Untitled.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3 layer Neural Network Architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.5, 0.5],\n",
       "       [0.5, 0.5]])"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def softmax(a):\n",
    "    e_pa = np.exp(a)\n",
    "    ans = e_pa/np.sum(e_pa,axis=1,keepdims=True)\n",
    "    return ans\n",
    "softmax([[10,10],[20,20]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "class NeuralNetwork:\n",
    "    def __init__(self,input_size,layer,output_size):\n",
    "        np.random.seed(0) # So everytime same datasets are generated\n",
    "        \n",
    "        ###DETERMINIG THE ARCHITECTURE OF THE NETWORK\n",
    "        \n",
    "        model = {}\n",
    "        \n",
    "        model['W1'] = np.random.randn(input_size,layer[0])\n",
    "        model['b1'] = np.zeros((1,layer[0]))\n",
    "        \n",
    "        model['W2'] = np.random.randn(layer[0],layer[1])\n",
    "        model['b2'] = np.zeros((1,layer[1]))\n",
    "        \n",
    "        model['W3'] = np.random.randn(layer[1],output_size)\n",
    "        model['b3'] = np.zeros((1,output_size))\n",
    "        \n",
    "        self.model = model\n",
    "        self.activation_outputs = None\n",
    "        \n",
    "        \n",
    "    def forward_propogation(self,x):\n",
    "        W1,W2,W3 = self.model['W1'],self.model['W2'],self.model['W3']\n",
    "        \n",
    "        b1,b2,b3 = self.model['b1'],self.model['b2'],self.model['b3']\n",
    "        \n",
    "        n = x.shape[0]\n",
    "        \n",
    "        z1 = np.dot(x,W1)+b1\n",
    "        a1 = np.tanh(z1)\n",
    "        \n",
    "        z2 = np.dot(a1,W2)+b2\n",
    "        a2 = np.tanh(z2)\n",
    "        \n",
    "        z3 = np.dot(a2,W3)+b3\n",
    "        y_ = softmax(z3)\n",
    "        \n",
    "        self.activation_outputs = (a1,a2,y_)\n",
    "        return y_\n",
    "        \n",
    "    def backward_propogation(self,x,y,learning_rate=0.001):\n",
    "        W1,W2,W3 = self.model['W1'],self.model['W2'],self.model['W3']\n",
    "        \n",
    "        b1,b2,b3 = self.model['b1'],self.model['b2'],self.model['b3']\n",
    "        \n",
    "        m = x.shape[0]\n",
    "        \n",
    "        a1,a2,y_ = self.activation_outputs\n",
    "        \n",
    "        \n",
    "        delta3 = (y_-y)\n",
    "        dw3 = np.dot(a2.T,delta3)\n",
    "        db3 = np.sum(delta3,axis=0)\n",
    "        \n",
    "        \n",
    "        delta2 = (1-np.square(a2))*np.dot(delta3,W3.T) ## (1-np.square(a2)) is the derivative of tanh(z)\n",
    "        dw2 = np.dot(a1.T,delta2)\n",
    "        db2 = np.sum(delta2,axis=0)\n",
    "        \n",
    "        delta1 = (1-np.square(a1))*np.dot(delta2,W2.T)\n",
    "        dw1 = np.dot(X.T,delta1)\n",
