9. Convolutional Neural Networks.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "9. Convolutional Neural Networks.ipynb",
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/DataScienceUB/DeepLearningMaster2019/blob/master/9.%20Convolutional%20Neural%20Networks.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RBrbcD8RcGVX",
        "colab_type": "text"
      },
      "source": [
        "# 9. Convolutional Neural Networks\n",
        "\n",
        "> A Convolutional Neural Networks is a NN which has at least one layer (``tf.nn.conv2d`` in ``tf``) that does a convolution between its input $f$ and a configurable kernel $k$, $(f*k)$. \n",
        "\n",
        "Convolutional Neural Networks have been some of the most influential innovations in the field of **computer vision** in the last years. \n",
        "\n",
        "When considering the analysis of an image with a computer we must define a computational representation of an image. To this end, images are represented as $n \\times m \\times 3$ array of numbers, called **pixels**.\n",
        "The 3 refers to RGB values and $n, m$ refers to the height and width of the image in pixels. Each number in this array is given a value from 0 to 255 which describes the pixel intensity at that point.  These numbers are the only inputs available to the computer. \n",
        "\n",
        "**What is to classify an image?** The idea is that you give the computer this array of numbers and it must output numbers that describe the probability of the image being a certain class.\n",
        "\n",
        "Of course, this kind of image representation is suited to be classified by a classical neural network composed of **dense layers**, but this approach has **several limitations**.\n",
        "\n",
        "The first one is that large images with a high number of pixels will need from **extremely large networks** to be analyzed. If an image has $256 \\times 256 = 65,536$ pixels, the first layer of a classical neural network needs to have $65,536 \\times 65,536 = 4,294,967,296$ different weights to consider all pixel interactions. Even in the case that this number of weights could be stored in the available memory, learning these weights would be very time consuming. But there are better alternatives.\n",
        "\n",
        "One possible solution for this problem is to consider **local receptive fields (LRF)** (an inspiration that comes from neuroscience). A local receptive field is a standard MLP with a local view of the image: its input is a submatrix $M \\times M$ of the original $N \\times N$, where $M << N$.\n",
        "\n",
        "Using LRF is like decomposing the problem of image analysis in a set of local *experts* that are specialized in certain parts of the image. The outcome of these specialists can be integrated at a high level of the model.\n",
        "\n",
        "> Could you propose a problem where a LRF architecture could be the best option?\n",
        "\n",
        "Let's suppose we are dealing with a common task: object detection.\n",
        "\n",
        "In the most general setting, images represent the field of view of a camera (robot, autonomous car, person, etc.). Hence, we cannot assume any prior information about the presence of the object of interest on the image (i.e. any prior information about its scale, point of view, etc. or even its localization with respect to image coordinates). \n",
        "\n",
        "> An object may appear at any image coordinates!\n",
        "\n",
        "In this case LRF are not suited for this task because in order to learn a detector we should provide the model with a lot of examples in order to consider all possible object locations!\n",
        "\n",
        "But, what about sharing LRF weights?\n",
        "\n",
        "Natural images are not a random combination of values in a $N \\times M$ array, but present **strong correlations** at different levels. \n",
        "\n",
        "At the most basic, it is evident that the **value of a pixel** is not independent of the values of its neighboring pixels. \n",
        "\n",
        "Moreover, natural images present another interesting property: **location invariance**. That means that visual structures, such as a cat or a dog, can be present on any place of the image at any scale. Image location is not important, what is important for attaching a meaning to an image are the **relative positions of geometric and photometric structures**. \n",
        "\n",
        "All this considerations leaded, partially inspired by biological models, to the proposal of a very special kind of layers: those based on **convolutions**. \n",
        "\n",
        "> A convolution is a mathematical operation that combines two input images to form a third one. One of the input images is the image we want to process. The other one, that is smaller, is called the kernel. \n",
        "\n",
        "Let's suppose that our kernel is this one:\n",
        "\n",
        "$$ Kernel = \\left( \\begin{array}{rrr} 0 & -1 & 0 \\\\ -1 & 5 & -1 \\\\  0 & -1 & 0 \\end{array} \\right)$$\n",
        "\n",
        "The output of image convolution is calculated as follows:\n",
        "\n",
        "- Flip the kernel both horizontally and vertically. As our selected kernel is symmetric, the flipped kernel is equal to the original.\n",
        "- Put the first element of the kernel at every element of the image matrix. Multiply each element of the kernel with its corresponding element of the image matrix (the one which is overlapped with it). \n",
        "- Sum up all product outputs and put the result at the same position in the output matrix as the center of kernel in image matrix.\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6dA1Hr8mcGVb",
        "colab_type": "text"
      },
      "source": [
        "## What is a convolution?\n",
        "\n",
        "Mathematically, given a convolution kernel $K$ represented by a $(M \\times N)$ array, the convolution of an image $I$ with $K$ is:\n",
        "\n",
        "$$\n",
        "output(x,y) = (I \\otimes K)(x,y) = \\sum_{m=0}^{M-1} \\sum_{n=1}^{N-1} K(m,n) I(x-n, y-m)\n",
        "$$\n",
        "\n",
        "The output of image convolution is another image that might represent some kind of information that was present in the image in a very subtle way. For example, the kernel we have used is called an **edge detector** because it highlights the edges of visual structures and attenuates smooth regions. \n",
        "\n",
        "> [Convolution visualization](http://deeplearning.net/software/theano_versions/dev/_images/numerical_no_padding_no_strides.gif)\n",
        "\n",
        "> Demo: http://setosa.io/ev/image-kernels/\n",
        "\n",
        "In **convolutional neural networks** the values of the kernel matrix are **free parameters** that must be learned to perform the optimal information extraction in order to classify the image. \n",
        "\n",
        "Convolutions are linear operators and because of this the application of successive convolutions can always be represented by a single convolution.\n",
        "\n",
        "$$\n",
        "((I \\otimes K_1) \\otimes K_2) = (I \\otimes (K_1 \\otimes K_2))\n",
        "$$\n",
        "\n",
        "But if we apply a non linear activation function after each convolution the application of successive convolution operators makes sense and results in a powerful image feature structure. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Jm8f73RWcGVq",
        "colab_type": "code",
        "outputId": "cd6bc9b9-a1d6-4c46-e0a1-62ffe6e67e95",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 287
        }
      },
      "source": [
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "from imageio import imread\n",
        "from skimage.transform import resize\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "\n",
        "sample_image = imread(\"https://raw.githubusercontent.com/DataScienceUB/DeepLearningMaster2019/master/images/result.png\")\n",
        "sample_image = sample_image.astype(float)\n",
        "\n",
        "size = sample_image.shape\n",
        "print(\"sample image shape: \"+ str(sample_image.shape))\n",
        "\n",
        "def show(image):\n",
        "    image = np.squeeze(image.astype(\"uint8\"))\n",
        "    plt.imshow(image, cmap=\"gray\")\n",
        "\n",
        "show(sample_image)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "sample image shape: (256, 209, 4)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAANkAAAD8CAYAAAD63wHzAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Pna4d08LcGVy",
