v2_ft_multi_mnist_test.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Improved CNN for FT of MNIST data. Also padded size."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "cellView": "both",
    "colab": {
     "autoexec": {
      "startup": false,
      "wait_interval": 0
     }
    },
    "colab_type": "code",
    "id": "apJbCsBHl-2A"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/mathew/anaconda/envs/egpu/lib/python2.7/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n",
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "#Choose GPUs and CPUs\n",
    "\n",
    "#If want to choose GPU to use\n",
    "#import os\n",
    "#os.environ[\"CUDA_VISIBLE_DEVICES\"]=\"1,4\"\n",
    "#This will make it so that only the 0th GPU is visible to TensorFlow.\n",
    "#In your case, you can choose any in the range [0, 3].\n",
    "#If you wanted, for example, the 0th and 2nd GPUs to both be visible, replace \"0\" with \"0,2\"\n",
    "\n",
    "import tensorflow as tf\n",
    "import keras\n",
    "from keras import backend as K\n",
    "from keras.datasets import mnist\n",
    "\n",
    "\n",
    "GPU=True\n",
    "CPU=False\n",
    "num_cores = 8\n",
    "\n",
    "if GPU:\n",
    "    num_GPU = 1\n",
    "    num_CPU = 2\n",
    "if CPU:\n",
    "    num_CPU = num_cores\n",
    "    num_GPU = 0\n",
    "\n",
    "config = tf.ConfigProto(intra_op_parallelism_threads=num_cores,\\\n",
    "        inter_op_parallelism_threads=num_cores, allow_soft_placement=True,\\\n",
    "        device_count = {'CPU' : num_CPU, 'GPU' : num_GPU},\\\n",
    "#        gpu_options=tf.GPUOptions(per_process_gpu_memory_fraction=0.5))\n",
    "                       )\n",
    "config.gpu_options.allow_growth = True\n",
    "\n",
    "session = tf.Session(config=config)\n",
    "K.set_session(session)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from __future__ import print_function\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "from six.moves.urllib.request import urlretrieve\n",
    "from six.moves import cPickle as pickle\n",
    "\n",
    "\n",
    "from keras.layers import Conv2D, MaxPool2D\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout, Activation, Flatten\n",
    "from keras.utils import to_categorical\n",
    "\n",
    "from tqdm import tqdm_notebook, tnrange\n",
    "\n",
    "import cv2\n",
    "from scipy.fftpack import fftn,fftshift\n",
    "\n",
    "\n",
    "# Config the matplotlib backend as plotting inline in IPython\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load MNIST data from Keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train_dataset shape: (60000, 28, 28, 1)\n",
      "60000 train samples\n",
      "10000 test samples\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# input image dimensions\n",
    "img_rows, img_cols = 28, 28\n",
    "num_classes=10\n",
    "\n",
    "# the data, split between train and test sets\n",
    "(x_train, y_train), (x_test, y_test) = mnist.load_data()\n",
    "\n",
    "if K.image_data_format() == 'channels_first':\n",
    "    x_train = x_train.reshape(x_train.shape[0], 1, img_rows, img_cols)\n",
    "    x_test = x_test.reshape(x_test.shape[0], 1, img_rows, img_cols)\n",
    "    input_shape = (1, img_rows, img_cols)\n",
    "else:\n",
    "    x_train = x_train.reshape(x_train.shape[0], img_rows, img_cols, 1)\n",
    "    x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1)\n",
    "    input_shape = (img_rows, img_cols, 1)\n",
    "\n",
    "train_dataset = x_train.astype('float32')\n",
    "test_dataset = x_test.astype('float32')\n",
    "train_dataset /= 255\n",
    "test_dataset /= 255\n",