    "        db1 = np.sum(delta1,axis=0)\n",
    "        \n",
    "        #Update the Model Parameters using Gradient Descent\n",
    "        self.model[\"W1\"]  -= learning_rate*dw1\n",
    "        self.model['b1']  -= learning_rate*db1\n",
    "        \n",
    "        self.model[\"W2\"]  -= learning_rate*dw2\n",
    "        self.model['b2']  -= learning_rate*db2\n",
    "        \n",
    "        self.model[\"W3\"]  -= learning_rate*dw3\n",
    "        self.model['b3']  -= learning_rate*db3\n",
    "        \n",
    "        \n",
    "    def predict(self,x):\n",
    "        y_ = self.forward_propogation(x)\n",
    "        return np.argmax(y_,axis=1)\n",
    "    \n",
    "    def summary(self):\n",
    "        W1,W2,W3 = self.model['W1'],self.model['W2'],self.model['W3']\n",
    "        a1,a2,y_ = self.activation_outputs\n",
    "        \n",
    "        print(\"W1 \",W1.shape)\n",
    "        print(\"A1 \",a1.shape)\n",
    "        \n",
    "        print(\"W2 \",W2.shape)\n",
    "        print(\"A2 \",a2.shape)\n",
    "        \n",
    "        print(\"W3 \",W3.shape)\n",
    "        print(\"Y_ \",y_.shape)\n",
    "        \n",
    "        \n",
    "    \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "def loss(y_oht,p):\n",
    "    l = -np.mean(y_oht*np.log(p))\n",
    "    return l\n",
    "\n",
    "\n",
    "def one_hot(y,depth):\n",
    "    m = y.shape[0]\n",
    "    oht = np.zeros((m,depth))\n",
    "    oht[np.arange(m),y]=1\n",
    "    return oht"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Generating data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.datasets import make_circles\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "#make_circles(n_samples=100, shuffle=True, noise=None, random_state=None, factor=0.8)\n",
    "X,y = make_circles(n_samples=500,shuffle=True,noise=0.2,random_state=1,factor=0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x1e3d4a76c50>"
      ]
     },
     "execution_count": 62,
     "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.scatter(X[:,0],X[:,1],c=y,cmap='Accent')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training of Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = NeuralNetwork(2,[10,5],2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(X,Y,model,epochs,learning_rate,logs=True):\n",
    "    training_loss = []\n",
    "    \n",
    "    classes = 2\n",
    "    Y_OHT = one_hot(Y,classes)\n",
    "    \n",
    "    for ix in range(epochs):\n",
    "        Y_ = model.forward_propogation(X)\n",
    "        l = loss(Y_OHT,Y_)\n",
    "        training_loss.append(l)\n",
    "        model.backward_propogation(X,Y_OHT,learning_rate)\n",
    "        \n",
    "        if(logs):\n",
    "            print(\"Epoch %d Loss %.4f\"%(ix,l))\n",
    "            \n",
    "    \n",