        "colab_type": "code",
        "outputId": "9cff5c60-830e-43ed-eceb-4812aae87e71",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "sample_image= sample_image[:,:,:-1]\n",
        "sample_image.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(256, 209, 3)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NomKnJcQcGV-",
        "colab_type": "text"
      },
      "source": [
        "#### Example\n",
        "\n",
        "Let's build a convolution filter that blurs the image using `tf.nn.depthwise_conv2d`. This function treats each channel separately. If not, it would be a 3D convolution!\n",
        "\n",
        "Given an input tensor of shape ``[batch,  in_height, in_width, in_channels]`` and a filter tensor of shape ``[filter_height, filter_width, in_channels, channel_multiplier]`` containing ``in_channels`` convolutional filters of depth 1, ``depthwise_conv2d`` applies a different filter to each input channel (expanding from 1 channel to ``channel_multiplier`` channels for each), then concatenates the results together. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9tvnQtibcGV_",
        "colab_type": "code",
        "outputId": "281934b5-cb9b-40ce-99b6-b7e44c259660",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 312
        }
      },
      "source": [
        "image = tf.placeholder(tf.float32, shape=(None, None, None, 3))\n",
        "kernel = tf.placeholder(tf.float32, shape=(5, 5, 3))\n",
        "\n",
        "def conv(x, k):\n",
        "    k = tf.reshape(k, shape=(5, 5, 3, 1))\n",
        "    return tf.nn.depthwise_conv2d(x, \n",
        "                                  k, \n",
        "                                  strides=(1, 1, 1, 1),\n",
        "                                  padding='SAME')\n",
        "    \n",
        "\n",
        "kernel_data = np.zeros(shape=(5, 5, 3)).astype(np.float32)\n",
        "kernel_data[:, :, :] = 1 / 25.0\n",
        "\n",
        "output_image = conv(image, kernel)\n",
        "\n",
        "\n",
        "# move the channel dimension to the first dimension to\n",
        "# make it easy to see the spacial organization of the kernel\n",
        "# on the last 2 dimensions with print:\n",
        "print(np.transpose(kernel_data, (2, 0, 1))) "
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[[0.04 0.04 0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04 0.04 0.04]]\n",
            "\n",
            " [[0.04 0.04 0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04 0.04 0.04]]\n",
            "\n",
            " [[0.04 0.04 0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04 0.04 0.04]]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8p_csSQ-MGLF",
        "colab_type": "code",
        "outputId": "10ca8236-a244-49de-86e3-3f8d9637c0fa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 520
        }
      },
      "source": [
        "print(kernel_data) "
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[[0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]]\n",
            "\n",
            " [[0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]]\n",
            "\n",
            " [[0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]]\n",
            "\n",
            " [[0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]]\n",
            "\n",
            " [[0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]\n",
            "  [0.04 0.04 0.04]]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "N3ZWx0WpcGWG",
        "colab_type": "code",
        "outputId": "15970207-faf6-4529-9714-8c5f29cfb843",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 287
        }
      },
      "source": [
        "with tf.Session() as sess:\n",
        "    feed_dict = {image: [sample_image], kernel: kernel_data}\n",
        "    conv_img = sess.run(output_image, feed_dict=feed_dict)\n",
        "    print(conv_img.shape)\n",
        "    show(conv_img[0])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(1, 256, 209, 3)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAANkAAAD8CAYAAAD63wHzAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\npg5kHXV0wtSBrKOOTpg6kHXU0QnT/wPwLWG5Q/gylwAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sZ-FpnsbcGWN",
        "colab_type": "text"
      },
      "source": [
        "#### Example\n",
        "\n",
        "``tf.nn.conv2d``: Given an input tensor of shape ``[batch, in_height, in_width, in_channels]`` and a filter / kernel tensor of shape ``[filter_height, filter_width, in_channels, out_channels]``, this op performs the following:\n",
        "\n",
        "+ Flattens the filter to a 2-D matrix with shape ``[filter_height * filter_width * in_channels, output_channels]``.\n",
        "+ Extracts image patches from the input tensor to form a virtual tensor of shape ``[batch, out_height, out_width, filter_height * filter_width * in_channels]``.\n",
        "+ For each patch, right-multiplies the filter matrix and the image patch vector."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H6sGnE95cGWO",
        "colab_type": "code",
        "outputId": "8efdbc9e-28aa-4209-ed28-e04c0f9fbeb4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 308
        }
      },
      "source": [
        "image = tf.placeholder(tf.float32, shape=(None, None, None, 3))\n",
        "kernel = tf.placeholder(tf.float32, shape=(5, 5, 3))\n",
        "\n",
        "def conv(x, k):\n",
        "    k = tf.reshape(k, shape=(5, 5, 3, 1))\n",
        "    return tf.nn.conv2d(x, \n",
        "                        k, \n",
        "                        strides=(1, 1, 1, 1),\n",
        "                        padding='SAME')\n",
        "    \n",
        "output_image = conv(image, kernel)\n",
        "kernel_data = np.zeros(shape=(5, 5, 3)).astype(np.float32)\n",
        "kernel_data[:, :, :] = 1 / 125.0\n",
        "\n",
        "# move the channel dimension to the first dimension to\n",
        "# make it easy to see the spacial organization of the kernel\n",
        "# on the last 2 dimensions with print:\n",
        "print(np.transpose(kernel_data, (2, 0, 1))) "
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[[[0.008 0.008 0.008 0.008 0.008]\n",
            "  [0.008 0.008 0.008 0.008 0.008]\n",
            "  [0.008 0.008 0.008 0.008 0.008]\n",
            "  [0.008 0.008 0.008 0.008 0.008]\n",
            "  [0.008 0.008 0.008 0.008 0.008]]\n",
            "\n",
            " [[0.008 0.008 0.008 0.008 0.008]\n",
            "  [0.008 0.008 0.008 0.008 0.008]\n",
            "  [0.008 0.008 0.008 0.008 0.008]\n",
            "  [0.008 0.008 0.008 0.008 0.008]\n",
            "  [0.008 0.008 0.008 0.008 0.008]]\n",
            "\n",
            " [[0.008 0.008 0.008 0.008 0.008]\n",
            "  [0.008 0.008 0.008 0.008 0.008]\n",
            "  [0.008 0.008 0.008 0.008 0.008]\n",
            "  [0.008 0.008 0.008 0.008 0.008]\n",
            "  [0.008 0.008 0.008 0.008 0.008]]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FlbG8XWlcGWW",
        "colab_type": "code",
        "outputId": "899124b9-1b1e-46d9-9461-c65cf92527fb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        }
      },
      "source": [
        "with tf.Session() as sess:\n",
        "    feed_dict = {image: [sample_image], kernel: kernel_data}\n",
        "    conv_img = sess.run(output_image, feed_dict=feed_dict)\n",
        "    print(conv_img.shape)\n",
        "    show(conv_img[0])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(1, 256, 209, 1)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAANkAAAD8CAYAAAD63wHzAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "m2yF8T-GcGWl",
        "colab_type": "text"
      },
      "source": [
        "### Exercise\n",
        "\n",
        "+ What happens if we build a 3x3 kernel with stride 2? What is the size of the output image?\n",
        "\n",
        "\n",
        "+ When a kernel is overlapped on an image it should be set to fit within the bounds of the image. Filling the missing area of the image is known as *padding* the image. The padding parameter must be ``SAME``  if we want the convolution output to be the same size as the input, and ``VALID`` if we want only the useful part of the output.   \n",
        "What happens if we change the padding to 'VALID'?"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oGBsCTy_cGWm",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# your code here\n",
