SVHN_Dataset_CV_DL.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "BApX9qgNsqV0",
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "import h5py\n",
    "import numpy as np\n",
    "\n",
    "# Open the file as readonly\n",
    "h5f = h5py.File('/content/drive/My Drive/SVHN_single_grey1.h5', 'r')\n",
    "\n",
    "# Load the training, test and validation set\n",
    "X_train = h5f['X_train'][:]\n",
    "y_train1 = h5f['y_train'][:]\n",
    "X_test = h5f['X_test'][:]\n",
    "y_test1 = h5f['y_test'][:]\n",
    "\n",
    "\n",
    "# Close this file\n",
    "h5f.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "f0vPUGcf26Og",
    "outputId": "50a7ac93-3658-45d9-ad0b-4539aec3ddc4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
     ]
    }
   ],
   "source": [
    "from google.colab import drive\n",
    "drive.mount('/content/drive')\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "aTeAO8HpsqV8",
    "outputId": "7c15bbda-05d2-42ed-923d-ed4450ceee7b"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "## Importing the required libraries\n",
    "import numpy as np\n",
    "from keras.datasets import mnist\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.utils import np_utils\n",
    "\n",
    "# fix random seed for reproducibility\n",
    "seed = 7\n",
    "np.random.seed(seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 115
    },
    "colab_type": "code",
    "id": "Bvsc8ytHsqWD",
    "outputId": "39e5f3c8-329c-4541-c2ac-9fe3598e7b23"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAk4AAABRCAYAAAAkRimSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
      "text/plain": [
       "<Figure size 720x72 with 10 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "label for each of the above image: [2 6 7 4 4 0 3 0 7 3]\n"
     ]
    }
   ],
   "source": [
    "# visualizing the first 10 images in the dataset and their labels\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "plt.figure(figsize=(10, 1))\n",
    "for i in range(10):\n",
    "    plt.subplot(1, 10, i+1)\n",
    "    plt.imshow(X_train[i], cmap=\"gray\")\n",
    "    plt.axis('off')\n",
    "plt.show()\n",
    "print('label for each of the above image: %s' % (y_train1[0:10]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 51
    },
    "colab_type": "code",
    "id": "D9YPwf9ysqWU",
    "outputId": "2224f4ca-aa8f-4e81-da5e-692e6d40cbe9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training set (42000, 1024) (42000,)\n",
      "Test set (18000, 1024) (18000,)\n"
     ]
    }
   ],
   "source": [
    "X_train = X_train.reshape(X_train.shape[0], 1024)\n",
    "X_test = X_test.reshape(X_test.shape[0], 1024)\n",
    "\n",
    "# # normalize inputs from 0-255 to 0-1\n",
    "X_train = X_train / 255.0\n",
    "X_test = X_test / 255.0\n",
    "\n",
    "print('Training set', X_train.shape, y_train1.shape)\n",
    "print('Test set', X_test.shape, y_test1.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 51
    },
    "colab_type": "code",
    "id": "H4kHHMqCsqWf",
    "outputId": "4533e6af-a696-4ac7-b989-2bfa193f962e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(18000, 1024)\n",
      "(18000,)\n"
     ]
    }
   ],
   "source": [
    "print(X_test.shape)\n",
    "print(y_test1.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "zL0lYER4sqWw"
   },
   "outputs": [],
   "source": [
    "# one hot encode outputs\n",
    "y_train = np_utils.to_categorical(y_train1)\n",
    "y_test = np_utils.to_categorical(y_test1)\n",
    "\n",
    "# no.of classes\n",
    "num_classes = y_test.shape[1]\n",
    "num_hidden = 32"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "EnOsi1o0sqW1",
    "outputId": "e08a127d-9231-4054-b5c8-1156e151dcdf"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10"