    "    return training_loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 0 Loss 0.3572\n",
      "Epoch 1 Loss 0.3391\n",
      "Epoch 2 Loss 0.2789\n",
      "Epoch 3 Loss 0.2609\n",
      "Epoch 4 Loss 0.2455\n",
      "Epoch 5 Loss 0.2334\n",
      "Epoch 6 Loss 0.2227\n",
      "Epoch 7 Loss 0.2129\n",
      "Epoch 8 Loss 0.2040\n",
      "Epoch 9 Loss 0.1957\n",
      "Epoch 10 Loss 0.1881\n",
      "Epoch 11 Loss 0.1811\n",
      "Epoch 12 Loss 0.1746\n",
      "Epoch 13 Loss 0.1686\n",
      "Epoch 14 Loss 0.1631\n",
      "Epoch 15 Loss 0.1580\n",
      "Epoch 16 Loss 0.1533\n",
      "Epoch 17 Loss 0.1490\n",
      "Epoch 18 Loss 0.1451\n",
      "Epoch 19 Loss 0.1414\n",
      "Epoch 20 Loss 0.1380\n",
      "Epoch 21 Loss 0.1349\n",
      "Epoch 22 Loss 0.1320\n",
      "Epoch 23 Loss 0.1293\n",
      "Epoch 24 Loss 0.1267\n",
      "Epoch 25 Loss 0.1244\n",
      "Epoch 26 Loss 0.1222\n",
      "Epoch 27 Loss 0.1201\n",
      "Epoch 28 Loss 0.1181\n",
      "Epoch 29 Loss 0.1163\n",
      "Epoch 30 Loss 0.1146\n",
      "Epoch 31 Loss 0.1130\n",
      "Epoch 32 Loss 0.1114\n",
      "Epoch 33 Loss 0.1100\n",
      "Epoch 34 Loss 0.1086\n",
      "Epoch 35 Loss 0.1073\n",
      "Epoch 36 Loss 0.1060\n",
      "Epoch 37 Loss 0.1049\n",
      "Epoch 38 Loss 0.1037\n",
      "Epoch 39 Loss 0.1027\n",
      "Epoch 40 Loss 0.1016\n",
      "Epoch 41 Loss 0.1007\n",
      "Epoch 42 Loss 0.0997\n",
      "Epoch 43 Loss 0.0988\n",
      "Epoch 44 Loss 0.0980\n",
      "Epoch 45 Loss 0.0972\n",
      "Epoch 46 Loss 0.0964\n",
      "Epoch 47 Loss 0.0956\n",
      "Epoch 48 Loss 0.0949\n",
      "Epoch 49 Loss 0.0942\n",
      "Epoch 50 Loss 0.0936\n",
      "Epoch 51 Loss 0.0929\n",
      "Epoch 52 Loss 0.0923\n",
      "Epoch 53 Loss 0.0917\n",
      "Epoch 54 Loss 0.0911\n",
      "Epoch 55 Loss 0.0906\n",
      "Epoch 56 Loss 0.0901\n",
      "Epoch 57 Loss 0.0895\n",
      "Epoch 58 Loss 0.0890\n",
      "Epoch 59 Loss 0.0886\n",
      "Epoch 60 Loss 0.0881\n",
      "Epoch 61 Loss 0.0877\n",
      "Epoch 62 Loss 0.0872\n",
      "Epoch 63 Loss 0.0868\n",
      "Epoch 64 Loss 0.0864\n",
      "Epoch 65 Loss 0.0860\n",
      "Epoch 66 Loss 0.0856\n",
      "Epoch 67 Loss 0.0852\n",
      "Epoch 68 Loss 0.0849\n",
      "Epoch 69 Loss 0.0845\n",
      "Epoch 70 Loss 0.0842\n",
      "Epoch 71 Loss 0.0839\n",
      "Epoch 72 Loss 0.0835\n",
      "Epoch 73 Loss 0.0832\n",
      "Epoch 74 Loss 0.0829\n",
      "Epoch 75 Loss 0.0826\n",
      "Epoch 76 Loss 0.0823\n",
      "Epoch 77 Loss 0.0821\n",
      "Epoch 78 Loss 0.0818\n",
      "Epoch 79 Loss 0.0815\n",
      "Epoch 80 Loss 0.0813\n",
      "Epoch 81 Loss 0.0810\n",
      "Epoch 82 Loss 0.0808\n",
      "Epoch 83 Loss 0.0805\n",
      "Epoch 84 Loss 0.0803\n",
      "Epoch 85 Loss 0.0801\n",
      "Epoch 86 Loss 0.0798\n",
      "Epoch 87 Loss 0.0796\n",