        "\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oxmDTba6cGWr",
        "colab_type": "text"
      },
      "source": [
        "## Convolutional Neural Netwoks\n",
        "\n",
        "The kernel weights can be learn from examples if we define a loss function.\n",
        "\n",
        "> CNN are composed of repetitive blocks of neurons that are applied across\n",
        "space (for images) or time (for audio signals).\n",
        "\n",
        "> For images, these blocks of neurons can be interpreted as\n",
        "2D convolutional kernels, repeatedly applied over each\n",
        "patch of the image.\n",
        "\n",
        "<img src=\"https://github.com/DataScienceUB/DeepLearningMaster2019/blob/master/images/conv.png?raw=1\" alt=\"https://community.arm.com/cfs-file/__key/communityserver-blogs-components-weblogfiles/00-00-00-20-66/4786.conv.png\" style=\"width: 500px;\"/>\n",
        "\n",
        "\n",
        "> For speech, they can be seen as the 1D convolutional\n",
        "kernels applied across time-windows.\n",
        "\n",
        "<img src=\"https://github.com/DataScienceUB/DeepLearningMaster2019/blob/master/images/spectrogram.png?raw=1\" alt=\"\" style=\"width: 500px;\"/>\n",
        "\n",
        "\n",
        "At training time, the weights for these repeated blocks are\n",
        "'shared', i.e. the weight gradients learned over various\n",
        "image patches are averaged."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "78h98OQ1cGWx",
        "colab_type": "text"
      },
      "source": [
        "CNN where originally proposed in early 90's by Y.LeCun:\n",
        "\n",
        "<img src=\"https://github.com/DataScienceUB/DeepLearningMaster2019/blob/master/images/conv2.png?raw=1\" alt=\"\" style=\"width: 700px;\"/>\n",
        "\n",
        "\n",
        "LeNet architecture used one additional ingredient: **max pooling**. Pooling is a way of sub-sampling, i.e. reducing the dimension of the input. It is usually done after convolutional layers. \n",
        "\n",
        "It is also useful since it provides a form of translation invariance. "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4Fa0xEZvcGWz",
        "colab_type": "text"
      },
      "source": [
        "## MNIST Convolutional Network \n",
        "\n",
        "A simple CNN architecture combines a convolutional layer (``tf.nn.conv2d``), a non-linear activation function (``tf.nn.relu``), a pooling layer (``tf.nn.max_pool``) and a fully connected layer (``tf.matmul``). \n",
        "\n",
        "<img src=\"https://github.com/DataScienceUB/DeepLearningMaster2019/blob/master/images/maxpool.jpeg?raw=1\" alt=\"\" style=\"width: 400px;\"/>\n",
        "<center>\n",
        "Max Pooling. Source: http://cs231n.github.io/convolutional-networks/\n",
        "</center>\n",
        "\n",
        "\n",
        "The ``tf`` input pipeline is designed to work with multiple images in a batch: `` [image_batch_size, image_height, image_width, image_channels] ``. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "J1VTGSC8cGW2",
        "colab_type": "code",
        "outputId": "e6df67b3-8863-49a9-8934-d905e799af3e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 87
        }
      },
      "source": [
        "# Adapted notebook from Author: Aymeric Damien\n",
        "# Project: https://github.com/aymericdamien/TensorFlow-Examples/\n",
        "\n",
        "# Import MINST data\n",
        "from tensorflow.examples.tutorials.mnist import input_data\n",
        "mnist = input_data.read_data_sets(\"tmp/data/\", one_hot=True)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Extracting tmp/data/train-images-idx3-ubyte.gz\n",
            "Extracting tmp/data/train-labels-idx1-ubyte.gz\n",
            "Extracting tmp/data/t10k-images-idx3-ubyte.gz\n",
            "Extracting tmp/data/t10k-labels-idx1-ubyte.gz\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u8NAYOlKcGW8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import tensorflow as tf\n",
        "\n",
        "# Parameters\n",
        "learning_rate = 0.01\n",
        "training_iters = 100000\n",
        "batch_size = 128\n",
        "display_step = 20"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lCd4WqFQcGW_",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Network Parameters\n",
        "n_input = 784 # MNIST data input (img shape: 28*28)\n",
        "n_classes = 10 # MNIST total classes (0-9 digits)\n",
        "dropout = 0.75 # Dropout, probability to keep units"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1dJl6TrJcGXD",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# tf Graph input\n",
        "x = tf.placeholder(tf.float32, [None, n_input])\n",
        "y = tf.placeholder(tf.float32, [None, n_classes])\n",
        "keep_prob = tf.placeholder(tf.float32) #dropout (keep probability)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OsEBDe0VcGXL",
        "colab_type": "text"
      },
      "source": [
        "The parameter ``strides`` causes a kernel to skip over pixels of an image and not to include them in the output. \n",
        "\n",
        "Strides are a way to adjust the dimensionality of input tensors (to reduce processing power). This allows the design of CNN with small kernels but getting information from large parts of the input. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "za8G8NGmcGXM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Create model\n",
        "def conv2d(img, w, b):\n",
        "    return tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(img, w, strides=[1, 1, 1, 1], \n",
        "                                                  padding='SAME'),b))\n",
        "\n",
        "def max_pool(img, k):\n",
        "    return tf.nn.max_pool(img, ksize=[1, k, k, 1], strides=[1, k, k, 1], padding='SAME')\n",
        "\n",
        "def conv_net(_X, _weights, _biases, _dropout):\n",
        "    # Reshape input picture\n",
        "    _X = tf.reshape(_X, shape=[-1, 28, 28, 1])\n",
        "\n",
        "    # Convolution Layer\n",
        "    conv1 = conv2d(_X, _weights['wc1'], _biases['bc1'])\n",
        "    # Max Pooling (down-sampling)\n",
        "    conv1 = max_pool(conv1, k=2)\n",
        "    # Apply Dropout\n",
        "    conv1 = tf.nn.dropout(conv1, _dropout)\n",
        "\n",
        "    # Convolution Layer\n",
        "    conv2 = conv2d(conv1, _weights['wc2'], _biases['bc2'])\n",
        "    # Max Pooling (down-sampling)\n",
        "    conv2 = max_pool(conv2, k=2)\n",
        "    # Apply Dropout\n",
        "    conv2 = tf.nn.dropout(conv2, _dropout)\n",
        "\n",
        "    # Fully connected layer\n",
        "    # Reshape conv2 output to fit dense layer input\n",
        "    dense1 = tf.reshape(conv2, [-1, _weights['wd1'].get_shape().as_list()[0]]) \n",
        "    # Relu activation\n",
        "    dense1 = tf.nn.relu(tf.add(tf.matmul(dense1, _weights['wd1']), _biases['bd1']))\n",
        "    # Apply Dropout\n",
        "    dense1 = tf.nn.dropout(dense1, _dropout) # Apply Dropout\n",
        "\n",
        "    # Output, class prediction\n",
        "    out = tf.add(tf.matmul(dense1, _weights['out']), _biases['out'])\n",
        "    return out"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OP6gwSLAcGXX",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Store layers weight & bias\n",
        "weights = {\n",
        "    # 5x5 conv, 1 input, 32 outputs\n",
        "    'wc1': tf.Variable(tf.random_normal([5, 5, 1, 32])), \n",
        "    # 5x5 conv, 32 inputs, 64 outputs\n",
        "    'wc2': tf.Variable(tf.random_normal([5, 5, 32, 64])), \n",
        "    # fully connected, 7*7*64 inputs, 512 outputs\n",
        "    'wd1': tf.Variable(tf.random_normal([7*7*64, 512])), \n",
        "    # 512 inputs, 10 outputs (class prediction)\n",
        "    'out': tf.Variable(tf.random_normal([512, n_classes])) \n",
        "}\n",
        "\n",
        "biases = {\n",
        "    'bc1': tf.Variable(tf.random_normal([32])),\n",
        "    'bc2': tf.Variable(tf.random_normal([64])),\n",
        "    'bd1': tf.Variable(tf.random_normal([512])),\n",
        "    'out': tf.Variable(tf.random_normal([n_classes]))\n",
        "}"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Vq-Ut_accGXc",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Construct model\n",
        "pred = conv_net(x, weights, biases, keep_prob)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OObJtIlhcGXg",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Define loss and optimizer\n",
