2.Cifar 10 Notebook with BN without dropout.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "kernelspec": {
      "display_name": "Python [conda env:tensorflow_p36]",
      "language": "python",
      "name": "conda-env-tensorflow_p36-py"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.6.4"
    },
    "colab": {
      "name": "2.+Cifar_10_Notebook_with_BN_without_dropout.ipynb",
      "provenance": []
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "wfarxqZ_XQCP",
        "colab_type": "code",
        "colab": {},
        "outputId": "bfd68d6b-39b7-4c11-fbc9-9c5de2433492"
      },
      "source": [
        "from __future__ import print_function\n",
        "import keras\n",
        "from keras.datasets import cifar10\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense, Dropout, Activation, Flatten\n",
        "from keras.layers.normalization import BatchNormalization\n",
        "from keras.layers import Conv2D, MaxPooling2D\n",
        "import numpy as np\n",
        "import os\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/home/ubuntu/anaconda3/envs/tensorflow_p36/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
            "  from ._conv import register_converters as _register_converters\n",
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kcJtUeckXQCd",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Defining the parameters\n",
        "batch_size = 32\n",
        "num_classes = 10\n",
        "epochs = 50"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "o5nvej9jXQCo",
        "colab_type": "code",
        "colab": {},
        "outputId": "97cac8d4-576c-40de-8804-96ea001fb7de"
      },
      "source": [
        "# Splitting the data between train and test\n",
        "(x_train, y_train), (x_test, y_test) = cifar10.load_data()\n",
        "print('x_train shape:', x_train.shape)\n",
        "print(x_train.shape[0], 'train samples')\n",
        "print(x_test.shape[0], 'test samples')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "x_train shape: (50000, 32, 32, 3)\n",
            "50000 train samples\n",
            "10000 test samples\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u5N4gaVEXQCx",
        "colab_type": "code",
        "colab": {},
        "outputId": "4b76cdbd-41a5-421e-e139-b6e7c23c88b1"
      },
      "source": [
        "# plotting some random 10 images\n",
        "\n",
        "class_names = ['airplane','automobile','bird','cat','deer',\n",
        "               'dog','frog','horse','ship','truck']\n",
        "\n",
        "fig = plt.figure(figsize=(8,3))\n",
        "for i in range(num_classes):\n",
        "    ax = fig.add_subplot(2, 5, 1 + i, xticks=[], yticks=[])\n",
        "    idx = np.where(y_train[:]==i)[0]\n",
        "    features_idx = x_train[idx,::]\n",
        "    img_num = np.random.randint(features_idx.shape[0])\n",
        "    im = (features_idx[img_num,::])\n",
        "    ax.set_title(class_names[i])\n",
        "    plt.imshow(im)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x216 with 10 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AhYgEsGnXQC6",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Convert class vectors to binary class matrices.\n",
        "y_train = keras.utils.to_categorical(y_train, num_classes)\n",
        "y_test = keras.utils.to_categorical(y_test, num_classes)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZMwL0uj8XQDE",
        "colab_type": "code",
        "colab": {},
        "outputId": "bee85d59-2e59-466c-ad9e-49048ed56c26"
      },
      "source": [
        "# Printing sample data\n",
        "print(y_train[:10])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-LuNl7QIXQDN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model = Sequential()\n",
        "model.add(Conv2D(32, (3, 3), padding='same',\n",
        "                 input_shape=x_train.shape[1:]))\n",
        "model.add(Activation('relu'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Conv2D(32, (3, 3)))\n",
        "model.add(Activation('relu'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "\n",
        "model.add(Conv2D(64, (3, 3), padding='same'))\n",
        "model.add(Activation('relu'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(Conv2D(64, (3, 3)))\n",
        "model.add(Activation('relu'))\n",
        "model.add(BatchNormalization())\n",
        "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "\n",
        "model.add(Flatten())\n",
        "model.add(Dense(512))\n",
        "model.add(Activation('relu'))\n",
        "model.add(Dropout(0.5))\n",
        "model.add(Dense(num_classes))\n",
        "model.add(Activation('softmax'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OUa0fvv9XQDX",
        "colab_type": "code",
        "colab": {},
        "outputId": "86d8602a-1be3-4d6b-eb50-a687dda0f519"
      },
      "source": [
        "# summary of the model\n",
        "print(model.summary())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_1 (Conv2D)            (None, 32, 32, 32)        896       \n",
            "_________________________________________________________________\n",
            "activation_1 (Activation)    (None, 32, 32, 32)        0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_1 (Batch (None, 32, 32, 32)        128       \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 30, 30, 32)        9248      \n",
            "_________________________________________________________________\n",
            "activation_2 (Activation)    (None, 30, 30, 32)        0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_2 (Batch (None, 30, 30, 32)        128       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 15, 15, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 15, 15, 64)        18496     \n",
            "_________________________________________________________________\n",
            "activation_3 (Activation)    (None, 15, 15, 64)        0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_3 (Batch (None, 15, 15, 64)        256       \n",
            "_________________________________________________________________\n",
            "conv2d_4 (Conv2D)            (None, 13, 13, 64)        36928     \n",
            "_________________________________________________________________\n",
            "activation_4 (Activation)    (None, 13, 13, 64)        0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_4 (Batch (None, 13, 13, 64)        256       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 6, 6, 64)          0         \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 2304)              0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 512)               1180160   \n",
            "_________________________________________________________________\n",
            "activation_5 (Activation)    (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 10)                5130      \n",
            "_________________________________________________________________\n",
            "activation_6 (Activation)    (None, 10)                0         \n",
            "=================================================================\n",
            "Total params: 1,251,626\n",
            "Trainable params: 1,251,242\n",
            "Non-trainable params: 384\n",
            "_________________________________________________________________\n",
            "None\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EkyU9GwCXQDe",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# compile\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer='sgd',\n",
        "              metrics=['accuracy'])\n",
        "\n",
        "x_train = x_train.astype('float32')\n",
        "x_test = x_test.astype('float32')\n",
        "\n",
        "# Normalizing the input image\n",
        "x_train /= 255\n",
        "x_test /= 255\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lH_Pc9WvXQDn",
        "colab_type": "code",
        "colab": {},
        "outputId": "d5973029-3174-4dbb-bade-6ff028915565"
      },
      "source": [
        "# Training the model\n",
        "model.fit(x_train, y_train,\n",
        "              batch_size=batch_size,\n",
        "              epochs=epochs,\n",
        "              validation_data=(x_test, y_test),\n",
        "              shuffle=True)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 50000 samples, validate on 10000 samples\n",
