UG_ASSIGN_4.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "PwHfjU1teRAq"
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn import *\n",
    "from sklearn.preprocessing import *\n",
    "from sklearn.metrics import *\n",
    "from sklearn.datasets import *\n",
    "import pandas as pd\n",
    "import cv2 #To test on custom images # To import images you can use Image.tk too "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "5PFACsc9exVW"
   },
   "source": [
    "# Downloading the Dataset \n",
    "## Normalising & Pre-Procesing\n",
    "### Declaration of Sigmoid, Loss , Forward Prop, Backward Prop Func\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "KXlvQckch7bB",
    "outputId": "9a51447c-6f55-46be-c053-52b5df7dfc7f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0. 0. 0. ... 0. 0. 0.]\n"
     ]
    }
   ],
   "source": [
    "X, y = fetch_openml('mnist_784', version=1, return_X_y=True)\n",
    "print(X[:,10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "zsHKRGUx99ml"
   },
   "outputs": [],
   "source": [
    "# Image Representation\n",
    "X_res = X[1].reshape(28,28)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 269
    },
    "colab_type": "code",
    "id": "tKp9ZJqn9-Rm",
    "outputId": "d3738c7c-7727-474d-8d28-cab933871663"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRfiBoAg/EBThB4Ii/EBQhB8I6v8AG8x2aarNGp8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(X_res)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "rAq-rpEyi9YI",
    "outputId": "3c52fecd-be6a-4ffa-81c1-87222d890e51"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['5' '0' '4' ... '4' '5' '6']\n"
     ]
    }
   ],
   "source": [
    "X = X/255\n",
    "print(y)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "qs9O0x2Mlst_",
    "outputId": "8497dda4-55b2-4c4a-8071-4cb760a301d0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "70000\n"
     ]
    }
   ],
   "source": [
    "# one-hot encode labels\n",
    "digits = 10\n",
    "print(y.shape[0])                #No. of dataset entries\n",
    "examples = y.shape[0]           \n",
    "y = y.reshape(1, examples)\n",
    "Y_new = np.eye(digits)[y.astype('int32')]\n",
    "Y_new = Y_new.T.reshape(digits, examples)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "VIBG-kYylt32",
    "outputId": "6e928594-3c0d-47d3-d675-f47f800ce886"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(784, 60000)\n"
     ]
    }
   ],
   "source": [
    "# split, reshape, shuffle\n",
    "m = 60000\n",
    "m_test = X.shape[0] - m\n",
    "X_train, X_test = X[:m].T, X[m:].T\n",
    "Y_train, Y_test = Y_new[:,:m], Y_new[:,m:]\n",
    "shuffle_index = np.random.permutation(m)\n",
    "X_train, Y_train = X_train[:, shuffle_index], Y_train[:, shuffle_index]\n",
    "print(X_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "vjlYYfoilvuI"
   },
   "outputs": [],
   "source": [
    "# Sigmoid Function\n",
    "def sigmoid(z):\n",
    "    s = 1. / (1. + np.exp(-z))\n",
    "    return s"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "t5jT98HpLIkM"
   },
   "outputs": [],
   "source": [
    "def softmax(z):\n",
    "    s = e ^ z / sum(e ^ z)\n",
    "    return s"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "b9zmUNtYnXkK"
   },
   "outputs": [],
   "source": [
    "# Compute Loss\n",
    "def compute_loss(Y, Y_hat):\n",
    "    L_sum = np.sum(np.multiply(Y, np.log(Y_hat)))\n",
    "    m = Y.shape[1]\n",
    "    L = -(1./m) * L_sum\n",