        "cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits_v2\n",
        "                      (logits= pred, labels= y))\n",
        "optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6gWP_yx7cGXn",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Evaluate model\n",
        "correct_pred = tf.equal(tf.argmax(pred,1), tf.argmax(y,1))\n",
        "accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ILiuDf-4cGXu",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Initializing the variables\n",
        "init = tf.global_variables_initializer()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U2IO5MM6cGX0",
        "colab_type": "code",
        "outputId": "b3235990-846b-400a-caa6-5288fb0a9d64",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 728
        }
      },
      "source": [
        "# Launch the graph\n",
        "with tf.Session() as sess:\n",
        "    sess.run(init)\n",
        "    step = 1\n",
        "    # Keep training until reach max iterations\n",
        "    while step * batch_size < training_iters:\n",
        "        batch_xs, batch_ys = mnist.train.next_batch(batch_size)\n",
        "        # Fit training using batch data\n",
        "        sess.run(optimizer, \n",
        "                 feed_dict={x: batch_xs, y: batch_ys, keep_prob: dropout})\n",
        "        if step % display_step == 0:\n",
        "            # Calculate batch accuracy\n",
        "            acc = sess.run(accuracy, \n",
        "                           feed_dict={x: batch_xs, y: batch_ys, keep_prob: 1.})\n",
        "            # Calculate batch loss\n",
        "            loss = sess.run(cost, \n",
        "                            feed_dict={x: batch_xs, y: batch_ys, keep_prob: 1.})\n",
        "            print(\"Iter \", str(step*batch_size),\n",
        "                  \"| Minibatch Loss= {:.6f}\".format(loss)  , \n",
        "                  \"| Training Accuracy= {:.5f}\".format(acc))\n",
        "        step += 1\n",
        "    print(\"Optimization Finished!\")\n",
        "    # Calculate accuracy for 256 mnist test images\n",
        "    print(\"Testing Accuracy:\", \n",
        "          sess.run(accuracy, \n",
        "                   feed_dict={x: mnist.test.images[:256], \n",
        "                   y: mnist.test.labels[:256], \n",
        "                   keep_prob: 1.}))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Iter  2560 | Minibatch Loss= 2423.555176 | Training Accuracy= 0.58594\n",
            "Iter  5120 | Minibatch Loss= 689.561096 | Training Accuracy= 0.73438\n",
            "Iter  7680 | Minibatch Loss= 395.890930 | Training Accuracy= 0.85156\n",
            "Iter  10240 | Minibatch Loss= 289.860046 | Training Accuracy= 0.85156\n",
            "Iter  12800 | Minibatch Loss= 243.940140 | Training Accuracy= 0.85156\n",
            "Iter  15360 | Minibatch Loss= 136.920807 | Training Accuracy= 0.89062\n",
            "Iter  17920 | Minibatch Loss= 175.171494 | Training Accuracy= 0.89062\n",
            "Iter  20480 | Minibatch Loss= 90.708237 | Training Accuracy= 0.94531\n",
            "Iter  23040 | Minibatch Loss= 229.978210 | Training Accuracy= 0.82031\n",
            "Iter  25600 | Minibatch Loss= 116.338692 | Training Accuracy= 0.90625\n",
            "Iter  28160 | Minibatch Loss= 278.184509 | Training Accuracy= 0.88281\n",
            "Iter  30720 | Minibatch Loss= 77.760223 | Training Accuracy= 0.93750\n",
            "Iter  33280 | Minibatch Loss= 198.099701 | Training Accuracy= 0.89062\n",
            "Iter  35840 | Minibatch Loss= 89.329224 | Training Accuracy= 0.91406\n",
            "Iter  38400 | Minibatch Loss= 51.937981 | Training Accuracy= 0.93750\n",
            "Iter  40960 | Minibatch Loss= 83.168900 | Training Accuracy= 0.96094\n",
            "Iter  43520 | Minibatch Loss= 104.349327 | Training Accuracy= 0.92969\n",
            "Iter  46080 | Minibatch Loss= 108.106598 | Training Accuracy= 0.92188\n",
            "Iter  48640 | Minibatch Loss= 38.267960 | Training Accuracy= 0.96094\n",
            "Iter  51200 | Minibatch Loss= 45.728611 | Training Accuracy= 0.96094\n",
            "Iter  53760 | Minibatch Loss= 70.274506 | Training Accuracy= 0.94531\n",
            "Iter  56320 | Minibatch Loss= 41.208733 | Training Accuracy= 0.97656\n",
            "Iter  58880 | Minibatch Loss= 53.943806 | Training Accuracy= 0.95312\n",
            "Iter  61440 | Minibatch Loss= 76.221741 | Training Accuracy= 0.92969\n",
            "Iter  64000 | Minibatch Loss= 21.152348 | Training Accuracy= 0.96094\n",
            "Iter  66560 | Minibatch Loss= 23.621267 | Training Accuracy= 0.97656\n",
            "Iter  69120 | Minibatch Loss= 25.055222 | Training Accuracy= 0.95312\n",
            "Iter  71680 | Minibatch Loss= 8.120686 | Training Accuracy= 0.97656\n",
            "Iter  74240 | Minibatch Loss= 54.463661 | Training Accuracy= 0.94531\n",
            "Iter  76800 | Minibatch Loss= 43.605202 | Training Accuracy= 0.96875\n",
            "Iter  79360 | Minibatch Loss= 50.153397 | Training Accuracy= 0.96094\n",
            "Iter  81920 | Minibatch Loss= 45.550892 | Training Accuracy= 0.95312\n",
            "Iter  84480 | Minibatch Loss= 28.427061 | Training Accuracy= 0.96875\n",
            "Iter  87040 | Minibatch Loss= 78.017303 | Training Accuracy= 0.92188\n",
            "Iter  89600 | Minibatch Loss= 29.417255 | Training Accuracy= 0.96094\n",
            "Iter  92160 | Minibatch Loss= 13.598746 | Training Accuracy= 0.96875\n",
            "Iter  94720 | Minibatch Loss= 34.922489 | Training Accuracy= 0.98438\n",
            "Iter  97280 | Minibatch Loss= 26.119286 | Training Accuracy= 0.94531\n",
            "Iter  99840 | Minibatch Loss= 32.713379 | Training Accuracy= 0.95312\n",
            "Optimization Finished!\n",
            "Testing Accuracy: 0.9765625\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rfdlX91acGYB",
        "colab_type": "text"
      },
      "source": [
        "Some CNN architectures use **normalization** layers. \n",
        "\n",
        "For example, ``tf.nn.local_response_normalization`` is a function which shapes the output based on a summation operation in the following way: within a given vector, each component is divided by the weighted, squared sum of inputs within a depth radius. \n",
        "\n",
        "> In neurobiology, there is a concept called “lateral inhibition”. This refers to the capacity of an excited neuron to subdue its neighbors.\n",
        "\n",
        "\n",
        "In fact, after a convolutional layer there are two kinds of non linear functions that are usually applied: non-linear activation functions such as sigmoids or **ReLU and pooling**. Pooling layers are used with the purpose to progressively reduce the spatial size of the image to achieve scale invariance.The most common layer is the maxpool layer. Basically a maxpool of $2 \\times 2$ causes a filter of 2 by 2 to traverse over the entire input array and pick the largest element from the window to be included in the next representation map. Pooling can also be implemented by using other criteria, such as averaging instead of taking the max element. \n",
        "\n",
        "A convolutional neural network is a neural network that is build by using several convolutional layers, each one formed by the concatenation of three different operators: **convolutions, non-linear activation and pooling**. This kind of networks are able of extracting powerful image descriptors when applied in sequence. The power of the method has been credited to the fact that these descriptors can be seen as hierarchical features that are suited to optimally represent visual structures in natural images. The last layers of a convolutional neural network are classical dense layers, which are connected to a classification or regression loss function.\n",
        "\n",
        "Finally, it is interesting to point out that convolutional layers are much **lighter**, in terms of number of weights, than fully connected layers, but more computationally demanding (The number of weights we must learn for a $(M \\times N)$ convolution kernel is only $(M \\times N)$, which is independent of the size of the image). \n",
        "\n",
        "In some sense, **convolutional layers trade weights for computation when extracting information**. \n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sjO2z_NfcGYE",
        "colab_type": "text"
      },
      "source": [
        "## Visualization of learned kernels.\n",
        "\n",
        "\n",
        "    \n",
