11-CNNMnist.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "kernelspec": {
      "display_name": "R",
      "language": "R",
      "name": "ir"
    },
    "language_info": {
      "codemirror_mode": "r",
      "file_extension": ".r",
      "mimetype": "text/x-r-source",
      "name": "R",
      "pygments_lexer": "r",
      "version": "3.3.1"
    },
    "colab": {
      "name": "UCT2019-CNNMnist_R.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": []
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YS9FclBXnTZL",
        "colab_type": "text"
      },
      "source": [
        "***Dr. Emmanuel Dufourq*** www.emmanueldufourq.com\n",
        "\n",
        "***African Institute for Mathematical Sciences***\n",
        "\n",
        "***2019***\n",
        "\n",
        "\n",
        "Credits: The data imports were extracted from https://keras.rstudio.com/articles/examples/mnist_cnn.html\n",
        "\n",
        "***NOTE***\n",
        "\n",
        "Be sure to use hardware acceleration to use the GPU. Click on `Runtime`, change `runtime type`, and select `GPU` for the *hardware accelerator* option."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fdUuuxhRiXzg",
        "colab_type": "code",
        "outputId": "f56753ab-4509-487f-c1d3-1d95e148efaa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "devtools::install_github(\"rstudio/keras\")"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Skipping install of 'keras' from a github remote, the SHA1 (34dc89ed) has not changed since last install.\n",
            "  Use `force = TRUE` to force installation\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fQteA-lKkVo-",
        "colab_type": "text"
      },
      "source": [
        "## Import the Keras library"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_iMaChtciX8e",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "library(keras)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XP5Uf7XsAQ3e",
        "colab_type": "text"
      },
      "source": [
        "## Load the MNIST dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XGk3t9776WoU",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "mnist <- dataset_mnist()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nB2p8bUSjTez",
        "colab_type": "text"
      },
      "source": [
        "## Load features"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Th0mYc28jTjS",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "x_train <- mnist$train$x\n",
        "x_test <- mnist$test$x\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ibdk4invjRlb",
        "colab_type": "text"
      },
      "source": [
        "## Load targets"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EBbgDsqAjRts",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "y_train <- mnist$train$y\n",
        "y_test <- mnist$test$y"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4YUwKP2fI1hy",
        "colab_type": "text"
      },
      "source": [
        "## Check dimension"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-XfY_xQrI1mH",
        "colab_type": "code",
        "outputId": "04551fbc-5567-4cb9-d69e-a7f17cb8ac2a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 96
        }
      },
      "source": [
        "dim(x_train)"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "[1] 60000    28    28"
            ],
            "text/latex": "\\begin{enumerate*}\n\\item 60000\n\\item 28\n\\item 28\n\\end{enumerate*}\n",
            "text/markdown": "1. 60000\n2. 28\n3. 28\n\n\n",
            "text/html": [
              "<ol class=list-inline>\n",
              "\t<li>60000</li>\n",
              "\t<li>28</li>\n",
              "\t<li>28</li>\n",
              "</ol>\n"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rF_VUlRMIDl5",
        "colab_type": "text"
      },
      "source": [
        "## Reshape\n",
        "\n",
        "Keras wants the dimension for CNNs to be the following: [batches, width, height, channels] so we need to reshape the data by adding the channel dimension."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "25KoZ4FwIDt3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "x_train <- array_reshape(x_train, c(nrow(x_train), 28, 28, 1))\n",
        "x_test <- array_reshape(x_test, c(nrow(x_test), 28, 28, 1))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7bZHFjjSJd0o",
        "colab_type": "text"
      },
      "source": [
        "## Take a look at a  training example\n",
        "\n",
        "The dimension is correct but take a look at the values in the data point. The values are large, between 0 and 255. Thus we will need to normalise the values."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KMJgcAncJd98",
        "colab_type": "code",
        "outputId": "c5de02d5-0b57-4fb0-9cfc-d829f9e75d75",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 79
        }
      },
      "source": [
        "dim(x_train[1,,,])"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "[1] 28 28"
            ],
            "text/latex": "\\begin{enumerate*}\n\\item 28\n\\item 28\n\\end{enumerate*}\n",
            "text/markdown": "1. 28\n2. 28\n\n\n",
            "text/html": [
              "<ol class=list-inline>\n",
              "\t<li>28</li>\n",
              "\t<li>28</li>\n",
              "</ol>\n"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7OuQ7eyFJKqB",
        "colab_type": "code",