    "    return L"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Gw0symkYnb7k"
   },
   "outputs": [],
   "source": [
    "# Feed forward Propagation\n",
    "def feed_forward(X, params):\n",
    "\n",
    "    cache = {}\n",
    "\n",
    "    cache[\"Z1\"] = np.matmul(params[\"W1\"], X) + params[\"b1\"]\n",
    "    cache[\"A1\"] = sigmoid(cache[\"Z1\"])\n",
    "    cache[\"Z2\"] = np.matmul(params[\"W2\"], cache[\"A1\"]) + params[\"b2\"]\n",
    "    cache[\"A2\"] = np.exp(cache[\"Z2\"]) / np.sum(np.exp(cache[\"Z2\"]), axis=0)\n",
    "    cache[\"Z3\"] = np.matmul(params[\"W3\"], cache[\"A2\"]) + params[\"b3\"]\n",
    "    cache[\"A3\"] = np.exp(cache[\"Z3\"]) / np.sum(np.exp(cache[\"Z3\"]), axis=0)\n",
    "    \n",
    "\n",
    "    return cache"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "YHa1MNhqnmcW"
   },
   "outputs": [],
   "source": [
    "# Back Propagation\n",
    "def back_propagate(X, Y, params, cache):\n",
    "\n",
    "    dZ3 = cache[\"A3\"] - Y\n",
    "    dW3 = (1./m_batch) * np.matmul(dZ3, cache[\"A2\"].T)\n",
    "    db3 = (1./m_batch) * np.sum(dZ3, axis=1, keepdims=True)\n",
    "    \n",
    "    \n",
    "    dA2 = np.matmul(params[\"W3\"].T, dZ3)\n",
    "    dZ2 = dA2 * sigmoid(cache[\"Z2\"]) * (1 - sigmoid(cache[\"Z2\"]))\n",
    "    dW2 = (1./m_batch) * np.matmul(dZ2, cache[\"A1\"].T)\n",
    "    db2 = (1./m_batch) * np.sum(dZ2, axis=1, keepdims=True)\n",
    "\n",
    "    \n",
    "    dA1 = np.matmul(params[\"W2\"].T, dZ2)\n",
    "    dZ1 = dA1 * sigmoid(cache[\"Z1\"]) * (1 - sigmoid(cache[\"Z1\"]))\n",
    "    dW1 = (1./m_batch) * np.matmul(dZ1, X.T)\n",
    "    db1 = (1./m_batch) * np.sum(dZ1, axis=1, keepdims=True)\n",
    "\n",
    "    grads = {\"dW1\": dW1, \"db1\": db1, \"dW2\": dW2, \"db2\": db2,\"dW3\": dW3, \"db3\": db3}\n",
    "\n",
    "    return grads\n",
    "  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "fIhSue0Yzc4w"
   },
   "source": [
    "# Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "bqFRjtPcnsLN"
   },
   "outputs": [],
   "source": [
    "### Initialisation of Parameter and other hyperparameter declaration #####\n",
    "np.random.seed(138)\n",
    "\n",
    "# hyperparameters\n",
    "n_x = X_train.shape[0]\n",
    "n_h1 = 64\n",
    "n_h2 = 32\n",
    "learning_rate = 0.08\n",
    "beta = .9\n",
    "batch_size = 128\n",
    "batches = -(-m // batch_size)\n",
    "\n",
    "# initialization\n",
    "params = { \"W1\": np.random.randn(n_h1, n_x) * np.sqrt(1. / n_x),\n",
    "           \"b1\": np.zeros((n_h1, 1)) * np.sqrt(1. / n_x),\n",
    "           \"W2\": np.random.randn(n_h2, n_h1) * np.sqrt(1. / n_h1),\n",
    "           \"b2\": np.zeros((n_h2, 1)) * np.sqrt(1. / n_h1),\n",
    "           \"W3\": np.random.randn(digits, n_h2) * np.sqrt(1. / n_h2),\n",
    "           \"b3\": np.zeros((digits, 1)) * np.sqrt(1. / n_h2) }\n",
    "\n",
    "V_dW1 = np.zeros(params[\"W1\"].shape)\n",
    "V_db1 = np.zeros(params[\"b1\"].shape)\n",
    "V_dW2 = np.zeros(params[\"W2\"].shape)\n",
    "V_db2 = np.zeros(params[\"b2\"].shape)\n",
    "V_dW3 = np.zeros(params[\"W3\"].shape)\n",
    "V_db3 = np.zeros(params[\"b3\"].shape)\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 2584
    },
    "colab_type": "code",
    "id": "v2eGOFdPRDHR",
    "outputId": "7603267a-e08d-4295-c47b-4b5760fad3ff"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "E 1: Training cost = 2.26977115583027, Test cost = 2.26880485137033\n",