        "<img src=\"https://github.com/DataScienceUB/DeepLearningMaster2019/blob/master/images/weights.jpeg?raw=1\" alt=\"\" style=\"width: 400px\" />\n",
        "\n",
        "Convolution kernels from the first AlexNet layer. <br>   \n",
        "Source: http://cs231n.github.io/convolutional-networks/\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "S2JAN0QRcGYI",
        "colab_type": "text"
      },
      "source": [
        "## Exercise:\n",
        "\n",
        "**VGGNet**. The runner-up in ILSVRC 2014 was the network from Karen Simonyan and Andrew Zisserman that became known as the VGGNet. Its main contribution was in showing that the depth of the network is a critical component for good performance. Their final best network contains 16 CONV/FC layers and, appealingly, features an extremely homogeneous architecture that only performs $3 \\times 3$ convolutions and $2 \\times 2$ pooling from the beginning to the end. \n",
        "\n",
        "The architecture is:\n",
        "\n",
        "+ INPUT: [224x224x3] \n",
        "+ CONV3-64: [224x224x64]  \n",
        "+ CONV3-64: [224x224x64]  \n",
        "+ POOL2: [112x112x64]  \n",
        "+ CONV3-128: [112x112x128]  \n",
        "+ CONV3-128: [112x112x128]  \n",
        "+ POOL2: [56x56x128]  \n",
        "+ CONV3-256: [56x56x256]  \n",
        "+ CONV3-256: [56x56x256]  \n",
        "+ CONV3-256: [56x56x256]  \n",
        "+ POOL2: [28x28x256]  \n",
        "+ CONV3-512: [28x28x512]  \n",
        "+ CONV3-512: [28x28x512]  \n",
        "+ CONV3-512: [28x28x512]  \n",
        "+ POOL2: [14x14x512]  \n",
        "+ CONV3-512: [14x14x512]  \n",
        "+ CONV3-512: [14x14x512]  \n",
        "+ CONV3-512: [14x14x512]  \n",
        "+ POOL2: [7x7x512]  \n",
        "+ FC1: [1x1x4096]  \n",
        "+ FC2: [1x1x4096]  \n",
        "+ FC3: [1x1x1000]  \n",
        "\n",
        "Which is the approximate number of parameters of this network? Which is the heaviest layer?"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "r7Ctr0MDcGYJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Your code here"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GAxFsYL2SkHt",
        "colab_type": "text"
      },
      "source": [
        "## VGG-16 in Keras"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OjKnrM1fSnYC",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.layers import Flatten, Dense, Dropout\n",
        "from tensorflow.keras.layers import Conv2D, MaxPooling2D, ZeroPadding2D\n",
        "from tensorflow.keras.optimizers import SGD\n",
        "\n",
        "model = Sequential()\n",
        "model.add(ZeroPadding2D((1,1), input_shape=(224, 224, 3), data_format='channels_last'))\n",
        "model.add(Conv2D(64, kernel_size=(3, 3), strides=1, activation='relu'))\n",
        "model.add(ZeroPadding2D((1,1)))\n",
        "model.add(Conv2D(64, kernel_size=(3, 3), strides=1, activation='relu'))\n",
        "model.add(MaxPooling2D((2,2), strides=(2,2), data_format='channels_last'))\n",
        "\n",
        "model.add(ZeroPadding2D((1,1)))\n",
        "model.add(Conv2D(128, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(ZeroPadding2D((1,1)))\n",
        "model.add(Conv2D(128, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(MaxPooling2D((2,2), strides=(2,2), data_format='channels_last'))\n",
        "\n",
        "model.add(ZeroPadding2D((1,1)))\n",
        "model.add(Conv2D(256, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(ZeroPadding2D((1,1)))\n",
        "model.add(Conv2D(256, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(ZeroPadding2D((1,1)))\n",
        "model.add(Conv2D(256, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(MaxPooling2D((2,2), strides=(2,2), data_format='channels_last'))\n",
        "\n",
        "model.add(ZeroPadding2D((1,1)))\n",
        "model.add(Conv2D(512, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(ZeroPadding2D((1,1)))\n",
        "model.add(Conv2D(512, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(ZeroPadding2D((1,1)))\n",
        "model.add(Conv2D(512, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(MaxPooling2D((2,2), strides=(2,2), data_format='channels_last'))\n",
        "\n",
        "model.add(ZeroPadding2D((1,1)))\n",
        "model.add(Conv2D(512, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(ZeroPadding2D((1,1)))\n",
        "model.add(Conv2D(512, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(ZeroPadding2D((1,1)))\n",
        "model.add(Conv2D(512, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(MaxPooling2D((2,2), strides=(2,2), data_format='channels_last'))\n",
        "\n",
        "model.add(Flatten())\n",
        "model.add(Dense(4096, activation='relu'))\n",
        "model.add(Dropout(0.5))\n",
        "model.add(Dense(4096, activation='relu'))\n",
        "model.add(Dropout(0.5))\n",
        "model.add(Dense(1000, activation='softmax'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ldmS1Q0IcGYO",
        "colab_type": "text"
      },
      "source": [
        "## Smile Detection using Convolutional Neural Networks"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JPlUydkxcGYP",
        "colab_type": "code",
        "outputId": "b034ae79-ac50-41d0-b13e-003ae22af094",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 87
        }
      },
      "source": [
        "# Adapted from Kyle McDonald\n",
        "# Project: https://github.com/kylemcdonald/SmileCNN\n",
        "\n",
        "!curl -L -O https://github.com/hromi/SMILEsmileD/archive/master.zip\n",
        "!unzip -q master.zip\n",
        "!rm master.zip"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current\n",
            "                                 Dload  Upload   Total   Spent    Left  Speed\n",
            "100   122    0   122    0     0    603      0 --:--:-- --:--:-- --:--:--   600\n",
            "100 39.6M    0 39.6M    0     0  13.9M      0 --:--:--  0:00:02 --:--:-- 21.8M\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SSt9PviocGac",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#util and imports\n",
        "import os\n",
        "import fnmatch\n",
        "\n",
        "def list_all_files(directory, extensions=None):\n",
        "    for root, dirnames, filenames in os.walk(directory):\n",
        "        for filename in filenames:\n",
        "            base, ext = os.path.splitext(filename)\n",
        "            joined = os.path.join(root, filename)\n",
        "            if extensions is None or ext.lower() in extensions:\n",
        "                yield joined\n",
        "                \n",
        "import numpy as np\n",
        "import math\n",
        "\n",
        "def find_rectangle(n, max_ratio=2):\n",
        "    sides = []\n",
        "    square = int(math.sqrt(n))\n",
        "    for w in range(square, max_ratio * square):\n",
        "        h = n / w\n",
        "        used = w * h\n",
        "        leftover = n - used\n",
        "        sides.append((leftover, (w, h)))\n",
        "    return sorted(sides)[0][1]\n",
        "\n",
        "# should work for 1d and 2d images, assumes images are square but can be overriden\n",
        "def make_mosaic(images, n=None, nx=None, ny=None, w=None, h=None):\n",
        "    if n is None and nx is None and ny is None:\n",
        "        nx, ny = find_rectangle(len(images))\n",
        "    else:\n",
        "        nx = n if nx is None else nx\n",
        "        ny = n if ny is None else ny\n",
        "    images = np.array(images)\n",
        "    if images.ndim == 2:\n",
        "        side = int(np.sqrt(len(images[0])))\n",
        "        h = side if h is None else h\n",
        "        w = side if w is None else w\n",
        "        images = images.reshape(-1, h, w)\n",
        "    else:\n",
        "        h = images.shape[1]\n",
        "        w = images.shape[2]\n",
        "    image_gen = iter(images)\n",
        "    mosaic = np.empty((h*ny, w*nx))\n",
        "    for i in range(ny):\n",
        "        ia = (i)*h\n",
        "        ib = (i+1)*h\n",
        "        for j in range(nx):\n",
        "            ja = j*w\n",
        "            jb = (j+1)*w\n",
        "            mosaic[ia:ib, ja:jb] = next(image_gen)\n",
        "    return mosaic\n",
        "\n",
        "import numpy as np\n",
        "from skimage.measure import block_reduce\n",
        "\n",
        "def crop_and_resize(img, target_size=32, zoom=1):\n",
        "    small_side = int(np.min(img.shape) * zoom)\n",
        "    reduce_factor = small_side / target_size\n",
