wgan-gp/W-GAN-GP-TF-v1.14.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "view-in-github"
   },
   "source": [
    "<a href=\"https://colab.research.google.com/github/kartikgill/The-GAN-Book/blob/main/Skill-07/W-GAN-GP-TF1.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import Useful Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:03.756187Z",
     "start_time": "2021-02-13T08:27:03.108447Z"
    },
    "collapsed": true,
    "id": "14h4ejQKFVe9"
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from tqdm import tqdm_notebook\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Installing TF-1.14"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "collapsed": true,
    "id": "rPD00N_hGtD5",
    "outputId": "d974b66f-e0aa-47f8-83ef-d2ff41bd8b4a"
   },
   "outputs": [],
   "source": [
    "pip install tensorflow==1.14"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "collapsed": true,
    "id": "IyT6iImGH31t",
    "outputId": "9fb84084-c26e-457b-f068-d246e85c191c"
   },
   "outputs": [],
   "source": [
    "pip install keras==2.3.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:13.152627Z",
     "start_time": "2021-02-13T08:27:11.516980Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "UnlKaodtFVe9",
    "outputId": "fd4d7274-dc08-4d67-ecb6-b16d960af6c4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.14.0\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import keras\n",
    "print (tf.__version__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:13.156766Z",
     "start_time": "2021-02-13T08:27:13.154094Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "1LLXEENMrVJc",
    "outputId": "7da4f05b-bc4f-4b72-fe51-ab81b806fc6d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.3.0\n"
     ]
    }
   ],
   "source": [
    "print (keras.__version__)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Download and show Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:26.499931Z",
     "start_time": "2021-02-13T08:27:24.434500Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 501
    },
    "id": "s0l5lTN_FVe-",
    "outputId": "be768ddf-4aab-4560-88ba-ade8fde16a84"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training data shapes: X=(60000, 28, 28), y=(60000,)\n",
      "Testing data shapes: X=(10000, 28, 28), y=(10000,)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAisAAAAyCAYAAACd6JtSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO2deXyU1fX/35PZl2xkgZAIYRNEEUSxuFSriHVFrLghrXtdqsWXrUrdqrXFBV+i1WqxIi+lUn0VlKK2RYoLi4pssu+QQNhCSCbJJJPZf388v3t4ZhICKCSTr/f9DySZTJ47z33uPfeczznHkkgk0Gg0Go1Go0lXMtr7AjQajUaj0WhaQxsrGo1Go9Fo0hptrGg0Go1Go0lrtLGi0Wg0Go0mrdHGikaj0Wg0mrRGGysajUaj0WjSGtshfp5uec2WNvxbP+SxwxGOf9q0aSxZsgSn00k8HgegqqqKbdu2cd555wFQVFTEnj17GDVqFP369TvS60nbex+Px8nIMOz+LVu2AHDxxRdz7rnnkpOTA8DSpUvp0aMHkydPTvrdaDSKzXaoxzB9x94G6LG3Dd9pzkciEex2u3z/5ZdfpqKiAoBnn332wJv//xIZFouFRCKBxXLIoaXt2NuAtF7r24AWx3/IVVJzeKiHN/VBXLRoER999BEA9fX1DBgwgFtvvVV+HovFkl6vNr2OhtPppKioiPLycnbu3AkYG/FJJ51EdnY2AP369SMajeJwONrzUr83sVgMq9UqX6t79sYbb3D33XcDkJ2dzeTJk2WRPu6441i7di3Dhg3j/fffl9eYDZXDNFw0mrQg1VB5/PHHmT17tsz56667jsGDB/Pggw/KGqfnuOa70jF3Ro1Go9FoND8YtIl7lFAnh1AohMvlAmDs2LHMmTNHTtJZWVnccccdLF26lFdffRUAq9Xa7KTekdi9ezcAZWVl9OzZkxUrVrB3714AevbsCcC3334LwOeff05paSk+n4/MzEwACgoK2uGqvxvqxKjuVTgcBuCJJ57g5ZdfJj8/n759+wKwb98+li5dyoABAwAYPHgwVquVtWvXSmjod7/7HePHj5f31ydOTUfCbrezdu1annvuOcAI8y5YsEC8LfPmzeMf//gHt9xyC2+++SZgzPGOvN5p2g/LIcrtd4hY1jHiiMceCoVwOp1Mnz4dgP/973/89a9/bfa63//+93g8HgAeeuihdNQtwGGOf8GCBcABQ2TLli0EAgHAMGA++eQTLr74YgAuu+wyiouL2b17txgpw4cPP9zraZd7r8J65gU2FApx/fXXs23bNgBqa2vJzs7GbrcTiUQAWL9+PWeccYbc582bN+NyuairqxNDrba2Fp/PxymnnALAiy++SF5eXkvXk9bz/hijx942HNbYlR5NhT7vvfdeHnzwQcAIdbbEU089JQbMuHHjCIfDSeGjg+hX0m7sbUharvVtiNasHEvi8ThOpxNAPCnTpk0DjDgtGCdyi8XCk08+yTXXXAMc2OhSF4GOQkNDAwD5+flkZGTQt29fGUN1dTUTJkxg8ODBgOGFOf7448nLy6O2trbdrvlISCQSJBKJpJPgddddx7fffktRUREALpeLRCJBQ0MDnTp1AoyFe/ny5aLXyc3Npa6ujkgkIoZpXl4e4XCYuXPnAnDeeeexcuXKthzeUaMl0eTf//53wNAzud1u8vPzyc/Pl+8FAgGZK8ojpUlvYrEYYKxTr7zyCi6XS4yUuro6PB6PzAOLxUJGRgaPPfYYv/jFL+Q9HA4Huied5kjRxspRIhKJ4HQ6mTZtmrj+FS15TdRp+t///jfXX3+9nMiVwdNR2LRpE2CMsaioiHg8Lp6Vs846i/LycsrKygAoKSnB5XLhcDjwer3tdclHhBJNA7z77rsAbNiwgS5dusjCnUgkRDhcV1cHGJ9HQUEB9fX1AJSXl5OZmUl+fr4YpqFQCLvdTpcuXQDYv38/zzzzDOPGjWvTMR4NWjod33XXXQByr3Nzc0Vc3atXLyorKykvLwegT58++Hw+Vq1axcCBAwGYOXNmW1y65jvy6aef8uSTT8rXPp/voIetqqoqwAgJDxo0iFAoBCAhc43mUHSsY7xGo9FoNJofHNqzcpRQJ4olS5YwevRo+X6qe1ylOA8dOhSAL774om0v9CijvEbqhLRlyxbmzZsHwLp163A4HFxwwQXyGiVADgaD7XPB3wF1/8aOHQtAjx49SCQS4iExhwDNbnLlcQEoLCzEarWSSCTkNXa7HYvFIu/v9Xr5+OOPO6RnBZrrGdxuN4B4lAoLC9m+fTsAXbp0wel0in7HYrHgcrmw2Wz4/f52uHpNS6h7mkgkyMjISNKaBIPBJC+yuu8thbSVt0x5kNWcj0Qi8jotutW0hjZWjhLqId6/f39S/D3VPa4ezF69egEwY8YMoOOFfxQqzJGVlYXNZiM3N5fi4mIAFi5cSCwWE12Lz+cjkUjQ1NREU1NTu13zkaCMzSeeeEJEwUpIqxZXdU9jsZj8PxQKEQ6HRTAbjUaTCmOp34tGo2LQeL1eysvL+eijj7jsssvabpDfE/UZmTepyspKGZcSkAcCATFqN2zYQFVVFeeccw5gaJ9UdtWwYcPaYRQaM8qobk34v2jRIrZs2SJrmaKlUNAnn3wCGLWWhgwZ0mHXO037kVbGSrqLTM0nZavV2uJ1vv3227zwwgvytfnkoN7DZrOJ0PCdd97hlVdeSXoP9Tko0vXzMBMOh6msrCQUCnHRRRcBMH36dMrKyjj++OMB2LNnDy6XC5/PR1ZWVnte7mFjsVgIh8N88skn4gUIhULiJVHE43ESiYRsuBkZGXg8HpkvSqhrvrcWi6XZZuB0OnnhhRc6jLGixmWxWJJOxsuXL5dTdHZ2NjU1NXi9Xk466SQAOnXqxOzZs0Vo7XK5qKqqIhKJ0K1bt7YfyA+ceDyeJIxNnZuBQIDt27eLkfHFF19gsVj4wx/+wKhRowDDq9rU1CS/53A4yMnJwW63i1ZpxowZlJaWitetV69e5OXlHW5VW007Yl67WtqTVq9eLf9Xz/mh3u9I9rb03wU1Go1Go9H8oGlXz4o6hVZWVlJSUtKilZXqOm8vlOVvjtm2RFFRUVKtjIO9XtXfaCk+3xE8KQpz+Csej1NQUCBjuvTSS5kzZ47UIznuuONoamoiHo93mGwggOeff56GhgbxBjU1NeF2u2X+qhOCWX+ivq++jsfjuN1uBgwYILVpVDhJnUTD4TBZWVns379fdC3pHsc3j9k89ilTpkjIx+fz4XA4kk7qq1evpqGhQTLH8vPz5bSt0r81bUdGRkbSnItGo9x8880y/2pra/F6vVLQcMmSJfTp04d169bxl7/8BYCuXbuSlZUla0JDQwNbt26lsbFR/s6mTZu48847xePo8/koLCwkEolw7bXXAnDHHXe0zaC/I6pFSkvrtGqt4vf7GT9+PA888AA333xzW1/iUUWtY6njNXvDdu7cyaRJk8T73K9fP0aPHp30zKd6zywWC4FAAJ/Pd1jXcdSNFbNbWF1QKsFgkHHjxsngA4EAf/vb35JeY17o2xs1ntraWl588UXAiMP/6le/EjHptddeywUXXED37t2l94/b7SYWiyUtwm63m/79+7Nx40YAhgwZwgMPPCChoJtuuomLLrqIjRs3yk0sKSnh8ssvb9MxHy5qMlZWVpKbm4vL5eLTTz8F4Mwzz2Tbtm0sW7YMMER2e/fupaampsOEgerr6/nwww/FuARjsQoEAuISV83ZzKTO/4yMDILBIAsWLJDPzGKxEIlEZJNwOp04HA4CgQDvvfceQJJYu6OwadMmZs6cKXVoAoEAVquVzp07s3nzZsDQ5xQUFIjRGggEqK6uxmq1snXr1na79tY4klBFuoe0W0KlE3s8HmbOnMnChQspLS0FjDCdy+WiuroaMMbl9Xqprq6WZ7mgoIB4PC7hv8bGRiwWC9XV1RL2zsnJwWKxyH0PBoMEg0Hq6+v517/+BaS/sdLSAeLhhx9m+vTpUsohKysLp9PJ/fffz8iRIwEjdd/MvHnzePTRR4lEIvznP/8BEGMwnUidw8FgELfbnfQsJBIJdu3aJULqjIwMJk6cSHV1NU8//TRwYC1Uh7z58+fjdDppbGzkwgsvPOR1fG9jpSVrqaUHWk3gl19+mauuuop+/foxa9Ys4EDWgJl0eMjNXp1QKER5eTlTpkwBjAf7vffek9PjGWecQVVVFU1NTcyePRswJqy5aZ86fYfDYRGYut1u3n33Xfr37w8YdQg++OADMjIy5DO4/PLL095Y8fv9BAIBampqpAT/fffdR3FxMTfccANgFI3yer34/X6pR5LuhEIhBg4cSFlZmdQEgeYLVjQaTbpnytg2zyGbzSaiWvUa8+8AovtRJ7R0N1ZUNVLzMz9ixIikdUAdYCoqKiSWnZOTQ0VFBfv37wcML1NxcTFbtmzhgw8+AA5kX7UnLR28VEbTmjVr2LdvHwUFBZx66qmAkfUFyetXR2neZ77mYDAoYngwnu+amhpZzxKJBEOHDuWzzz6Tz2Pfvn04nc6k1iMWi4XMzEwxehKJBOXl5fK+sViMSCRCTk6O1GNqT8zdpNXnEY1GmwmCp0+fzvjx41m+fDlgzGen0ynd5JWuzVwo8rbbbmPixImSVHHfffeRnZ3N3r17ee211wCjBUc6YNaTNDU1MWvWLBlbWVkZHo9HsjzBOFC/8847YnDG43E2bdqE1+uV8V5xxRWsWLFCiqWedtppkgWoDu9K39gSrT5Bh3OSSP15MBhk9erVDBkyBIAVK1awYsUKqWB41llnsWzZMu666y6ZnMqNaGbXrl08/PDDUun1kksuafU6jjVOp5Pdu3fLKUIVN1Nf79mzhy5durB582YJA6lThDkUFIvFCIfDsnjFYjE8Hk9SKffS0lIKCwvZsWMHYNzkdEVN6MbGRoLBIFVVVeIR6ty5M507dxZDdePGjZSWllJbW0uPHj3a7ZqPhPz8fGmZoLxf48aNo6ioSO69MlTMKCNEnSJsNpuIcNVrVZ8U1UspGAxy6623MmbMmLSq6Gr2GpkNEEguePib3/wGMBazbt26ybwPhUJkZmYSDoclJJCTk8PAgQPFbexwONi7dy+lpaVJRmF7Y17f9uzZw/Tp02VzVkXtAoEA+/btA2DHjh1cddVV/PrXvwbg7LPP7hCGCiQXaKutrSUzM1NKDFRXV8tzDIZxOXPmTAnjgbGpqc0HjM/O7XYTjUalP5jD4SAUCslhxel04vV6aWpqkmrPjY2NSZ7MY405U0/NWfOardbmmpoawPCQ22w2srKypHqvzWbD6XTK55VIJIhEIjgcDsmOfOONN3jjjTfkfUtLS7Hb7Xg8Hj788EMgPYwV1VpE3aORI0fSu3dvKdJYVFTE3XffzY033siECRMAw0h3uVwSynvzzTdZtGgRI0eOFONk0aJF2Gw2cU74/X6Ki4vZt28fu3btAlo3VtrffaHRaDQajUbTCoc0+dXJ0BzOMLNjxw6xLufMmcMFF1zAa6+9xn//+18ARo0axdSpU1m7di0AX375Jffccw9TpkzhrLPOAuDKK69k0KBBcuLcvXs3mZmZkuYKbe9ZUa5rQOKuymUJxufStWtX6Xvz1ltv8fzzzzNs2DAZ++uvv86yZcskXmu1WsWroMbzyCOPMHXqVO69917AEKV26tSJxYsXS3w4ncWo5vBVOBymtraWzp07J71GtRaorq6muLgYu93e7DXpitkdes8998i/jz/+OO+88w5weIJQpWEwC9WampoIhUJy7x999NGjfv3HCrOHxWKxMGvWLNFzdevWDZ/PJ+Gu6upqGhsbGTVqlHhS1q9fj9/vlxh9MBgkJyeHUCgkcf+NGze2etI62hxMd/Tzn/8cMJp2+v1+TjvtNMAQlXu9XoYOHSo1eL744gsWLVokYdva2loWLlzIGWeckfR3WtLkmdec1PDgscbcqHPr1q188MEHWK1W0bGMGjWKkpIScdfbbDaampoktK3ew9yeIhqNynxXYZ9wOIzH45H33b17N1VVVXi9XvFYbdmypVnLkmPJwTxfKq1+/vz5TJo0SUKznTt3xuPxyPjBmL9qrwTj/jkcDqLRqHgSzKEw9Tvqc1m0aNHRH9gRYN7r1P9ViwS73c7rr78ubUH27NkDGGEsFfZM5ZZbbuGKK67gyiuvlOelW7duFBYWitdsw4YNVFdX85Of/KRZrZ6WaNVYsVgsSUaKEtYol97rr78OHHCPT5w4kfr6esaMGSMuvddee41TTjlFcrAvvPBCtm3bRs+ePUVsN2LECIYNG0bXrl0BQ6w6depUYrGYxARbKj50LInH40m6hLy8PJ588klxcQYCAWpra/n4448Bw5jLz8/nxhtv5Ec/+hEAzz77LG63W95HaV8SiYSIyCoqKhg0aJD8nS1bthAKhZIqvCrRUjqixKEul4vi4mK+