    "print('train_dataset shape:', x_train.shape)\n",
    "print(x_train.shape[0], 'train samples')\n",
    "print(x_test.shape[0], 'test samples')\n",
    "\n",
    "# convert class vectors to binary class matrices\n",
    "train_labels = keras.utils.to_categorical(y_train, num_classes)\n",
    "test_labels = keras.utils.to_categorical(y_test, num_classes)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Pad numpy array \n",
    "Want get 36x36 which is how much the multiMNIST will be"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(60000, 36, 36, 1)\n",
      "(10000, 36, 36, 1)\n"
     ]
    }
   ],
   "source": [
    "train_dataset=np.pad(train_dataset, ((0,0),(4,4),(4,4),(0,0)), 'constant', constant_values=(0,0))\n",
    "print (train_dataset.shape)\n",
    "\n",
    "test_dataset=np.pad(test_dataset, ((0,0),(4,4),(4,4),(0,0)), 'constant', constant_values=(0,0))\n",
    "print (test_dataset.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# FT the training and test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "6edfc4361caf4487976e74903c58ec48",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=60000), HTML(value=u'')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1a49a612b2b6404d80ed365723bc5ec5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=10000), HTML(value=u'')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(60000, 36, 36, 1)\n",
      "(10000, 36, 36, 1)\n"
     ]
    }
   ],
   "source": [
    "dims=train_dataset.shape\n",
    "ft_train=[]\n",
    "for i in tnrange(dims[0]):\n",
    "    ft_train.append(np.abs(fftshift(fftn(train_dataset[i,:,:]))))\n",
    "\n",
    "dims=test_dataset.shape\n",
    "ft_test=[]\n",
    "for i in tqdm_notebook(range(dims[0])):\n",
    "    ft_test.append(np.abs(fftshift(fftn(test_dataset[i,:,:])))) #Don't forget to take abs!\n",
    "\n",
    "ft_train=np.asarray(ft_train)\n",
    "ft_test=np.asarray(ft_test)\n",
    "print (ft_train.shape)\n",
    "print (ft_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.colorbar.Colorbar at 0x1561adf50>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfQAAADtCAYAAABJeIdtAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzt3X+QZWV95/H35/b8AgYYZgYnI7CAgY1FWStkJ0TLVBYhZkm0AqmyKHU3SzZsxt3SFFmtXYl/RMyPLUzFEKpikYxKGGtVYFEXitUYQrBYUwYZEBEhUSCwzmSYAZnBGYWZ7r7f/eOelp4+33PvuT/6/urPq+pW93363HOfc2/3ffqc8znfRxGBmZmZTbbGqDtgZmZm/fOAbmZmNgU8oJuZmU0BD+hmZmZTwAO6mZnZFPCAbmZmNgU8oJuZmU0BD+hmZmZTwAO6mZnZFFjVz4MlXQrcAMwAn4iI69otv0ZrYx0n9POUZivCIQ48HxGnjrofZjY5eh7QJc0AHwPeAuwGHpB0Z0Q8VvWYdZzAz+qSXp/SbMX4m7j9mVH3wcwmSz+H3C8EnoiIpyLiKHALcNlgumVmZmbd6GdAPw343qL7u4u2Y0jaLmmXpF2zHOnj6czMzKzKsofiImJHRGyLiG2rWbvcT2dmZrYi9TOg7wHOWHT/9KLNzMzMhqyfAf0B4FxJZ0taA7wDuHMw3TIzM7Nu9Jxyj4g5Se8FvkzrsrWbIuLbA+uZmZmZ1dbXdegR8UXgiwPqi5mNULd1JcxWun/75hPi+y/Mt13mwUeOfDkiLh1Gf/oa0M1sOnRbV2JFFInSYBesvToAdbX0sSLqLzroJeuvcCJ0KvD0/Avz3P/l09uuY/XWJzcPvGMVPKCbGSyqKwEgaaGuRDqg91Ukqp/BCkDDqVitRs1+Zv1JHqtutrvRxzY2m7UXjWzwbyZtUW+dkT12OdTsT/Xj6/Wzc4GnYL7fvgyQB3Qzg7yuxM8uXkDSdmA7wDqOH17PzMZUAHO0P+Q+TB7QzayWiNgB7AA4SRun7OCqWfeCYL6LUxzLzQO6mUEvdSWWHkJODj3XPmzdhfSwbt3DnhWH69N+Lseh9JpivuZeX83XovJQePa61R2g6m73cpwiydbZz7b0oTlGwQEP6GYGi+pK0BrI3wG8a7RdMhtvAczic+hmNkZcV8KsewE+5G5m48d1Jcy6Nz775x7QzczMehIRHPUeuplNNIFmZpb9aWpf17wcgbwsZJVk1aJu6KvP/qThu5lyW0QS3Fv+t6qlm+vQ64b3hhHcgzxo1yGbGHgP3czMbOIFYjb5B2pUPKCbmZn1aL67or7LygO6mZlZDwIP6GZmZhMvgNkYztwCdXhAN7PuRRfVzEqP7TMVnAWd6gbTqpYbdEW7LNxV8XrVDf5Fn9XwBm45JiXJ+p62Df6pe9meQMzjAd3MzGziNR2KMzMzm2yBOBrDuiawMw/oZmZmPWhdh+5D7mZmZhMtYor20CU9DRyiVU9nLiK2DaJTZmZmk6A5ZZetvTkinh/AesxskvSapu6mGmdWXrbunOSNPg+FNvtIca+uv9eW9j27EiDpT8xnZVGz5bq4IqHfKwbGSHflf5P3rMOvQOs69P5fB0kbgE8ArytW+xvAPwK3AmcBTwNXRMSBdusZ/3fEzMxsDLVKv65qe6vpBuCvIuK1wOuBx4FrgHsi4lzgnuJ+W/0O6AH8taQHJW3PFpC0XdIuSbtmOdLn05mZmY2P+VDbWyeSTgZ+HvgkQEQcjYiDwGXAzmKxncDlndbV7yH3n4uIPZJeBdwt6R8i4r7FC0TEDmAHwEnaOD7zzJmZmfWhZmGZzZJ2Lbq/oxgXF5wNPAf8paTXAw8CVwNbImJvscyzwJZOT9TXgB4Re4qv+yV9AbgQuK/9o8zMzCZfq/Rrx2H0+Q6B8VXATwO/FRH3S7qBJYfXIyIkddwh7nlAl3QC0IiIQ8X3vwj8Xq/rM7PRGtpVK12U2Iy5enNhp590Sdiscg73bD71mWTPKw3pJctlgbqqkrdJYC3m5sptNUvEZtuoNX1eWlVznvLa85kPSXRTnbiHkF9Q77B6B7uB3RFxf3H/dloD+j5JWyNir6StwP5OK+pnD30L8IUiobkK+ExE/FUf6zOz0fNVK2Y1RdTaQ++wjnhW0vck/VRE/CNwCfBYcbsSuK74ekendfXck4h4ilYaz8zMbAXSoK5D/y3g05LWAE8B/5FWaP02SVcBzwBXdFqJK8WZ2YKFq1YC+IslwR2KK1m2A6zj+BF0z2y8BDA/gOlTI+JhIDvFdUk36/GAbmYL2l614itWzI7Vug59Skq/mtn06OqqFakcvhr0nOJUVFFbBpEF7bJg2tHZ8oObSfKqkQTTVucft2mIbe3aclvd8F2mai72upXmxizsVltF0C2tIJctWyNUN07zoY9PT8xsZCSdIOnEhe9pXbXy6Gh7ZTbeAmhGo+1tmLyHbmbgq1bMuuZD7mY2dnzVillv5qdstjUzM7MVJ0JDP6zejgd0M+teRHlKzppV3bIKbq32upXZ+tgjqqq2lgW80upxyWMrwm51xdGjSXfqhdDSyndZyK7iNU+r4WXRqux9zCrcdROe62f61WydaSW+/LmjmT13N2XlivWAD7nb4Mz81Dmlttf8z++ly/7ZafeX2rZ96L+ky276+Nf665iZ2dTTQK5DHxQP6GZmZj3wHrqZmdkUCESz/8lZBsYDupmZWY+aY1TOxQO6mXVNjQaNE5bUc8+qlmXBqappRDNZyKpuGCuRVghr/aB+n2r0p5vn0Kryx3AadssqxdVVVVEuey2T6nGlAGQ3unlta1af06rV5ba6091C+lqmwcEX2/cjAmbTgN1oeEA3MzPrQeuQuwd0G5B9/+bUUtvtWz+TLpuGNzzFhplZz1xYxszMbMIFYq7plLuZmdnEa3oP3cwm3pKgVBqcSkNXFQGrLCyXTUOaBJ0aa5LzmN2EyJLQWO1wWN0qc42KqmVZhb1sWtNM3ap5NUODrQ4lFemWYWrcNMiYVuerORVtFmrr4negl+BfKxTnPXQzM7OJNnHXoUu6CXgbsD8iXle0bQRuBc4CngauiIgDy9dNq5T8Lq1OC05DI1n4ax/+s3TZt33iX/fVLTOzlWAQh9wlPQ0colVQfi4itvUyztY5HnEzcOmStmuAeyLiXOCe4r6ZmdmKEcBcc6btrQtvjojzI2Jbcb/rcbbjgB4R9wEvLGm+DNhZfL8TuLx2l83MzKZBtA65t7v1oetxttcr4rdExN7i+2eBLVULStouaZekXbMc6fHpzMzMxkvQOuTe7gZsXhgDi9v2ilX9taQHF/289ji7oO9QXESEpMoIZUTsAHYAnKSNLmNiNkKDysREMh96Wjpz7dpyHyqSx1VzdmfPXZKl6Wfnym1zs/lzr1lTbltX7ntjTbnkaJbE72rO9rpziM8l25Ml8bP51au2O0uQZ69FUp62sqzqUhXp8SxVHsk2Zo9vvvRSebkuSsymqf0eyv+2Drl3fNzziw6jV/m5iNgj6VXA3ZL+4Zjn6TDOLuh1D32fpK0Axdf9Pa7HzIbrZpyJMRuYQRxyj4g9xdf9wBeAC+lhnO11QL8TuLL4/krgjh7XY/2K8m025tNbkyjdqpa16eRMjNngBGIuGm1vnUg6QdKJC98Dvwg8Sg/jbJ3L1j4LXETrPMBu4EPAdcBtkq4CngGu6NhrMxtXtc7VFef2tgOs4/hsEbOVJRjEdehbgC8Up5xWAZ+JiL+S9ABdjrMdB/SIeGfFjy6p318zmwTtztUdk4dpbHIexla8oP8BPSKeAl6ftH+fLsdZV4ozs32StkbE3rrn6iSVwlN1Q21V0rDbbDnMlQWnsjYlgbzG8fmRhbSUaCYL2s2Xr96pG+6CvARqGlbL5vteXQ7pNY5bV17uxPXpc2fBtux9rBtE5GjyfiUhPYBmsizN8jrT+eKz4F4W0uvzd5K86z/WmpxlfKZPHZ+emNmoOBNj1qMItb0Nkwd0sxWkyMR8DfgpSbuL83PXAW+R9F3gF4r7ZlZDjevQh8aH3Cddn7Xcq5Y98H/OTdtPeet36/fNxo4zMWaDEwHzY3TI3QO6mZlZTyZstjUzs6Wi2aR5+PAxbWkoqW41MahdMS2dRzur1pYtVxVMy0JfNefXzgNs9be7duAsm7M9qwqXPUnVfOh153LPAnk15ynXccelTz1zQs1LH+v+XmRzyFe9311UEWwn8B66mZnZ5IvWYfdx4QHdzMysR8MOvrXjAX3SZZfuVpRuzQJwVcsO+3ILM7NJE8iH3M3MzKaBD7mb2UTTTIOZ9Sf29Ng0BAZpVbjm0aRyWDL1ZVo5LJvqtGqKzCxklQWnXi5XhWsmwatGMvVqVrkOKqYmrTn9qrLQVxYOq5I9TxIwTENxdauwVbzfaYgtqRSXBveyoFwSGqyShvey35fD5aalxulopgd0MzOzHvg6dDMzsynhQ+5mZmZTwIfcbXCWqfRrxQyaZmZWCFeKM7OJ14xylbKkilr+2Dy8lIXlGllQKXueZJ3Nwz8sP0dFxbRsylElU602NibBtrpTr1ZULeNIUu3tRy+V25LHp2HAteU2suBdleR9iGyq09mk39n0qUm4sPWDelXq6gYEs6Bc9pq1niZZZzdVDX/8BN5DNzMzmwrR9IBuZmY28cYpFDc+eXszM7MJErQOube71SVpRtI3JN1V3D9b0v2SnpB0q6TkXMqxOg7okm6StF/So4varpW0R9LDxe2Xa/fazMxsGkTrkHu7WxeuBh5fdP8jwPURcQ5wALiq0wrqHHK/Gfgz4FNL2q+PiD+u109bLlvu3V9qu/o/vSld9oZX/12pzbXcVxZJNwFvA/ZHxOuKtmuB3wSeKxb7YER8sd16gnKlrzhSrqKWTsVZESJLpxxNqpGlyyXrbKxOAnXdTGs6l1SKO3SovGASxtJJ5Sp6zQ15Zb350zeV2uZOKG9Pc1XyN5kc7p05Wg6bzfwoD6bNHCq/Z43DPyovmIXisoBh8n5lgUMAksBjWjWvbqCv7jS0QCRVCbMqgPWet7eHLSbpdOCtwB8C75Mk4GLgXcUiO4FrgRvbrafjHnpE3Ae80E9nzWxs3AxcmrRfHxHnF7e2g7mZLWh/uL3YMdosadei2/ZkRX8K/Hdg4b+xTcDBiFj4T2w3cFqn3vQTinuvpP8A7ALeHxEHsoWKzm8HWEfNCe3NbFlExH2Szhp1P8ymQtRKuT8fEduqfihp4YjZg5Iu6qc7vYbibgR+Ejgf2At8tGrBiNgREdsiYttq8skJzGzk3ivpkSIzc0q2gKTtC3sZs/HysPtnNp6iw62zNwG/Iulp4BZah9pvADZIWtjpPh3Y02lFPQ3oEbEvIuYjogl8HLiwl/WY2Vio9Q/6Mf+cq+K8qNlKE2p/6/TwiN+JiNMj4izgHcDfRsS/A+4F3l4sdiVwR6d19XTIXdLWiNhb3P1V4NF2y9vymf/Ok6W2XfvPTZdtvNqlX60sIvYtfC/p48BdI+yO2WRZvo/KDwC3SPoD4BvAJzs9oOOALumzwEW0TuzvBj4EXCTpfFqb8jTw7t77bGaj1Ms/6JLQ0jm/X0rmFM8S4BVJ83S+8LpJ6ESaUv9BklIHmofLE1/PbNpYXvBVm8uPPbmcDZpfl81xnj41SlLp63b/oLzg/u+X27L0+U+cWmqa27w+fe4jry63R6Ocxm/Mlvu46lC59OvMgXK5XZ5L+g00Xyy/F42NG0pt2VURUbPcruYqSr/OZiWFk9/LTvOhB7X2wuuKiK8AXym+f4ouj353fFUi4p1Jc8f/FMxs/PgfdLPBykrSj4pLv5qtIP4H3WzAxqhmhwd0MzOzHo1T3MgDupmZWS9C4NnWbDlVlW1tJnFMl361XkREuVRmUua1kYTD6obagDT0FS+Vr4Evzc1OHopLg3fAzMbk0vtsXvC95VLLjeeSsrOvflWp7YevOTl97hfPLj/+8JnHldd5WvlSwVdvLIfnNq4rt81FElYDnj1cDsA9t/+kUtua3eV5QU58ptx28j+V29bN5yeZlQX6srbk90XNZJ1ZKdqkxCuMX+nXQfGAbmZm1iuH4szMzCbcgC9b65cHdDMzsx45FGdmZjYNPKDbcqoq29rApV9tQCKIirmmj1lsLg8lZbIKcmmALpnnXGvKYaysLZs3HYBTyoG1o2eUg3IHX1MO1f3gnPLqVp1broJ28ZnfTJ/6V055qNT2M2tfLHdxplyR7sB8ee7yB46Ut+XOAz+dPvcTz5Ur361+NgnAPV1+7IanyiGyNd8rT7oZLyZV76h4b5PfqWZW3S8LtWWVAWv8jr7Sod7mKhunj0oP6GZmZr3yOXQzM7MJFzjlbmZmNg18yN3MzGwaeA/dzCaaVAqxKZvqdOkUq4CSUBuQht1oJOcnkykx40cvlduOlENbjc3JlKjAy2eW2188uxwOO3Rm+bGzP1GuUrfhuPJz/78fJdXogD9/6aJS2/94+YRS2z+/UK7g1txTDsqtf6Yc7jr5n8qBMYCzniqH7/jnfyg1pVX3kvc7U1WBLXt/sqBcGm48rlxJLz2TXRGKSwN0s8lr1CFTp/Aeui0zl341MxuSMfqs9IBuZmbWI/mQu5mZ2RQYo0PuvV1Jb2ZmttLFK+fRq26dSFon6euSvinp25I+XLSfLel+SU9IulVSEiY4lvfQzVYQSWcAnwK20Nq32BERN0jaCNwKnAU8DVwREeWyX6+spxR407ry9J5ZKK6qWlusyqsWltbZSMJLWcjpcHnK0Hgh36S1M+V9m02HysG0bHrQ5uryY9UsV2t7+XA5wAaw6vnDpbbjn32u1HbOmqRi2qs2lZrmTi4HxmJNvu82u7Hcp5njTy+1NV4sV6TjhYOlpvnvv1B+7Pr16XM3NpWDiHWn1s1CbelUu9lyQDMJ6ikLYNbR/yH3I8DFEXFY0mrgq5K+BLwPuD4ibpH058BVwI3tVtRxD13SGZLulfRY8d/D1UX7Rkl3S/pu8TWPcJrZOJkD3h8R5wFvAN4j6TzgGuCeiDgXuKe4b2Yd9LuHHi0L/9WtLm4BXAzcXrTvBC7vtK46/w4tfAA8JOlE4EFJdwO/TusD4DpJ19D6APhAjfXZMtv4tu+k7Y099Wu5//0Ft6Tt5/zFu0tt//LdD3TROxuliNgL7C2+PyTpceA04DLgomKxncBX8N+z2SBslrRr0f0dEbFj8QKSZoAHgXOAjwFPAgcjYuEQw25af6dtdRzQ/QFgNp0knQVcANwPbCn+1gGepXVIfuny24HtAOtUPhxttuJErZT78xGxre1qIuaB8yVtAL4AvLaX7nQViuv2A6B4zHZJuyTtmiUvMGBmwyVpPfA54Lcj4pjpsCIiSLK7EbEjIrZFxLY1Kp8vN1uRosOtm1VFHATuBd4IbJC0sNN9OrCn0+Nrh+KWfgBIrxy+jYhQxXybxaGFHQAnaeMYBfzNVqYiePM54NMR8fmieZ+krRGxV9JWYH+7dUREuQJYM9lVaZYLF1VVisumT0X1gkpKlmucfGJ5wSQ4BRD/vK/UNrOv3J+ZtUnluxPKIbTm+nLYbP7EJCAIHDljQ3nZn0wq2iUvRWOu/JG66oflINiq75cr6QE0DpaDdpFNV5oFxpLXYtWrt6bPk0qquMXLL5eXSyq4pdOiRvm16Gb61K6mWi2I/q9Dl3QqMBsRByUdB7wF+Aitgf3twC3AlcAdndZVaw+93QdA8fOOHwBmNnpqjXyfBB6PiD9Z9KM7aX1oQM0PDzNjEHvoW4F7JT0CPADcHRF30TqF/T5JTwCbaP3dttVxD73GB8B1+ANgInRT+rUqLPftt36s1ParXNhfx2yY3gT8GvAtSQ8XbR+k9Xd8m6SrgGeAK0bUP7PJMYBa7hHxCK1T2Uvbn4LuPlzrHHL3B4DZlIiIr1IxjwVwyTD7YjYVJqn0qz8AzMzMcp5tzczMbBp4QDezSSZAS8qlpnNMv1i+VDWa+SdgmnKvW44zW2ckx0KV54CVlaOt+9xJCjtLj2dtAKuz1+1IeY71OFpuy14zHV9O3bO2ogx4Nod9Nmd8ljRPyqc2D5bnV2++lCTXyUutpqVfG8l7llxREUnKPbv6ASpKEmfK1YOXPKlnWzMzM5sO3kO3UWgkUYiqNHu2bNXyR37pZ9Jl137JJWHNbLp5D93MzGzS9VANbjl5QDczM+uBqL4EbBQ8oJtZ9xpCa5YErbLSmVloq2qdSYCpdlAuXS557mTe86rnTmXlRbNysknJ2zS4VyXpp9bXmxAn5pNjwD8oz7kOEEeSYFs2V3j2PmaBuiRcOLPmpPS5s9Bi9lpmYcAsWNlI+kNVmeG673cNPuRuZmY2DXzI3czMbAp4QLdRGEQt92z5H76nfO0pwNovddE5M7NJ4+vQzczMpoNLv5rZRIv5JvOH8spni2lVveAUkFcEywJwFdXeSpJgWlbNrvWDJKCVVEdL5/DOQmhdVKmrW5EuDXJlr1kWYKuqmJZUZptZX27LqrCl1drSgGC+C5u9bnXnJM+qzOXvQzlQBxVHyXuYDx28h25mZjb5fB26mZnZlPCAbqPw/r1vKLV9dOvfp8t2U/r10EOb0mWTKR7MzKaG8CF3MzOzqaAsXzAiHtDNVghJZwCfArbQOlC4IyJukHQt8JvAc8WiH4yIL7ZdV6NBY/36Yxsrwk+1JaGkZlIlLA2mZWG3pFJcWk0MylXvAB23rrxgEjhrJIGzNERWFbrKpk/NAl6zSRW1mtOaZiEyIA/QJUG5tGJfVbhxaX+ycCGkVfeYLb9GaX+yaVaTqnB9V4TLr8h9xQDOobf5u9wI3AqcBTwNXBERB9qtq2Zc1MymwBzw/og4D3gD8B5J5xU/uz4izi9ubQdzM3uFmu1vNVT9XV4D3BMR5wL3FPfb8oButkJExN6IeKj4/hDwOHDaaHtlNtkU7W+dtPm7vAzYWSy2E7i807o6DuiSzpB0r6THJH1b0tVF+7WS9kh6uLj9cueum9k4kHQWcAFwf9H0XkmPSLpJ0ikVj9kuaZekXUfj5SH11GyMRa099M0LfzfFbXvV6pb8XW6JiL3Fj56ldUi+rTonQRYOBzwk6UTgQUl3Fz+7PiL+uMY6bAz83Y5tpbbZ3/27dNluSr+e+btf669jNlSS1gOfA347In4g6Ubg92mdv/t94KPAbyx9XETsAHYAnDyzeXySQGaj1Pkv4fmIKH/4LpH8Xb7yFBEhdd7f7zigF/8h7C2+PyTJh+nMJpSk1bQ+ND4dEZ8HiIh9i37+ceCuTuuJiPK0ltn0oFkFtm6mEU0en07luXZtua1uuAvy6VezYFpSCa2ZhdqyKT8rQnFpIC9rW1dvG9OKdDWr0QH5+5hV3as71WkWbCSvItg4/vjygl1Uvis9d1UCPQtw9lApTgym9Gv2dwnsk7Q1IvZK2grs77Sers6h93KYzszGg1qfgp8EHo+IP1nUvnXRYr8KPDrsvplNKjWj7a3j4yv+LoE7gSuL768E7ui0rtoD+tLDAcCNwE8C59Pag/9oxeN+fN5tlvLlFGY2NG8Cfg24eEn25Y8kfUvSI8Cbgf860l6aTYqoceus6u/yOuAtkr4L/EJxv61aFxL2c5hu8Xm3k7TR593MRiQivgppCUBfpmbWo34rxbX5uwS4pJt11Um5+zCdmZlZYgDXoQ9MnT30hcMB35L0cNH2QeCdks6ndVDhaeDdy9JDW1ZVafZuarnbyiOpHDBbVW96z8o4UxZUSkJNaXW0mm2VFdMGLQmmaVX9yFIaLqsbJsymbq1ctF6QMd2eLKw2kyx33HG1+5NOyZq8j80swJYE96iaNjYLQVYFJtsJ8op3I1In5e7DdGZmZolBpNwHxbXczczMeuDZ1szMzKZBxGQdcjczM7Oc99BtJLZ8pVxo6I3N96bLXvjub6TtX/+LC0ptm3DpVzNbmXwO3cwm35JUcFaOMy29WTVvejbPeVJ+NebKJUdTSTK7q7KzNdVNe1emqJMrAWqrWcI0nV8dUCNrT/qZlZNNpM+TzVVP/fciuzKhn/nZoaJ0bK8p92X4neqVB3QzM7Me+ZC7mZnZNHAozszMbPL5HLqZmdmEU1BrRrVh8YC+gsx/58lS26akDeDJT+TrcKLdAKLZpPnDH/X64Ir25IMxCy/VDGh19dxZadOsTGwfc3NXzbcdWWgsDbZlpU1773fxg4r2JfopO1vxfml1H2V4k/5kJWKr3u80rFn3fSz1pbeHLQcP6GZmZr3wHrqZmdk0cKU4MzOzqeBQnJmZ2aQL0Pz4jOge0M1WEEnrgPuAtbT+/m+PiA9JOhu4BdgEPAj8WkSUJ+V+ZUV5ta6lklBSNKtCbTXTRV3M911b1s8sw5adL81CaOn84RXhsFXJx3DSpqyiXN055KsCeVmQLJtXvJGEAQcdGuzGTLLO5Zjr/uUay4zRIfc+4qJmNoGOABdHxOuB84FLJb0B+AhwfUScAxwArhphH80mR3S4dSDpJkn7JT26qG2jpLslfbf4ekqdrnhAN1tBouVwcXd1cQvgYuD2on0ncPkIumc2cdRstr3VcDNw6ZK2a4B7IuJc4J7ifkce0M1WGEkzkh4G9gN3A08CByNi4fjrbuC05HHbJe2StGs26hyLNJtyQetMUbtbp1VE3Ae8sKT5Mlr/WEMX/2D7HLrZChMR88D5kjYAXwBeW/NxO4AdACc1No3PiUOzERFRdy+8W1siYm/x/bPAljoP8oButkJFxEFJ9wJvBDZIWlXspZ8O7Onw4Mqg1bHLjVEZrXbqVlzLqp7VDGOl1ckAjpang6312lIxtWgWqKsIMKaBvLrhu7rT3abpwoptrBsw6zdo10+1waU693mzpF2L7u8o/jmuufoIqd7FcR23StI6SV+X9E1J35b04aL9bEn3S3pC0q2S1tTtoJmNhqRTiz1zJB0HvAV4HLgXeHux2JXAHaPpodkEqXfI/fmI2LboVmcw3ydpK0DxdX+d7tT5N8WpWLPpsRW4V9IjwAPA3RFxF/AB4H2SnqB16donR9hHs4kxgFBc5k5a/1hDF/9gdzzkHq3jRFWp2HcV7TuBa4Eba3fXzIYuIh4BLkjanwIuHH6PzCZZ/6Wz/dUGAAAHsklEQVRfJX0WuIjWofndwIeA64DbJF0FPANcUWddtc6hS5qhVWziHOBj1EzFFo/dDmwHWMfxdZ7OzMxs/AV9D+gR8c6KH13S7bpqDei9pmKLx76SjNVGJ2PNplQaIqs7PScQYzRrFVT1JwlyZZmvLsKAWWAtb8sCeUlbFhirGnSy9tlySK8ura6fs05jbckh6jRMmE2fWjNIOGgTW/q1r1SsmZnZtJmk0q9OxZqZmSUiYL7Z/jZEdfbQtwI7i/PoDeC2iLhL0mPALZL+APgGTsWamdlKM0Z76HVS7k7FmpmZZSZpQDczSy0JfqUFwQZZkatby1Glru72dLHdafgukupxyUyndZ8nDyxWPL6faUizYFpV2DGdWjd7LepNEduVQU3pGpFv84h4QDczM+uV99DNzMwmXFB9BGIEPKCbmZn1anlmW+uJB3QzM7Oe9F/6dZA8oJuZmfUi8B66mU04lcuT1k4oV+lnLuxsPvNsDvBuUvf9pL37pJop7Mo51uvK3rN+38dBy96z9FdgGd6vOpvtAd3MzGzShUNxZmZmEy9GNylMxgO6mZlZr8YoFDfCMk5mNmyS1kn6uqRvSvq2pA8X7TdL+idJDxe380fdV7OxF9E6h97uNkTeQzdbWY4AF0fEYUmrga9K+lLxs/8WEbfXWkt2qHE5ynH2UTq29nzmrYVrrnSwwb1Wc3nZ6KMka91AHQAzdcN3yXI1w3OV89zXfc2zwOMyBOAq+9npcT7kbmajEK1Y9OHi7uriNj7HDM0mynhdh+5D7mYrjKQZSQ8D+4G7I+L+4kd/KOkRSddLWps8brukXZJ2zXJkqH02G0sLpV/b3YbIA7rZChMR8xFxPnA6cKGk1wG/A7wW+BlgI/CB5HE7ImJbRGxbTWm8N1txgtYh93a3YfKAbrZCRcRB4F7g0ojYGy1HgL8ELhxt78wmQEQrC9DuNkQ+h262gkg6FZiNiIOSjgPeAnxE0taI2KtWoupy4NF26znEgef/pvm/ninubgae77tz43Mqsv/t6WZblncnbjDvzfhobc/wdnzP7LSAQ3FmNipbgZ2SZmgdobstIu6S9LfFYC/gYeA/t1tJRJy68L2kXRGxbTk7PUzTtD3TtC0wfttziANf/pu4fXOHxYb2D5UHdLMVJCIeAS5I2i8eQXfMJlpEXDrqPizW8Ry6C1GYmZmNvzp76IMpRGFm02rHqDswYNO0PdO0LTB92zNQHQd0F6Iws3YiYqo+ZKdpe6ZpW2D6tmfQal221mshiuKxLkZhZma2zGoN6L0Woige62IUZmZmy6yrwjIuRGFmi0m6VNI/SnpC0jWj7k83JN0kab+kRxe1bZR0t6TvFl9PGWUf65J0hqR7JT1WhJevLtondXuqwthnS7q/+H27VdKaUfd1nCg6FJZPClH8NfAR4MFFhSiuB16OiLZ/0JKeAwZbjGJ8efsm3yi38czF13qPo+Ja9u/QKk6zG3gAeGdEPDbSjtUk6edp5YM+FRGvK9r+CHghIq4r/kE5JSLSo4/jRNJWYGtEPCTpROBBWgWCfp3J3B4BJywOYwNXA+8DPh8Rt0j6c+CbEXHjKPs6Tuqk3AdSiAKmuxjFUt6+ybcStrFPFwJPRMRTAJJuAS4DJmJAj4j7JJ21pPky4KLi+53AV6g4nThOImIvsLf4/pCkx4HTmNztqQpjXwy8q2jfCVwLeEAv1Em5uxCFmWVOA7636P5u4GdH1JdB2VIMjgDPAltG2ZleFP+kXADczwRvT7ET+SBwDvAx4EngYETMFYvspvU7aAVPzmJmlij2EifqEl1J64HPAb8dET9Y/LNJ256lYWxaIWxrY5QD+rRfT+jtm3wrYRv7sQc4Y9H904u2SbavOB+9cF56/4j7U1txrvlzwKcj4vNF88Ruz4JFYew3AhskLRxZnobft4Ea2YA+7QUCvH2TbyVsY58eAM4tksdrgHcAd464T/26E7iy+P5K4I4R9qW2IkT2SeDxiPiTRT+a1O05VdKG4vuFWQEfpzWwv71YbGK2Z1g6ptzNzKpI+mXgT4EZ4KaI+MMRd6k2SZ+lFRjbDOwDPgT8b+A24F/QuiLnioh4YVR9rEvSzwH/F/gWsDAJ9wdpnUefxO35V7RCb4vD2L8n6TXALbRqn3wD+PfFpdOGB3QzM7OpMPRD7pNciKLKNBWoyExb0YqlXMTCzKbBUAf04jKEjwG/BJwHvFPSecPswzK5GVg6L+41wD0RcS5wT3F/Us0B74+I84A3AO8p3rdp2caFGQVfD5wPXCrpDbQKKF0fEecAB4CrRthHM7O2hr2H/uNCFBFxlNa5kMuG3IeBi4j7gKXnpS6jdQ6I4uvlQ+3UABVlfh8qvj9EK5yyULRi4rexKGFcVcRiYXrgid0+M1sZhj2gZ4UoprUwwMQWdGhnWopWLKUlMwriIhZmNmFcWGYIJq2gQ5VpKlqxlItYmNmkG/aAPo2FKKpMfEGHxaa1aMVSLmJhZpNq2AP6NBaiqDKRBR0y01a0YikXsTCzaTD069AnuRBFlWkqUJGZtqIVS7mIhZlNAxeWMTMzmwIOxZmZmU0BD+hmZmZTwAO6mZnZFPCAbmZmNgU8oJuZmU0BD+hmZmZTwAO6mZnZFPj/LP9GOrej0XIAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 576x432 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "i=200\n",
    "plt.viridis()\n",
    "f,axarr=plt.subplots(1,2,figsize=(8, 6))\n",
    "axarr[0].imshow(train_dataset[i].squeeze())\n",
    "img=axarr[1].imshow(ft_train[i].squeeze())\n",
    "f.colorbar(img, fraction=0.046, pad=0.04)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# O.K now try CNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(60000, 36, 36, 1)\n",
      "(60000, 36, 36, 1)\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_67 (Conv2D)           (None, 36, 36, 32)        320       \n",
      "_________________________________________________________________\n",
      "conv2d_68 (Conv2D)           (None, 36, 36, 32)        9248      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_32 (MaxPooling (None, 18, 18, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_69 (Conv2D)           (None, 18, 18, 64)        18496     \n",
      "_________________________________________________________________\n",
      "conv2d_70 (Conv2D)           (None, 18, 18, 64)        36928     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_33 (MaxPooling (None, 9, 9, 64)          0         \n",
      "_________________________________________________________________\n",
      "conv2d_71 (Conv2D)           (None, 9, 9, 128)         73856     \n",
      "_________________________________________________________________\n",
      "conv2d_72 (Conv2D)           (None, 9, 9, 128)         147584    \n",
      "_________________________________________________________________\n",
      "max_pooling2d_34 (MaxPooling (None, 4, 4, 128)         0         \n",
      "_________________________________________________________________\n",
      "flatten_14 (Flatten)         (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_27 (Dense)             (None, 600)               1229400   \n",
      "_________________________________________________________________\n",
      "dropout_13 (Dropout)         (None, 600)               0         \n",
      "_________________________________________________________________\n",
      "dense_28 (Dense)             (None, 10)                6010      \n",
      "=================================================================\n",
      "Total params: 1,521,842\n",
      "Trainable params: 1,521,842\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "Train on 54000 samples, validate on 6000 samples\n",