      ]
     },
     "execution_count": 9,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ars5ie87jCgL"
   },
   "outputs": [],
   "source": [
    "x_tr = []\n",
    "for i in range(42000):\n",
    "    x_tr.append(X_train[i,:].flatten())\n",
    "x_te = []\n",
    "for i in range(18000):\n",
    "    x_te.append(X_test[i,:].flatten())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "mzTBd-KIlTDu"
   },
   "outputs": [],
   "source": [
    "x_tr =x_tr[:2000]\n",
    "x_te =x_te[:2000]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "zkl53rqKlva2"
   },
   "outputs": [],
   "source": [
    "y_tr = y_train1[0:2000]\n",
    "y_te = y_test1[0:2000]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 85
    },
    "colab_type": "code",
    "id": "Hu7wmUi6mAZl",
    "outputId": "a4cd46d9-0ecd-41b4-cc53-18093e46a290"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2000\n",
      "2000\n",
      "2000\n",
      "2000\n"
     ]
    }
   ],
   "source": [
    "print(len(x_tr))\n",
    "print(len(y_tr))\n",
    "print(len(x_te))\n",
    "print(len(y_tr))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "UJDUoaEj1d6e"
   },
   "source": [
    "## Neural Network Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "MmKU7JeWz23r"
   },
   "outputs": [],
   "source": [
    "from keras.layers import BatchNormalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1394
    },
    "colab_type": "code",
    "id": "Cmi81Gr5sqW-",
    "outputId": "f49eeb6f-87ce-4e09-fd82-426842c7a79e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 42000 samples, validate on 18000 samples\n",
      "Epoch 1/30\n",
      " - 2s - loss: 1.5298 - acc: 0.4847 - val_loss: 1.0465 - val_acc: 0.6742\n",
      "Epoch 2/30\n",
      " - 2s - loss: 0.9724 - acc: 0.6959 - val_loss: 0.9098 - val_acc: 0.7159\n",
      "Epoch 3/30\n",
      " - 2s - loss: 0.8173 - acc: 0.7486 - val_loss: 0.8234 - val_acc: 0.7464\n",
      "Epoch 4/30\n",
      " - 2s - loss: 0.7333 - acc: 0.7725 - val_loss: 0.7403 - val_acc: 0.7778\n",
      "Epoch 5/30\n",
      " - 2s - loss: 0.6619 - acc: 0.7968 - val_loss: 0.7220 - val_acc: 0.7846\n",
      "Epoch 6/30\n",
      " - 2s - loss: 0.6083 - acc: 0.8104 - val_loss: 0.7070 - val_acc: 0.7962\n",
      "Epoch 7/30\n",
      " - 2s - loss: 0.5786 - acc: 0.8202 - val_loss: 0.6481 - val_acc: 0.8099\n",
      "Epoch 8/30\n",
      " - 2s - loss: 0.5403 - acc: 0.8329 - val_loss: 0.6509 - val_acc: 0.8126\n",
      "Epoch 9/30\n",
      " - 2s - loss: 0.5109 - acc: 0.8408 - val_loss: 0.6444 - val_acc: 0.8160\n",
      "Epoch 10/30\n",
      " - 2s - loss: 0.4877 - acc: 0.8485 - val_loss: 0.6588 - val_acc: 0.8126\n",
      "Epoch 11/30\n",
      " - 2s - loss: 0.4643 - acc: 0.8529 - val_loss: 0.6406 - val_acc: 0.8177\n",
      "Epoch 12/30\n",
      " - 2s - loss: 0.4480 - acc: 0.8608 - val_loss: 0.6453 - val_acc: 0.8187\n",
      "Epoch 13/30\n",
      " - 2s - loss: 0.4318 - acc: 0.8632 - val_loss: 0.6726 - val_acc: 0.8109\n",
      "Epoch 14/30\n",
      " - 2s - loss: 0.4124 - acc: 0.8691 - val_loss: 0.6311 - val_acc: 0.8205\n",
      "Epoch 15/30\n",
      " - 2s - loss: 0.3985 - acc: 0.8751 - val_loss: 0.6291 - val_acc: 0.8218\n",
      "Epoch 16/30\n",
      " - 2s - loss: 0.3805 - acc: 0.8801 - val_loss: 0.6128 - val_acc: 0.8287\n",
      "Epoch 17/30\n",
      " - 2s - loss: 0.3805 - acc: 0.8801 - val_loss: 0.6206 - val_acc: 0.8312\n",
      "Epoch 18/30\n",
      " - 2s - loss: 0.3615 - acc: 0.8860 - val_loss: 0.6350 - val_acc: 0.8289\n",
      "Epoch 19/30\n",
      " - 2s - loss: 0.3604 - acc: 0.8860 - val_loss: 0.6182 - val_acc: 0.8368\n",
      "Epoch 20/30\n",
      " - 2s - loss: 0.3407 - acc: 0.8925 - val_loss: 0.6386 - val_acc: 0.8293\n",
      "Epoch 21/30\n",