      "E 2: Training cost = 2.1452407909609197, Test cost = 2.141133101068511\n",
      "E 3: Training cost = 1.9788950116027493, Test cost = 1.9724470942251817\n",
      "E 4: Training cost = 1.8084613438130683, Test cost = 1.8013085982386956\n",
      "E 5: Training cost = 1.6456338742978507, Test cost = 1.6411806967548537\n",
      "E 6: Training cost = 1.496076742942292, Test cost = 1.4941974577921204\n",
      "E 7: Training cost = 1.348713726193463, Test cost = 1.347649850407181\n",
      "E 8: Training cost = 1.213360603685921, Test cost = 1.211834235577573\n",
      "E 9: Training cost = 1.1172340884701328, Test cost = 1.1141832364643869\n",
      "E 10: Training cost = 1.0609396604720729, Test cost = 1.0540312059024288\n",
      "E 11: Training cost = 1.000056593621028, Test cost = 0.9925320319934123\n",
      "E 12: Training cost = 0.9374500425002059, Test cost = 0.9309936341374334\n",
      "E 13: Training cost = 0.9003214120050205, Test cost = 0.8934737580521162\n",
      "E 14: Training cost = 0.8674798871024412, Test cost = 0.8603919526546905\n",
      "E 15: Training cost = 0.8420646804392669, Test cost = 0.835231779343754\n",
      "E 16: Training cost = 0.8200922334773333, Test cost = 0.8135493918079949\n",
      "E 17: Training cost = 0.8021113934659118, Test cost = 0.7964950400478809\n",
      "E 18: Training cost = 0.7932065618805387, Test cost = 0.7891687453857457\n",
      "E 19: Training cost = 0.7849090778311409, Test cost = 0.7807703113084409\n",
      "E 20: Training cost = 0.7818099961552248, Test cost = 0.7787100247774078\n",
      "E 21: Training cost = 0.7631049998011812, Test cost = 0.7599109868858633\n",
      "E 22: Training cost = 0.7549025020224658, Test cost = 0.7523905712637116\n",
      "E 23: Training cost = 0.7430247752764438, Test cost = 0.7424480187062579\n",
      "E 24: Training cost = 0.7325464432581008, Test cost = 0.7297552310237729\n",
      "E 25: Training cost = 0.7293453890420103, Test cost = 0.7269548188266471\n",
      "E 26: Training cost = 0.7174285529185777, Test cost = 0.7161078183650353\n",
      "E 27: Training cost = 0.7077977360828394, Test cost = 0.7069841159118527\n",
      "E 28: Training cost = 0.7020339135800553, Test cost = 0.7014804810751687\n",
      "E 29: Training cost = 0.6920661284492989, Test cost = 0.6905903248325119\n",
      "E 30: Training cost = 0.6818070301184794, Test cost = 0.6793802488420199\n",
      "E 31: Training cost = 0.6749093654369616, Test cost = 0.6734934077469255\n",
      "E 32: Training cost = 0.6669204233222987, Test cost = 0.6659167300656217\n",
      "E 33: Training cost = 0.6597675638307737, Test cost = 0.6562186201859797\n",
      "E 34: Training cost = 0.6510955765379958, Test cost = 0.6476430224944537\n",
      "E 35: Training cost = 0.6528871996394741, Test cost = 0.6490603686398054\n",
      "E 36: Training cost = 0.6411116915184313, Test cost = 0.6359618309401543\n",
      "E 37: Training cost = 0.638234489974643, Test cost = 0.6334420341872019\n",
      "E 38: Training cost = 0.628426295993326, Test cost = 0.6236492509971728\n",
      "E 39: Training cost = 0.6277666147800729, Test cost = 0.6222899281023239\n",
      "E 40: Training cost = 0.6144018964290103, Test cost = 0.6087963370343522\n",
      "E 41: Training cost = 0.6186887855294902, Test cost = 0.612516441094273\n",
      "E 42: Training cost = 0.6070689992416488, Test cost = 0.6004396731304157\n",
      "E 43: Training cost = 0.6026455878646744, Test cost = 0.5968558378836383\n",
