neural-network-for-apple-detection.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 427,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import torch\n",
    "import numpy as np\n",
    "from torch.autograd import Variable\n",
    "import torch.nn as nn\n",
    "\n",
    "images = np.load(\"./images.npy\")\n",
    "labels = np.load(\"./labels.npy\")\n",
    "\n",
    "input_labels = (labels > 0).astype(dtype=np.float)\n",
    "images = np.reshape(images, (images.shape[0], images.shape[1] * images.shape[2]))\n",
    "images = images.astype(dtype=np.float)\n",
    "images = images/255\n",
    "\n",
    "train_images = torch.Tensor(images[0:40000, : ])\n",
    "cv_images = torch.Tensor(images[40000:45000, : ])\n",
    "test_images = torch.Tensor(images[45000:50000, :])\n",
    "\n",
    "#split the labels in to train, validation and test data\n",
    "train_labels = torch.FloatTensor(input_labels[0:40000])\n",
    "cv_labels = torch.FloatTensor(input_labels[40000:45000])\n",
    "test_labels = torch .FloatTensor(input_labels[45000:50000])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "def accuracy(y,y_hat):\n",
    "    count = 0\n",
    "    for i in range(y.shape[0]):\n",
    "        if y[i] == y_hat[i]:\n",
    "            count += 1\n",
    "    return count/y.shape[0]\n",
    "\n",
    "def accuracy_random_train():\n",
    "\n",
    "    import random\n",
    "    import numpy as np\n",
    "\n",
    "\n",
    "    x =random.sample(range(0,40000),1000)\n",
    "    y = np.empty(1000)\n",
    "    y_hat = np.empty(1000)\n",
    "\n",
    "\n",
    "    index=0\n",
    "    print(w1)\n",
    "    \n",
    "    for i in x:\n",
    "\n",
    "        images_random = Variable(train_images[i].view(1,676), requires_grad=False)\n",
    "        y[index] = train_labels[i]\n",
    "\n",
    "        y_pred = torch.sigmoid(images_random.mm(w1).clamp(0).mm(w2))\n",
    "        #y_pred = torch.sigmoid(a.mm(w1).clamp(0).mm(w2))\n",
    "        \n",
    "        \n",
    "\n",
    "        res = y_pred.data[0][0]\n",
    "    \n",
    "\n",
    "        if res <= 0.5:\n",
    "            act_label = 0\n",
    "        else:\n",
    "            act_label = 1\n",
    "\n",
    "        y_hat[index] = act_label\n",
    "        index += 1\n",
    "\n",
    "    acc = accuracy(y,y_hat)\n",
    "   \n",
    "  \n",
    "    return acc\n",
    "\n",
    "\n",
    "    \n",
    "\n",
    "def accuracy_random_validation():\n",
    "\n",
    "    import random\n",
    "    import numpy as np\n",
    "\n",
    "\n",
    "    x =random.sample(range(0,5000),1000)\n",
    "    y = np.empty(1000)\n",
    "    y_hat = np.empty(1000)\n",
    "\n",
    "\n",
    "    index=0\n",
    "    \n",
    "    for i in x:\n",
    "\n",
    "        images_random = Variable(cv_images[i].view(1,676), requires_grad=False)\n",
    "        y[index] = cv_labels[i]\n",
    "\n",
    "       \n",
    "        y_pred = torch.sigmoid(images_random.mm(w1).clamp(0).mm(w2))\n",
    "\n",
    "\n",
    "        res = y_pred.data[0][0]\n",
    "    \n",
    "\n",
    "        if res <= 0.5:\n",
    "            act_label = 0\n",
    "        else:\n",
    "            act_label = 1\n",
    "\n",
    "        y_hat[index] = act_label\n",
    "        index += 1\n",
    "\n",
    "    acc = accuracy(y,y_hat)\n",
    "  \n",
    "    return acc\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "dtype = torch.FloatTensor\n",
    "\n",
    "# N is batch size; D_in is input dimension;\n",
    "# H is hidden dimension; D_out is output dimension.\n",
    "N, D_in, H, D_out = 64, 676, 100, 1\n",
    "online_training_iterations = 4\n",
    "input_size = 40000\n",
    "size = (online_training_iterations * input_size) / 100\n",
    "index = 0\n",
    "\n",
    "# N dimensional arrays to store training and validation accuracies for every 100 steps\n",
    "accuracy_train = []\n",
    "accuracy_valid = []\n",
    "w1 = Variable(torch.randn(D_in, H).type(dtype), requires_grad=True)\n",
    "w2 = Variable(torch.randn(H, D_out).type(dtype), requires_grad=True)\n",
    "\n",
    "learning_rate = 0.001\n",
    "for i in range(online_training_iterations):\n",
    "    for t in range(40000):\n",
    "\n",
    "        x = Variable(train_images[t].view(1,676), requires_grad=False)\n",
    "\n",
    "        y = train_labels[t]\n",
    "        \n",
    "        \n",
    "        y_pred = torch.sigmoid(x.mm(w1).clamp(min = 0).mm(w2).clamp(min= 0.001, max= 0.999))\n",
    "        \n",
    "        loss = -(y * torch.log(y_pred) + \n",
    "                 (1-y) * torch.log(1 - y_pred)) \n",
    "        \n",
    "        if t%100 == 0:\n",
    "            accuracy_train.append(accuracy_random_train())\n",
    "            accuracy_valid.append(accuracy_random_validation())\n",
    "\n",
    "        loss.backward()\n",
    "        w1.data -= learning_rate * w1.grad.data\n",
    "        w2.data -= learning_rate * w2.grad.data\n",
    "        \n",
    "        w1.data = w1.data/torch.norm(w1.data)\n",
    "        w2.data = w2.data/torch.norm(w2.data)\n",
    "\n",
    "        w1.grad.data.zero_()\n",
    "        w2.grad.data.zero_()    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 441,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa388db2e80>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fa388f00630>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "t = np.arange(0,400,1)\n",
    "\n",
    "s = accuracy_train[0:400]\n",
    "k = accuracy_valid[0:400]\n",
    "plt.plot(t, s, t, k)\n",
    "\n",
    "plt.xlabel('number of iteration')\n",
    "plt.ylabel('accuracy')\n",
    "plt.title('Training/validation accuracy')\n",
    "plt.grid(True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Is this Overfitting:\n",
    "\n",
    "From the plot above, there is no over fitting. The accuracies of training and cross validation data sets differ only by 1 percentage points"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Kaggle submission username : villa1"
   ]
  }
 ],
 "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": "3.5.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}