hands-on/060_mnist_cnn_complete.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "060_mnist_cnn_complete.ipynb",
      "version": "0.3.2",
      "provenance": []
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.3"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xX05QJuPqxDU",
        "colab_type": "text"
      },
      "source": [
        "# MNIST: convolutional neural networks"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FeBuk0jpqxDX",
        "colab_type": "text"
      },
      "source": [
        "We define and train a convolutional neural network (CNN) model to recognize handwritten digits.  Although CNNs can be used in many contexts, they are probably most often used in the context of image processing."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dsSGP2BzqxDY",
        "colab_type": "text"
      },
      "source": [
        "## Required imports"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vntuf5EzqxDZ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import keras\n",
        "from keras import backend as K\n",
        "from keras.datasets import mnist\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "import numpy as np\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.pipeline import Pipeline\n",
        "from sklearn.preprocessing import FunctionTransformer, OneHotEncoder"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hah5lmFZqxDh",
        "colab_type": "text"
      },
      "source": [
        "## Data preparation"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "42NZmGBzqxDh",
        "colab_type": "text"
      },
      "source": [
        "Depending on the backend keras uses, the input format slightly differs. Either the number of color channels is the first, or the last dimension.  Our input data is grayscale, so there is only a single color channel"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TsnEMyYyqxDi",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "img_channels, img_rows, img_cols = 1, 28, 28\n",
        "\n",
        "if K.image_data_format() == 'channels_first':\n",
        "    shape_ord = (img_channels, img_rows, img_cols)\n",
        "else:  # channel_last\n",
        "    shape_ord = (img_rows, img_cols, img_channels)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ylfYzQarqxDn",
        "colab_type": "text"
      },
      "source": [
        "Load the data again."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "11Jz611EqxDo",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "54e8ead8-841e-4efb-8158-0b951252efc5"
      },
      "source": [
        "(x_train, y_train), (x_test, y_test) = mnist.load_data()"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://s3.amazonaws.com/img-datasets/mnist.npz\n",
            "11493376/11490434 [==============================] - 1s 0us/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LBUQOrUAqxDr",
        "colab_type": "text"
      },
      "source": [
        "### Data preprocessing"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cg4Mp1wpqxDs",
        "colab_type": "text"
      },
      "source": [
        "First, reshape the input data to take into account the color channels of the images."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "06kp5tg0qxDt",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "input_pipeline = FunctionTransformer(lambda x: x.reshape((x.shape[0], ) + shape_ord).astype(np.float32)/255.0,\n",
        "                                     validate=False)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "udYsweBCqxDy",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "x_train = input_pipeline.fit_transform(x_train)\n",
        "x_test = input_pipeline.transform(x_test)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xYA9075OqxD6",
        "colab_type": "text"
      },
      "source": [
        "Most learning algorithms work better when the input data is normalized to the interval $[0, 1]$."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "csQt9IsVqxD7",
        "colab_type": "text"
      },
      "source": [
        "The output can be transformed using the pipeline defined for the multilayer network and can be imported from the `mnist_utils` module."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "F2VFCAP1qxD-",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "output_reshaper = FunctionTransformer(lambda x: x.reshape(-1, 1),                                                                                                                                        \n",
        "                                      validate=False)                                                                                                                                                    \n",
        "output_encoder = OneHotEncoder(categories='auto')\n",
        "output_type_changer = FunctionTransformer(lambda x: x.astype(np.float32),\n",
        "                                          validate=False)\n",
        "output_pipeline = Pipeline([\n",
        "    ('reshaper', output_reshaper),\n",
        "    ('binarizer', output_encoder),\n",
        "    ('type_changer', output_type_changer),\n",
        "])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jMcnEXJlqxEF",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "y_train = output_pipeline.fit_transform(y_train)\n",
        "y_test = output_pipeline.transform(y_test)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DnsXw5Z3qxEK",
        "colab_type": "text"
      },
      "source": [
        "### Validation data"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "W6AX8PY2qxEL",
        "colab_type": "text"
      },
      "source": [
        "In order to make this reproducible, we have to seed the random number generator."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1Q8hDaQTqxEM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "np.random.seed(1234)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "azWUAholqxEP",