      "Epoch 88 Loss 0.0794\n",
      "Epoch 89 Loss 0.0792\n",
      "Epoch 90 Loss 0.0790\n",
      "Epoch 91 Loss 0.0788\n",
      "Epoch 92 Loss 0.0786\n",
      "Epoch 93 Loss 0.0784\n",
      "Epoch 94 Loss 0.0782\n",
      "Epoch 95 Loss 0.0780\n",
      "Epoch 96 Loss 0.0779\n",
      "Epoch 97 Loss 0.0777\n",
      "Epoch 98 Loss 0.0775\n",
      "Epoch 99 Loss 0.0773\n",
      "Epoch 100 Loss 0.0772\n",
      "Epoch 101 Loss 0.0770\n",
      "Epoch 102 Loss 0.0769\n",
      "Epoch 103 Loss 0.0767\n",
      "Epoch 104 Loss 0.0765\n",
      "Epoch 105 Loss 0.0764\n",
      "Epoch 106 Loss 0.0762\n",
      "Epoch 107 Loss 0.0761\n",
      "Epoch 108 Loss 0.0760\n",
      "Epoch 109 Loss 0.0758\n",
      "Epoch 110 Loss 0.0757\n",
      "Epoch 111 Loss 0.0755\n",
      "Epoch 112 Loss 0.0754\n",
      "Epoch 113 Loss 0.0753\n",
      "Epoch 114 Loss 0.0752\n",
      "Epoch 115 Loss 0.0750\n",
      "Epoch 116 Loss 0.0749\n",
      "Epoch 117 Loss 0.0748\n",
      "Epoch 118 Loss 0.0747\n",
      "Epoch 119 Loss 0.0746\n",
      "Epoch 120 Loss 0.0744\n",
      "Epoch 121 Loss 0.0743\n",
      "Epoch 122 Loss 0.0742\n",
      "Epoch 123 Loss 0.0741\n",
      "Epoch 124 Loss 0.0740\n",
      "Epoch 125 Loss 0.0739\n",
      "Epoch 126 Loss 0.0738\n",
      "Epoch 127 Loss 0.0737\n",
      "Epoch 128 Loss 0.0736\n",
      "Epoch 129 Loss 0.0735\n",
      "Epoch 130 Loss 0.0734\n",
      "Epoch 131 Loss 0.0733\n",
      "Epoch 132 Loss 0.0732\n",
      "Epoch 133 Loss 0.0731\n",
      "Epoch 134 Loss 0.0730\n",
      "Epoch 135 Loss 0.0729\n",
      "Epoch 136 Loss 0.0728\n",
      "Epoch 137 Loss 0.0728\n",
      "Epoch 138 Loss 0.0727\n",
      "Epoch 139 Loss 0.0726\n",
      "Epoch 140 Loss 0.0725\n",
      "Epoch 141 Loss 0.0724\n",
      "Epoch 142 Loss 0.0723\n",
      "Epoch 143 Loss 0.0723\n",
      "Epoch 144 Loss 0.0722\n",
      "Epoch 145 Loss 0.0721\n",
      "Epoch 146 Loss 0.0720\n",
      "Epoch 147 Loss 0.0719\n",
      "Epoch 148 Loss 0.0719\n",
      "Epoch 149 Loss 0.0718\n",
      "Epoch 150 Loss 0.0717\n",
      "Epoch 151 Loss 0.0717\n",
      "Epoch 152 Loss 0.0716\n",
      "Epoch 153 Loss 0.0715\n",
      "Epoch 154 Loss 0.0714\n",
      "Epoch 155 Loss 0.0714\n",
      "Epoch 156 Loss 0.0713\n",
      "Epoch 157 Loss 0.0712\n",
      "Epoch 158 Loss 0.0712\n",
      "Epoch 159 Loss 0.0711\n",
      "Epoch 160 Loss 0.0711\n",
      "Epoch 161 Loss 0.0710\n",
      "Epoch 162 Loss 0.0709\n",
      "Epoch 163 Loss 0.0709\n",
      "Epoch 164 Loss 0.0708\n",
      "Epoch 165 Loss 0.0708\n",
      "Epoch 166 Loss 0.0707\n",
      "Epoch 167 Loss 0.0706\n",
      "Epoch 168 Loss 0.0706\n",
      "Epoch 169 Loss 0.0705\n",
      "Epoch 170 Loss 0.0705\n",
      "Epoch 171 Loss 0.0704\n",
      "Epoch 172 Loss 0.0704\n",
      "Epoch 173 Loss 0.0703\n",
