Deep Neural Networks - Fashion MNIST.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "c1ea675c",
   "metadata": {},
   "source": [
    "# Deep Neural Networks\n",
    "They have neurons in the input layer equal to the no of different inputs needed by the NN and output layer has it generally 1.\n",
    "\n",
    "$y = \\vec{w}.\\vec{x}+b$\n",
    "\n",
    "$y = \\sum_{i=0}^n{w_i.x_i}+b$\n",
    "\n",
    "Activation func can be anything we want but we have some standard ones like ReLU, tanh, sigmoid.\n",
    "\n",
    "Loss/Cost functions: MAE, MSE, RMSE, Hinge loss.\n",
    "\n",
    "Gradient descent - used to calc optimum weights and biases for the network to move in direction to minimise the losses.\n",
    "\n",
    "Back Propagation - used to calc the gradient used in the GD step.\n",
    "\n",
    "Optimiser - func implimenting the back-propagation algo. Eg GD, mini-batch GD, stochastic GD, Adam's optimiser, momentum, nesterov accelerated gradient."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8e7e7815",
   "metadata": {},
   "source": [
    "## Analysing and processing the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "b94e68e8",
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow.keras import layers\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "b6081a5b",
   "metadata": {},
   "outputs": [],
   "source": [
    "fashion = tf.keras.datasets.fashion_mnist\n",
    "(train, y_train), (test, y_test) = fashion.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "4a5756c6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[[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]]], dtype=uint8)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "4cadccc8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(60000, 28, 28)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#60000 images of 28x28 pixels\n",
    "train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "762750bd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10000, 28, 28)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "cdc2fdfe",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "188"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Pixel value of 1st img at 23,22 posn\n",
    "train[0,23,22]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "9917d69f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([9, 0, 0, 3, 0, 2, 7, 2, 5, 5], dtype=uint8)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#no betn 0 and 9 to indicate 10 diff classes\n",
    "class_name = ['Tshirt or top', 'Trouser', 'Pull-over', 'Dress', 'Coat', 'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']\n",
    "y_train[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "ebbacaef",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Visualise the image\n",
    "plt.imshow(test[250])\n",
    "plt.colorbar()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "696ad271",
   "metadata": {},
   "source": [
    "#### To make it easy for the NN to process the pixel values, we squeeze them to 0 to 1 by dividing by 255."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "34725205",
   "metadata": {},
   "outputs": [],
   "source": [
    "train = train/255\n",
    "test = test/255"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "484ef9f2",
   "metadata": {},
   "source": [
    "## Building the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "8c40160c",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = tf.keras.Sequential([\n",
    "    layers.Flatten(input_shape = (28,28)),            #Flattens the 28*28 matrix into a single array of len = 784 to ease computation\n",
    "    layers.Dense(units = 128, activation = 'relu'),   #hidden layer\n",
    "    layers.Dense(units = 10, activation = 'softmax')  # as 10 outputs are possible\n",
    "])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "382f9861",
   "metadata": {},
   "source": [
    "## Compling the model\n",
    "This is hyperparameter tuning where we observe model's performance on changing the optimizers, loss functions, epochs, etc."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "0a3aa35a",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(\n",
    "    optimizer = 'adam',\n",
    "    loss = 'sparse_categorical_crossentropy',\n",
    "    metrics = ['accuracy']\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5f5a9aa1",
   "metadata": {},
   "source": [
    "## Training the model\n",
    "Accuracy given here is the training accuracy and is generally high as the model sees the same data many times by epochs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "cbdd7b59",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.4990 - accuracy: 0.8255\n",
      "Epoch 2/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.3743 - accuracy: 0.8662\n",
      "Epoch 3/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.3360 - accuracy: 0.8784\n",
      "Epoch 4/10\n",
      "1875/1875 [==============================] - 7s 3ms/step - loss: 0.3115 - accuracy: 0.8852\n",
      "Epoch 5/10\n",
      "1875/1875 [==============================] - 6s 3ms/step - loss: 0.2936 - accuracy: 0.8913\n",
      "Epoch 6/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.2789 - accuracy: 0.8964\n",
      "Epoch 7/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.2671 - accuracy: 0.9004\n",
      "Epoch 8/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.2557 - accuracy: 0.9056\n",
      "Epoch 9/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.2476 - accuracy: 0.9083\n",
      "Epoch 10/10\n",
      "1875/1875 [==============================] - 7s 4ms/step - loss: 0.2374 - accuracy: 0.9111\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x1a337b59ca0>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(train, y_train, epochs = 10)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "86628814",
   "metadata": {},
   "source": [
    "## Testing the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "2c16fb64",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 1s 2ms/step - loss: 0.3261 - accuracy: 0.8857\n",
      "0.885699987411499\n"
     ]
    }
   ],
   "source": [
    "test_loss, test_acc = model.evaluate(test, y_test)\n",
    "print(test_acc)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1495ee7d",
   "metadata": {},
   "source": [
    "## Prediction by model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "17adc1bb",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "313/313 [==============================] - 1s 2ms/step\n"
     ]
    }
   ],
   "source": [
    "prediction = model.predict(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "df5f57ae",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted: Ankle boot\n",
      "Actual: Ankle boot\n"
     ]
    }
   ],
   "source": [
    "print(\"Predicted:\",class_name[np.argmax(prediction[250])])\n",
    "print(\"Actual:\", class_name[y_test[250]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fa9578c9",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}