning Rate Decay -------------------------------------------
    learningrate = Math.exp(-0.5 * testrun) * learningrate_initial;
    nn.setLearningRate ( learningrate );
    console.log("Decreasing Learning Rate to: " + nn.learning_rate.toFixed(5))
    
    testrun++;
    
    test_index = 0;
    total_tests = 0;
    total_correct = 0;
    shuffleTrainingData()
  }
}




//--- find no.1 (and maybe no.2) output nodes ---------------------------------------
// (restriction) assumes array values start at 0 (which is true for output nodes) 


function find12 (a)         // return array showing indexes of no.1 and no.2 values in array 
{
  let no1 = 0;
  let no2 = 0;
  let no1value = 0;     
  let no2value = 0;
  
  for (let i = 0; i < a.length; i++) 
  {
    if (a[i] > no1value) 
    {
      no1 = i;
      no1value = a[i];
    }
    else if (a[i] > no2value) 
    {
      no2 = i;
      no2value = a[i];
    }
  }
  
  var b = [ no1, no2 ];
  return b;
}


// just get the maximum - separate function for speed - done many times 
// find our guess - the max of the output nodes array

function findMax (a)        
{
  let no1 = 0;
  let no1value = 0;     
  
  for (let i = 0; i < a.length; i++) 
  {
    if (a[i] > no1value) 
    {
      no1 = i;
      no1value = a[i];
    }
  }
  
  return no1;
}




// --- the draw function -------------------------------------------------------------
// every step:
 
function draw() 
{
  // check if libraries and data loaded yet:
  if ( typeof mnist == 'undefined' ) return;


// how can we get white doodle on black background on yellow canvas?
//        background('#ffffcc');    doodle.background('black');

      background ('grey');
    
if ( do_training )    
{
    
  // do some training per step 
    for (let i = 0; i < TRAINPERSTEP; i++) 
    {
      if (i == 0)    trainit(true);    // show only one per step - still flashes by  
      else           trainit(false);
    }
    
  // do some testing per step 
    for (let i = 0; i < TESTPERSTEP; i++) 
      testit();
}

  // keep drawing demo and doodle images 
  // and keep guessing - we will update our guess as time goes on 
  
  if ( demo_exists )
  {
    drawDemo();
    guessDemo();
  }
  if ( doodle_exists ) 
  {
    drawDoodle();
    guessDoodle();
  }


// detect doodle drawing 
// (restriction) the following assumes doodle starts at 0,0 

  if ( mouseIsPressed )         // gets called when we click buttons, as well as if in doodle corner  
  {
     var MAX = ZOOMPIXELS + 20;     // can draw up to this pixels in corner 
     if ( (mouseX < MAX) && (mouseY < MAX) && (pmouseX < MAX) && (pmouseY < MAX) )
     {
        mousedrag = true;       // start a mouse drag 
        doodle_exists = true;
        doodle.stroke('white');
        doodle.strokeWeight( DOODLE_THICK );
        doodle.line(mouseX, mouseY, pmouseX, pmouseY);      
     }
  }
  else 
  {
      // are we exiting a drawing
      if ( mousedrag )
      {
            mousedrag = false;
            doodle.filter (BLUR, DOODLE_BLUR);    // just blur once
      }
  }
}




//--- demo -------------------------------------------------------------
// demo some test image and predict it
// get it from test set so have not used it in training


function makeDemo()
{
    demo_exists = true;
    var  i = AB.randomIntAtoB ( 0, NOTEST - 1 );  
    
    demo        = mnist.test_images[i];     
    var label   = mnist.test_labels[i];
    
   thehtml =  "Test image no: " + i + "<br>" + 
            "Classification: " + label + "<br>" ;
   AB.msg ( thehtml, 8 );
   
   // type "demo" in console to see raw data 
}


function drawDemo()
{
    var theimage = getImage ( demo );
     //  console.log (theimage);
     
    image ( theimage,   0,                canvasheight - ZOOMPIXELS,    ZOOMPIXELS,     ZOOMPIXELS  );      // magnified 
    image ( theimage,   ZOOMPIXELS+50,    canvasheight - ZOOMPIXELS,    PIXELS,         PIXELS      );      // original
}


function guessDemo()
{
  let inputs = getInputs ( demo ); 
   
  demo_inputs = inputs;  // can inspect in console 
  
  // User Code: ----------------------------------------------------------------
  // If we have loaded the pretrained model use it
  
  // Model in training
  if(!ModelLoad)
  {
      let prediction = nn.predict(inputs);
      let guess         = findMax(prediction);      
      thehtml =   " We classify it as: " + greenspan + guess + "</span>" ;
      AB.msg ( thehtml, 9 );
  } 
  
  // Pretrained Model
  else 
  {
      let prediction = trained_nn.predict(inputs); 
      let guess         = findMax(prediction);     
      thehtml =   " We classify it as: " + greenspan + guess + "</span>" ;
      AB.msg ( thehtml, 9 );
  }

}




//--- doodle -------------------------------------------------------------

function drawDoodle()
{
    // doodle is createGraphics not createImage
    let theimage = doodle.get();
    // console.log (theimage);
    
    image ( theimage,   0,                0,    ZOOMPIXELS,     ZOOMPIXELS  );      // original 
    image ( theimage,   ZOOMPIXELS+50,    0,    PIXELS,         PIXELS      );      // shrunk
}
      
      
function guessDoodle() 
{
   // doodle is createGraphics not createImage
   let img = doodle.get();
   img
   img.resize ( PIXELS, PIXELS );     
   img.loadPixels();

  // set up inputs   
  let inputs = [];
  for (let i = 0; i < PIXELSSQUARED ; i++) 
  {
     inputs[i] = img.pixels[i * 4] / 255;
  }
  
  doodle_inputs = inputs;     // can inspect in console 

  // User Code: ----------------------------------------------------------------
  // If we have loaded the pretrained model use it
  
  // Model in training
  if(!ModelLoad)
  {
      let prediction    = nn.predict(inputs);       // array of outputs 
      let b             = find12(prediction);       // get no.1 and no.2 guesses  
      thehtml =   " We classify it as: " + greenspan + b[0] + "</span> <br>" +
            " No.2 guess is: " + greenspan + b[1] + "</span>";
      AB.msg ( thehtml, 2 );
  }
  // Pretrained Model
  else
  {
      let prediction    = trained_nn.predict(inputs);       // array of outputs 
      let b             = find12(prediction);       // get no.1 and no.2 guesses  
      thehtml =   " We classify it as: " + greenspan + b[0] + "</span> <br>" +
            " No.2 guess is: " + greenspan + b[1] + "</span>";
      AB.msg ( thehtml, 2 );   
  }
}

// User Code: ----------------------------------------------------------------
// Modified so the background doesn't disappear after wiping, but need to wipe
// before drawing

function wipeDoodle()    
{
    doodle_exists = false;
    doodle.background('black');
    doodle_exists = true;
}




// --- debugging --------------------------------------------------
// in console
// showInputs(demo_inputs);
// showInputs(doodle_inputs);


function showInputs ( inputs )
// display inputs row by row, corresponding to square of pixels 
{
    var str = "";
    for (let i = 0; i < inputs.length; i++) 
    {
      if ( i % PIXELS == 0 )    str = str + "\n";                                   // new line for each row of pixels 
      var value = inputs[i];
      str = str + " " + value.toFixed(2) ; 
    }
    console.log (str);
}