            "Epoch 1/50\n",
            "50000/50000 [==============================] - 21s 418us/step - loss: 1.5669 - acc: 0.4556 - val_loss: 1.2642 - val_acc: 0.5416\n",
            "Epoch 2/50\n",
            "50000/50000 [==============================] - 20s 392us/step - loss: 1.1605 - acc: 0.5898 - val_loss: 1.1346 - val_acc: 0.5885\n",
            "Epoch 3/50\n",
            "50000/50000 [==============================] - 20s 391us/step - loss: 0.9843 - acc: 0.6516 - val_loss: 0.8799 - val_acc: 0.6878\n",
            "Epoch 4/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.8683 - acc: 0.6946 - val_loss: 0.8654 - val_acc: 0.6950\n",
            "Epoch 5/50\n",
            "50000/50000 [==============================] - 19s 389us/step - loss: 0.7806 - acc: 0.7267 - val_loss: 0.7691 - val_acc: 0.7332\n",
            "Epoch 6/50\n",
            "50000/50000 [==============================] - 19s 389us/step - loss: 0.7106 - acc: 0.7499 - val_loss: 0.7544 - val_acc: 0.7362\n",
            "Epoch 7/50\n",
            "50000/50000 [==============================] - 19s 389us/step - loss: 0.6559 - acc: 0.7688 - val_loss: 0.7481 - val_acc: 0.7402\n",
            "Epoch 8/50\n",
            "50000/50000 [==============================] - 20s 393us/step - loss: 0.6082 - acc: 0.7871 - val_loss: 0.6894 - val_acc: 0.7618\n",
            "Epoch 9/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.5629 - acc: 0.8015 - val_loss: 0.6972 - val_acc: 0.7582\n",
            "Epoch 10/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.5239 - acc: 0.8140 - val_loss: 0.7060 - val_acc: 0.7569\n",
            "Epoch 11/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.4842 - acc: 0.8297 - val_loss: 0.6678 - val_acc: 0.7719\n",
            "Epoch 12/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.4525 - acc: 0.8389 - val_loss: 0.6628 - val_acc: 0.7722\n",
            "Epoch 13/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.4188 - acc: 0.8529 - val_loss: 0.6738 - val_acc: 0.7755\n",
            "Epoch 14/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.3897 - acc: 0.8609 - val_loss: 0.7095 - val_acc: 0.7645\n",
            "Epoch 15/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.3573 - acc: 0.8747 - val_loss: 0.6839 - val_acc: 0.7749\n",
            "Epoch 16/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.3361 - acc: 0.8808 - val_loss: 0.6542 - val_acc: 0.7875\n",
            "Epoch 17/50\n",
            "50000/50000 [==============================] - 19s 390us/step - loss: 0.3108 - acc: 0.8883 - val_loss: 0.6900 - val_acc: 0.7772\n",
            "Epoch 18/50\n",
            "50000/50000 [==============================] - 19s 389us/step - loss: 0.2897 - acc: 0.8970 - val_loss: 0.6985 - val_acc: 0.7774\n",
            "Epoch 19/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.2669 - acc: 0.9040 - val_loss: 0.7531 - val_acc: 0.7721\n",
            "Epoch 20/50\n",
            "50000/50000 [==============================] - 19s 390us/step - loss: 0.2447 - acc: 0.9121 - val_loss: 0.6995 - val_acc: 0.7876\n",
            "Epoch 21/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.2260 - acc: 0.9208 - val_loss: 0.7171 - val_acc: 0.7879\n",
            "Epoch 22/50\n",
            "50000/50000 [==============================] - 19s 390us/step - loss: 0.2112 - acc: 0.9257 - val_loss: 0.7539 - val_acc: 0.7852\n",
            "Epoch 23/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.1959 - acc: 0.9307 - val_loss: 0.7350 - val_acc: 0.7923\n",
            "Epoch 24/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.1802 - acc: 0.9355 - val_loss: 0.7393 - val_acc: 0.7888\n",
            "Epoch 25/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.1711 - acc: 0.9399 - val_loss: 0.7224 - val_acc: 0.7967\n",
            "Epoch 26/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.1600 - acc: 0.9423 - val_loss: 0.7619 - val_acc: 0.7877\n",
            "Epoch 27/50\n",
            "50000/50000 [==============================] - 19s 390us/step - loss: 0.1498 - acc: 0.9487 - val_loss: 0.7753 - val_acc: 0.7865\n",