        "outputId": "f5fe5ae9-eecd-4ccc-c868-bf701eea6a02",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "print(x_train[1,,,1])"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "      [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10] [,11] [,12] [,13]\n",
            " [1,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            " [2,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            " [3,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            " [4,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            " [5,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            " [6,]    0    0    0    0    0    0    0    0    0     0     0     0     3\n",
            " [7,]    0    0    0    0    0    0    0    0   30    36    94   154   170\n",
            " [8,]    0    0    0    0    0    0    0   49  238   253   253   253   253\n",
            " [9,]    0    0    0    0    0    0    0   18  219   253   253   253   253\n",
            "[10,]    0    0    0    0    0    0    0    0   80   156   107   253   253\n",
            "[11,]    0    0    0    0    0    0    0    0    0    14     1   154   253\n",
            "[12,]    0    0    0    0    0    0    0    0    0     0     0   139   253\n",
            "[13,]    0    0    0    0    0    0    0    0    0     0     0    11   190\n",
            "[14,]    0    0    0    0    0    0    0    0    0     0     0     0    35\n",
            "[15,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            "[16,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            "[17,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            "[18,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            "[19,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            "[20,]    0    0    0    0    0    0    0    0    0     0     0     0    39\n",
            "[21,]    0    0    0    0    0    0    0    0    0     0    24   114   221\n",
            "[22,]    0    0    0    0    0    0    0    0   23    66   213   253   253\n",
            "[23,]    0    0    0    0    0    0   18  171  219   253   253   253   253\n",
            "[24,]    0    0    0    0   55  172  226  253  253   253   253   244   133\n",
            "[25,]    0    0    0    0  136  253  253  253  212   135   132    16     0\n",
            "[26,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            "[27,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            "[28,]    0    0    0    0    0    0    0    0    0     0     0     0     0\n",
            "      [,14] [,15] [,16] [,17] [,18] [,19] [,20] [,21] [,22] [,23] [,24] [,25]\n",
            " [1,]     0     0     0     0     0     0     0     0     0     0     0     0\n",
            " [2,]     0     0     0     0     0     0     0     0     0     0     0     0\n",
            " [3,]     0     0     0     0     0     0     0     0     0     0     0     0\n",
            " [4,]     0     0     0     0     0     0     0     0     0     0     0     0\n",
            " [5,]     0     0     0     0     0     0     0     0     0     0     0     0\n",
            " [6,]    18    18    18   126   136   175    26   166   255   247   127     0\n",
            " [7,]   253   253   253   253   253   225   172   253   242   195    64     0\n",
            " [8,]   253   253   253   253   251    93    82    82    56    39     0     0\n",
            " [9,]   253   198   182   247   241     0     0     0     0     0     0     0\n",
            "[10,]   205    11     0    43   154     0     0     0     0     0     0     0\n",
            "[11,]    90     0     0     0     0     0     0     0     0     0     0     0\n",
            "[12,]   190     2     0     0     0     0     0     0     0     0     0     0\n",
            "[13,]   253    70     0     0     0     0     0     0     0     0     0     0\n",
            "[14,]   241   225   160   108     1     0     0     0     0     0     0     0\n",
            "[15,]    81   240   253   253   119    25     0     0     0     0     0     0\n",
            "[16,]     0    45   186   253   253   150    27     0     0     0     0     0\n",
            "[17,]     0     0    16    93   252   253   187     0     0     0     0     0\n",
            "[18,]     0     0     0     0   249   253   249    64     0     0     0     0\n",
            "[19,]     0    46   130   183   253   253   207     2     0     0     0     0\n",
            "[20,]   148   229   253   253   253   250   182     0     0     0     0     0\n",
            "[21,]   253   253   253   253   201    78     0     0     0     0     0     0\n",
            "[22,]   253   253   198    81     2     0     0     0     0     0     0     0\n",
            "[23,]   195    80     9     0     0     0     0     0     0     0     0     0\n",
            "[24,]    11     0     0     0     0     0     0     0     0     0     0     0\n",
            "[25,]     0     0     0     0     0     0     0     0     0     0     0     0\n",
            "[26,]     0     0     0     0     0     0     0     0     0     0     0     0\n",
            "[27,]     0     0     0     0     0     0     0     0     0     0     0     0\n",
            "[28,]     0     0     0     0     0     0     0     0     0     0     0     0\n",
            "      [,26] [,27] [,28]\n",
            " [1,]     0     0     0\n",
            " [2,]     0     0     0\n",
            " [3,]     0     0     0\n",
            " [4,]     0     0     0\n",
            " [5,]     0     0     0\n",
            " [6,]     0     0     0\n",