        "colab_type": "text"
      },
      "source": [
        "During the training, we will require a validation set, so we split the training data into two sets, one for actual training, the other for validation.  Note, we don't touch the test data set at all during the training process.  The default is using 75 % of the data for training, 25 % for validation. This function will also shuffle the data set prior to splitting (hence seeding the random number generator)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "W464pK-sqxEQ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "x_train, x_val, y_train, y_val = train_test_split(x_train, y_train)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "05cdUzOMqxES",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        },
        "outputId": "5f2caffc-5f3b-4611-e537-4dfa5f728a87"
      },
      "source": [
        "x_train.shape, x_val.shape, x_train.dtype, y_train.shape, y_val.shape, y_train.dtype"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((45000, 28, 28, 1),\n",
              " (15000, 28, 28, 1),\n",
              " dtype('float32'),\n",
              " (45000, 10),\n",
              " (15000, 10),\n",
              " dtype('float32'))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LMyrPzoHqxEY",
        "colab_type": "text"
      },
      "source": [
        "### Verification"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EzRTwDngqxEZ",
        "colab_type": "text"
      },
      "source": [
        "Since we've done quite some transformations, let's verify whether we didn't mess up."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "v_Ysyp8wqxEa",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 248
        },
        "outputId": "e4c9069f-05c0-4910-9cc2-054ed147d96d"
      },
      "source": [
        "frame = plt.gca()\n",
        "frame.axes.get_xaxis().set_visible(False)\n",
        "frame.axes.get_yaxis().set_visible(False)\n",
        "plt.imshow(x_train[0, :, :, 0], cmap='gray');"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAOcAAADnCAYAAADl9EEgAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhBInhPo/p4v5pGZVQDMAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u7P_nqh1qxEd",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "3c56d844-372a-4545-bc23-e9f7254eb037"
      },
      "source": [
        "print(y_train[0])"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "  (0, 6)\t1.0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VhxpRIIFqxEh",
        "colab_type": "text"
      },
      "source": [
        "## Simple CNN"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "i5xRRWjfqxEi",
        "colab_type": "text"
      },
      "source": [
        "We start with a very straightforward network with a single convolutional layer."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_99OaC-iqxEi",
        "colab_type": "text"
      },
      "source": [
        "### Required imports"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pXEYDCjhqxEj",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from keras.models import Sequential\n",
        "from keras.layers import Activation, Conv2D, Dense, Dropout, Flatten\n",
        "from keras.optimizers import SGD\n",
        "import tensorflow as tf"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CVq32ELKqxEr",
        "colab_type": "text"
      },
      "source": [
        "Seed TensorFlow's random number generator."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AaZS1YfNqxEs",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "tf.set_random_seed(40580)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IOlV7_JWqxEv",
        "colab_type": "text"
      },
      "source": [
        "### Model definition"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UT_I9oGsqxEv",
        "colab_type": "text"
      },
      "source": [
        "The network has 32 filters, each 3 $\\times$ 3 pixels in its convolutional layer, and a ReLU activation function. For the second, dense layer, the output of the convolutional layer is first flattened. The dense layer's activation function is SoftMax to determine the category for the image."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FipbpBIzqxEw",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        },
        "outputId": "1f315761-95cc-4fc3-fea9-821a232e70c1"
      },
      "source": [
        "nr_filters = 32\n",
        "nr_classes = 10\n",
        "conv_x_size, conv_y_size = 3, 3\n",
        "cnn_model = Sequential()\n",
        "cnn_model.add(Conv2D(nr_filters, (conv_x_size, conv_y_size), padding='valid',\n",
        "                     input_shape=shape_ord))\n",
        "cnn_model.add(Activation('relu'))\n",
        "cnn_model.add(Flatten())\n",
        "cnn_model.add(Dense(nr_classes))\n",
        "cnn_model.add(Activation('softmax'))"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Colocations handled automatically by placer.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MV5FFC-IqxE0",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 306
        },
        "outputId": "1b4d9328-30c7-4e5b-af40-b17298cbf8dc"
      },
      "source": [
        "cnn_model.summary()"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_1 (Conv2D)            (None, 26, 26, 32)        320       \n",
            "_________________________________________________________________\n",
            "activation_1 (Activation)    (None, 26, 26, 32)        0         \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 21632)             0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 10)                216330    \n",
            "_________________________________________________________________\n",