      "Epoch 1/10\n",
      "53504/54000 [============================>.] - ETA: 0s - loss: 0.6204 - acc: 0.7907Epoch 00000: saving model to tmp_models/weights.00.hdf5\n",
      "54000/54000 [==============================] - 9s - loss: 0.6167 - acc: 0.7921 - val_loss: 0.1833 - val_acc: 0.9372\n",
      "Epoch 2/10\n",
      "53504/54000 [============================>.] - ETA: 0s - loss: 0.2003 - acc: 0.9353Epoch 00001: saving model to tmp_models/weights.01.hdf5\n",
      "54000/54000 [==============================] - 6s - loss: 0.1998 - acc: 0.9354 - val_loss: 0.1554 - val_acc: 0.9475\n",
      "Epoch 3/10\n",
      "53504/54000 [============================>.] - ETA: 0s - loss: 0.1515 - acc: 0.9501Epoch 00002: saving model to tmp_models/weights.02.hdf5\n",
      "54000/54000 [==============================] - 6s - loss: 0.1511 - acc: 0.9502 - val_loss: 0.1261 - val_acc: 0.9572\n",
      "Epoch 4/10\n",
      "53504/54000 [============================>.] - ETA: 0s - loss: 0.1318 - acc: 0.9558Epoch 00003: saving model to tmp_models/weights.03.hdf5\n",
      "54000/54000 [==============================] - 6s - loss: 0.1316 - acc: 0.9559 - val_loss: 0.1223 - val_acc: 0.9592\n",
      "Epoch 5/10\n",
      "53504/54000 [============================>.] - ETA: 0s - loss: 0.1091 - acc: 0.9641Epoch 00004: saving model to tmp_models/weights.04.hdf5\n",
      "54000/54000 [==============================] - 6s - loss: 0.1088 - acc: 0.9642 - val_loss: 0.1232 - val_acc: 0.9563\n",
      "Epoch 6/10\n",
      "53504/54000 [============================>.] - ETA: 0s - loss: 0.1002 - acc: 0.9662Epoch 00005: saving model to tmp_models/weights.05.hdf5\n",
      "54000/54000 [==============================] - 6s - loss: 0.1000 - acc: 0.9663 - val_loss: 0.1190 - val_acc: 0.9592\n",
      "Epoch 7/10\n",
      "53504/54000 [============================>.] - ETA: 0s - loss: 0.0827 - acc: 0.9721Epoch 00006: saving model to tmp_models/weights.06.hdf5\n",
      "54000/54000 [==============================] - 6s - loss: 0.0830 - acc: 0.9720 - val_loss: 0.1299 - val_acc: 0.9565\n",
      "Epoch 8/10\n",
      "53504/54000 [============================>.] - ETA: 0s - loss: 0.0789 - acc: 0.9724Epoch 00007: saving model to tmp_models/weights.07.hdf5\n",
      "54000/54000 [==============================] - 6s - loss: 0.0790 - acc: 0.9724 - val_loss: 0.1287 - val_acc: 0.9608\n",
      "Epoch 9/10\n",
      "53504/54000 [============================>.] - ETA: 0s - loss: 0.0678 - acc: 0.9767Epoch 00008: saving model to tmp_models/weights.08.hdf5\n",
      "54000/54000 [==============================] - 6s - loss: 0.0682 - acc: 0.9765 - val_loss: 0.1251 - val_acc: 0.9590\n",
      "Epoch 10/10\n",
      "53504/54000 [============================>.] - ETA: 0s - loss: 0.0623 - acc: 0.9785Epoch 00009: saving model to tmp_models/weights.09.hdf5\n",
      "54000/54000 [==============================] - 6s - loss: 0.0623 - acc: 0.9785 - val_loss: 0.1431 - val_acc: 0.9562\n"
     ]
    }
   ],
   "source": [
    "dims=ft_train.shape\n",
    "print (dims)\n",
    "trainee=ft_train.reshape(dims[0],dims[1],dims[2],1)\n",
    "dims=test_dataset.shape\n",
    "testee=ft_test.reshape(dims[0],dims[1],dims[2],1)\n",
    "print (trainee.shape)\n",
    "\n",
    "w_size=3\n",
    "\n",
    "np.random.seed(123)\n",
    "model=Sequential()\n",
    "\n",
    "model.add(Conv2D(32,(w_size,w_size), activation='relu', input_shape=(36,36,1),\n",
    "                     data_format='channels_last', padding='same'))\n",
    "model.add(Conv2D(32,(w_size,w_size), activation='relu', padding='same'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "model.add(Conv2D(64,(w_size,w_size), activation='relu', padding='same'))\n",