      " - 2s - loss: 0.3297 - acc: 0.8954 - val_loss: 0.6016 - val_acc: 0.8418\n",
      "Epoch 22/30\n",
      " - 2s - loss: 0.3177 - acc: 0.8996 - val_loss: 0.6316 - val_acc: 0.8340\n",
      "Epoch 23/30\n",
      " - 2s - loss: 0.3191 - acc: 0.8987 - val_loss: 0.6385 - val_acc: 0.8288\n",
      "Epoch 24/30\n",
      " - 2s - loss: 0.3109 - acc: 0.9017 - val_loss: 0.6350 - val_acc: 0.8348\n",
      "Epoch 25/30\n",
      " - 2s - loss: 0.3013 - acc: 0.9042 - val_loss: 0.6312 - val_acc: 0.8364\n",
      "Epoch 26/30\n",
      " - 2s - loss: 0.2984 - acc: 0.9055 - val_loss: 0.6335 - val_acc: 0.8366\n",
      "Epoch 27/30\n",
      " - 2s - loss: 0.2920 - acc: 0.9078 - val_loss: 0.6278 - val_acc: 0.8401\n",
      "Epoch 28/30\n",
      " - 2s - loss: 0.2762 - acc: 0.9127 - val_loss: 0.6196 - val_acc: 0.8439\n",
      "Epoch 29/30\n",
      " - 2s - loss: 0.2785 - acc: 0.9110 - val_loss: 0.6385 - val_acc: 0.8411\n",
      "Epoch 30/30\n",
      " - 2s - loss: 0.2697 - acc: 0.9141 - val_loss: 0.6711 - val_acc: 0.8357\n",
      "Error: 16.43%\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "batch_normalization_3 (Batch (None, 1024)              4096      \n",
      "_________________________________________________________________\n",
      "dense_11 (Dense)             (None, 256)               262400    \n",
      "_________________________________________________________________\n",
      "dense_12 (Dense)             (None, 64)                16448     \n",
      "_________________________________________________________________\n",
      "dense_13 (Dense)             (None, 64)                4160      \n",
      "_________________________________________________________________\n",
      "dense_14 (Dense)             (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "dense_15 (Dense)             (None, 10)                330       \n",
      "=================================================================\n",
      "Total params: 289,514\n",
      "Trainable params: 287,466\n",
      "Non-trainable params: 2,048\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# define model\n",
    "import keras\n",
    "from keras import losses\n",
    "from keras import optimizers\n",
    "from keras.layers import Dropout, MaxPooling2D\n",
    "\n",
    "\n",
    "\n",
    "def nn_model():\n",
    "    # create model\n",
    "    model = Sequential()  \n",
    "    model.add(BatchNormalization(input_shape = (1024,)))  ### Batch_Norm Layer\n",
    "    \n",
    "    model.add(Dense(256, activation='relu')) ###Multiple Dense units with Relu activation\n",
    "    model.add(Dense(64, activation='relu'))\n",
    "    model.add(Dense(64, activation='relu'))\n",
    "    model.add(Dense(32, activation='relu'))\n",
    "    model.add(Dense(num_classes, activation='softmax'))\n",
    "    \n",
    "    # Compile model\n",
    "    sgd = optimizers.Adam(lr=1e-3)\n",
    "    model.compile(loss=losses.categorical_crossentropy, optimizer=sgd, metrics=['accuracy']) ### Loss function = Categorical cross entropy\n",
    "    return model\n",
    "\n",
    "# build the model\n",
    "model = nn_model()\n",
    "\n",
    "# Fit the model\n",
    "model.fit(X_train, y_train, validation_data=(X_test, y_test), epochs=30, batch_size=200, verbose=2)\n",
    "\n",
    "# Final evaluation of the model\n",
    "scores = model.evaluate(X_test, y_test, verbose=0)\n",
    "\n",
    "print(\"Error: %.2f%%\" % (100-scores[1]*100))\n",
    "\n",
    "model.summary()"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "SVHN_NN_Solution.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "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"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}