      "Epoch 174 Loss 0.0703\n",
      "Epoch 175 Loss 0.0702\n",
      "Epoch 176 Loss 0.0702\n",
      "Epoch 177 Loss 0.0701\n",
      "Epoch 178 Loss 0.0701\n",
      "Epoch 179 Loss 0.0700\n",
      "Epoch 180 Loss 0.0700\n",
      "Epoch 181 Loss 0.0699\n",
      "Epoch 182 Loss 0.0699\n",
      "Epoch 183 Loss 0.0698\n",
      "Epoch 184 Loss 0.0698\n",
      "Epoch 185 Loss 0.0697\n",
      "Epoch 186 Loss 0.0697\n",
      "Epoch 187 Loss 0.0696\n",
      "Epoch 188 Loss 0.0696\n",
      "Epoch 189 Loss 0.0695\n",
      "Epoch 190 Loss 0.0695\n",
      "Epoch 191 Loss 0.0695\n",
      "Epoch 192 Loss 0.0694\n",
      "Epoch 193 Loss 0.0694\n",
      "Epoch 194 Loss 0.0693\n",
      "Epoch 195 Loss 0.0693\n",
      "Epoch 196 Loss 0.0692\n",
      "Epoch 197 Loss 0.0692\n",
      "Epoch 198 Loss 0.0692\n",
      "Epoch 199 Loss 0.0691\n",
      "Epoch 200 Loss 0.0691\n",
      "Epoch 201 Loss 0.0691\n",
      "Epoch 202 Loss 0.0690\n",
      "Epoch 203 Loss 0.0690\n",
      "Epoch 204 Loss 0.0689\n",
      "Epoch 205 Loss 0.0689\n",
      "Epoch 206 Loss 0.0689\n",
      "Epoch 207 Loss 0.0688\n",
      "Epoch 208 Loss 0.0688\n",
      "Epoch 209 Loss 0.0688\n",
      "Epoch 210 Loss 0.0687\n",
      "Epoch 211 Loss 0.0687\n",
      "Epoch 212 Loss 0.0687\n",
      "Epoch 213 Loss 0.0686\n",
      "Epoch 214 Loss 0.0686\n",
      "Epoch 215 Loss 0.0686\n",
      "Epoch 216 Loss 0.0685\n",
      "Epoch 217 Loss 0.0685\n",
      "Epoch 218 Loss 0.0685\n",
      "Epoch 219 Loss 0.0684\n",
      "Epoch 220 Loss 0.0684\n",
      "Epoch 221 Loss 0.0684\n",
      "Epoch 222 Loss 0.0683\n",
      "Epoch 223 Loss 0.0683\n",
      "Epoch 224 Loss 0.0683\n",
      "Epoch 225 Loss 0.0682\n",
      "Epoch 226 Loss 0.0682\n",
      "Epoch 227 Loss 0.0682\n",
      "Epoch 228 Loss 0.0682\n",
      "Epoch 229 Loss 0.0681\n",
      "Epoch 230 Loss 0.0681\n",
      "Epoch 231 Loss 0.0681\n",
      "Epoch 232 Loss 0.0680\n",
      "Epoch 233 Loss 0.0680\n",
      "Epoch 234 Loss 0.0680\n",
      "Epoch 235 Loss 0.0680\n",
      "Epoch 236 Loss 0.0679\n",
      "Epoch 237 Loss 0.0679\n",
      "Epoch 238 Loss 0.0679\n",
      "Epoch 239 Loss 0.0678\n",
      "Epoch 240 Loss 0.0678\n",
      "Epoch 241 Loss 0.0678\n",
      "Epoch 242 Loss 0.0678\n",
      "Epoch 243 Loss 0.0677\n",
      "Epoch 244 Loss 0.0677\n",
      "Epoch 245 Loss 0.0677\n",
      "Epoch 246 Loss 0.0677\n",
      "Epoch 247 Loss 0.0676\n",
      "Epoch 248 Loss 0.0676\n",
      "Epoch 249 Loss 0.0676\n",
      "Epoch 250 Loss 0.0676\n",
      "Epoch 251 Loss 0.0675\n",
      "Epoch 252 Loss 0.0675\n",
      "Epoch 253 Loss 0.0675\n",
      "Epoch 254 Loss 0.0675\n",
      "Epoch 255 Loss 0.0674\n",
      "Epoch 256 Loss 0.0674\n",
      "Epoch 257 Loss 0.0674\n",
      "Epoch 258 Loss 0.0674\n",
      "Epoch 259 Loss 0.0674\n",
      "Epoch 260 Loss 0.0673\n",