            "Epoch 28/50\n",
            "50000/50000 [==============================] - 19s 390us/step - loss: 0.1424 - acc: 0.9502 - val_loss: 0.7850 - val_acc: 0.7885\n",
            "Epoch 29/50\n",
            "50000/50000 [==============================] - 19s 390us/step - loss: 0.1345 - acc: 0.9518 - val_loss: 0.8050 - val_acc: 0.7925\n",
            "Epoch 30/50\n",
            "50000/50000 [==============================] - 19s 390us/step - loss: 0.1242 - acc: 0.9566 - val_loss: 0.7938 - val_acc: 0.7901\n",
            "Epoch 31/50\n",
            "50000/50000 [==============================] - 20s 391us/step - loss: 0.1239 - acc: 0.9558 - val_loss: 0.8703 - val_acc: 0.7742\n",
            "Epoch 32/50\n",
            "50000/50000 [==============================] - 19s 390us/step - loss: 0.1121 - acc: 0.9615 - val_loss: 0.8173 - val_acc: 0.7913\n",
            "Epoch 33/50\n",
            "50000/50000 [==============================] - 19s 390us/step - loss: 0.1093 - acc: 0.9632 - val_loss: 0.8185 - val_acc: 0.7948\n",
            "Epoch 34/50\n",
            "50000/50000 [==============================] - 19s 390us/step - loss: 0.1023 - acc: 0.9636 - val_loss: 0.8599 - val_acc: 0.7834\n",
            "Epoch 35/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.0956 - acc: 0.9664 - val_loss: 0.8692 - val_acc: 0.7814\n",
            "Epoch 36/50\n",
            "50000/50000 [==============================] - 19s 390us/step - loss: 0.0927 - acc: 0.9683 - val_loss: 0.8758 - val_acc: 0.7865\n",
            "Epoch 37/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.0863 - acc: 0.9699 - val_loss: 0.8669 - val_acc: 0.7942\n",
            "Epoch 38/50\n",
            "50000/50000 [==============================] - 20s 391us/step - loss: 0.0852 - acc: 0.9707 - val_loss: 0.8649 - val_acc: 0.7921\n",
            "Epoch 39/50\n",
            "50000/50000 [==============================] - 20s 391us/step - loss: 0.0813 - acc: 0.9721 - val_loss: 0.8827 - val_acc: 0.7916\n",
            "Epoch 40/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.0800 - acc: 0.9724 - val_loss: 0.8649 - val_acc: 0.7986\n",
            "Epoch 41/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.0742 - acc: 0.9746 - val_loss: 0.9434 - val_acc: 0.7878\n",
            "Epoch 42/50\n",
            "50000/50000 [==============================] - 20s 391us/step - loss: 0.0730 - acc: 0.9746 - val_loss: 0.8730 - val_acc: 0.7954\n",
            "Epoch 43/50\n",
            "50000/50000 [==============================] - 20s 391us/step - loss: 0.0700 - acc: 0.9757 - val_loss: 0.8952 - val_acc: 0.7971\n",
            "Epoch 44/50\n",
            "50000/50000 [==============================] - 20s 391us/step - loss: 0.0669 - acc: 0.9777 - val_loss: 0.9063 - val_acc: 0.7979\n",
            "Epoch 45/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.0672 - acc: 0.9768 - val_loss: 0.9189 - val_acc: 0.7982\n",
            "Epoch 46/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.0649 - acc: 0.9777 - val_loss: 0.9207 - val_acc: 0.7979\n",
            "Epoch 47/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.0590 - acc: 0.9800 - val_loss: 0.8991 - val_acc: 0.7991\n",
            "Epoch 48/50\n",
            "50000/50000 [==============================] - 20s 391us/step - loss: 0.0583 - acc: 0.9802 - val_loss: 0.9539 - val_acc: 0.7943\n",
            "Epoch 49/50\n",
            "50000/50000 [==============================] - 20s 390us/step - loss: 0.0553 - acc: 0.9813 - val_loss: 0.9931 - val_acc: 0.7932\n",
            "Epoch 50/50\n",
            "50000/50000 [==============================] - 19s 390us/step - loss: 0.0544 - acc: 0.9817 - val_loss: 0.9499 - val_acc: 0.7950\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7fd798247748>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2mtT4GMMXQDu",
        "colab_type": "text"
      },
      "source": [
        "Result: The training accuracy is increasing because of BN but the difference between train accuracy and validation accuracy is high."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Abl0MAOsXQDv",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}