/fZbevfunfQatfCsWrUKv99PKBTqMI0MzQXf1FgdDgfXXHMNzz33HGC4QFWM+lCY65FEo1GampqSjJR4PJ5kILWndksZIuYFLDU0nJGRwa5du7jhhhvEaHM6nWRkZIix7Xa7KS0tZfHixSK6Xb58Oeeccw5r1qwBjE1r2LBhScUC29pYgeZh7VmzZkk2m/pXhak+/PBD3nvvvaQ6Sf369eOEE06QzXj79u0iNFebVWpNmnTAPM9KSko4/fTTJTQBsHTpUubPn09FRQVghPFS56aaJ2ZDVmXOKIMgEongdrtlPqnMwEQiIYZs9+7dj91AD8LixYsBwyjr2rUry5cvl/lcU1ODzWaTcI7dbicYDCYVdDQ3cIQDBqlqcAoHQmBqjquvXS5Xs9pabU2qtg6gZ8+egGGUbN++nfvuuw+AP/7xj9x99914vV5uvPFGwDism98HjD3z0Ucf5U9/+hNgJJOsX79ejLpOnToxevTogzpCUmnVWKmtreXzzz8H4Pzzz5fJpLwkiUSCbt26MX78eAAGDRrExx9/TDwe5+KLLwbgoYcewul0MnnyZACuvvpqevbsyapVq+QEVVlZySmnnCKDePDBBxkzZgy9e/cW42DVqlVtaqyYqaqqYuTIkTz++OO8+eabgJFyF4/HZSJ6PB5qa2t56623JBuosbExSXwYDoflZi5ZsgQw4pjXX3+9CLiamprYvHkz8XhcHpb2nsitoU6UlZWVrFy5km3btom3SaH0F19//TVr1qyhoaGh2WvSHdUxW7Fnzx769OkjX7ek7Uo9qavuzeaS/Kmp6y0VimtPDlU6YO/evZxwwgkSe4cDYkzFkCFDcLlcLF26VPQ5JSUlfPPNN7Jpn3feecTjcRF2AiK+bStaGuPKlSubGUxqM73nnnvE26YOY3feeSezZ8+W3+nUqRPRaJThw4fL2lFRUYHD4ZADicfjkUw6tSG2dWE889gdDgc/+9nPmD59umg3qqurk/QpwWBQDJHU90k1ZlMLKAaDQXlfm80mSQfKi97WmYKhUIhf/vKX8v+Kigo6deokxlhhYSHxeFzmtM/nEyNDGaV+vx+r1Zp0EDFXdQZj3OaWBapNh1krVF1d3S6p+6laNIvFQmVlJWA0cL366qvl2V23bh2rVq2irq4uKS3djKp+PGDAANFefvnll3Tt2lWe62uvvVY6cJs9igdDa1Y0Go1Go9GkNa0e4bKyssR1uWTJEnJzcwmFQrz99tuA0Yxs7dq1Yl35/X7OOOMMbDabeBw2btzIgAEDJO2pV69ejBw5khEjRnDiiScC8Nlnn7FhwwZp6ufz+aToljppmJXobUEgEJBQ1pQpU8RbolzAAwYMIDc3V+J6Xbt2pbCwkDFjxkh6pt1ul6whMLwmjY2NZGVlSfyvuLiYHj16JDV069y5M1VVVWJl1tfXp2X+PRw4ZWZkZLBz504CgUCzTB/V/2Xq1KkkEgkqKyvTKtvlcEmtnaE8JGaXtjlLxnzyVL9rzjxIJBLNQmap4ce2xhyCUi59cxsB5QFcv349gDzvWVlZSSGcjIwM8brt2bNHlP/qlBWLxWhsbJR5UFhYyKeffiondmj7QpBTp06VGkEFBQUUFBSwY8cOOQk+9dRTLFu2TE7CKsRrtVplrOeffz7dunVj3bp1gJHO6/V6qampkWypTp06UVtbm6RjUlmDytvy2GOPMWLEiLYbPAfWWLvdjs/no7y8XLzZXq+XSCQic9fr9Yr3wExq2X6r1SqhTTiQcWR+BlQbk/YqBvjnP/9Zavt0795ddDfqmtXzqDw/6rluqUyH8ooqj5H5WVavVfuB0gnFYjFZ3xcsWNDm9918bWaUHmXIkCHMnTuXFStWAHD//fczadKkJImDuTAmHPgcunTpwk033QQYmT52u13GquQBByt3kkqrxkpNTY0spkrk2djYKAJSMIwR9WDm5OTQu3dvfvvb38pkHDZsGHV1dXz11VcA3HDDDVx99dXs3r1bRJYDBw6ktLSUq666CjDCJaeffjqAuKJOPvnkQw7maGKuqvfVV1/xwAMPsGHDBlnILRZLUu797t27ueuuu7Db7VKxcN++fWRlZYnhFg6HCYfDbNiwQVzCl112GY888ojEPmtqaiTtU00Es/s13VAPdH19PaFQCJvNJvdOoeZQMBjEZrPhcrnSekyHg9kwAWNzPtxChqmx7XSiJTdsRkZG0lx/6aWX+P3vfw8YG5vX623m2jfXk1Hzv7y8nBNOOAEwQswnnngiJSUlgHEYUtWN1Wdi1nK1BaeeeqoUMNy+fTtr164lGo2K8TR37twkEWU8HmfRokU4HA7ZgFRFTrUpB4NBvF4vHo9H0ptDoRDdu3dPWheUEauMnrbWdM2bN49zzz0XMELuxcXFknoLRkjOXC8mGo22GJ5O1Telzu9QKITdbk9aR9Wmfajq4MeKK6+8kqVLlwLGHP3666+Tfm6xWMjNzU0SxquN2Wxwpo41Ho8nHVhUleDUJq6qkjUYYcf2MFYU+/fvx+FwsHPnTkmcUXKG4cOHA7Bs2TJuvvlm1q5dKynHn332GcOGDWtW5sBisUgizcSJEyksLEwqCLh//34KCgpED5aZmSkOilRaNVZ8Pp+chFTxp8zMTPGkZGdnc/zxxyfFdF999VVuv/122bD8fj8bNmyQ03Vubi7RaJTc3Fx5IF999VV69OghF7l48WJ2795NdXW1nGpcLpdk1bQFVqtVboQSClZUVMgNVKIoJTi77rrrcDqdfPnll3Iz+vbtm3QaURa0zWaTkus1NTX4fD7xYC1btoy6ujpsNltS2ep0RXmaVH0Vl8vVYpE/MMrtKyW52sg6KmohggMPZWuGh/q5uc5KS6ST8WK+zvfffx+ACRMmsG7dOtEVqEwPtaCDYaQ3NTXJhp2Tk8OZZ55JdXW1CGzXrFlDIBCQg05NTQ1Op5PKykrRM7X1AaV///5MnDgx6Xu7du2SdaqhoUHaSoCxPsZiMaLRqGw2eXl5eDweeQaysrLkvqvvOZ1ObDZb0r1OJBLttlmDoSdQ9OnTR55TNdZYLIbT6RRjTWX5pBokqQVCU//fUkNPu91OKBQ6bKHl0aZ37968++678rXf72f9+vWy/peVlVFWVibGptmLoNaxrl274nQ6k+q0pLaYgAOeczDmT2ZmJk6nUzZrVUCzLZk0aZIcItQ+rSIhcGA+/PSnPwXg6aefpkePHiQSCdGxbN68mcGDB7Nq1SoAqZ3idDpFRF9bW4vf7xfv4aWXXkowGExqCjx8+HAee+yxFq+zVWPF4XCIGzCRSFBVVUV9fb08mDt27BB3FxgbfGVlJVlZWdx2222AUVQrPz8/aUKrU5pyB9lsNnbt2iVF4kpKSiguLsbpdIoB0x4KcaVwVpZuIBBIegAjkYiEb8rKyvD7/USjUTHmVq9ejcPhkE3NjPKaKFGiqvoKB0IB6n3q6+uP9tCOGsqYVEXPWtuIlXg0Go12eM+KOVxhFqUdKnsn1fMSjUZFpJdaJbO92blzJ8899xwzZsyQsSohrTn05XQ6JcsDjA07JydHnu9IJMKuXbvYvHmzeFiLi4uTiiN6PB4pDHf77be39VAPSteuXSVL8f8ydrtdRL1OpzPpgAUH1iu1GSujoyVvQmoYUf1fvU9qdeMj8Uq2BTk5OQwdOrRN/+aPf/zjNv17ZoqKiqTLu9Vqpbi4mJNOOknCQNXV1cyZM0cSYlauXMnXX39NOByWez1v3jzuvPNOec/c3Fyys7Olsi8YYVOv1ytGaefOnTn55JM58cQTOfPMMwFadUhoga1Go9FoNJq05rBzJC0Wi4jOFIFAgHA4LGEKu93O2LFjyc3NZcyYMUd0IbFYTMr7pjZ8ai/mzJkDIMW/VJoVHGhMZ07XrKyslDQ3QIRjLRV/Mse+E4mEWJuq2FBOTo642NIZZVnn5ubi9/tbLfbm8/nYu3cvWVlZae0tOhwaGhqS4u6p/V9ST53meWN2gZs9ceniWRk3bhxgzP9gMEheXp6crPfv34/Vak3SsFmtVoLBoJzEBg4cyMqVKyWEvGfPHmKxGJdeeilPPfUUYJzgTj75ZBGjq7oVJSUlHX5udEROO+00KcoGB7SCZg9LLBZr5kVMFUe29LXZ29hSqnMsFksKq2ralhEjRsjeu2HDBhYtWsTChQvF219QUMAdd9whUYSbbroJr9dLp06dZP/r3r27/PxY8b0KOigR6tFQcVut1rQxUsDIdlCGmYqzq1AHGIZZfX19sz4Y5gfO4XC0GBpRlR4V5s0rMzOThoYGunTpIu91rCfB98HsFlat5Q9G586d2bRpU5tndh0LWqp22lJdldSvW4rvmz+zdNCsqIyYnTt3kp2dTSQSEUPM6XQSCoXkHjqdTjweD4WFhUndVL/55hvpkfXSSy9Jjy9FRUUFTU1NYvSoxnFmQ0jTdihxLRj6RLPmBowmpGZRrRJCm+sPHcxQac2YUc+EMlg07Uvfvn3p27ev9PJLJ9Kn+lSasWTJkmYl/lOL15gb2MXjcVG5m5sUmstPK8wPsHrgzTHdurq6pHLbbdnU67uixmMulJZaUEzpj6xWa1IhpI6IuaItHKhYacZs3B7MCFHeN0U6dBt/5plnAPjoo49Ys2YNa9asEQP6X4AAAARBSURBVOPE4/EkdcxWRvW+ffsk/RMM0bzKilNEIhExbpuamggEAixcuFC+zszMJJFI8M033xzT8Wla5p///CdgVB2/4oorsNvtIipNfZZVWm5rXpPDbRmgMqG0saJpjfZfGTUajUaj0WhaQXtWDkIoFJIePwqbzZakim9oaEg6cdjtdklZA5qdqs01G8wnbXOoSJWe9vv98hpVnC4dUSdu5Wkwe1ZSiyq5XC6pM5AOHoTvQ2Nj4yHDPkCS5w2Sw0UWi4VIJCJlvIuKito9IyIej3P++ecDyL9m6urq2Llzp+hswuEwZWVlZGZmSqFD1bjM/J6p7v+8vDzGjh0rfa8sFguZmZlUVla2eT8gjcGoUaMAozBZjx49KCgoEP1QS5lu8Xgcj8eTFApXNXaguSe6Jcx1dTr6mqA5tmhj5SAUFRUxd+5cABEBut3upM3HLIyNx+PYbDZsNpt8z9zMSpGaypcqKguFQlJIToV/zAXq0g0VwlACOVWHBpqLSfv06YPdbqeoqOigPSU6CuFwuFl12pbSklPrS6iKnnAgTKTEjKqDdXtyqA0jKyurWe8Wc0fh1t7TLEDOzc2VLs2a9secZnz22WczePBgtm7dKoaI0iOZ9WaqUrn6PdWQsCURrrnSqfp76jVOp5Pq6up2rTOjSX+0sXIQLrnkEukqPW3aNEaPHk3v3r2Tqu/BgQwOVYHRXF5Z1Y5QBkksFkuqVgnJAjMwFoBIJEI4HJaNP509K2rx8Xq9uN3uVptN9u/fH6vVyv79+9N6TAcj1RhRi6tajFNFtqkxe+VBM3dUtlgsSYZbOghsNT88Uts8uN1uLBaLHJicTidutzspcy0YDFJUVCRi3Pr6ejwejxg24XCYUChELBaT9TJVo9WjRw+i0Shutzuti19q2h9trLSC6sQ5Y8YMXnrpJbZs2SKCs2g0SlZWlrjw7XZ7Uv8XOLjATBUBM7/GXCmye/fuVFZWJr0uXVHF3VRZafNClCqYCwaD5OfnEwwG07bX0eFSXl4ulU1dLheVlZVJLnB1T83hvVRvWm5uLo2NjezcuVPeVxsrmvbALPK3WCwMGjSIsrIyqd6d2h3c5/MRCASYPHkyF1544ff626lp/xpNS+ggoUaj0Wg0mrRGm7MHwRzDVQ0W9+7dKyK0zMxMcnNz5USgQjnmk/HBUvtSaxNAclzfbrdTV1fH/Pnzj+UQjwqqw7Lb7SYQCCQVhUvVrKiu3dnZ2WldO+ZgmO/lOeecI56V0tJS8bApb1hTU1OzlGQlMlXu9szMTPbt25fUsE+nb2rag1StySuvvMKECROkP87GjRupqKgQzYrf7ycvL+97e1WAZl6VQ7Xt0PwwsWi3s0aj0Wg0mnRGm68ajUaj0WjSGm2saDQajUajSWu0saLRaDQajSat0caKRqPRaDSatEYbKxqNRqPRaNIabaxoNBqNRqNJa/4fgVSXTVoBM7gAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 720x504 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x504 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from tensorflow.keras.datasets import fashion_mnist, mnist\n",
    "\n",
    "(trainX, trainY), (testX, testY) = fashion_mnist.load_data()\n",
    "\n",
    "print('Training data shapes: X=%s, y=%s' % (trainX.shape, trainY.shape))\n",
    "print('Testing data shapes: X=%s, y=%s' % (testX.shape, testY.shape))\n",
    "\n",
    "\n",
    "for k in range(9):\n",
    "    plt.figure(figsize=(10, 7))\n",
    "    for j in range(9):\n",
    "        i = np.random.randint(0, 10000)\n",
    "        plt.subplot(990 + 1 + j)\n",
    "        plt.imshow(trainX[i], cmap='gray_r')\n",
    "        #plt.title(trainY[i])\n",
    "        plt.axis('off')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:26.915059Z",
     "start_time": "2021-02-13T08:27:26.906921Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "752PqjAI-N7m",
    "outputId": "2bf1802c-be0d-46ec-d7cb-e9731b0e523e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0, 1, 2, 3, 4, 5, 6, 7, 8, 9}"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Ten classes\n",
    "set(trainY)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:28.049636Z",
     "start_time": "2021-02-13T08:27:27.228648Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "hqqMY5NQFVe-",
    "outputId": "df0722f2-8a92-4469-f0ea-45282f890c30"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(60000, 28, 28, 1) (10000, 28, 28, 1) (60000,) (10000,)\n"
     ]
    }
   ],
   "source": [
    "trainX = [(image-127.5)/127.5 for image in trainX]\n",
    "testX = [(image-127.5)/127.5 for image in testX]\n",
    "\n",
    "trainX = np.reshape(trainX, (60000, 28, 28, 1))\n",
    "testX = np.reshape(testX, (10000, 28, 28, 1))\n",
    "\n",
    "print (trainX.shape, testX.shape, trainY.shape, testY.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "kxFTsuaGFVe-"
   },
   "source": [
    "# Define Generator Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:28.893118Z",
     "start_time": "2021-02-13T08:27:28.885940Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "collapsed": true,
    "id": "SaVxAgAaFVe-",
    "outputId": "77e002d7-7d65-43af-bbc3-415d0d3fb1ba"
   },
   "outputs": [],
   "source": [
    "def make_generator(random_input): \n",
    "    x = keras.layers.Dense(7*7*128)(random_input)\n",
    "    x = keras.layers.Reshape((7, 7, 128))(x)\n",
    "\n",
    "    x = keras.layers.Conv2DTranspose(filters=128, kernel_size=(4,4), strides=2, padding='same')(x)\n",