            " [7,]     0     0     0\n",
            " [8,]     0     0     0\n",
            " [9,]     0     0     0\n",
            "[10,]     0     0     0\n",
            "[11,]     0     0     0\n",
            "[12,]     0     0     0\n",
            "[13,]     0     0     0\n",
            "[14,]     0     0     0\n",
            "[15,]     0     0     0\n",
            "[16,]     0     0     0\n",
            "[17,]     0     0     0\n",
            "[18,]     0     0     0\n",
            "[19,]     0     0     0\n",
            "[20,]     0     0     0\n",
            "[21,]     0     0     0\n",
            "[22,]     0     0     0\n",
            "[23,]     0     0     0\n",
            "[24,]     0     0     0\n",
            "[25,]     0     0     0\n",
            "[26,]     0     0     0\n",
            "[27,]     0     0     0\n",
            "[28,]     0     0     0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "25rl1ReBLJVY",
        "colab_type": "text"
      },
      "source": [
        "## Normalise"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-b-wMlDSKyYK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "x_train <- x_train / 255\n",
        "x_test <- x_test / 255"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VNa9k3I6LO5S",
        "colab_type": "text"
      },
      "source": [
        "Now we can check the values. This looks better and is suitable for a CNN."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8T1veQYrLI5G",
        "colab_type": "code",
        "outputId": "589f6589-b86f-4590-895e-79f6a903c433",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "print(x_train[1,,,1])"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "      [,1] [,2] [,3] [,4]      [,5]      [,6]       [,7]       [,8]       [,9]\n",
            " [1,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            " [2,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            " [3,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            " [4,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            " [5,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            " [6,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            " [7,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.11764706\n",
            " [8,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.19215686 0.93333333\n",
            " [9,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.07058824 0.85882353\n",
            "[10,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.31372549\n",
            "[11,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[12,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[13,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[14,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[15,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[16,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[17,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[18,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[19,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[20,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[21,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[22,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.09019608\n",
            "[23,]    0    0    0    0 0.0000000 0.0000000 0.07058824 0.67058824 0.85882353\n",
            "[24,]    0    0    0    0 0.2156863 0.6745098 0.88627451 0.99215686 0.99215686\n",
            "[25,]    0    0    0    0 0.5333333 0.9921569 0.99215686 0.99215686 0.83137255\n",
            "[26,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[27,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "[28,]    0    0    0    0 0.0000000 0.0000000 0.00000000 0.00000000 0.00000000\n",
            "           [,10]       [,11]      [,12]      [,13]      [,14]       [,15]\n",
            " [1,] 0.00000000 0.000000000 0.00000000 0.00000000 0.00000000 0.000000000\n",
            " [2,] 0.00000000 0.000000000 0.00000000 0.00000000 0.00000000 0.000000000\n",
            " [3,] 0.00000000 0.000000000 0.00000000 0.00000000 0.00000000 0.000000000\n",
            " [4,] 0.00000000 0.000000000 0.00000000 0.00000000 0.00000000 0.000000000\n",
            " [5,] 0.00000000 0.000000000 0.00000000 0.00000000 0.00000000 0.000000000\n",
            " [6,] 0.00000000 0.000000000 0.00000000 0.01176471 0.07058824 0.070588235\n",
            " [7,] 0.14117647 0.368627451 0.60392157 0.66666667 0.99215686 0.992156863\n",
            " [8,] 0.99215686 0.992156863 0.99215686 0.99215686 0.99215686 0.992156863\n",
            " [9,] 0.99215686 0.992156863 0.99215686 0.99215686 0.99215686 0.776470588\n",
            "[10,] 0.61176471 0.419607843 0.99215686 0.99215686 0.80392157 0.043137255\n",
            "[11,] 0.05490196 0.003921569 0.60392157 0.99215686 0.35294118 0.000000000\n",
            "[12,] 0.00000000 0.000000000 0.54509804 0.99215686 0.74509804 0.007843137\n",
            "[13,] 0.00000000 0.000000000 0.04313725 0.74509804 0.99215686 0.274509804\n",
            "[14,] 0.00000000 0.000000000 0.00000000 0.13725490 0.94509804 0.882352941\n",
            "[15,] 0.00000000 0.000000000 0.00000000 0.00000000 0.31764706 0.941176471\n",
            "[16,] 0.00000000 0.000000000 0.00000000 0.00000000 0.00000000 0.176470588\n",
            "[17,] 0.00000000 0.000000000 0.00000000 0.00000000 0.00000000 0.000000000\n",
            "[18,] 0.00000000 0.000000000 0.00000000 0.00000000 0.00000000 0.000000000\n",