      "Epoch 261 Loss 0.0673\n",
      "Epoch 262 Loss 0.0673\n",
      "Epoch 263 Loss 0.0673\n",
      "Epoch 264 Loss 0.0672\n",
      "Epoch 265 Loss 0.0672\n",
      "Epoch 266 Loss 0.0672\n",
      "Epoch 267 Loss 0.0672\n",
      "Epoch 268 Loss 0.0672\n",
      "Epoch 269 Loss 0.0671\n",
      "Epoch 270 Loss 0.0671\n",
      "Epoch 271 Loss 0.0671\n",
      "Epoch 272 Loss 0.0671\n",
      "Epoch 273 Loss 0.0671\n",
      "Epoch 274 Loss 0.0670\n",
      "Epoch 275 Loss 0.0670\n",
      "Epoch 276 Loss 0.0670\n",
      "Epoch 277 Loss 0.0670\n",
      "Epoch 278 Loss 0.0670\n",
      "Epoch 279 Loss 0.0669\n",
      "Epoch 280 Loss 0.0669\n",
      "Epoch 281 Loss 0.0669\n",
      "Epoch 282 Loss 0.0669\n",
      "Epoch 283 Loss 0.0669\n",
      "Epoch 284 Loss 0.0668\n",
      "Epoch 285 Loss 0.0668\n",
      "Epoch 286 Loss 0.0668\n",
      "Epoch 287 Loss 0.0668\n",
      "Epoch 288 Loss 0.0668\n",
      "Epoch 289 Loss 0.0667\n",
      "Epoch 290 Loss 0.0667\n",
      "Epoch 291 Loss 0.0667\n",
      "Epoch 292 Loss 0.0667\n",
      "Epoch 293 Loss 0.0667\n",
      "Epoch 294 Loss 0.0667\n",
      "Epoch 295 Loss 0.0666\n",
      "Epoch 296 Loss 0.0666\n",
      "Epoch 297 Loss 0.0666\n",
      "Epoch 298 Loss 0.0666\n",
      "Epoch 299 Loss 0.0666\n",
      "Epoch 300 Loss 0.0666\n",
      "Epoch 301 Loss 0.0665\n",
      "Epoch 302 Loss 0.0665\n",
      "Epoch 303 Loss 0.0665\n",
      "Epoch 304 Loss 0.0665\n",
      "Epoch 305 Loss 0.0665\n",
      "Epoch 306 Loss 0.0665\n",
      "Epoch 307 Loss 0.0664\n",
      "Epoch 308 Loss 0.0664\n",
      "Epoch 309 Loss 0.0664\n",
      "Epoch 310 Loss 0.0664\n",
      "Epoch 311 Loss 0.0664\n",
      "Epoch 312 Loss 0.0664\n",
      "Epoch 313 Loss 0.0663\n",
      "Epoch 314 Loss 0.0663\n",
      "Epoch 315 Loss 0.0663\n",
      "Epoch 316 Loss 0.0663\n",
      "Epoch 317 Loss 0.0663\n",
      "Epoch 318 Loss 0.0663\n",
      "Epoch 319 Loss 0.0662\n",
      "Epoch 320 Loss 0.0662\n",
      "Epoch 321 Loss 0.0662\n",
      "Epoch 322 Loss 0.0662\n",
      "Epoch 323 Loss 0.0662\n",
      "Epoch 324 Loss 0.0662\n",
      "Epoch 325 Loss 0.0662\n",
      "Epoch 326 Loss 0.0661\n",
      "Epoch 327 Loss 0.0661\n",
      "Epoch 328 Loss 0.0661\n",
      "Epoch 329 Loss 0.0661\n",
      "Epoch 330 Loss 0.0661\n",
      "Epoch 331 Loss 0.0661\n",
      "Epoch 332 Loss 0.0661\n",
      "Epoch 333 Loss 0.0660\n",
      "Epoch 334 Loss 0.0660\n",
      "Epoch 335 Loss 0.0660\n",
      "Epoch 336 Loss 0.0660\n",
      "Epoch 337 Loss 0.0660\n",
      "Epoch 338 Loss 0.0660\n",
      "Epoch 339 Loss 0.0660\n",
      "Epoch 340 Loss 0.0659\n",
      "Epoch 341 Loss 0.0659\n",
      "Epoch 342 Loss 0.0659\n",
      "Epoch 343 Loss 0.0659\n",
      "Epoch 344 Loss 0.0659\n",
      "Epoch 345 Loss 0.0659\n",
      "Epoch 346 Loss 0.0659\n",