      "E 44: Training cost = 0.6074561457840495, Test cost = 0.6023431776167185\n",
      "E 45: Training cost = 0.5952160338394167, Test cost = 0.5900992730888364\n",
      "E 46: Training cost = 0.5805518107386921, Test cost = 0.5760318273241983\n",
      "E 47: Training cost = 0.5750199047854189, Test cost = 0.5707822648780106\n",
      "E 48: Training cost = 0.5687496198222631, Test cost = 0.5632358360629235\n",
      "E 49: Training cost = 0.5579431119803057, Test cost = 0.5530852990893087\n",
      "E 50: Training cost = 0.5589864273969509, Test cost = 0.5542014112392583\n",
      "E 51: Training cost = 0.5498939616247732, Test cost = 0.5456862435468105\n",
      "E 52: Training cost = 0.5409219183829161, Test cost = 0.5352013885342333\n",
      "E 53: Training cost = 0.5371701794020058, Test cost = 0.5312705429736818\n",
      "E 54: Training cost = 0.5329115366904241, Test cost = 0.5284029547853425\n",
      "E 55: Training cost = 0.5270183449144806, Test cost = 0.5211470517463193\n",
      "E 56: Training cost = 0.5183697881738425, Test cost = 0.51343889221808\n",
      "E 57: Training cost = 0.5124603143126514, Test cost = 0.506217603092398\n",
      "E 58: Training cost = 0.5111998569211351, Test cost = 0.5056442663024996\n",
      "E 59: Training cost = 0.5001649030460156, Test cost = 0.4940621672536059\n",
      "E 60: Training cost = 0.49497836099887627, Test cost = 0.4895153094255825\n",
      "E 61: Training cost = 0.4888820644003066, Test cost = 0.4827821399016359\n",
      "E 62: Training cost = 0.48298613196727047, Test cost = 0.47705668947149066\n",
      "E 63: Training cost = 0.4794263915834811, Test cost = 0.4722353629447919\n",
      "E 64: Training cost = 0.47576396920498576, Test cost = 0.4695799446845944\n",
      "E 65: Training cost = 0.46671873895280497, Test cost = 0.46094317275042596\n",
      "E 66: Training cost = 0.4646845581494732, Test cost = 0.45899457645674474\n",
      "E 67: Training cost = 0.460884937725844, Test cost = 0.4540655570933222\n",
      "E 68: Training cost = 0.46163832567169333, Test cost = 0.45603875215866585\n",
      "E 69: Training cost = 0.4643897137866579, Test cost = 0.45867043480426284\n",
      "E 70: Training cost = 0.4502486142965501, Test cost = 0.44384038380228585\n",
      "E 71: Training cost = 0.44930720350698033, Test cost = 0.4439832474620469\n",
      "E 72: Training cost = 0.4432704050472225, Test cost = 0.43682845576678003\n",
      "E 73: Training cost = 0.44280452110391416, Test cost = 0.4376530483670911\n",
      "E 74: Training cost = 0.4407252128491301, Test cost = 0.43436161398325807\n",
      "E 75: Training cost = 0.4381360168695162, Test cost = 0.43182935227706276\n",
      "E 76: Training cost = 0.42957810246414174, Test cost = 0.42340767749585634\n",
      "E 77: Training cost = 0.4294649675753888, Test cost = 0.4255316100502494\n",
      "E 78: Training cost = 0.43267887947423145, Test cost = 0.42700087563666095\n",
      "E 79: Training cost = 0.4311835287424502, Test cost = 0.4257886933347512\n",
      "E 80: Training cost = 0.426137430593559, Test cost = 0.4223366072292599\n",
      "E 81: Training cost = 0.4299692097583262, Test cost = 0.4255356148164044\n",
      "E 82: Training cost = 0.4244944395959516, Test cost = 0.42015098995302264\n",
      "E 83: Training cost = 0.42257831647923155, Test cost = 0.4178334120376283\n",
      "E 84: Training cost = 0.42206928767993585, Test cost = 0.4176457436461462\n",
