NN_MNIST_Digit_Recognition.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "NN_MNIST_Digit_Recognition.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/girisairam450/Episodes/blob/master/NN_MNIST_Digit_Recognition.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "metadata": {
        "id": "TaTeUTSiGc0L",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "e6a3a027-28bd-42b6-e648-fe53839c038d"
      },
      "cell_type": "code",
      "source": [
        "! pip install msgpack"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: msgpack in /usr/local/lib/python3.6/dist-packages (0.5.6)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "MRTKcd7QGtpS",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "c8e6f097-a691-4cde-e5b9-d437be3ff33d"
      },
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import pickle\n",
        "import matplotlib.pyplot as plt\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import h5py\n",
        "import sklearn\n",
        "import sklearn.datasets\n",
        "import scipy\n",
        "\n",
        "from PIL import Image\n",
        "from scipy import ndimage\n",
        "\n",
        "import tensorflow as tf\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense, Dropout, BatchNormalization, Activation\n",
        "from keras import regularizers\n",
        "from keras.datasets import mnist\n",
        "\n",
        "np.random.seed(7)\n",
        "\n",
        "\n",
        "%matplotlib inline"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "metadata": {
        "id": "mpKfUti_G8b9",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "#(x_train, y_train), (x_test, y_test)\n",
        "(x_train, y_train), (x_test, y_test) = mnist.load_data()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "yxlGj6DcNP61",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def one_hot(j):\n",
        "    # input is the target dataset of shape (1, m) where m is the number of data points\n",
        "    # returns a 2 dimensional array of shape (10, m) where each target value is converted to a one hot encoding\n",
        "    # Look at the next block of code for a better understanding of one hot encoding\n",
        "    n = j.shape[0]\n",
        "    new_array = np.zeros((10, n))\n",
        "    index = 0\n",
        "    for res in j:\n",
        "        new_array[res][index] = 1.0\n",
        "        index = index + 1\n",
        "    return new_array"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "M5p4Gd35NiXc",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        },
        "outputId": "e94e2746-c7a1-4d6b-af4c-09044d8ee13f"
      },
      "cell_type": "code",
      "source": [
        "data = np.array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])\n",
        "one_hot(data)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[1., 0., 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., 1., 0., 0., 0., 0., 0., 0.],\n",
              "       [0., 0., 0., 0., 1., 0., 0., 0., 0., 0.],\n",
              "       [0., 0., 0., 0., 0., 1., 0., 0., 0., 0.],\n",
              "       [0., 0., 0., 0., 0., 0., 1., 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., 0., 0., 0., 0., 0., 0., 1.]])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "metadata": {
        "id": "IbtbQ4MJOMLQ",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def data_wrapper():\n",
        "    (x_train, y_train), (x_test, y_test) = mnist.load_data()\n",
        "    \n",
        "    training_inputs = x_train.T\n",
        "    training_results = y_train\n",
        "    train_set_y = one_hot(training_results)\n",
        "    \n",
        "    validation_inputs = x_test.T\n",
        "    validation_results = y_test\n",
        "    validation_set_y = one_hot(validation_results)\n",
        "    \n",
        "    return (training_inputs, train_set_y, validation_inputs, validation_set_y)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "GqOsLURnPkov",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "train_set_x, train_set_y, test_set_x, test_set_y = data_wrapper()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "wDSG4YobPq4z",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "train_set_x = train_set_x.T\n",
        "train_set_y = train_set_y.T\n",
        "test_set_x = test_set_x.T\n",
        "test_set_y = test_set_y.T"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "dlb-UA-qP4zQ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        },
        "outputId": "34c8beac-d3a6-4dd0-a13a-e19a19db33d6"
      },
      "cell_type": "code",
      "source": [
        "print (\"train_set_x shape: \" + str(train_set_x.shape))\n",
        "print (\"train_set_y shape: \" + str(train_set_y.shape))\n",
        "print (\"test_set_x shape: \" + str(test_set_x.shape))\n",
        "print (\"test_set_y shape: \" + str(test_set_y.shape))\n"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "train_set_x shape: (60000, 28, 28)\n",