      "Epoch 347 Loss 0.0659\n",
      "Epoch 348 Loss 0.0658\n",
      "Epoch 349 Loss 0.0658\n",
      "Epoch 350 Loss 0.0658\n",
      "Epoch 351 Loss 0.0658\n",
      "Epoch 352 Loss 0.0658\n",
      "Epoch 353 Loss 0.0658\n",
      "Epoch 354 Loss 0.0658\n",
      "Epoch 355 Loss 0.0658\n",
      "Epoch 356 Loss 0.0657\n",
      "Epoch 357 Loss 0.0657\n",
      "Epoch 358 Loss 0.0657\n",
      "Epoch 359 Loss 0.0657\n",
      "Epoch 360 Loss 0.0657\n",
      "Epoch 361 Loss 0.0657\n",
      "Epoch 362 Loss 0.0657\n",
      "Epoch 363 Loss 0.0657\n",
      "Epoch 364 Loss 0.0656\n",
      "Epoch 365 Loss 0.0656\n",
      "Epoch 366 Loss 0.0656\n",
      "Epoch 367 Loss 0.0656\n",
      "Epoch 368 Loss 0.0656\n",
      "Epoch 369 Loss 0.0656\n",
      "Epoch 370 Loss 0.0656\n",
      "Epoch 371 Loss 0.0656\n",
      "Epoch 372 Loss 0.0656\n",
      "Epoch 373 Loss 0.0655\n",
      "Epoch 374 Loss 0.0655\n",
      "Epoch 375 Loss 0.0655\n",
      "Epoch 376 Loss 0.0655\n",
      "Epoch 377 Loss 0.0655\n",
      "Epoch 378 Loss 0.0655\n",
      "Epoch 379 Loss 0.0655\n",
      "Epoch 380 Loss 0.0655\n",
      "Epoch 381 Loss 0.0654\n",
      "Epoch 382 Loss 0.0654\n",
      "Epoch 383 Loss 0.0654\n",
      "Epoch 384 Loss 0.0654\n",
      "Epoch 385 Loss 0.0654\n",
      "Epoch 386 Loss 0.0654\n",
      "Epoch 387 Loss 0.0654\n",
      "Epoch 388 Loss 0.0654\n",
      "Epoch 389 Loss 0.0654\n",
      "Epoch 390 Loss 0.0654\n",
      "Epoch 391 Loss 0.0653\n",
      "Epoch 392 Loss 0.0653\n",
      "Epoch 393 Loss 0.0653\n",
      "Epoch 394 Loss 0.0653\n",
      "Epoch 395 Loss 0.0653\n",
      "Epoch 396 Loss 0.0653\n",
      "Epoch 397 Loss 0.0653\n",
      "Epoch 398 Loss 0.0653\n",
      "Epoch 399 Loss 0.0653\n",
      "Epoch 400 Loss 0.0653\n",
      "Epoch 401 Loss 0.0652\n",
      "Epoch 402 Loss 0.0652\n",
      "Epoch 403 Loss 0.0652\n",
      "Epoch 404 Loss 0.0652\n",
      "Epoch 405 Loss 0.0652\n",
      "Epoch 406 Loss 0.0652\n",
      "Epoch 407 Loss 0.0652\n",
      "Epoch 408 Loss 0.0652\n",
      "Epoch 409 Loss 0.0652\n",
      "Epoch 410 Loss 0.0652\n",
      "Epoch 411 Loss 0.0651\n",
      "Epoch 412 Loss 0.0651\n",
      "Epoch 413 Loss 0.0651\n",
      "Epoch 414 Loss 0.0651\n",
      "Epoch 415 Loss 0.0651\n",
      "Epoch 416 Loss 0.0651\n",
      "Epoch 417 Loss 0.0651\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 418 Loss 0.0651\n",
      "Epoch 419 Loss 0.0651\n",
      "Epoch 420 Loss 0.0651\n",
      "Epoch 421 Loss 0.0650\n",
      "Epoch 422 Loss 0.0650\n",
      "Epoch 423 Loss 0.0650\n",
      "Epoch 424 Loss 0.0650\n",
      "Epoch 425 Loss 0.0650\n",
      "Epoch 426 Loss 0.0650\n",
      "Epoch 427 Loss 0.0650\n",
      "Epoch 428 Loss 0.0650\n",
      "Epoch 429 Loss 0.0650\n",
      "Epoch 430 Loss 0.0650\n",
      "Epoch 431 Loss 0.0650\n",
      "Epoch 432 Loss 0.0649\n",