      "E 85: Training cost = 0.4191281944358771, Test cost = 0.4154541463184228\n",
      "E 86: Training cost = 0.4287840055740964, Test cost = 0.4256675909494744\n",
      "E 87: Training cost = 0.41790635958846384, Test cost = 0.4146814804370999\n",
      "E 88: Training cost = 0.425547156456487, Test cost = 0.4220548412327907\n",
      "E 89: Training cost = 0.4205004954884442, Test cost = 0.41737315545354314\n",
      "E 90: Training cost = 0.4201821874194649, Test cost = 0.41628226098906873\n",
      "E 91: Training cost = 0.4124593652435847, Test cost = 0.40860561708498877\n",
      "E 92: Training cost = 0.4096105923475256, Test cost = 0.40640404209076014\n",
      "E 93: Training cost = 0.41564541414495787, Test cost = 0.41398317841235577\n",
      "E 94: Training cost = 0.4202655237989926, Test cost = 0.4186757205946434\n",
      "E 95: Training cost = 0.4166302394538928, Test cost = 0.41495030910129294\n",
      "E 96: Training cost = 0.40761226176793813, Test cost = 0.4068184588585663\n",
      "E 97: Training cost = 0.41227014918401766, Test cost = 0.4098236491147691\n",
      "E 98: Training cost = 0.4003856821573546, Test cost = 0.4001666991966991\n",
      "E 99: Training cost = 0.40862384706972876, Test cost = 0.40789842748452054\n",
      "E 100: Training cost = 0.4088104999514338, Test cost = 0.40966721218717367\n",
      "E 101: Training cost = 0.4102461945950398, Test cost = 0.4082170495869654\n",
      "E 102: Training cost = 0.40495970692228317, Test cost = 0.4069657077117891\n",
      "E 103: Training cost = 0.4063260754819307, Test cost = 0.4083235627210793\n",
      "E 104: Training cost = 0.4000780248756612, Test cost = 0.4011033604900442\n",
      "E 105: Training cost = 0.4023056821676515, Test cost = 0.4033619256018141\n",
      "E 106: Training cost = 0.3977666332201565, Test cost = 0.40098421960532965\n",
      "E 107: Training cost = 0.3993411386776496, Test cost = 0.40360494879532194\n",
      "E 108: Training cost = 0.40271836770294606, Test cost = 0.4078098831362123\n",
      "E 109: Training cost = 0.3986257705862609, Test cost = 0.40248220277946567\n",
      "E 110: Training cost = 0.4006442089310286, Test cost = 0.4052022427932187\n",
      "E 111: Training cost = 0.39762794405400337, Test cost = 0.4016322090738943\n",
      "E 112: Training cost = 0.4061033153633453, Test cost = 0.4118815021983241\n",
      "E 113: Training cost = 0.3975952626427913, Test cost = 0.3992554429894106\n",
      "E 114: Training cost = 0.39100392693720315, Test cost = 0.39632165905202177\n",
      "E 115: Training cost = 0.39207313463286536, Test cost = 0.39735592100089184\n",
      "E 116: Training cost = 0.3864123759925327, Test cost = 0.39106917023408444\n",
      "E 117: Training cost = 0.405009707853655, Test cost = 0.40971501408281485\n",
      "E 118: Training cost = 0.3926011748322743, Test cost = 0.3999330164484476\n",
      "E 119: Training cost = 0.40034733501833325, Test cost = 0.4069505843588239\n",
      "E 120: Training cost = 0.3901307672522966, Test cost = 0.39584184409478235\n",
      "E 121: Training cost = 0.38630289171400545, Test cost = 0.394458905169076\n",
      "E 122: Training cost = 0.3854480406087247, Test cost = 0.3925018324212312\n",
      "E 123: Training cost = 0.3879065452579934, Test cost = 0.39681723899776367\n",
      "E 124: Training cost = 0.39653192329532944, Test cost = 0.4050146797170575\n",
      "E 125: Training cost = 0.3866413810631643, Test cost = 0.3951985596627589\n",