    "    x = keras.layers.BatchNormalization(momentum=0.8)(x)\n",
    "    x = keras.layers.Activation('relu')(x)\n",
    "\n",
    "    x = keras.layers.Conv2DTranspose(filters=64, kernel_size=(4,4), strides=2, padding='same')(x)\n",
    "    x = keras.layers.BatchNormalization(momentum=0.8)(x)\n",
    "    x = keras.layers.Activation('relu')(x)\n",
    "\n",
    "    x = keras.layers.Conv2DTranspose(filters=1, kernel_size=(4,4), padding='same')(x)\n",
    "    generated_image = keras.layers.Activation('tanh')(x)\n",
    "\n",
    "    generator_network = keras.models.Model(inputs=random_input, outputs=generated_image)\n",
    "    print (generator_network.summary())\n",
    "    return generator_network"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "CKXvIJKaFVe-"
   },
   "source": [
    "# Define Critic"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:29.705357Z",
     "start_time": "2021-02-13T08:27:29.695112Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "collapsed": true,
    "id": "Am8xrShkFVe-",
    "outputId": "06d6a0ca-ce5f-459a-8a1c-662bc37512ce"
   },
   "outputs": [],
   "source": [
    "def make_critic(): \n",
    "    image_input = keras.layers.Input(shape=(28, 28, 1))\n",
    "\n",
    "    x = keras.layers.Conv2D(filters=16, kernel_size=(3,3), strides=2, padding='same')(image_input)\n",
    "    x = keras.layers.LeakyReLU(alpha=0.2)(x)\n",
    "    x = keras.layers.Dropout(0.25)(x)\n",
    "\n",
    "    x = keras.layers.Conv2D(filters=32, kernel_size=(3,3), strides=2, padding='same')(x)\n",
    "    x = keras.layers.BatchNormalization(momentum=0.8)(x)\n",
    "    x = keras.layers.LeakyReLU(alpha=0.2)(x)\n",
    "    x = keras.layers.Dropout(0.25)(x)\n",
    "\n",
    "    x = keras.layers.Conv2D(filters=64, kernel_size=(3,3), strides=2, padding='same')(x)\n",
    "    x = keras.layers.BatchNormalization(momentum=0.8)(x)\n",
    "    x = keras.layers.LeakyReLU(alpha=0.2)(x)\n",
    "    x = keras.layers.Dropout(0.25)(x)\n",
    "\n",
    "    x = keras.layers.Conv2D(filters=128, kernel_size=(3,3), padding='same')(x)\n",
    "    x = keras.layers.BatchNormalization(momentum=0.8)(x)\n",
    "    x = keras.layers.LeakyReLU(alpha=0.2)(x)\n",
    "    x = keras.layers.Dropout(0.25)(x)\n",
    "\n",
    "    x = keras.layers.Flatten()(x)\n",
    "\n",
    "    c_out = keras.layers.Dense(1)(x)\n",
    "\n",
    "    critic_network = keras.models.Model(inputs=image_input, outputs=c_out)\n",
    "\n",
    "    print (critic_network.summary())\n",
    "    return critic_network"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Define Wasserstein Loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:31.580870Z",
     "start_time": "2021-02-13T08:27:31.577922Z"
    },
    "collapsed": true,
    "id": "EDfo3viEnFkc"
   },
   "outputs": [],
   "source": [
    "# custom loss function\n",
    "def wasserstein_loss(y_true, y_pred):\n",
    "    return keras.backend.mean(y_true * y_pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Define Gradient Penalty Loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:32.919557Z",
     "start_time": "2021-02-13T08:27:32.912798Z"
    },
    "collapsed": true,
    "id": "Eq_xDgolOR_B"
   },
   "outputs": [],
   "source": [
    "# This cell code is taken from: https://github.com/eriklindernoren/Keras-GAN/blob/master/wgan_gp/wgan_gp.py\n",
    "import keras\n",
    "from functools import partial\n",
    "import keras.backend as K\n",
    "from keras.layers.merge import _Merge\n",
    "\n",
    "class RandomWeightedAverage(_Merge):\n",
    "    \"\"\"Provides a (random) weighted average between real and generated image samples\"\"\"\n",
    "    def _merge_function(self, inputs):\n",
    "        alpha = K.random_uniform((32, 1, 1, 1))\n",
    "        return (alpha * inputs[0]) + ((1 - alpha) * inputs[1])\n",
    "\n",
    "def gradient_penalty_loss(y_true, y_pred, averaged_samples):\n",
    "    \"\"\"\n",
    "    Computes gradient penalty based on prediction and weighted real / fake samples\n",
    "    \"\"\"\n",
    "    gradients = K.gradients(y_pred, averaged_samples)[0]\n",
    "    # compute the euclidean norm by squaring ...\n",
    "    gradients_sqr = K.square(gradients)\n",
    "    #   ... summing over the rows ...\n",
    "    gradients_sqr_sum = K.sum(gradients_sqr,\n",
    "                              axis=np.arange(1, len(gradients_sqr.shape)))\n",