            "train_set_y shape: (60000, 10)\n",
            "test_set_x shape: (10000, 28, 28)\n",
            "test_set_y shape: (10000, 10)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "7bSiKEaHTJl1",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "train_set_x = train_set_x.reshape(60000,784)\n",
        "test_set_x  =test_set_x.reshape(10000,784)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "d0iwCsG5QOOl",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        },
        "outputId": "a5fac83a-141e-41c1-db02-0149189e093c"
      },
      "cell_type": "code",
      "source": [
        "index  = 1000\n",
        "k = train_set_x[index,:]\n",
        "k = k.reshape((28, 28))\n",
        "plt.title('Label is {label}'.format(label= y_train))\n",
        "plt.imshow(k, cmap='gray')"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fcf9b64c550>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPoAAAEHCAYAAACHl1tOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "_Oppb8mSQ6tz",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# create model\n",
        "nn_model = Sequential()\n",
        "nn_model.add(Dense(35, input_dim=784, activation=tf.nn.relu))\n",
        "nn_model.add(Dropout(0.3))\n",
        "nn_model.add(Dense(21, activation = 'relu'))\n",
        "nn_model.add(Dropout(0.1))\n",
        "nn_model.add(Dense(16, activation = 'relu'))\n",
        "nn_model.add(Dense(10, activation='softmax'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "Seg5QhGeSaaJ",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "nn_model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "OIy8zrzAW7Rs",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 340
        },
        "outputId": "715963ea-0691-4ff6-ad4d-7edd41b28ddb"
      },
      "cell_type": "code",
      "source": [
        "nn_model.summary()"
      ],
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "dense_16 (Dense)             (None, 35)                27475     \n",
            "_________________________________________________________________\n",
            "dropout_7 (Dropout)          (None, 35)                0         \n",
            "_________________________________________________________________\n",
            "dense_17 (Dense)             (None, 21)                756       \n",
            "_________________________________________________________________\n",
            "dropout_8 (Dropout)          (None, 21)                0         \n",
            "_________________________________________________________________\n",
            "dense_18 (Dense)             (None, 16)                352       \n",
            "_________________________________________________________________\n",
            "dense_19 (Dense)             (None, 10)                170       \n",
            "=================================================================\n",
            "Total params: 28,753\n",
            "Trainable params: 28,753\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "XN-r98bUSgs1",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1054
        },
        "outputId": "8fb01ad5-ead1-4d87-84a7-3021bcb286b0"
      },
      "cell_type": "code",
      "source": [
        "nn_model.fit(train_set_x, train_set_y, epochs=30, batch_size=10)"
      ],
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/30\n",
            "60000/60000 [==============================] - 10s 164us/step - loss: 8.9124 - acc: 0.4388\n",
            "Epoch 2/30\n",
            "60000/60000 [==============================] - 9s 153us/step - loss: 6.4534 - acc: 0.5931\n",
            "Epoch 3/30\n",
            "60000/60000 [==============================] - 9s 152us/step - loss: 5.9488 - acc: 0.6264\n",
            "Epoch 4/30\n",
            "60000/60000 [==============================] - 9s 152us/step - loss: 5.5471 - acc: 0.6521\n",
            "Epoch 5/30\n",
            "60000/60000 [==============================] - 9s 153us/step - loss: 5.1700 - acc: 0.6739\n",
            "Epoch 6/30\n",
            "60000/60000 [==============================] - 9s 155us/step - loss: 3.9273 - acc: 0.6846\n",
            "Epoch 7/30\n",
            "60000/60000 [==============================] - 9s 153us/step - loss: 1.8249 - acc: 0.5571\n",
            "Epoch 8/30\n",
            "60000/60000 [==============================] - 9s 153us/step - loss: 1.5827 - acc: 0.5439\n",
            "Epoch 9/30\n",
            "60000/60000 [==============================] - 10s 165us/step - loss: 1.4068 - acc: 0.5519\n",
            "Epoch 10/30\n",
            "60000/60000 [==============================] - 9s 158us/step - loss: 1.2811 - acc: 0.6005\n",
            "Epoch 11/30\n",
            "60000/60000 [==============================] - 9s 155us/step - loss: 1.0022 - acc: 0.7048\n",
            "Epoch 12/30\n",
            "60000/60000 [==============================] - 9s 153us/step - loss: 0.8149 - acc: 0.7588\n",
            "Epoch 13/30\n",
