mnist_dataset_classification_tfnn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### MNIST dataset classification using NN TensorFlow"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.1.0\n"
     ]
    }
   ],
   "source": [
    "#Feel free to add any other module needed\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "print(tf.__version__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "mnist = tf.keras.datasets.mnist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "(training_images, training_labels), (test_images, test_labels) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of training images set: (60000, 28, 28)\n",
      "Shape of training labels set: (60000,)\n",
      "Shape of test images set: (10000, 28, 28)\n",
      "Shape of test labels set: (10000,)\n"
     ]
    }
   ],
   "source": [
    "print(\"Shape of training images set: {}\".format(training_images.shape))\n",
    "print(\"Shape of training labels set: {}\".format(training_labels.shape))\n",
    "print(\"Shape of test images set: {}\".format(test_images.shape))\n",
    "print(\"Shape of test labels set: {}\".format(test_labels.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x245d702b708>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "### Visualize samples of  images using matplotlib ###\n",
    "\n",
    "#display(training_images)\n",
    "%matplotlib inline\n",
    "plt.imshow(training_images[10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of training images set: (60000, 28, 28, 1)\n",
      "Shape of test images set: (10000, 28, 28, 1)\n",
      "Training images type:float32 \n",
      "Test images type: float32 \n",
      "Test images type: float32 \n"
     ]
    }
   ],
   "source": [
    "#Grouping all data in 4D ARRAY \n",
    "training_images = training_images.reshape(training_images.shape[0], 28, 28, 1)\n",
    "print(\"Shape of training images set: {}\".format(training_images.shape))\n",
    "\n",
    "test_images = test_images.reshape(test_images.shape[0], 28, 28, 1)\n",
    "print(\"Shape of test images set: {}\".format(test_images.shape))\n",
    "\n",
    "print(\"Training images type:{} \".format(training_images.dtype))\n",
    "print(\"Test images type: {} \".format(test_images.dtype))\n",
    "\n",
    "test_images = test_images.astype('float32')\n",
    "print(\"Test images type: {} \".format(test_images.dtype))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x245d534bb08>"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "### Normalize images ###\n",
    "training_images = tf.keras.utils.normalize(training_images, axis=1)\n",
    "test_images = tf.keras.utils.normalize(test_images, axis=1)\n",
    "\n",
    "plt.imshow(training_images[10])\n",
    "plt.imshow(test_images[10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 60000 samples\n",
      "Epoch 1/3\n",
      "60000/60000 [==============================] - 8s 132us/sample - loss: 0.2641 - accuracy: 0.9217\n",
      "Epoch 2/3\n",
      "60000/60000 [==============================] - 5s 87us/sample - loss: 0.1089 - accuracy: 0.9666\n",
      "Epoch 3/3\n",
      "60000/60000 [==============================] - 5s 82us/sample - loss: 0.0750 - accuracy: 0.9765\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x245d5529c48>"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "### Create , compile and run model ###\n",
    "model = tf.keras.models.Sequential()\n",
    "model.add(tf.keras.layers.Flatten())\n",
    "model.add(tf.keras.layers.Dense(128, activation=tf.nn.relu))\n",
    "model.add(tf.keras.layers.Dense(128, activation=tf.nn.relu))\n",
    "model.add(tf.keras.layers.Dense(10, activation=tf.nn.softmax))\n",
    "\n",
    "model.compile(optimizer='adam',\n",
    "              loss='sparse_categorical_crossentropy',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "model.fit(training_images, training_labels, epochs=3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 0s 40us/sample - loss: 0.0954 - accuracy: 0.9709\n",
      "Validation loss: 0.09544917915351689\n",
      "Validation accuracy: 0.9708999991416931\n"
     ]
    }
   ],
   "source": [
    "### Evaluate model ###\n",
    "val_loss, val_accuracy = model.evaluate(test_images, test_labels)\n",
    "print(\"Validation loss: {}\".format(val_loss))\n",
    "print(\"Validation accuracy: {}\".format(val_accuracy))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: mnist_nn.model\\assets\n"
     ]
    }
   ],
   "source": [
    "### Do prediction on a single sample from test set ###\n",
    "model.save('mnist_nn.model')\n",
    "number_reader = tf.keras.models.load_model('mnist_nn.model')\n",
    "prediction = number_reader.predict(test_images)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[4.8469087e-09 2.1479902e-09 1.9899832e-07 2.7437202e-06 8.1017276e-10\n",
      " 1.9784034e-08 3.2546604e-14 9.9999559e-01 1.5548256e-08 1.4419777e-06]\n"
     ]
    }
   ],
   "source": [
    "print(prediction[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "7\n"
     ]
    }
   ],
   "source": [
    "print(np.argmax(prediction[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x245d823d108>"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(test_images[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}