        "    crop_size = target_size * reduce_factor\n",
        "    mid = np.array(img.shape) / 2\n",
        "    half_crop = crop_size / 2\n",
        "    center = img[mid[0]-half_crop:mid[0]+half_crop,\n",
        "    \tmid[1]-half_crop:mid[1]+half_crop]\n",
        "    return block_reduce(center, (reduce_factor, reduce_factor), np.mean)\n",
        "\n",
        "from io import BytesIO\n",
        "import numpy as np\n",
        "import PIL.Image\n",
        "import IPython.display\n",
        "import shutil\n",
        "\n",
        "def show_array(a, fmt='png', filename=None):\n",
        "    a = np.squeeze(a)\n",
        "    a = np.uint8(np.clip(a, 0, 255))\n",
        "    image_data = BytesIO()\n",
        "    PIL.Image.fromarray(a).save(image_data, fmt)\n",
        "    if filename is None:\n",
        "        IPython.display.display(IPython.display.Image(data=image_data.getvalue()))\n",
        "    else:\n",
        "        with open(filename, 'w') as f:\n",
        "            image_data.seek(0)\n",
        "            shutil.copyfileobj(image_data, f)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6XxOgKtecGai",
        "colab_type": "code",
        "outputId": "0765298a-e7fe-4beb-c8e3-a101fd8a7fc8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "negative_paths = list(list_all_files('SMILEsmileD-master/SMILEs/negatives/negatives7/', ['.jpg']))\n",
        "print('loaded', len(negative_paths), 'negative examples')\n",
        "positive_paths = list(list_all_files('SMILEsmileD-master/SMILEs/positives/positives7/', ['.jpg']))\n",
        "print('loaded', len(positive_paths), 'positive examples')\n",
        "examples = [(path, 0) for path in negative_paths] + [(path, 1) for path in positive_paths]"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "loaded 9475 negative examples\n",
            "loaded 3690 positive examples\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GrZqMUWKcGas",
        "colab_type": "code",
        "outputId": "4319303b-ae4d-4fd2-9dae-6c7eb1f6c379",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "import numpy as np\n",
        "from skimage.measure import block_reduce\n",
        "from skimage.io import imread\n",
        "\n",
        "def examples_to_dataset(examples, block_size=2):\n",
        "    X = []\n",
        "    y = []\n",
        "    for path, label in examples:\n",
        "        img = imread(path, as_grey=True)\n",
        "        img = block_reduce(img, block_size=(block_size, block_size), func=np.mean)\n",
        "        X.append(img)\n",
        "        y.append(label)\n",
        "    return np.asarray(X), np.asarray(y)\n",
        "\n",
        "%time X, y = examples_to_dataset(examples)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "CPU times: user 7.36 s, sys: 488 ms, total: 7.84 s\n",
            "Wall time: 7.85 s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yEXeERWkcGa2",
        "colab_type": "code",
        "outputId": "296a079d-0efe-4610-a0f1-ffa22d6b01d6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 52
        }
      },
      "source": [
        "X = X.astype(np.float32) / 255.\n",
        "y = y.astype(np.int32)\n",
        "print(X.dtype, X.min(), X.max(), X.shape)\n",
        "print(y.dtype, y.min(), y.max(), y.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "float32 0.0 1.0 (13165, 32, 32)\n",
            "int32 0 1 (13165,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1MtGvg0ucGbL",
        "colab_type": "code",
        "outputId": "32452ae4-9fe6-4888-956c-c2f5b7300521",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 529
        }
      },
      "source": [
        "show_array(255 * make_mosaic(X[:len(negative_paths)], 8), fmt='jpeg') # negative at the beginning\n",
        "show_array(255 * make_mosaic(X[-len(positive_paths):], 8), fmt='jpeg') # positive at the end"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/jpeg": "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0a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\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/jpeg": "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0a\nHBwgJC4nICIsIxwcKDcpLDAxNDQ0Hyc5PTgyPC4zNDL/wAALCAEAAQABAREA/8QAHwAAAQUBAQEB\nAQEAAAAAAAAAAAECAwQFBgcICQoL/8QAtRAAAgEDAwIEAwUFBAQAAAF9AQIDAAQRBRIhMUEGE1Fh\nByJxFDKBkaEII0KxwRVS0fAkM2JyggkKFhcYGRolJicoKSo0NTY3ODk6Q0RFRkdISUpTVFVWV1hZ\nWmNkZWZnaGlqc3R1dnd4eXqDhIWGh4iJipKTlJWWl5iZmqKjpKWmp6ipqrKztLW2t7i5usLDxMXG\nx8jJytLT1NXW19jZ2uHi4+Tl5ufo6erx8vP09fb3+Pn6/9oACAEBAAA/AOW8Kae1joUDG2u5WlOW\nML7Qme5rsPCmqyT3MsJmuCinb5dyvzKfY9xWxrOpTWqZg+yjsTO5UZrGjnvtSiYyQ28sLfKz20m7\nH1B5rP0+efSbATW8Ks0jZz+gFb/h7xfJeTtBcWwTBwCrZBPp9a1NW8U2mmD94jlv7qDJrL/4TCwv\nF+QSrkfxLjFcEvh8aZajEMk0mP4BT7L/AEg/PbSRbf7wqzfRosgQLu4zjFZEgaeQ5sp0QcbioxWh\n8Jby2srLUZ5pEgWPaXmk6c8fy/nXrWm61aT2bzxXqXUI53rjj2OKxtc8QfYrZNQtdGfUTKu4jrsX\nOBj3rCs9efU9RjK2NxZSq4LIy4GPft0q54z1m70WOBLWTymcZLKm4158Nfm1S8zc3bzP2DRhf5VF\nqU0IwHupIyeyVBZj590dw00f8Qc5Irb8NahcTN5n38ttUH9TXc2fjWJLoQ2mLwof3gVSAv412E3i\nG1Omi88sqgHI9/SuJXxbaXuoM0waJzlIl2nBP19agl1CSwXFunz4+9s3fpWLH4w1WS7CLNlM8iSD\nb7cc10l74xk0NbVpY1f7RgqjvtDDvzU194iMkbSQQm3kuAGZQ2dvGOorWCWWkbZI47K2lHALjBqa\n08VXjzKrQW8kLHiWNutamq67DpNoJrgElh8qg4yfrXm2paWfiD4hst8kMIjEnCnJwCOPc80eHZYX\ntpQXwoPGOgra07y7q5MkYOxP4sdakutNgubrMoEjocrzytLpPh+x026M1pD5LsMEIxww9x0qbTrK\nKS0jj7AVJFZ29vcCK2hRfm3MVUDJ/Cqps4tQ1C5aSFWcHHPYVnxeEoLWZnAcBmLNl8g+2MVsXFnZ\nSptlAXIxkVjQaXZPemNJFESEbj0Gew+tZmq2cMOoSeUQUXuD0pPJgWEFdvzelYfwr0a31jSdQivF\ndrczxnYpxkrz/hXY340uwjXTtG/cvJIdyQws43AgHcQOv1rodDm+0aX+43xbXaNkkGOQecH0rL1m\n6e0J3Bcj+LHNL4mubGbT5I7pS8xGE2jkH1zXD6NocE0lxdTEhIELO7dAewrFuLGOe4yDg56YzVy3\nstg2hcEjkhcVd8FyQwtskRTsJyDXpem6TYx2ZeGGKKOTnIwCcdqrxkS28oABhEwbbjin32lafaKk\nrwR7UbzVUAY3etU7YQ3EG3enI6mqsmlWEUoJdHnb7qAkn8q15/D9p4g0q1hIiEtq5KCReSvcc++K\n57XbVbPVGgQYRUTav935RxXot1olhrEYaaCN3Ax8wzUcXh6G0jICIFHQAdMdKkuLCDU4YlljQlDg\nFlBrlfGEaaFLpd3aBUlilJyoxuB6j8a82UT2MmowqzDyWJ2j+7n/AANdR4c1Z4YfNRsRSAFQ6kVt\nXUsDgXYnjSZf7rAA+1bemSLIqEj5mrmLFp00iFlkbasSlyOTjFNW7u1l82yn4CYCFc5P1qCzvb+K\nZmuZFQlskgcmtg6q0tv86/MOv+NZt9dOxkweFUniq2mQ2F1pyi5glnuC3mbYkLlfQkD6GqF7p/2a\n4MklheZ5/eSREcCqsEhZJMfcByPanfBm7i/sjULdRmeOVWx/skdf0r0mIXE9vPNbRRCNWK+bIwVc\ng/N+tY7315YaikUkaSwSFQJLd94BJ4H6j86f4gsBMmWbOVzisLxYPs2q3UKsdqNtRj1Az1/KuYsG\nE+4Wl3GSrsHhdwCSPVajvIJLZ1a5lA3vtUAg8n0rRglBVPMIyepqjpkttc6jNJZhkj3YIPFdzFPP\nbCNZIpJkdSVVBx+NEqCGHcttqaoxDPFjj2OapaheXd3p++KNkic4AcYIqliRFeSElnI+VSeKz4tQ\nFxESnnpegkbsbXB+npzVrRtQukYwSXkguWcMqyMSzDPPWrmqXTXuozTsckkD8uP6V6HZausSgclj\n0AFT3VzeyossQUpzujJxn8axbPxHIlrcmaymgaJyPm5Ax6etcX4/8TLe3NjHCTiLJkwcckdKg1m0\nMV2L1PuTL5UmPXsam0XTr/T4WS0neNWbcNp4zjGcEGtePwz9pjWS9nluZV+75jEhT6iums9kLqP7\noxXBeCtSnutGh3DBC7MnuB0rYFlbxklYmTJydgBGahfSbaaZfMEpQclQdufrin31zCjLDEAoUDOK\nraQ8eqWC3AxmVOR6HuKkfw34ks7QXPh/VDbtJhpY/LUk46YJ7e1c/eWfj6+kC3+qXKIpPzGNVH6D\nmrDWQs7HyjIZGOAzHqT3NcT8J9X/ALL8ZxROf3V2hib69Qf0/WvTfEfhmXWNUaG4uP8AiWcyrCGI\nJduTyO365qHR/C+n6NcCSGz8t4/mQu7MQfU5NaesaqPsZnkYZC888ZrnvEF5Hqd1cKpxMpwR6kGu\ncm02wIaSTCPJyxK9/aoYtPsrV/Ni+eTGFY9qv2jLLfQW/Vc5b6DmqPhwNHeMMZBUF/r1r0a3iTUd\nMFrKMlTgfTrTbjw7NHbqj3szQ9ozKxB9sZqhqd5l4raH/VQLtHue9YWkatFNsDybXB71032K31S3\nLW8sKy5wwYd6kvNFh0jTortltmuWJRZIgM8isLIr0dtMgu7Ly23K443IxVh7gispIrmxeSF7q+ZA\nMq6S5P0IxVzQBcXV5ItxLNImAE83AI9TgCvJPFNqbTWb6A7iYrl1BIycZOOfpirZ1G30lfJaSBR0\nVXBOT+FTab4otbyUwlAHHHynIP0q3P4iiswIxDuf1Y4A9yaHul1fT54llt5AyNnys5BxXd3TywSJ\nJEnmHt2GfekOvXtiokmS2uFJx5cTZYVsw66DZG7lhkiRQDgqSRTtN8SRzSF4IpSrqRkrgV49Fqkl\niIL9rO7kNyhMbxAFQoyec961tP8AEj6lgCF1OcDem0mrF1qwtFIlVQVGSXOAK5u71+yllPk3MJfO\ndiE8H8a7nS0vJoVdJo4Iz93K5JHqa6WxvBB+5umSQ9VdOhrM1G6tJLmS5vbx4oVOI40JCgepqOdo\n1gJgmMsEinBPOK8r8oqcSi6lUdl7Vb0ySO3uRLaSzxvnlZCa9Fm167sNItp43jWWVCVZu2OM1geI\ndV1S/wDCrfbRCRLOGMsTcYx0NebXuhnwv4nCRBp0EIkjB7knGK7bw34nmlu4ra6sFiVjgSK+4fpx\nXQ6x4qXRnZX0+eYbQVKj72azYvFUd/KsUtnPaF/uGVcBj6ZrXOqRWiRwtdPHIw/dxQpucj16VJo+\nvJqIeFLpp5FOPnXaw+tZus3wnla0Z5AAwDCM4LH0rktU1W2s7ttObTpopgdpJOTmta1u7DQbZYpp\nZTPtGdhPH4DtW7o2pTzTLHI0nlyANGZB83PvUmu3jxMIi0hRPmkWPqR6ZFVtO1q2vCsUMTLkcEqR\nkfjWLaWtnfoPtITeOjADIp8GkWC3iLbKhdW5YAZ/SrN9olrc3WZHRZCOAR1FQ2mk2ejny7SINLLx\n8g616bb+Xc24DIoB55qxLbxWtsGJCxDkDiqdptMzQmMneO44rN1q7Xwlo91qMForvLIiMobCjJ5Y\niuZ8ORw3fhyxjm+79nTHP+yK0oLO0juAkLbyDk5OcVWns7a6vpkdoxIxypYA4IrPuPCumwqbi5gi\nkkznO3v61sWdtFcaYiTDzIsAbc4z+VWBEqX0EECbVVSTGDUkSJHdvFIEDHJUSAMKvSw+Xaybxg9a\n4OK0DxK79faq9xbxK+1Fwx6nNddp2kJ4h8OpBK+JLWTCkjIwfUVleNtHttC8O2ltHCEnuJRuEZJD\ngZ5A+pqle6ZFqXiaZbhAUa2RFPtk5rW07QrO1mt7IRLtDmU4FaGpaTb6qXt51Ug9PUe30qhF4Zsb\nAAoXwBgISdv5HvWl/ZFtdrFeRq0dyibfNjba2O4yKTTNFtdPld4Y8SNksxOSa5xPLXxBcJOFIdsg\nt2NWbjRbGKRrkQJ5nXcRzV