            "[19,] 0.00000000 0.000000000 0.00000000 0.00000000 0.00000000 0.180392157\n",
            "[20,] 0.00000000 0.000000000 0.00000000 0.15294118 0.58039216 0.898039216\n",
            "[21,] 0.00000000 0.094117647 0.44705882 0.86666667 0.99215686 0.992156863\n",
            "[22,] 0.25882353 0.835294118 0.99215686 0.99215686 0.99215686 0.992156863\n",
            "[23,] 0.99215686 0.992156863 0.99215686 0.99215686 0.76470588 0.313725490\n",
            "[24,] 0.99215686 0.992156863 0.95686275 0.52156863 0.04313725 0.000000000\n",
            "[25,] 0.52941176 0.517647059 0.06274510 0.00000000 0.00000000 0.000000000\n",
            "[26,] 0.00000000 0.000000000 0.00000000 0.00000000 0.00000000 0.000000000\n",
            "[27,] 0.00000000 0.000000000 0.00000000 0.00000000 0.00000000 0.000000000\n",
            "[28,] 0.00000000 0.000000000 0.00000000 0.00000000 0.00000000 0.000000000\n",
            "           [,16]     [,17]       [,18]      [,19]     [,20]       [,21]\n",
            " [1,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            " [2,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            " [3,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            " [4,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            " [5,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            " [6,] 0.07058824 0.4941176 0.533333333 0.68627451 0.1019608 0.650980392\n",
            " [7,] 0.99215686 0.9921569 0.992156863 0.88235294 0.6745098 0.992156863\n",
            " [8,] 0.99215686 0.9921569 0.984313725 0.36470588 0.3215686 0.321568627\n",
            " [9,] 0.71372549 0.9686275 0.945098039 0.00000000 0.0000000 0.000000000\n",
            "[10,] 0.00000000 0.1686275 0.603921569 0.00000000 0.0000000 0.000000000\n",
            "[11,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            "[12,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            "[13,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            "[14,] 0.62745098 0.4235294 0.003921569 0.00000000 0.0000000 0.000000000\n",
            "[15,] 0.99215686 0.9921569 0.466666667 0.09803922 0.0000000 0.000000000\n",
            "[16,] 0.72941176 0.9921569 0.992156863 0.58823529 0.1058824 0.000000000\n",
            "[17,] 0.06274510 0.3647059 0.988235294 0.99215686 0.7333333 0.000000000\n",
            "[18,] 0.00000000 0.0000000 0.976470588 0.99215686 0.9764706 0.250980392\n",
            "[19,] 0.50980392 0.7176471 0.992156863 0.99215686 0.8117647 0.007843137\n",
            "[20,] 0.99215686 0.9921569 0.992156863 0.98039216 0.7137255 0.000000000\n",
            "[21,] 0.99215686 0.9921569 0.788235294 0.30588235 0.0000000 0.000000000\n",
            "[22,] 0.77647059 0.3176471 0.007843137 0.00000000 0.0000000 0.000000000\n",
            "[23,] 0.03529412 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            "[24,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            "[25,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            "[26,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            "[27,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            "[28,] 0.00000000 0.0000000 0.000000000 0.00000000 0.0000000 0.000000000\n",
            "          [,22]     [,23]     [,24] [,25] [,26] [,27] [,28]\n",
            " [1,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            " [2,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            " [3,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            " [4,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            " [5,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            " [6,] 1.0000000 0.9686275 0.4980392     0     0     0     0\n",
            " [7,] 0.9490196 0.7647059 0.2509804     0     0     0     0\n",
            " [8,] 0.2196078 0.1529412 0.0000000     0     0     0     0\n",
            " [9,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[10,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[11,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[12,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[13,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[14,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[15,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[16,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[17,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[18,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[19,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[20,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[21,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[22,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[23,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[24,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[25,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[26,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[27,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n",
            "[28,] 0.0000000 0.0000000 0.0000000     0     0     0     0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2UJ0_RnvkbQ-",
        "colab_type": "text"
      },
      "source": [
        "## Convert targets/labels to their one-hot encoded equivalent"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tzmu19wwkbVV",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "y_train <- to_categorical(y_train, 10)\n",
        "y_test <- to_categorical(y_test, 10)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qthOKREbl0wA",