    "model.add(Conv2D(64,(w_size,w_size), activation='relu', padding='same'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "model.add(Conv2D(128,(w_size,w_size), activation='relu', padding='same'))\n",
    "model.add(Conv2D(128,(w_size,w_size), activation='relu', padding='same'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "model.add(Flatten())\n",
    "model.add(Dense(600, activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(10, activation='softmax'))\n",
    "\n",
    "#parallel_model = ModelMGPU(model, gpus=num_GPU)\n",
    "model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "\n",
    "print (model.summary())\n",
    "#print (parallel_model.summary())\n",
    "\n",
    "\n",
    "\n",
    "checkpoints=keras.callbacks.ModelCheckpoint('ft_models/weights.{epoch:02d}.hdf5', \n",
    "                                            monitor='val_loss', verbose=1, \n",
    "                                            save_best_only=False, \n",
    "                                            save_weights_only=False, \n",
    "                                            mode='auto', period=1)\n",
    "\n",
    "history=model.fit(trainee, train_labels, batch_size=256, verbose=1, epochs=10, \n",
    "                  validation_split = 0.1, callbacks=[checkpoints])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "hist=history\n",
    "epochs=np.asarray(history.epoch)+1\n",
    "\n",
    "%matplotlib inline\n",
    "import matplotlib\n",
    "plt.style.use('seaborn-white')\n",
    "matplotlib.rc('font',family='Times New Roman')\n",
    "matplotlib.rcParams['font.size'] = 20\n",
    "\n",
    "f, axarr = plt.subplots(2, sharex=True, figsize=(8, 6))\n",
    "\n",
    "axarr[0].set(ylabel='Loss')\n",
    "axarr[0].plot(epochs,hist.history['loss'], 'C3o', label='Training')\n",
    "axarr[0].plot(epochs,hist.history['val_loss'], 'C3-', label='Validation')\n",
    "axarr[0].grid()\n",
    "axarr[0].legend(loc='center right', bbox_to_anchor=(1.0, 0.5))\n",
    "\n",
    "\n",
    "axarr[1].plot(epochs,hist.history['acc'], 'C0o', label='Training')\n",
    "axarr[1].plot(epochs,hist.history['val_acc'], 'C0-', label='Validation')\n",
    "plt.legend(loc='center right', bbox_to_anchor=(1.0, 0.5))\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.tight_layout()\n",
    "plt.grid()\n",
    "plt.tick_params(which='both', width=2)\n",
    "plt.tick_params(which='major', length=7)\n",
    "plt.tick_params(which='minor', length=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 2s     \n",
      "[0.11876363634229638, 0.9621]\n"
     ]
    }
   ],
   "source": [
    "from keras.models import Model, load_model\n",
    "model=load_model('tmp_models/weights.08.hdf5')\n",
    "model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "score=model.evaluate(testee, test_labels, verbose=1)\n",
    "print (score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "preds=(model.predict(testee))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "i=100\n",
    "f,axarr=plt.subplots(1,2,figsize=(8, 6))\n",
    "axarr[0].imshow(test_dataset[i].squeeze(),cmap='viridis')\n",
    "axarr[1].imshow(testee[i].squeeze(), cmap='viridis')\n",
    "print (np.argmax(preds[i]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "default_view": {},
   "name": "1_notmnist.ipynb",
   "provenance": [],
   "version": "0.3.2",
   "views": {}
  },
  "kernelspec": {
   "display_name": "Python [conda env:egpu]",
   "language": "python",
   "name": "conda-env-egpu-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}