            "60000/60000 [==============================] - 9s 154us/step - loss: 0.7179 - acc: 0.7854\n",
            "Epoch 14/30\n",
            "60000/60000 [==============================] - 9s 152us/step - loss: 0.6588 - acc: 0.8004\n",
            "Epoch 15/30\n",
            "60000/60000 [==============================] - 9s 151us/step - loss: 0.6384 - acc: 0.8097\n",
            "Epoch 16/30\n",
            "60000/60000 [==============================] - 9s 154us/step - loss: 0.6213 - acc: 0.8164\n",
            "Epoch 17/30\n",
            "60000/60000 [==============================] - 9s 152us/step - loss: 0.5987 - acc: 0.8256\n",
            "Epoch 18/30\n",
            "60000/60000 [==============================] - 9s 156us/step - loss: 0.5954 - acc: 0.8250\n",
            "Epoch 19/30\n",
            "60000/60000 [==============================] - 9s 155us/step - loss: 0.5838 - acc: 0.8290\n",
            "Epoch 20/30\n",
            "60000/60000 [==============================] - 9s 157us/step - loss: 0.5829 - acc: 0.8282\n",
            "Epoch 21/30\n",
            "60000/60000 [==============================] - 9s 154us/step - loss: 0.5753 - acc: 0.8318\n",
            "Epoch 22/30\n",
            "60000/60000 [==============================] - 9s 153us/step - loss: 0.5696 - acc: 0.8362\n",
            "Epoch 23/30\n",
            "60000/60000 [==============================] - 9s 155us/step - loss: 0.5598 - acc: 0.8386\n",
            "Epoch 24/30\n",
            "60000/60000 [==============================] - 9s 154us/step - loss: 0.5648 - acc: 0.8373\n",
            "Epoch 25/30\n",
            "60000/60000 [==============================] - 10s 165us/step - loss: 0.5522 - acc: 0.8382\n",
            "Epoch 26/30\n",
            "60000/60000 [==============================] - 10s 173us/step - loss: 0.5531 - acc: 0.8400\n",
            "Epoch 27/30\n",
            "60000/60000 [==============================] - 9s 153us/step - loss: 0.5518 - acc: 0.8381\n",
            "Epoch 28/30\n",
            "60000/60000 [==============================] - 10s 160us/step - loss: 0.5414 - acc: 0.8428\n",
            "Epoch 29/30\n",
            "60000/60000 [==============================] - 9s 151us/step - loss: 0.5375 - acc: 0.8425\n",
            "Epoch 30/30\n",
            "60000/60000 [==============================] - 9s 152us/step - loss: 0.5418 - acc: 0.8423\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7fcf94085240>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 33
        }
      ]
    },
    {
      "metadata": {
        "id": "V3_slVGwUOZH",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "outputId": "71841920-59f6-46b7-9243-26386251251f"
      },
      "cell_type": "code",
      "source": [
        "scores_train = nn_model.evaluate(train_set_x, train_set_y)\n",
        "print(\"\\n%s: %.2f%%\" % (nn_model.metrics_names[1], scores_train[1]*100))"
      ],
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "60000/60000 [==============================] - 2s 30us/step\n",
            "\n",
            "acc: 92.55%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "sGOh1thxUl3q",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "2378d2c2-f516-4f3c-bb3b-bed101bdfa5d"
      },
      "cell_type": "code",
      "source": [
        "predictions = nn_model.predict(test_set_x)\n",
        "predictions = np.argmax(predictions, axis = 1)\n",
        "predictions"
      ],
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([7, 2, 1, ..., 4, 8, 6])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 35
        }
      ]
    },
    {
      "metadata": {
        "id": "3YF9cXUJU63q",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "outputId": "44ef5712-a1fe-4147-f9a6-07a27f7db9f1"
      },
      "cell_type": "code",
      "source": [
        "scores_test = nn_model.evaluate(test_set_x, test_set_y)\n",
        "print(\"\\n%s: %.2f%%\" % (nn_model.metrics_names[1], scores_test[1]*100))"
      ],
      "execution_count": 36,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "10000/10000 [==============================] - 0s 28us/step\n",
            "\n",
            "acc: 92.09%\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "rE3wgwt1VFmb",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        },
        "outputId": "b821e900-7555-4382-f51a-c4c87bf7c56d"
      },
      "cell_type": "code",
      "source": [
        "index  = 9999\n",
        "k = test_set_x[index, :]\n",
        "k = k.reshape((28, 28))\n",
        "plt.title('Label is {label}'.format(label=(predictions[index], np.argmax(test_set_y, axis = 1)[index])))\n",
        "plt.imshow(k, cmap='gray')"
      ],
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fcf92600a20>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 37
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPoAAAEHCAYAAACHl1tOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nKOgiAVDQRQKgoIsE4P8BfqKbDBc206QAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}