      "E 126: Training cost = 0.3982359609764343, Test cost = 0.40684940147103166\n",
      "E 127: Training cost = 0.38908128980386664, Test cost = 0.39893316166194454\n",
      "E 128: Training cost = 0.39304304751480784, Test cost = 0.40273052034916035\n",
      "E 129: Training cost = 0.39990546366373425, Test cost = 0.40963526650046544\n",
      "E 130: Training cost = 0.393565299294676, Test cost = 0.4035265454870426\n",
      "E 131: Training cost = 0.3987342430767503, Test cost = 0.41027837645550064\n",
      "E 132: Training cost = 0.3966574451992998, Test cost = 0.4072631600073976\n",
      "E 133: Training cost = 0.396203176895159, Test cost = 0.4061373618179153\n",
      "E 134: Training cost = 0.3923612546075803, Test cost = 0.4036990594156648\n",
      "E 135: Training cost = 0.3938399453014341, Test cost = 0.4065141150685239\n",
      "E 136: Training cost = 0.3907852492706465, Test cost = 0.40192821334956846\n",
      "E 137: Training cost = 0.4039657905767947, Test cost = 0.4166389110587203\n",
      "E 138: Training cost = 0.397461485739174, Test cost = 0.41081614087034896\n",
      "E 139: Training cost = 0.4056238209146284, Test cost = 0.4173606661075049\n",
      "E 140: Training cost = 0.39376639208791625, Test cost = 0.4048082753694892\n",
      "E 141: Training cost = 0.40153190068095124, Test cost = 0.4154378890600564\n",
      "E 142: Training cost = 0.39632990844465227, Test cost = 0.40887997646505386\n",
      "E 143: Training cost = 0.39675269466124224, Test cost = 0.4098146095137322\n",
      "E 144: Training cost = 0.39800325555304056, Test cost = 0.41003748136571494\n",
      "E 145: Training cost = 0.4037126682064477, Test cost = 0.41544246253987316\n",
      "E 146: Training cost = 0.3927241921118802, Test cost = 0.40531502018860416\n",
      "E 147: Training cost = 0.412585706466679, Test cost = 0.4265884505346432\n",
      "E 148: Training cost = 0.39687562411910793, Test cost = 0.4085986139840907\n",
      "E 149: Training cost = 0.3997382739297689, Test cost = 0.416620246173481\n",
      "E 150: Training cost = 0.3995542617848686, Test cost = 0.41443712075505956\n",
      "Done.\n"
     ]
    }
   ],
   "source": [
    "# train\n",
    "train = []\n",
    "test = []\n",
    "for i in range(150):\n",
    "\n",
    "    permutation = np.random.permutation(X_train.shape[1])\n",
    "    X_train_shuffled = X_train[:, permutation]\n",
    "    Y_train_shuffled = Y_train[:, permutation]\n",
    "\n",
    "    for j in range(batches):\n",
    "\n",
    "        begin = j * batch_size\n",
    "        end = min(begin + batch_size, X_train.shape[1] - 1)\n",
    "        X = X_train_shuffled[:, begin:end]\n",
    "        Y = Y_train_shuffled[:, begin:end]\n",
    "        m_batch = end - begin\n",
    "\n",
    "        cache = feed_forward(X, params)\n",
    "        grads = back_propagate(X, Y, params, cache)\n",
    "\n",
    "        V_dW1 = (beta * V_dW1 + (1. - beta) * grads[\"dW1\"])\n",
    "        V_db1 = (beta * V_db1 + (1. - beta) * grads[\"db1\"])\n",
    "        V_dW2 = (beta * V_dW2 + (1. - beta) * grads[\"dW2\"])\n",
    "        V_db2 = (beta * V_db2 + (1. - beta) * grads[\"db2\"])\n",
    "        V_dW3 = (beta * V_dW3 + (1. - beta) * grads[\"dW3\"])\n",
    "        V_db3 = (beta * V_db3 + (1. - beta) * grads[\"db3\"])\n",
    "\n",
    "        params[\"W1\"] = params[\"W1\"] - learning_rate * V_dW1\n",
    "        params[\"b1\"] = params[\"b1\"] - learning_rate * V_db1\n",