            "activation_2 (Activation)    (None, 10)                0         \n",
            "=================================================================\n",
            "Total params: 216,650\n",
            "Trainable params: 216,650\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HES92ClZqxE6",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "cnn_hist = cnn_model.compile(loss='categorical_crossentropy',\n",
        "                             optimizer=SGD(lr=0.1, decay=1.0e-6, \n",
        "                                           momentum=0.9, nesterov=True),\n",
        "                             metrics=['accuracy'])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4pVlMkXYqxE9",
        "colab_type": "text"
      },
      "source": [
        "### Training"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "scrolled": false,
        "id": "0Xwz6yHPqxE-",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 445
        },
        "outputId": "e9e0d701-b892-4457-9f51-f787ba746ffb"
      },
      "source": [
        "cnn_hist = cnn_model.fit(x_train, y_train, batch_size=32, epochs=10,\n",
        "                         verbose=1, validation_data=(x_val, y_val))"
      ],
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.cast instead.\n",
            "Train on 45000 samples, validate on 15000 samples\n",
            "Epoch 1/10\n",
            "45000/45000 [==============================] - 11s 239us/step - loss: 0.1902 - acc: 0.9426 - val_loss: 0.1158 - val_acc: 0.9663\n",
            "Epoch 2/10\n",
            "45000/45000 [==============================] - 5s 122us/step - loss: 0.0866 - acc: 0.9729 - val_loss: 0.1407 - val_acc: 0.9615\n",
            "Epoch 3/10\n",
            "45000/45000 [==============================] - 5s 121us/step - loss: 0.0606 - acc: 0.9811 - val_loss: 0.1298 - val_acc: 0.9688\n",
            "Epoch 4/10\n",
            "45000/45000 [==============================] - 5s 122us/step - loss: 0.0425 - acc: 0.9868 - val_loss: 0.1076 - val_acc: 0.9701\n",
            "Epoch 5/10\n",
            "45000/45000 [==============================] - 5s 121us/step - loss: 0.0343 - acc: 0.9886 - val_loss: 0.1287 - val_acc: 0.9705\n",
            "Epoch 6/10\n",
            "45000/45000 [==============================] - 5s 122us/step - loss: 0.0289 - acc: 0.9908 - val_loss: 0.1297 - val_acc: 0.9711\n",
            "Epoch 7/10\n",
            "45000/45000 [==============================] - 5s 121us/step - loss: 0.0233 - acc: 0.9928 - val_loss: 0.1433 - val_acc: 0.9733\n",
            "Epoch 8/10\n",
            "45000/45000 [==============================] - 5s 121us/step - loss: 0.0202 - acc: 0.9940 - val_loss: 0.1451 - val_acc: 0.9721\n",
            "Epoch 9/10\n",
            "45000/45000 [==============================] - 5s 121us/step - loss: 0.0163 - acc: 0.9948 - val_loss: 0.1514 - val_acc: 0.9705\n",
            "Epoch 10/10\n",
            "45000/45000 [==============================] - 6s 122us/step - loss: 0.0179 - acc: 0.9942 - val_loss: 0.1469 - val_acc: 0.9751\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eZtRIW2qqxFB",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def plot_history(network_history):\n",
        "    plt.figure()\n",
        "    plt.xlabel('Epochs')\n",
        "    plt.ylabel('Loss')\n",
        "    plt.plot(network_history.history['loss'])\n",
        "    plt.plot(network_history.history['val_loss'])\n",
        "    plt.legend(['Training', 'Validation'])\n",
        "    plt.figure()\n",
        "    plt.xlabel('Epochs')\n",
        "    plt.ylabel('Accuracy')\n",
        "    plt.plot(network_history.history['acc'])\n",
        "    plt.plot(network_history.history['val_acc'])\n",
        "    plt.legend(['Training', 'Validation'], loc='lower right')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BjTXFIJ2qxFE",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 549
        },
        "outputId": "dfc8ddcd-038e-418a-abbf-6702bc92b467"
      },
      "source": [
        "plot_history(cnn_hist)"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZIAAAEKCAYAAAA4t9PUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nS723G5MTAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEKCAYAAADjDHn2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nY4xpl4WFMcaYdllYGGOMadf/B7vhdNujJDLMAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fkJuL7MoqxFG",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "f6fabc84-85d3-434d-8e8d-a0947e5e2e4c"
      },
      "source": [
        "cnn_model.evaluate(x_test, y_test)"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "10000/10000 [==============================] - 1s 63us/step\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.14751214716108985, 0.974]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pAvvRVrDqxFK",
        "colab_type": "text"
      },
      "source": [
        "Although the test accuracy is quite acceptable, the training accuracy is a lot higher, so most likely there is some overfitting."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ooIrcGprqxFL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "cnn_model.save('mnist_cnn.h5')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WEQaS5VrqxFN",
        "colab_type": "text"
      },
      "source": [
        "## More layers & dropout"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JXwuZrIgqxFO",
        "colab_type": "text"
      },
      "source": [
        "This model has an additional hidden layer to improve accuracy."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RTdZNqJgqxFO",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        },
        "outputId": "38f7b9a8-9576-4c1a-fec2-0667601b6f4d"
      },
      "source": [
        "nr_filters = 32\n",
        "nr_classes = 10\n",