      "Epoch 433 Loss 0.0649\n",
      "Epoch 434 Loss 0.0649\n",
      "Epoch 435 Loss 0.0649\n",
      "Epoch 436 Loss 0.0649\n",
      "Epoch 437 Loss 0.0649\n",
      "Epoch 438 Loss 0.0649\n",
      "Epoch 439 Loss 0.0649\n",
      "Epoch 440 Loss 0.0649\n",
      "Epoch 441 Loss 0.0649\n",
      "Epoch 442 Loss 0.0649\n",
      "Epoch 443 Loss 0.0649\n",
      "Epoch 444 Loss 0.0648\n",
      "Epoch 445 Loss 0.0648\n",
      "Epoch 446 Loss 0.0648\n",
      "Epoch 447 Loss 0.0648\n",
      "Epoch 448 Loss 0.0648\n",
      "Epoch 449 Loss 0.0648\n",
      "Epoch 450 Loss 0.0648\n",
      "Epoch 451 Loss 0.0648\n",
      "Epoch 452 Loss 0.0648\n",
      "Epoch 453 Loss 0.0648\n",
      "Epoch 454 Loss 0.0648\n",
      "Epoch 455 Loss 0.0648\n",
      "Epoch 456 Loss 0.0647\n",
      "Epoch 457 Loss 0.0647\n",
      "Epoch 458 Loss 0.0647\n",
      "Epoch 459 Loss 0.0647\n",
      "Epoch 460 Loss 0.0647\n",
      "Epoch 461 Loss 0.0647\n",
      "Epoch 462 Loss 0.0647\n",
      "Epoch 463 Loss 0.0647\n",
      "Epoch 464 Loss 0.0647\n",
      "Epoch 465 Loss 0.0647\n",
      "Epoch 466 Loss 0.0647\n",
      "Epoch 467 Loss 0.0647\n",
      "Epoch 468 Loss 0.0647\n",
      "Epoch 469 Loss 0.0646\n",
      "Epoch 470 Loss 0.0646\n",
      "Epoch 471 Loss 0.0646\n",
      "Epoch 472 Loss 0.0646\n",
      "Epoch 473 Loss 0.0646\n",
      "Epoch 474 Loss 0.0646\n",
      "Epoch 475 Loss 0.0646\n",
      "Epoch 476 Loss 0.0646\n",
      "Epoch 477 Loss 0.0646\n",
      "Epoch 478 Loss 0.0646\n",
      "Epoch 479 Loss 0.0646\n",
      "Epoch 480 Loss 0.0646\n",
      "Epoch 481 Loss 0.0646\n",
      "Epoch 482 Loss 0.0645\n",
      "Epoch 483 Loss 0.0645\n",
      "Epoch 484 Loss 0.0645\n",
      "Epoch 485 Loss 0.0645\n",
      "Epoch 486 Loss 0.0645\n",
      "Epoch 487 Loss 0.0645\n",
      "Epoch 488 Loss 0.0645\n",
      "Epoch 489 Loss 0.0645\n",
      "Epoch 490 Loss 0.0645\n",
      "Epoch 491 Loss 0.0645\n",
      "Epoch 492 Loss 0.0645\n",
      "Epoch 493 Loss 0.0645\n",
      "Epoch 494 Loss 0.0645\n",
      "Epoch 495 Loss 0.0644\n",
      "Epoch 496 Loss 0.0644\n",
      "Epoch 497 Loss 0.0644\n",
      "Epoch 498 Loss 0.0644\n",
      "Epoch 499 Loss 0.0644\n"
     ]
    }
   ],
   "source": [
    "l1 = train(X,y,model,epochs=500,learning_rate=0.001,logs=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "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": [
    "plt.plot(l1)\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_ = model.predict(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy 95.2 %\n"
     ]
    }
   ],
   "source": [
    "print('accuracy',np.sum(y_==y)*100/len(y),'%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "x = "
   ]
  }
 ],
 "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": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}