/+xYbrEnlIrYwTtHIp/wBiEd9Agc5BzzVy4sI5Zm3PjJ5NOtbJLYsQ\nM46cCvJb9p9N8zcXEanDFBk1as7zVbZ4Z9P0+8MLAEq8WVOeAQRS6vrM03lG9Se3uyf3WYio+lXd\nK1ERtFc6nIkMSnkyHAz2r1KIPH5ZTkN2qQ3FpexPbXNwm5hjbuxj/wCvWRHZ6zYayrW16LjTz0eV\ngSntnqas6st3f6Hf2hjjlZgwBccYNebaSb5tDgjt5ApjXYw78HHFbmn3xtECwNE8gyJEkO1t31NV\nLi+lfMlybe3+bcf3g3ZqvJr09zbFUzKvQOvINdbojMsB3DIXP86ileCO9eaZ9r5Clt4Xr0GSQKfH\nNarcKQ8e44yRKrtj8CTW95e63KlsjBAz6V5gLwI3kyPtOPlUnGaaGlVifL8wMeMDn8K7TT3ubbwv\nc3VvHMjoyu2QV4HX61z/AMRPEMkg0F4QJGWJpWU++MfyNTXplh1i0kjxiQFPqetUnvpL3VI5o9R+\nyyR/Iy57e4Nb0MqxKzC8a4fHBZwSp9qkn1CWe3VmGD0P1qbS9WzpQuIgXjkQMnvmpFvNkS7rtVlb\n5iNw5rjdTSAXss02pbix3Mg4AxV86ut7pTNC2/Z8pIOa1IdVuEtmDRlWXg/41C1ytyY5Le/iDDBJ\n3A1bEnlq7T3okB+bC9BRY6q5RwUeWIfcfGA1YfiPTEF0xwPLmXp796wbK/1zSV+z26SGJflXb8ww\nOnBPFWWgub+T+0NSX50UhAxyR7n/AAqG8it7iDybhQc9vSvU9LvFu9OsrgHOW2k1eudKtmfzYf3E\nxOWkSNST9cg1XjsftNyq3MsTxq25fLh8ps++OtaUyRCERDo3HFeK6VqJgu4ppEWO1lkMLKp4Bzw3\n4k11R8nT7iUywJLFMMkOm4E4xnjkHFUb1bXUD5dvYwoGPzGKIr6dz9O1U9amh0qzjgQKDj7q10mj\n3yRyvFIQWJL89wf/AK/86tRR28lz9oaNGLH+7nJq5JGjyY8vaQP7oGKJNQWCyZQcuflUCuAubdbm\nAFkXcBxkVs6FawBo/Nt5InAGPIl6/ga67WrFLTRbi81G/lFvwVj3/LGnp7k/1rxrWdSOqag9xt8u\nPYqxp/dXqB+WK3/FGswWr20H7wXSyq8ZUcCrVkTdsuoLEN8gKvhipOD6itlLKC4AlvIfuchWJbH5\n9KqXF0jkKvyoOgFZvw916Ge0/sueTcyDMDHuvQr9R/I109rpyaVqstzFbpPFMBuSQbgp9R3rN8S3\nE+oxukVjBEpyCyoB/OsZPI0fRUtggUs29/UmrvgrxFDrWnxmXb5wXy5VPqO9bQ082Fyfs8SGB23F\nCowauyQvdHy44RDG3XaAM/lRqBjsrUIij5F3YHtVF7qeMbbaMGXGQWOAadZeIr1Ci3axDcOVQZ28\n96072/hgjEko/wBYAFx3rl9UvLG8tpmiZQ6dQCKobINMlWSS1kuQfYtgfSuh0TW7XzmezEsB6NDI\nTg+4z0rrrrVPIt4nIB8wce9clfyy3pkY31ltPRFJJzXGaeYIImnuJNi5AweM1rLexQYVY2GecMP1\nqA6khuExHkSru4NVZb62ZTKtuww209jn6VynhaSTS9SuJ0aIMsW3MmcDJ9vpXrHh3xLJqkbIyp5i\nqSCn3TiqeoeLdRtHVXa0itycbnBJ/SpbXWHukw7xzKeQ8dXnufsy+YSqqOdzHisjUvE8NoYWjluH\nEx+R1h+U+uKvTvKNMa9mO4KM5xg15d4xvvM3wXMbrcEhkGcKo/rUl5pt4hH2icnH3VPOfyqbSZ43\nYI6rnOM4q7qkawxbAq7iM9elS+EtEUa5a3l1J/o7SKpMfJyTgY/Guvvnu9L2vplpu3cswJP4VraX\ncS4Es8SxSsedvQ1T1zVljiXzGbymfbhQcmuF8ZalFBaWz2mXilzweoPeuwGqx2sPmJIsbZwRL6+m\nKr6f45N3dta4TIbaWQ5H69Kn1fxdJplwbaOISyenQD60lhqct9bXF0jCUrE28RHcOR04963ktoru\nBjuA2jIqmlrbBSQBv6dcZqrfRHyI2b5zG/APvXO6vGIbGe4Eax7QWAA713Gjae01qGEYwBmq+qxx\nwHaI0WQnG8nJptxPFObe3fc0SptI7802TwjpQiaZLZVPXzFyDXI2VsjXEcvliTZg7W6Z9amvVTbk\nR4dvfJNQPZrHbpcLKDKhyFJ/SrgQS26yKq7SAcgVzXw70yPVZNRklVXYbAu7oOpr0Ww0uOwEqxqF\nCxkfKMDmmtosF5BiQZAGMEcEHsaswaPBaruSFEOMDaMcClgjW5twjIGOO4zVGfR7Sa7jM0S/IeMK\nBita6tElsprcchk4Ga8m8b6ZCmmJMkQEokHIHOD2rtrzTElQOjgexArHOkQiYMhBIOTitS70FJ5I\n5QVP7sAgjrXS+F9Lit9UsI2ih8oNvwowAQOP1qfTCDaiQ45GarzyrLqCoWUHsCcVXFvBOZoZVWQI\n+4EH7przzxxFawYRVCiPkD3Jrtf7AFxi4WRdy9CyFyv05yKj03wxY287SJIHlEgZyqAAH0wKs+Id\nKWTUlu0uxBNKOpUFScAc1h3sEfhzQpIoGUTXRUZhOFHOSf0relkeDZGN2xhk4pwzIFmCqUQdF6VU\nvtRt0sSN7+eW4UrjHpXKeL9Txpawn78x5PsK7+K7ns7AAMwUjGQKz5J7a/dWhlWNsgbiuQfrU144\n85TPc24KAY8nj9a07HVIVhe1HzKxLhidx561xNsHgcEDg+lLcKsjfPIEJOSc4xUD2ViIuJlBHzKQ\n38VR28rx7og4Yex61ifDPUPs09/EOZGRXA9hnP8AOuyfU9QmMotmbYxAPHIPetW1uJLeNhdzsZGU\nbWAwBVmXVpHiCOgDD+IdDSh5LRdwPCjJxUVu63su+S4iRQfuFwG/H0qDUNUe0DJbzxMx+UDAOT6V\ny+rrPcXEVkYw8w+Y8cA+pqzeXUogyu4qMltvJNVFuJGijntN4Q8nKHDfWrlzqTSwJHNL5cp+4vfP\ntXYeBJpLi7Z7nkxR5/HNV0la0hiIVpEbCgBgOfxqZ7qJ4m+2WcseOcsvX6EVn2+p2PnSQWY+bGWX\nHP415L8Q74nV2jVvmP3lz0x0/rXoGqahPawhY5jFFISGfP3RSJc2CWK/YtWWOXGTsk+8ff1qG41O\nVrKRrjVYZJc5SONh17fWqWrXL3NhAZTiTdkL7Y5/pXcRRiaFkZsEEjPcUwQ/Zrd2uAjNnAKHaxHr\n71gJPb3Gqu1x5n2aJc4kbOWrgPE+pJq+qMUA8mP5YwOles6RqUslgomXhlH4GpEsZIr1Z4SArH7h\nOAabqtnPqMyidESPuqkHP6VbtoLS0hVVjQNjBwOlcpFNGcEkcZzzVaS1ZrrzVmG7phwCCP6ValMu\nwAmINtxwtZNvbeRfMzuXZzk56VxPg29ay8T2jgFlYlGX1BFexW1vZw6rJJLEHhZd21mYYPtg1Yub\nWzvLgLDDGqYz8rsSP1xTLrbbRRwxKxI685IFaiGOeINkEHg064srWZFMsSswHDY5rlLy1FnfR3CK\nMo+QNuKry3sWmQy3srKZ3OXk6hfYepqaCHJ2k8GpliuLRj5Q4PXB61JNC0y/aJ48SY2hj1xXQeHU\nbT7R3YYeUg49AOlUYALnT3hlGVORzUWzUrS2EVtcyKO2CCPyNUFlfR9H1XU5R5tyFLse5IFeI37N\nft9oJZpXJZmY+vNey2Z8+4QSKpjHIz61FPZxW0zsmlvhjyYMr/Kqf9nW9vFJeGzaJ5CAglOWOPr0\nFZtzcC5cEMCqjAI6Vt3OqNYoGnvUgyflHlk5rZ0bxBPLsffA6no8eQfyNa+q6/fWdqTuA4yWfkAe\ntcQ2oWmpSCS4upWmkOxGEO1M+laFpPqHhfSxbQWxn/eSOcD/AGjj+lO0Lxpea1f/AGVrN4nwWwVx\nwPWlv/GsmnXxtXtzKw4KheRVmx8W22rgJFBgn1GDXONcQwZadyoH+1irFnqrm6jexv5TGDymcj9a\n6O+1a705InVkjd+cyDIIrP8AF2sTXXhg+d5ZZ2Ubojwa4LTnm0a8nnM10DFAWQ7vmOSo6/Wul8L+\nM9XvJjZzNLLK4LJvOc4681ka74712DUmSK6lhCnaUArY0XxPc6vi1vpS8vqy4YfjVOw1WxsJHElu\nxZsclh+QrrdA8c27agtrHBKd/AXG45+grpF8cabZXDRXLOjqMEbec/StOHxFZ6iwSISAsPlLLgGu\nZGzS7JYoNoUDq7YFY8ms3v2jyluLKdW7K/zCp7uUW9sHuEIYqSFrhp7htT1W2toYniuBcxmPeuM/\nMOa6ywN4dkVtEjMByZDgVvWd46t9nvbdYpc4BRsg0281F4nMUIgRuN8kzYCcVRk8yfgXsF4CMnyS\nCB+Vazava6Z5Iut0chQMxC5AzUsXibTNSmVLa5WWUAjbgg/lVbUfE1tGAGZ2I6qi9K4+816JriT/\nAEaUgnjPUVtQW1pqcMQniVyOhI6Vq3Gm2umWgCKgLgDHUqKkZomk8q4+ZHjA5FVLzSLK3iiuYYo2\nzkqR/Ca2JpLFg8NxKMHkbTjH40mk2liJXubQBxyvmYzn8aqXWlWFxd77nbHI5+VuhNTfYrPT4v3b\nLK+OGYDj6Yrm7OzsdUiCzoufWpZtNsNJQRwmPc3P0FaEtlYeItFht/OQyIu3k81mal4OtdN8NvG8\n0wiMis+Du289RUthoaf2uDLbpGjwtGXIBGcqR1+hrc8PeFdN0vUJbyKf7VMqlQzAAIT2AFYGp+FN\nO1LUG8+RbedjuRhjJH0PWnLoVno8YfelzMP4yoB/Ss/wnpVlPG11M4SVW4bAOPzzW5oUNlqHjR7i\n1hUi2TDS7QDIfXj6Vt654U0nWL9Z2RFvTyQSQHHv/jWlZ6IlqkZMZjCcqvmFwPoT0rElgjngBdFb\n6isSTRbOa9DtEkZU5+QYz9ag8VW7XFrayLI6+W+3KemK5lIL2LUbdzcM7I4aN5BypHPbtXUQ6Q+o\nWqOGbb3VZNn6ipf7Ljs7mIpuB35I3k/zqybOC5vpY5UzuYMuRntVxNAgtlaVIoo3x/Am3P41b1nQ\nYtUl8m4XfGEDfK5HGMc+tZWnaLY6SSLaKJZI89OSBjvWDFGr6/50gV0bI2N0/KreseGrJUinighX\nPIKj5j/hVXTZZolUhSy+orVmnmvIyqsiP1G8ZpzfaFWKS+mgAQYwp61WN05yqMHhb7rA5x7VJ4mt\nJBcbLcBFCA4HHHtWTb6ve2kIFq0sWwbXhkUgA+tNe8le1kk1G+Zz1XcQAh9q0dOuJZbXzJM4x8rH\n+KsHSpLko3lH5lHQ0kt4l2xt7q0mlcHkkYz/APWrTtBpljCZUS5guNpSPcpxn04o1nxLPL4TkSVW\nV5cRjI/P9M12HiiGQaBcTWxKyxgMD+n9a8qjub1tE/0W8ujeeYXnVUcquexI74przvFo2+6vLz7b\nGco0kT4HsCauaZqV5KqC5ZuR0NZw1OaztXaKTAcnIHYetaXhDWtQsr4myML+YOfNbaMfWu+8QT3T\naPDerfRPf27CRDHhQR3UDNbWg+JJNVsIWlUhmBBHoR1rLZh09FrJt5on85/tCFy2MhuV9qwdY1W6\nAS1Vw0Ak5y25qtaFfQReIbVrlwiEFQx6AkcZ9K0rLUnisgEHIqFLmZrrzfKaZcjJHUH2q7fzS3V1\n50dq8aHaS0gwTj0FOTUJQnliQmM5+VjnFSX/AIgbVbJr7TCWWR/LUjopzjmnWiR6faOWXzSRiRx1\nLHqTXGajP5t65gRt6NjJBBWtS31SW601reRSZtpCe5qHQbkrI8TDcqNwfaq+uWEraqs8EjqM8r1U\n/UU+9gu7mFISLcqe0MOD+JJNblpHBZWUUO1eB2Fb2rWn2m2SVR+8h5PuveqEkUjRjbAsiEe1Ri2W\nUgy2iKw5+7VWbEWVXhRya5XSroxIHYducVo3k0MWnXF6ihpIoy6oeM4ql4U8ZpfX0dnfwKvmHCSB\niQp9waoeMb8SaktrE+YohyFPG4//AFq9puLcXFtJDIAQ6lT7Zrz/AFL7fptr5VpYNMFJSQxsyEY9\n1P8AOqlp9onheW6s/KkPOHYt+rEnua57UrlYbjCkbyeFFY+jRS6iFtkA3FtoJ6cV1nh3TIIjl5bi\nJw+35VSROvowOK7658K6ZqVjvuXmuJFXMbhFh2/ggGfxp2nwRaYiRKMBFwT6k9axo7s4YHuMUxpP\nIgLQIySL0ZOM/X1rlb6Qz6glxcSM3l5bB4ya3PBEkFxqdyxdDKF/1Z64POf5Vn6DeLfWzQs4E8R2\nsD1yP8R/KtrSJ7O3uXW5szOCc7mlcEfTBGK09SvbQgRwafASwxukkaUqPYk4H5VmvFb29n8o2cEZ\n/rWX4F3abayWV0f3Nwcrn+Fq6RbKGw1GWSWON4pcE+YzYBHpjpWNq7wSbhbWkcasMFlJLfmao2Ec\nGnQv+72uBwS2duevWrOia1bwrHA6BSFxkOCK2tW1a50qyWdEVYj1dgT+Q70ab4jsb+1UC6nW4bOC\n8eEbHXA/+vTdXsodQs3gukDBxkMOx9asaFYDRfBUCLdLZb08ySVsZJPbJ4qOwvzqLmNNQ+1qDjgK\nefXIpbuSe4+021iiGZJGDs4yAOK8/wBXa8tp3iuxE5PGYs8fUGtnw/DZaLpkbzFRLjJ3dzWvY+Lz\nc3RgWBFGcKVJzn3zVy/8YXGlzrGIFc9w524FVPFOo23iDwjdsIgs8MZkMYOeB1IqzB4mt7CK1s5J\nYZrgxgeWpbcTj6YrUg1pLiAyRIRgYZeuKwL3xxFaylDGcZwXc/KPyrP1DXI9RdFgaF3Az8oOQKmj\nMMcbiQ3EuAWIjOCAKv6DdR/LJb3dw8LceVcdRWxqF+sNuY4mQM44Oeg9a4aW3svtLRtel7g/NgJg\nfr1qrrk8ti1kcb/m3hXG78PwzW54O157jVZIzFI2/I2xZCdzuwRgDFU9T8USLrCzXgleGLdut1Yq\njc4GD9OelTzeJ9Oudas1sY5FgaMLN5qLksRzyOorz+w8WJoHiC5vYd7ybPKUYACDPbr6V6H4V+I8\nmrrIs8E80igt+6jBIAGecfzrlta+LWpm9ljsk8uFWwA45/GtLw547v8AV3Md1EZsKf3ca5bpkEH8\nK57RJLHTjNLeE7A3kxspPHcmu68O6zcQ3IEV39ptCdqtj5kPYGt7VfGGqW7G2t4Y5VI+bzOAv1rn\nbz7bco0k1zY73GfKj9Prk1Q8OeHILS1W5MpczRqpBUHOcGu9u7a3v9PNvL80SqOO/FVNK0qzt0Ih\nt0Cpk7iMmq2pXYL4XoDgCus1iNXtw0caGPG3yygIIrJ0fTtPg3ywQxwuAQETAArjprjd4hu7clvJ\nlfLKCRmqd/4ft7ODzoGcxgHAZyxP510H/CP22pxbwN0oXGGdlU/lSaZ4Zs9HdnzCsjcrHHkge+TW\ntrvhy21WK2u5JI4ptoTLjKsfem2fhixtNOmiBV5XRkZkzgA9uSavS6TZ3CRuIo4yqjawUdMetJa2\nEUEEzwgFXBAPrWNDpFvLKUYJuOSQVHNST2FpZIZBEu4fxbQKfp8MZjyoX5jznk06/jSC3YRoAx9h\nVGcEXNmJFDBoypDDvUV/a20zNK0KmXgb+/HSqiWy6pqVooKFkd2O88EenH+eK7Pwzpmj6DLNLLLG\nLmTjnooPYVS1rwnZX85e18gncSVJ4OfpXOXXhVtMhaSVolAGQE5JrznRPBz+JtUnTzlt2VN5bG4N\nzj8DXq3g7wZbeFrG8ZJfOnmUKzsMYFYOp/DHSdVuWu4p5baWRssgxtzWvovgyz8KKzxyPLM4xlsc\nVW07QLa6M8bWwjDOzlHwcZ9a0LLT7Ozu4re3hCgctjpgHPSormCDWbq6spCwDvuG04yR2NYUnw/F\ngDdmaZQh4O/r7YqtoXiEyaFaXEPl/uCFlVzgDjHWuwstTtbpM3F9EhPJXcoBo+2C2keO3uY542HO\nxslR71UcGQbsd66jWLp2sIRbOP35AVuwBGc/lUYVbGxRIJoWIXHIBJJ71wWrQSxasZmkiWTzA25M\n1Jr98zWioifeHGOmaluNeeygKJkMeABRE2rLZG7jmj82TlhIM4HoKW31fV9Rt57eWeOOPbghgOfp\n6VBo2vXa3ps53LY43eorQ03XLnVPDtq8bfMYwkjDnBAwa1F1pLbTRFDKsuxdozwQfeuefVGUie4v\nYxNnEaRnJ57VJeapcPaEldpx3rU0mZ2jMgbCjge5ovojqJHl3E0ZUY3oe/41UZ9QW+hkvLpZUQYU\nBMH86TU7kbA6cK1Y1jcvFdQSq2MSgE57Nkf1rrv7VUX/ANnbT2lliwc5ALA+ntV298R2DLHbpBNa\nXIPCeXkH8V4rM1d5JLGVnJICE15