        "conv_x_size, conv_y_size = 3, 3\n",
        "cnn_do_model = Sequential()\n",
        "cnn_do_model.add(Conv2D(nr_filters, (conv_x_size, conv_y_size), padding='valid',\n",
        "                     input_shape=shape_ord))\n",
        "cnn_do_model.add(Activation('relu'))\n",
        "cnn_do_model.add(Flatten())\n",
        "cnn_do_model.add(Dense(128))\n",
        "cnn_do_model.add(Activation('relu'))\n",
        "cnn_do_model.add(Dropout(0.1))\n",
        "cnn_do_model.add(Dense(nr_classes))\n",
        "cnn_do_model.add(Activation('softmax'))"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kqufPbwhqxFR",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 408
        },
        "outputId": "fbe13086-19e1-42df-b364-b66398d25fea"
      },
      "source": [
        "cnn_do_model.summary()"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_2 (Conv2D)            (None, 26, 26, 32)        320       \n",
            "_________________________________________________________________\n",
            "activation_3 (Activation)    (None, 26, 26, 32)        0         \n",
            "_________________________________________________________________\n",
            "flatten_2 (Flatten)          (None, 21632)             0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 128)               2769024   \n",
            "_________________________________________________________________\n",
            "activation_4 (Activation)    (None, 128)               0         \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 128)               0         \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 10)                1290      \n",
            "_________________________________________________________________\n",
            "activation_5 (Activation)    (None, 10)                0         \n",
            "=================================================================\n",
            "Total params: 2,770,634\n",
            "Trainable params: 2,770,634\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "O-96Tt2-qxFU",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "cnn_do_model.compile(loss='categorical_crossentropy',\n",
        "                     optimizer=SGD(),\n",
        "                     metrics=['accuracy'])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IkvW4CAMqxFX",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 374
        },
        "outputId": "e1bdb278-6f08-4bb0-c66d-01a1da6b8946"
      },
      "source": [
        "cnn_do_hist = cnn_do_model.fit(x_train, y_train, batch_size=32, epochs=10,\n",
        "                               verbose=1, validation_data=(x_val, y_val))"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 45000 samples, validate on 15000 samples\n",
            "Epoch 1/10\n",
            "45000/45000 [==============================] - 7s 147us/step - loss: 0.4521 - acc: 0.8761 - val_loss: 0.2451 - val_acc: 0.9266\n",
            "Epoch 2/10\n",
            "45000/45000 [==============================] - 6s 139us/step - loss: 0.2088 - acc: 0.9395 - val_loss: 0.2071 - val_acc: 0.9371\n",
            "Epoch 3/10\n",
            "45000/45000 [==============================] - 7s 151us/step - loss: 0.1592 - acc: 0.9540 - val_loss: 0.1383 - val_acc: 0.9597\n",
            "Epoch 4/10\n",
            "45000/45000 [==============================] - 7s 155us/step - loss: 0.1357 - acc: 0.9599 - val_loss: 0.1281 - val_acc: 0.9613\n",
            "Epoch 5/10\n",
            "45000/45000 [==============================] - 6s 139us/step - loss: 0.1172 - acc: 0.9645 - val_loss: 0.1095 - val_acc: 0.9678\n",
            "Epoch 6/10\n",
            "45000/45000 [==============================] - 7s 150us/step - loss: 0.1043 - acc: 0.9685 - val_loss: 0.1065 - val_acc: 0.9676\n",
            "Epoch 7/10\n",
            "45000/45000 [==============================] - 7s 145us/step - loss: 0.0943 - acc: 0.9715 - val_loss: 0.0915 - val_acc: 0.9723\n",
            "Epoch 8/10\n",
            "45000/45000 [==============================] - 6s 139us/step - loss: 0.0863 - acc: 0.9738 - val_loss: 0.0885 - val_acc: 0.9739\n",
            "Epoch 9/10\n",
            "45000/45000 [==============================] - 6s 139us/step - loss: 0.0814 - acc: 0.9751 - val_loss: 0.0822 - val_acc: 0.9748\n",
            "Epoch 10/10\n",
            "45000/45000 [==============================] - 6s 139us/step - loss: 0.0738 - acc: 0.9768 - val_loss: 0.0826 - val_acc: 0.9749\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DDRwiauZqxFe",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 549
        },
        "outputId": "90d889e0-dee5-43bb-a4d1-ee5f9fc4cd7f"
      },
      "source": [
        "plot_history(cnn_do_hist)"
      ],
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEKCAYAAADjDHn2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEKCAYAAADjDHn2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "scrolled": true,
        "id": "k3-V9sFuqxFh",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "49a6da9e-5859-4db6-a23b-f342680f695c"
      },
      "source": [
        "cnn_do_model.evaluate(x_test, y_test)"
      ],
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "10000/10000 [==============================] - 1s 56us/step\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.0775438034596853, 0.9767]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 33
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nwud0IKlqxFk",
        "colab_type": "text"
      },
      "source": [
        "Now the accuracies for training, validation and test data are more in line."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4feXt0MbqxFk",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "cnn_do_model.save('mnist_cnn_do.h5')"
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}