        "colab_type": "text"
      },
      "source": [
        "## Create a model\n",
        "\n",
        "We specify an input shape of `c(28, 28, 1)`.\n",
        "\n",
        "The architecture is as follows: two convolutional layers, followed by max pooling and dropout, then a fully connected layer with dropout and finally the output fully connected layer. The last layer has 10 units (one for each digit 0 to 9). Note that between the max pooling layer and the first fully connected layer is the flattening operation `layer_flatten()`. This is how we go from convolutional based layers to fully connected ones."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Lpe8mJN8UNnk",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model <- keras_model_sequential() %>%\n",
        "  layer_conv_2d(filters = 32, kernel_size = c(3,3), activation = 'relu',input_shape = c(28,28,1)) %>% \n",
        "  layer_conv_2d(filters = 64, kernel_size = c(3,3), activation = 'relu') %>% \n",
        "  layer_max_pooling_2d(pool_size = c(2, 2)) %>% \n",
        "  layer_dropout(rate = 0.25) %>% \n",
        "  layer_flatten() %>%  # We use the flatten function to go from the CNN structure to the MLP\n",
        "  layer_dense(units = 128, activation = 'relu') %>% # the MLP starts here\n",
        "  layer_dropout(rate = 0.5) %>% \n",
        "  layer_dense(units = 10, activation = 'softmax')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ap-4xGUvbqCJ",
        "colab_type": "text"
      },
      "source": [
        "## Compile the model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TWwCI-yJUylq",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model %>% compile(\n",
        "  loss = loss_categorical_crossentropy,\n",
        "  optimizer = optimizer_adadelta(),\n",
        "  metrics = c('accuracy')\n",
        ")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IMAd3c8zbsNu",
        "colab_type": "text"
      },
      "source": [
        "## Allow printing to the screen"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sKi3HWp7WrUI",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "print_dot_callback <- callback_lambda(\n",
        "  on_epoch_end = function(epoch, logs) {\n",
        "    if (epoch %% 2 == 0) cat(epoch, '\\n')\n",
        "  }\n",
        ")   "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iQGEEfa6bvlu",
        "colab_type": "text"
      },
      "source": [
        "## Early stopping to avoid over-fitting"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ke_QJs3FXIwO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# The patience parameter is the amount of epochs to check for improvement.\n",
        "early_stop <- callback_early_stopping(monitor = \"val_loss\", patience = 6)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "b8UxbwiKbyWP",
        "colab_type": "text"
      },
      "source": [
        "## Train the model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5gGCDUXBWzL9",
        "colab_type": "code",
        "outputId": "3dd6d55b-091c-4517-a847-7ad0c29c1e11",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 136
        }
      },
      "source": [
        "history <- model %>% fit(\n",
        "  x_train,\n",
        "  y_train,\n",
        "  epochs = 14,\n",
        "  validation_split = 0.2,\n",
        "  verbose = 0,\n",
        "  batch_size = 128,\n",
        "  callbacks = list(early_stop, print_dot_callback)\n",
        ")"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0 \n",
            "2 \n",
            "4 \n",
            "6 \n",
            "8 \n",
            "10 \n",
            "12 \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lmu7cOgRk2zR",
        "colab_type": "text"
      },
      "source": [
        "### Plot the performance"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CuM3-FyUWtqq",
        "colab_type": "code",
        "outputId": "8623f544-3c6d-42f7-88f8-820f5ac27bee",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 857
        }
      },
      "source": [
        "plot(history)"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94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",
            "text/plain": [
              "plot without title"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nxIn0FARk4wh",
        "colab_type": "text"
      },
      "source": [
        "## Evaluate the performance"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Gbs8pa6GWtyw",
        "colab_type": "code",
        "outputId": "6fae2fa2-d4d6-4ca6-8e47-9219c29d5a94",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 113
        }
      },
      "source": [
        "model %>% evaluate(x_test, y_test)"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "$loss\n",
              "[1] 0.03411421\n",
              "\n",
              "$acc\n",
              "[1] 0.9895\n"
            ],
            "text/latex": "\\begin{description}\n\\item[\\$loss] 0.0341142140480395\n\\item[\\$acc] 0.989499986171722\n\\end{description}\n",
            "text/markdown": "$loss\n:   0.0341142140480395\n$acc\n:   0.989499986171722\n\n\n",
            "text/html": [
              "<dl>\n",
              "\t<dt>$loss</dt>\n",
              "\t\t<dd>0.0341142140480395</dd>\n",
              "\t<dt>$acc</dt>\n",
              "\t\t<dd>0.989499986171722</dd>\n",
              "</dl>\n"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5Y2vjmFBXBgh",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}