    "        params[\"W2\"] = params[\"W2\"] - learning_rate * V_dW2\n",
    "        params[\"b2\"] = params[\"b2\"] - learning_rate * V_db2\n",
    "        params[\"W3\"] = params[\"W3\"] - learning_rate * V_dW3\n",
    "        params[\"b3\"] = params[\"b3\"] - learning_rate * V_db3\n",
    "\n",
    "    cache = feed_forward(X_train, params)\n",
    "    train_cost = compute_loss(Y_train, cache[\"A3\"])\n",
    "    cache = feed_forward(X_test, params)\n",
    "    test_cost = compute_loss(Y_test, cache[\"A3\"])\n",
    "    train.append(train_cost)\n",
    "    test.append(test_cost)\n",
    "    print(\"E {}: Training cost = {}, Test cost = {}\".format(i+1 ,train_cost, test_cost))\n",
    "\n",
    "print(\"Done.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "OBD0GHEtoT4J"
   },
   "source": [
    "# Results\n",
    "## Visualisation\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 306
    },
    "colab_type": "code",
    "id": "x6sudtdLnwRL",
    "outputId": "074e02bf-340a-4273-e33c-de6c6e85f9d9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.98      0.87      0.92      1111\n",
      "           1       0.97      0.97      0.97      1139\n",
      "           2       0.87      0.91      0.89       987\n",
      "           3       0.73      0.97      0.83       755\n",
      "           4       0.93      0.83      0.88      1091\n",
      "           5       0.85      0.69      0.76      1099\n",
      "           6       0.91      0.91      0.91       954\n",
      "           7       0.88      0.91      0.89       997\n",
      "           8       0.88      0.88      0.88       973\n",
      "           9       0.81      0.92      0.86       894\n",
      "\n",
      "    accuracy                           0.88     10000\n",
      "   macro avg       0.88      0.88      0.88     10000\n",
      "weighted avg       0.89      0.88      0.88     10000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Results\n",
    "cache = feed_forward(X_test, params)\n",
    "predictions = np.argmax(cache[\"A3\"], axis=0)\n",
    "labels = np.argmax(Y_test, axis=0)\n",
    "print(classification_report(predictions, labels))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 283
    },
    "colab_type": "code",
    "id": "vN-BJuBtoWqb",
    "outputId": "8b5df681-50d0-4f2c-cb7c-6884f4217fa2"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEKCAYAAAAB0GKPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQ0REoqLAEBGRqPx/Gf+NYZNaX8gAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plotting Train cost\n",
    "plt.plot(train)\n",
    "plt.ylabel(\"Train_Loss\")\n",
    "plt.xlabel(\"Iterations\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 283
    },
    "colab_type": "code",
    "id": "C2XJP9Gn6JSd",
    "outputId": "1776ea19-2ad5-40d1-d9e1-b90bb7c8de0e"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEKCAYAAAAB0GKPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhoiIREWBISIiUVFgiIhIVP4/PFh5/GloB+UAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plotting Test Cost\n",
    "plt.plot(test)\n",
    "plt.ylabel(\"Train_Loss\")\n",
    "plt.xlabel(\"Iterations\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "E5Dp4Uw36RVY"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "38T_M2PL6Yfs"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "MNIST_Assign.ipynb",
   "provenance": [],
   "version": "0.3.2"
  },
  "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}