    "    #   ... and sqrt\n",
    "    gradient_l2_norm = K.sqrt(gradients_sqr_sum)\n",
    "    # compute lambda * (1 - ||grad||)^2 still for each single sample\n",
    "    gradient_penalty = K.square(1 - gradient_l2_norm)\n",
    "    # return the mean as loss over all the batch samples\n",
    "    return K.mean(gradient_penalty)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Define Final Critic and WGAN-GP Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:33.302788Z",
     "start_time": "2021-02-13T08:27:33.293223Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "collapsed": true,
    "id": "tMc1CyNNFVe-",
    "outputId": "5a4583f0-8b19-4726-d475-d0f9e3921c64"
   },
   "outputs": [],
   "source": [
    "def make_critic_model():\n",
    "    real_img = keras.Input(shape=(28, 28, 1))\n",
    "    random_input = keras.layers.Input(shape=(100,))\n",
    "    \n",
    "    generator_network = make_generator(random_input)\n",
    "    critic_network = make_critic()\n",
    "    \n",
    "    generator_network.trainable=False\n",
    "    critic_network.trainable=True\n",
    "    \n",
    "    fake_img = generator_network(random_input)\n",
    "\n",
    "    interpolated_img = RandomWeightedAverage()([real_img, fake_img])\n",
    "    c_out_interpolated = critic_network(interpolated_img)\n",
    "\n",
    "    partial_gp_loss = partial(gradient_penalty_loss,\n",
    "                              averaged_samples=interpolated_img)\n",
    "    partial_gp_loss.__name__ = 'gradient_penalty' # Keras requires function names\n",
    "\n",
    "    fake = critic_network(fake_img)\n",
    "    real = critic_network(real_img)\n",
    "\n",
    "    critic_model = keras.models.Model(inputs=[real_img, random_input], outputs=[real, fake, c_out_interpolated])\n",
    "    \n",
    "    # compiling model\n",
    "    optimizer = keras.optimizers.RMSprop(lr=0.00005)\n",
    "\n",
    "    critic_model.compile(loss=[wasserstein_loss, wasserstein_loss, partial_gp_loss], \\\n",
    "                         optimizer=optimizer, metrics=['accuracy'], loss_weights=[1,1,10])\n",
    "    \n",
    "    print (critic_model.summary())\n",
    "    return critic_model, critic_network, generator_network\n",
    "\n",
    "def make_wgan_gp(critic_network, generator_network):\n",
    "\n",
    "    #Defining Combined Model\n",
    "    critic_network.trainable=False\n",
    "    generator_network.trainable=True\n",
    "\n",
    "    noise_input = keras.layers.Input(shape=(100,))\n",
    "    \n",
    "    g_output = generator_network(noise_input)\n",
    "    c_output = critic_network(g_output)\n",
    "\n",
    "    wgan_gp_model = keras.models.Model(inputs = noise_input, outputs = c_output)\n",
    "    \n",
    "    # compiling model\n",
    "    optimizer = keras.optimizers.RMSprop(lr=0.00005)\n",
    "\n",
    "    wgan_gp_model.compile(loss=wasserstein_loss, optimizer=optimizer)\n",
    "    \n",
    "    print (wgan_gp_model.summary())\n",
    "    \n",
    "    return wgan_gp_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:36.840186Z",
     "start_time": "2021-02-13T08:27:35.579045Z"
    },
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_2 (InputLayer)         (None, 100)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 6272)              633472    \n",
      "_________________________________________________________________\n",
      "reshape_1 (Reshape)          (None, 7, 7, 128)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_1 (Conv2DTr (None, 14, 14, 128)       262272    \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 14, 14, 128)       512       \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 14, 14, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_2 (Conv2DTr (None, 28, 28, 64)        131136    \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 28, 28, 64)        256       \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 28, 28, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_3 (Conv2DTr (None, 28, 28, 1)         1025      \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 28, 28, 1)         0         \n",
      "=================================================================\n",
      "Total params: 1,028,673\n",
      "Trainable params: 1,028,289\n",
      "Non-trainable params: 384\n",
