mnist-FFNN.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "mnist = tf.keras.datasets.mnist\n",
    "\n",
    "(X_train, Y_train), (X_test, Y_test) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0  67 232  39   0   0   0   0   0]\n",
      " [  0   0   0   0  62  81   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0 120 180  39   0   0   0   0   0]\n",
      " [  0   0   0   0 126 163   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   2 153 210  40   0   0   0   0   0]\n",
      " [  0   0   0   0 220 163   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0  27 254 162   0   0   0   0   0   0]\n",
      " [  0   0   0   0 222 163   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0 183 254 125   0   0   0   0   0   0]\n",
      " [  0   0   0  46 245 163   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0 198 254  56   0   0   0   0   0   0]\n",
      " [  0   0   0 120 254 163   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "   23 231 254  29   0   0   0   0   0   0]\n",
      " [  0   0   0 159 254 120   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "  163 254 216  16   0   0   0   0   0   0]\n",
      " [  0   0   0 159 254  67   0   0   0   0   0   0   0   0   0  14  86 178\n",
      "  248 254  91   0   0   0   0   0   0   0]\n",
      " [  0   0   0 159 254  85   0   0   0  47  49 116 144 150 241 243 234 179\n",
      "  241 252  40   0   0   0   0   0   0   0]\n",
      " [  0   0   0 150 253 237 207 207 207 253 254 250 240 198 143  91  28   5\n",
      "  233 250   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0 119 177 177 177 177 177  98  56   0   0   0   0   0 102\n",
      "  254 220   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0 169\n",
      "  254 137   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0 169\n",
      "  254  57   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0 169\n",
      "  254  57   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0 169\n",
      "  255  94   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0 169\n",
      "  254  96   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0 169\n",
      "  254 153   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0 169\n",
      "  255 153   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0  96\n",
      "  254 153   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]\n",
      " [  0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0\n",
      "    0   0   0   0   0   0   0   0   0   0]]\n",
      "4\n"
     ]
    }
   ],
   "source": [
    "print(X_train[2])\n",
    "print(Y_train[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.imshow(X_train[2], cmap = plt.cm.binary)\n",
    "#plt.imshow(X_train[0])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = tf.keras.models.Sequential()\n",
    "\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', loss = 'sparse_categorical_crossentropy', metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/6\n",
      "60000/60000 [==============================] - 6s 97us/step - loss: 9.1407 - acc: 0.4326\n",
      "Epoch 2/6\n",
      "60000/60000 [==============================] - 6s 97us/step - loss: 8.9874 - acc: 0.4422\n",
      "Epoch 3/6\n",
      "60000/60000 [==============================] - 6s 98us/step - loss: 9.0934 - acc: 0.4356\n",
      "Epoch 4/6\n",
      "60000/60000 [==============================] - 6s 98us/step - loss: 9.1131 - acc: 0.4345\n",
      "Epoch 5/6\n",
      "60000/60000 [==============================] - 6s 100us/step - loss: 8.9562 - acc: 0.4442\n",
      "Epoch 6/6\n",
      "60000/60000 [==============================] - 6s 101us/step - loss: 8.7184 - acc: 0.4590\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7fc1c6709fd0>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train, Y_train, epochs=6)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 0s 35us/step\n",
      "8.549460787200928\n"
     ]
    }
   ],
   "source": [
    "test_loss, test_accuracy = model.evaluate(X_test, Y_test)\n",
    "print(test_loss)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The model has fit the data properly. It has low variance but the error is too high. We'll train another model after we normalize the data to optimize our model properly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = tf.keras.utils.normalize(X_train, axis = 1)\n",
    "X_test = tf.keras.utils.normalize(X_test, axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/4\n",
      "60000/60000 [==============================] - 7s 125us/step - loss: 0.2600 - acc: 0.9234\n",
      "Epoch 2/4\n",
      "60000/60000 [==============================] - 6s 101us/step - loss: 0.1040 - acc: 0.9681\n",
      "Epoch 3/4\n",
      "60000/60000 [==============================] - 6s 102us/step - loss: 0.0714 - acc: 0.9779\n",
      "Epoch 4/4\n",
      "60000/60000 [==============================] - 6s 107us/step - loss: 0.0524 - acc: 0.9835\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7f14d0291fd0>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model2 = tf.keras.models.Sequential()\n",
    "\n",
    "model2.add(tf.keras.layers.Flatten())\n",
    "model2.add(tf.keras.layers.Dense(128, activation = tf.nn.relu))\n",
    "model2.add(tf.keras.layers.Dense(128, activation = tf.nn.relu))\n",
    "model2.add(tf.keras.layers.Dense(10, activation = tf.nn.softmax))\n",
    "\n",
    "model2.compile(optimizer = 'adam', loss = 'sparse_categorical_crossentropy', metrics = ['accuracy'])\n",
    "\n",
    "model2.fit(X_train, Y_train, epochs = 4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 0s 42us/step\n",
      "0.9699\n"
     ]
    }
   ],
   "source": [
    "test_loss, test_accuracy = model2.evaluate(X_test, Y_test)\n",
    "print(test_accuracy)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "No overfitting detected. The test error is more than training error but that is expected. It's a harmful difference."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 7200x5040 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot = plt.figure(figsize = (100,70))\n",
    "plot.add_subplot(30,30,1)\n",
    "plt.imshow(X_test[5], cmap = plt.cm.binary)\n",
    "plot.add_subplot(30,30,2)\n",
    "plt.imshow(X_test[6], cmap = plt.cm.binary)\n",
    "plot.add_subplot(30,30,3)\n",
    "plt.imshow(X_test[7], cmap = plt.cm.binary)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "num = model2.predict(X_test[5:8])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[8.9046949e-08 9.9989522e-01 2.6051091e-06 3.4436830e-08 2.9415410e-06\n",
      "  7.8818796e-10 1.5178776e-07 9.5183175e-05 3.5320684e-06 1.6700412e-07]\n",
      " [4.4840309e-10 8.8778997e-06 1.8013806e-08 1.8856183e-07 9.9959868e-01\n",
      "  4.6425615e-07 1.5149750e-07 6.9540863e-07 1.2955310e-04 2.6136127e-04]\n",
      " [4.0571027e-08 3.1746233e-06 1.3804527e-07 6.2747611e-05 7.8816738e-06\n",
      "  1.5440586e-04 6.8416885e-09 6.2489624e-07 1.0442433e-06 9.9976987e-01]]\n"
     ]
    }
   ],
   "source": [
    "print(num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "print(np.argmax(num))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1, 4, 9]\n"
     ]
    }
   ],
   "source": [
    "l = [np.argmax(i) for i in num]\n",
    "print(l)"
   ]
  },
  {
   "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}