Day 30 Getting Started with TensorFlow.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### What is Tensor?\n",
    "A tensor is a physical entity that can be characterized by magnitude and multiple directions. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "# importing the required libraries\n",
    "\n",
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "# initializing the constants\n",
    "x1 = tf.constant([1, 2, 3, 4])\n",
    "x2 = tf.constant([5, 6, 7, 8])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tf.Tensor([ 5 12 21 32], shape=(4,), dtype=int32)\n"
     ]
    }
   ],
   "source": [
    "#multiply\n",
    "result = tf.multiply(x1, x2)\n",
    "\n",
    "print (result)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## A little bit complex example"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Loading th MNIST dataset, which is already present in the keras libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[[0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  ...\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]]\n",
      "\n",
      " [[0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  ...\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]]\n",
      "\n",
      " [[0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  ...\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]]\n",
      "\n",
      " ...\n",
      "\n",
      " [[0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  ...\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]]\n",
      "\n",
      " [[0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  ...\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]]\n",
      "\n",
      " [[0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  ...\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]]] [[[0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  ...\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]]\n",
      "\n",
      " [[0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  ...\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]]\n",
      "\n",
      " [[0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  ...\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]]\n",
      "\n",
      " ...\n",
      "\n",
      " [[0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  ...\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]]\n",
      "\n",
      " [[0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  ...\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]]\n",
      "\n",
      " [[0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  ...\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]\n",
      "  [0. 0. 0. ... 0. 0. 0.]]]\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.keras.datasets import fashion_mnist\n",
    "\n",
    "mnist = fashion_mnist\n",
    "\n",
    "(x_train, y_train), (x_test, y_test) = mnist.load_data()\n",
    "\n",
    "#converting the samples from integers to floating-point numbers\n",
    "\n",
    "x_train, x_test = x_train/255.0, x_test/255.0\n",
    "\n",
    "print (x_train, x_test)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Building the sequential model by stacking layers by choosing optimizer and loss function for training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = tf.keras.models.Sequential(\n",
    "    [tf.keras.layers.Flatten(input_shape=(28, 28)), \n",
    "    tf.keras.layers.Dense(128, activation = 'relu'),\n",
    "    tf.keras.layers.Dropout(0.2),\n",
    "    tf.keras.layers.Dense(10)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.6215282 ,  0.05453612,  0.7274628 , -0.5026709 ,  0.46836096,\n",
       "        -1.2439388 ,  0.9960353 , -0.26266104, -0.00821025, -0.06901443]],\n",
       "      dtype=float32)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predictions = model(x_train[:1]).numpy()\n",
    "predictions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Converting above logits to probabilities for each class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.1445492 , 0.08199251, 0.16070247, 0.04696586, 0.12402123,\n",
       "        0.02237967, 0.21021421, 0.05970589, 0.07700586, 0.0724631 ]],\n",
       "      dtype=float32)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tf.nn.softmax(predictions).numpy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Loss Function using cross entropy calculation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "loss_fn = tf.keras.losses.SparseCategoricalCrossentropy(from_logits = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.624678"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loss_fn(y_train[:1], predictions).numpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(optimizer = 'adam', loss = loss_fn, metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "1875/1875 [==============================] - 5s 2ms/step - loss: 0.5357 - accuracy: 0.8126\n",
      "Epoch 2/5\n",
      "1875/1875 [==============================] - 4s 2ms/step - loss: 0.4026 - accuracy: 0.8549\n",
      "Epoch 3/5\n",
      "1875/1875 [==============================] - 4s 2ms/step - loss: 0.3652 - accuracy: 0.8675\n",
      "Epoch 4/5\n",
      "1875/1875 [==============================] - 4s 2ms/step - loss: 0.3460 - accuracy: 0.8734\n",
      "Epoch 5/5\n",
      "1875/1875 [==============================] - 4s 2ms/step - loss: 0.3285 - accuracy: 0.8796\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7f03a46b65b0>"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(x_train, y_train, epochs = 5)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Testing the performance of the model on the test-set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 - 0s - loss: 0.3510 - accuracy: 0.8724\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.35100677609443665, 0.8723999857902527]"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(x_test, y_test, verbose =2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Returning the probability of the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "probability_model = tf.keras.Sequential([model, tf.keras.layers.Softmax()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tf.Tensor: shape=(5, 10), dtype=float32, numpy=\n",
       "array([[1.70984913e-05, 1.35960107e-07, 6.81261440e-07, 1.00518139e-06,\n",
       "        2.12835266e-06, 4.46370617e-02, 6.84958195e-06, 4.78531271e-02,\n",
       "        1.04025014e-04, 9.07377779e-01],\n",
       "       [7.20335636e-04, 3.33749561e-08, 9.77513254e-01, 3.36423199e-07,\n",
       "        8.30146018e-03, 6.98189946e-15, 1.34626022e-02, 4.31440594e-15,\n",
       "        2.10654071e-06, 1.42399032e-13],\n",
       "       [1.06809432e-06, 9.99998331e-01, 8.18383938e-08, 4.05405956e-07,\n",
       "        9.40936005e-08, 1.60161578e-10, 1.29451827e-08, 3.55355971e-16,\n",
       "        1.02453171e-10, 2.83876773e-14],\n",
       "       [3.16607071e-07, 9.99985218e-01, 1.09788594e-07, 1.42809204e-05,\n",
       "        4.26788596e-08, 4.87422269e-10, 1.07925118e-08, 2.16605317e-17,\n",
       "        4.31165346e-11, 3.85055757e-14],\n",
       "       [1.54354021e-01, 1.39318974e-04, 5.29151149e-02, 1.07390189e-03,\n",
       "        6.82443706e-03, 6.07825714e-05, 7.81798720e-01, 4.04262573e-05,\n",
       "        2.77643278e-03, 1.67646176e-05]], dtype=float32)>"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "probability_model(x_test[:5])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# A Simple Model of ANN in Tensor Flow"
   ]
  },
  {
   "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}