      "_________________________________________________________________\n",
      "None\n",
      "Model: \"model_2\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_3 (InputLayer)         (None, 28, 28, 1)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 14, 14, 16)        160       \n",
      "_________________________________________________________________\n",
      "leaky_re_lu_1 (LeakyReLU)    (None, 14, 14, 16)        0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 14, 14, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 7, 7, 32)          4640      \n",
      "_________________________________________________________________\n",
      "batch_normalization_3 (Batch (None, 7, 7, 32)          128       \n",
      "_________________________________________________________________\n",
      "leaky_re_lu_2 (LeakyReLU)    (None, 7, 7, 32)          0         \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 7, 7, 32)          0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 4, 4, 64)          18496     \n",
      "_________________________________________________________________\n",
      "batch_normalization_4 (Batch (None, 4, 4, 64)          256       \n",
      "_________________________________________________________________\n",
      "leaky_re_lu_3 (LeakyReLU)    (None, 4, 4, 64)          0         \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 4, 4, 64)          0         \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 4, 4, 128)         73856     \n",
      "_________________________________________________________________\n",
      "batch_normalization_5 (Batch (None, 4, 4, 128)         512       \n",
      "_________________________________________________________________\n",
      "leaky_re_lu_4 (LeakyReLU)    (None, 4, 4, 128)         0         \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 4, 4, 128)         0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 2049      \n",
      "=================================================================\n",
      "Total params: 100,097\n",
      "Trainable params: 99,649\n",
      "Non-trainable params: 448\n",
      "_________________________________________________________________\n",
      "None\n",
      "Model: \"model_3\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_2 (InputLayer)            (None, 100)          0                                            \n",
      "__________________________________________________________________________________________________\n",
      "input_1 (InputLayer)            (None, 28, 28, 1)    0                                            \n",
      "__________________________________________________________________________________________________\n",
      "model_1 (Model)                 (None, 28, 28, 1)    1028673     input_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "random_weighted_average_1 (Rand (None, 28, 28, 1)    0           input_1[0][0]                    \n",
      "                                                                 model_1[1][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "model_2 (Model)                 (None, 1)            100097      random_weighted_average_1[0][0]  \n",
      "                                                                 model_1[1][0]                    \n",
      "                                                                 input_1[0][0]                    \n",
      "==================================================================================================\n",
      "Total params: 1,128,770\n",
      "Trainable params: 99,649\n",
      "Non-trainable params: 1,029,121\n",
      "__________________________________________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "critic_model, critic_network, generator_network = make_critic_model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:38.066358Z",
     "start_time": "2021-02-13T08:27:37.712944Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_4\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_4 (InputLayer)         (None, 100)               0         \n",
      "_________________________________________________________________\n",
      "model_1 (Model)              (None, 28, 28, 1)         1028673   \n",
      "_________________________________________________________________\n",
      "model_2 (Model)              (None, 1)                 100097    \n",
      "=================================================================\n",
      "Total params: 1,128,770\n",
      "Trainable params: 1,028,289\n",
      "Non-trainable params: 100,481\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "wgan_gp_model = make_wgan_gp(critic_network, generator_network)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "5RpVWMO8FVe_"
   },
   "source": [
    "# Define Data Generators"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T08:27:40.225489Z",
     "start_time": "2021-02-13T08:27:40.213666Z"
    },
    "collapsed": true,
    "id": "pTSivi-dFVe_"
   },
   "outputs": [],
   "source": [
    "indices = [i for i in range(0, len(trainX))]\n",
    "\n",
    "def get_random_noise(batch_size, noise_size):\n",
    "    random_noise_batches = np.random.normal(0, 1, (batch_size, noise_size))\n",
    "    return random_noise_batches\n",
    "\n",
    "def get_fake_samples(generator_network, batch_size, noise_size):\n",
    "    random_noise_batches = get_random_noise(batch_size, noise_size) \n",
    "    fake_samples = generator_network.predict_on_batch(random_noise_batches)\n",
    "    return fake_samples\n",
    "\n",
    "def get_real_samples(batch_size):\n",
    "    random_indices = np.random.choice(indices, size=batch_size)\n",
    "    real_images = trainX[np.array(random_indices),:]\n",
    "    return real_images\n",
    "\n",
    "def show_generator_results(generator_network):\n",
    "    for k in range(9):\n",
    "        plt.figure(figsize=(9, 6))\n",
    "        random_noise_batches = get_random_noise(10, noise_size) \n",
    "        fake_samples = generator_network.predict_on_batch(random_noise_batches)\n",
    "        for j in range(9):\n",
    "            i = j\n",
    "            plt.subplot(990 + 1 + j)\n",
    "            plt.imshow(fake_samples[i,:,:,-1], cmap='gray_r')\n",
    "            plt.axis('off')\n",
    "        plt.show()\n",
    "    return"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "fEOpY_BZFVe_"
   },
   "source": [
    "# Training W-GAN-GP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-13T09:53:34.336991Z",
     "start_time": "2021-02-13T08:27:57.589987Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "E_nihCaQFVe_",
    "outputId": "92e5b5c0-7fab-4d51-a70f-2363e79b44a8",
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "epochs = 500\n",
    "batch_size = 32\n",
    "steps = 500\n",
    "noise_size = 100\n",
    "\n",
    "for i in range(0, epochs):\n",
    "    for j in range(steps):\n",
    "        # With Number of Critics=5\n",
    "        for _ in range(5):\n",
    "            random_noise_batch = get_random_noise(batch_size, noise_size)\n",
    "            real_samples = get_real_samples(batch_size=batch_size)\n",
    "\n",
    "            fake_y = np.ones((batch_size, 1))\n",
    "            real_y = -np.ones((batch_size, 1))\n",
    "            dummy = np.zeros((batch_size, 1))\n",
    "\n",
    "            # Updating Critic weights            \n",
    "            loss_c = critic_model.train_on_batch([real_samples,random_noise_batch], [real_y, fake_y, dummy])\n",
    "        \n",
    "        noise_batch = get_random_noise(batch_size, noise_size)\n",
    "        wgan_input = noise_batch\n",
    "        \n",
    "        # Make the Discriminator belive that these are real samples and calculate loss to train the generator\n",
    "        wgan_output = -np.ones((batch_size, 1))\n",
    "        \n",
    "        # Updating Generator weights\n",
    "        loss_g = wgan_gp_model.train_on_batch(wgan_input, wgan_output)\n",
    "        \n",
    "        if j%10 == 0:\n",
    "            print (\"Epoch:%.0f, Step:%.0f, C-Loss:%.3f G-Loss:%.3f\"%(i,j,loss_c[0],loss_g))\n",
    "            if j%100==0:\n",
    "                show_generator_results(generator_network)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "id": "OtkUT-YzKOFU"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true,
    "id": "KIqucdSLnFkf"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "include_colab_link": true,
   "name": "Generative Adversarial Network.ipynb",
   "provenance": []
  },
  "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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}