v4V1E2PiaCON/lmBi+vcfyr09rm3nmNuBctMo3OsaEjHvVpru\n1t0EUkEsGTjBjxz+FT3EUnlhnbcB0+lYU88tpqUgUHDjIqK11TZfn7NqESXGCBHKo2k/Wq1rMYtT\nW5e/guHQkskSbc57deaua3r3nW/lhdo/u+9eU6JcjT44M4MV5+5mRuR1+VvwNekeH4rpI47b7Hbx\nxAY3hRuP49a6VrSKCFhtG5uSe5rG1a9hsLV5GIG0Z+tZ/wAMvEx1vRYtJvgftVgAFLD78fRT9R0/\nKt7WtNSGQzW0ewucuqovJ9eQa4iXTZLjVt1x8kKnLKGwD9cVBrusJIVit14HyxKBkn1b/Crb+XfQ\n85DD7rDsaWy08WhYaiPtMOco0krDj0PanXFhY6pbta2ENrBLI2C8ZZyq55x2BqtdmHwtZTSlSWRt\nqburHHFN8G2t9ayCCeznt7X7yNyAT716DNYSuhktwmT1DCsW50uYO1zOsYKDhYxwKxtWnddPk2DP\nFbOhszaaSeMOQDVm1tvKLb5pZW3HDeaUx7YAxVHU0u5JEjgddueRKwOPoQAajmiC2iK7Aknjn0rG\ntYWfT53bIbjA9ODXaaVCut6db3Zk2SkbH5xyK6A2VvY2pLSqxxy7HNchq2oRTQyxRuCDxx6V48jT\nB0urc4kjcFfY9RXsuhXf9v6dFeQvscgLIv8AdYdR/n1rejsEt23uAW7sarajqS+Ube3/AHkuNqqt\nZ+pjzLxOxIxkVN4UuNJ0+0ubKcWRmMzPsvYSQc4wQ45/A1Q8VNYX37qKXTrW6VvkXT7d2c/8COB+\nlZV2wt44jM2WCksx74FUNH16203Rlldd83PBYDZz0PvW7pnxHtFRFuQkYccMPmAPPHqDxVTUPiK8\n04WJUSAcqzqfnX1GKjfxHDqkJi27SwIBHQ46j60vhzy9B8I2M73cNp56ebym4uTz/Kut0bVE1GEs\ntxHNgZDIuM1h6zr/AJEszmeVI4uqRICSPqaqCa317RpAvnEMMbZ1wfqPWvPZruFXJS5l684LEVNb\naqGZU3M75ABNXpda+wuQryJMpG10JGO/1rqNC8ZXIuIhfX00iH++jDH51o+JtXW2g3qB9ng2qMED\nJPT9BWDpmqR3/mBYXR1G4A9cVSvPFNzC8cS2oESMQGAyxP0p/wDaY1KAhZd+OxXawp2h60unhp4Z\nrbzG42znacexr0jwr4gbVCSFAkAw6qcj8DS6z4qu7OeSNBboicGSU4ArNbXraeNN+oCSR+8ZG38B\nXGx2w07TYZPtf2ckmT7m4cnuPxrc0rU5J3NvdJFKw5WROMii61mZJTHBb26IOC0shHNZ1zNdud8j\nQumeiHOK542thAdggdiDjcc4zWxo9wLdzBJGRg465FXtculWBY40Y5Gcg4xWFb6da320O0oeQZQs\ncivUb3U72z0NpoXVHZiN78Y6VxUetztO5nvFuCTglT0qzf37Wq4JKyMAAT2rn51mmfJuoHbrsJw1\nXNK0iLUbNJvsbz/aVScN8vGRnvx361vaPpCwapARaiFo3O3O0sWwR/DwByeKWTSU3lPs8TJuOxjJ\n5bKSenQgj0/Ko7rQxZDzmRAAmCFYNknrz+FbtlolnPoVlDPEjGK3RMsoPG0A9afZWen6WjrbeWgO\ncgcZrPgjsbi4D7kbJI3K2efSr12kUEG1ABXlT28JUkZHqo6UzT7WCa/jkOIowcBiOPrXV6zoGmWs\nIvZpftEBUYkhGCCeOfoabpF3DDNAkTq6o4KNMTx/n61J4j0YSalNCVJWJ1WMMAVYhe4PXrWdYafJ\nY5CnhRsJAxUcukrMwP3snGM4yPetA6VBYgbYAkjDJwOKo+FtA03WtGt7l1VpsEP3xz0rtvDVvFYa\nhceSgVAvYY9quy6dDqCSmRFcsSeeozVa18L6ZZsswtI96DhmUZFZVxYjbsdAwwB0zVdYlUvLGgDA\nbeBRZaXaahbZmLcd+v50v9hRWsrSxNhQMBccVR+yvOnChR6gVXFskEyhTls/lWlNaJcbRJwQOKmS\n1aEANGpA6HHIrc0K6N9oJUopZGKqSM4x3qhb6ClxqIjkliLn5ysaYwPU+lc3qe6bV58ONochAfSn\nm2DcmFS2Mbm64rK8C6xPdaQmnRZWaAGMMTnIJJH9a2Lq5uY7lYdLkvo9RgGTbm3YiTPc8fqKW1uL\nm2Uy6ja6lc3M3yBPKO3J9B2q5qH2i3tlSXzIyVz5cjBiB9RWjYajdPoFpLsYM0CE7hg8gc4qNdXt\n5Q8EdjNNJB8rsVCY7nqc1QutahgQ4sZY2PRUZGJ/AGp0uZ7mzDyBk7YYc/jXnM0r5MfQHqTSqtwp\nURSrtH8OM1pW819PA8U1ziFxgxgYFV4XayuVRjuRm4xzivR/FOowq6yQrmUuznH04/lXLm61B7Yf\nvoYUlfAbgN7n3oleztonD3zSSD7snHBH0qQ6o9xaAyA7iMBux+led+F/EV5pE7rASyvwU9fevVfC\nGvXE/wBpWW1dpWIxlgMdeK62CF2G/wA1Y5OoCjv71JPdSSwlXXa68MKwtQuWUqox0xmqSXF0iNCr\nQuCNuTjIptosljOSSSx6FRgH61d1C+aW2woG0j0xis1rxYLcD8sVRCtPKHTKDOc56mrrM7Bmm3EE\nY+mKuWt6s1rsbqpwfaqWl6i9toMoQ872bA69qXR9RuIYrydJh5kq7Sp71kTebcStJKgTJ6d6nt7w\nqhhyWUdCeTXEeGor7TbpbofLG4+Y+mOQf8+tex2R/tWOC7jlVJigGeoNXmtFtWE9xMryKOCAAK4L\nXb+TUNVKQSbmbgY6AVU+HniN5S+mXszOdv7kyHPAGNtehNZw3J8wxoxxjOKpy2EEbFgiKfXFZWo3\nSQwbUOSeFA7157cKkv7uReV4INNtLcWfyhFlQnK5JBFaSxNd27ReXDCGGGkGSce2ahis4LSdPLB2\nKOM9z610mrX/ANi8RKs3zQuMeozn/wDXT7CYWN/cG2trWSOZgxEibsH1HpW5OJrkKZdscJH+rRQK\no31vBbxGV8ADkDsDXhdtKYrlHBxzXpfhaZ4ZVnmml8qdyzCMhSB06+lekyX2nR2u2zglmnYYDm5J\nwcnnAJ9quqrRQRieTzJAoDMfWuJGorc311ps2GeOQ4HcrV20eO1LRxwo8ZbI8xA2PpUcscaztKZG\nUk5AHAH4dKzta1xI4vKhGXI2ge9XmijntSj5B7EHBqC2QQbkkhjl4OGfOfxwatywLeBVWNIU4z5W\n4H6ZJJqwIYoE/djGa53SpyJpoQMqckfyrSj0iACO8dD5ig8AcH6iqt6N6ERwxgngfKeKhg05Yo9x\nYl2PJzW9ZRrYx+dLLHGnQb66LR9XglmMSz282PvBBgr9avaveJBCP3iw46sE3EjGelc7cPba/o9x\nBDOJVKEjKbfmxweaTX/Edlo4MJgSQ46O+0Vzdp8QYrmY2408IPWNww/xqjr19aCNpGGC3OAOawES\nw1aEL+8il7bkODXq2lTz6RYQosVqX2jbG8u1j+ldx4f8QSTYintfJk/u7s1n+KPFF3EGgtIYAF5a\naZiAg9a5AXtrqkZil1M3M82duxQEPsvr+dbFtaiW0EMjlIv4sHBb8ai83SLW7EUUrNJjACuWwag1\nu6S3VYZB5kkyhtoHGOnNcRrWnaePOFuwjvVQ71HGfY0/Xdat7GOK4njMm0YQdskD/CsPS/EaPesg\nkYRS9EJJwfbNb2oas2l2qs+0H+HecCpNK1ddQjdluo2Owh40Xt6g5rTufE1xoqIizWlsScIssbu7\n+4C9q6rwx42vdYUWsxTzs4Vo0KA/gelc34z8VaqbySzspLhEiOJWiUZJ9ielcHd3D6vps4C33nQ4\nLm4mLqc9Mdvypmmrp8EMQuxIoHzOVOSRjmuugvreztozbtJ5J+4sw5IPQj1rQk1YRtHAjorFck4z\niuf159OuolY6gZJATuVBtI56Grl/4ptLOVo/NCsR3FUdP8TSXtxtDIyA4Bxg1f1bUp9OiBXIZxkV\nw+u68l88SyxPJIgPzuD+WDXrlnYQagqNMgcD7oIq1cWFpbfvI4kikc43gYJ/xq2LZJPKMgzxxuPf\n61ZeySGNpPKCnHTFZGo6NZ6vGTLFAH7O6g/zrEsPB+lWd55hdHcHO1EAx+VUvHmiW9/qMCqFUFMA\nZ68d8Vhaf4ZTTJPNkdlH91XyDXpV54JsdcvI7xt4IKsQh9M8e2c810U8MNlewP8AdkHbAGPwFNt4\n7LUrieC4hDlsg9ehpq+E9M0yGM28IVYSWjU5wpPUj0rJjiW8tAWG5QchfWqEemQxX3mmNIG+8Agw\nT71leIRHsgvPtkXD7NobJrO1K3Eln9rfY7MPvgDJPrSG3tZovKmRXQ9QwzVX+ydPWdVt448gg4A6\nVf1zQ7W/hiFwqZA+UGm6R4W07TFaaCP96643ZPFdna+HtP1+BPOt0Z4x8smMMPxrSj0qz0ZoIbZf\nnZwWJPJxWXbwWd5rl7aXMSFWfcoYZ61n+M9MttK0p4raNFQjLBRjJ9a43w8LfycvAkk3UlhnNJqa\nzaldpEkKoisMiIdPrTri0uLHU4Lgp8uMHdyMe/pS6vZWFzays8Bt5id5Yd2/r9a0n0TTb2UMiqj+\npQVIuhWtrIFjKvJwSABW3q2lwS2cLzgCPaMt6GuK8T+HrRNMM0T7ig3bvau40lpLcKVGcjIp9xcx\nXM8bPcJG6kjr+dXjLFDEonvlZE/vcDHvT5rh/snzMPL5IOf4a5TW7q4IZLUkKPlBHaqamKDSD9l1\ndEuW++VkUk+3Nc9JLdy3DPJdKwC4A3A8+uavWM81yyRScg8EV6v4e1WGGzBYhz2I5B+lJf3u7UEu\nNiNt4CscVAJ5l1Xz/LjidXDHaecY6VqXuuGa1ZPLzx1Arm7GUxwKUPHWqCPJcO6u7NLOSdiqWYgd\nABVDXNGcQHzbSfCJvBwuVGcZxmubine40mZFcmNCoVsYyDUmqTvbWzSxqWx1C9cVkWepXM8S3NpZ\n3JUnl0O78CBWw2upPaGO/WeBVHDOmAD2JNbejTGS23M+5Rn5+xGM5ra0u7vYbcxWpGM/K5/hFWf7\nUt0limttRglmiP7xJD99h1yexrnbzVS+sSX19cQ2kII/dxtvJ92YdPoKPFVxK2kJI8h8qTkF+DtH\nPNch4evRE4RlyWG0DOK6EaXFPciT7SImYggqcbTU99pE80oEl8JXUDhuOP5Vj6/exRWPkuNsqcEd\nc0w6lLCNwyCKbDqt3b757e5XzJeGzg1qxatcz2bC5v4442UqRuzn/CsDVLuRfD10JGIUfIDnqO+P\nwrtPD2rpLaRqxy8ZHWttbV49QF1aTGNj1X+Fgfar14s97gXcxeIEHysAAnrz+NYXifU7e20mZJpT\nDEU8tnVdxUdOB3NZ1swlcKw3BjUWpiygVhHaFXzkvFxn69jXNPaWdxL5rxM2zkb8cfgAK19Ltx5X\n2gjG7hfp61Z+HWvQ3OmJYXDj7RbjGOmU7EfSusWCGHUJZLqKW4hc/KyyEGM/hS6gti/7qztnmnbg\nSNMxC/XgVNJcRaXpQhdgSByzHJrmfDesR6lpMEyMCWQZHoe4pNY0RNSEdzHNJDPCx2skhTIOMjj6\nVz154esLazYz3FzNcdQzTM34darXGoi08O4ZQh3F+B/Co4/U1BYaksty1jPnePmQt0daemh3Vpdt\nPpzFAxzhXKEfl1rXg0a91AIdUmkkSP5lieQsCR37CnDXdMt7yTRFkzMkZ3BR1OPuj3q34J1m5SUx\nXSBrYEeTcr9yRT0+n9Old4beSO+N1ZRx7n+aSIgYc+v1rM1Syvb6aE3EKQwowOxcYY/QfSuB8fXs\n2orcWluxZbaHDbfzx/n0rG0YpMqsvXGVrqtJEUF0sU9lFNHnd5jOwdvqf/1U7XjBKqR2lv5Ls2Wn\nEpZl+nauL8RyRk4dyWX7xPoOlavmRu23ghuDVixji0yU7fkhk5KMu5Qfp/hWx51neoYkMT87tiRB\nRn1JrmtWsH13WLXQVmFtCyGV325J7YH6/nW8mrapYqEe/sIH/ht5sjH411vh7X7i6jVLqGJ3I+9A\n+4H6VDrfiO7hdorSwWXBHDyBf51zGsalqP2cNc2kXlMRkxSBiv1rkDeT2NwFQMGz3jOMfWuk0bxB\nLdq0aoS6/ewKbe+Np7K5EKbc5wcgnH5Vs6b4qF7HhvLMgHXJH86t6OkiXom1C9W2t3bATIBx9a7K\n01uyhLxwzmREzkhs5A4yOOelXLnW4fLdWOECB3YrjAPTHHJri9XlsL3K2GqIWPUFs4PpUGkSRCJp\nTIqxp6jrWqNVhRASwaLGdy54HvmrkN3bMjAsq46tnjOM4rL1OSF4CyMCOuc1Utra5s59lxdrAvZi\nCefwrobDXbi1lWKeSKeEj5JYwwJH0I5refV7WFN8rhQRkbuM1z2reIIGRikMzKDy2w4/OsHUbnyJ\n3bKDaejtjNLp+rxzyhJIij9ucg1q3WpW9vGTIHxjsMmorK4S/gZ0R1UnjeMGqGgwHTrFYTNDEUVV\nLOer45P0ya37LU5i4inMTbm2h4TlaXV9Wks96h4o9oBeSXomen1rntQnsdbihZb21u7i3kEmAOGH\nRh+INcOfErrHw0Y7Y6GtDS9fubhXjGcEemaUeLZ7O55ydpxycc/jXU6J44N6RBc+VhuMgitqHT4Z\ngHwFkI5IqW2soYbjfGijZ/dFZ66faapdzedDHJljwyBv0NVLzw1YWDvLFGqZHMaLtX646VzMkNrP\nCpeNDgd6taBDAdUVUQKhHJA61eu9F0yO62zYXfkqcYz+NalhY2drGViIx1BwM1fs9PguGE0iwmTG\ncS4/kag1C2t4HKxIoz1KcBee2KbLO5uV82LfG0e0g+nqKt/2fpzW43fZk4zjA3Zqpo9vG8aq7Kqg\ndfU1T1d4Uk8tG8xiRlieAM/rU2Y21AxOFdcKVBwVzipL2zjtbWRDEi71JUADH5V0NtarcjcAC59R\nTb7ywSDKhkHBwelVruBJpbaKQCQFCADWY+hKgxGZ0jBJZCx2n257VleI9HXUGWQSFVJ3DB4JxWfD\nYm2MXJyuAPer91FJqGU3YPQYq3pkYsYzbt0RMFs9TnmrUfh6z1qKOedFkyvKsSVJPfHrzWgmlWmk\nxxwwxog3Lwo4wDxVue2gu7h450GThsMoOfTrUa6JZ2irIlvANgONkQXr9K8kSzgltFnaWNcDnegJ\nH410Xhazg+0pKgDoRg7lwPY1J4q8PaXYS+fIHUO33/L3ICe1S+HtN06IeYjW/PGRGB+vNXDcSLGr\nPKwjJ5x/LNTxXl7BJutZYpIyDiNmBxUEJvIZWnu5Nj8ldnGDTr3U5JrcOwLMR1rgsyvBuRiYxycd\nxUdvq80LCW0kDp0GAeKvXWqLNbwi8u2adziNFQ8H24q7p13eeTh3IVWAJPpXcabeLHZGXCsxXC7u\neTVW6ciferLIwHIboDTdQuJi0aypAPl6qa0rW+2W/lO4dMcMeoP1qituY7R2BIwu4gdxWVaeTeT7\nZ7i1BPVJZMYHvxxT7wx2tzFseFg7BQIpA2M+vfFW9QSVLBnYn5VOKuxapLBbxrNIIkI5k6UkzyRh\nrmGISW5wXC8tj1HrRda5b3d5A1mrFUXBcrtCn8anvNciFn+8OJPukAdayGvAkflsdxQ5XPpVG4nk\nkmjYRtKB1VMZp0N2/wBrJSF4h2Dikvr5ZHSNcKfutj9a1tC1SSKxSLYcqoGRz7H9RUmoXccyqX1b\n7NM3KgYY9eAB1pEuEt91zqGrTTgjkmIpGPTnH9atW2pO9qUjPmQIMLIvIIxxXh0d1J5axykhAeSe\n9bdhrt3YENayCWMgAKf4av3ut3OqW+dQujs6iHoBVPTNSkguPJR9yMOCprtvD1xDrOlzW8yg703D\n61e0j7ZBEbaTTrC8SPKq8ibXx/tH17VWvoZr/UIrZtPtbGLduZolO4j0HP61S8T3qWZeCBRuRfLR\nf9rHJP0GK4HR7wwSiJmO09M1tRKLSZmWWSJWOSExyfXkVsWcVpdy+aY2ebbgPIcn/wCtWX4oaaw0\nm6e3OGC5/Wu1tY0NioUlXQ49s1UW80m0Rxqkl5HN6xkEMfYYrndR8RGeUmCKdl6K0kgzj6AVnS67\neTRlEkcDuCa7qzvxcaWuT8y/JJ9OxpLe1jMpMaNvI5ZeP1pr29sJ2lRWE2cEsOSfrVvUc3rQWKcN\nKRux2Xuau2MaTQSxugaNXZcMM5FUraxlgWX7LeRKit8kcgII9gR1/HNRRWcl7dGS+kICDoj8E+pr\nA8ReMdI0C5jgKebM38K87E9T7n+VSiUalbqY32uOD7VU0yTy55Y7tDK4Y7T5xTAp+pXiLFi2SRJO\ni/vy+T+IqM2062yT3UgZ8AcDHua2dE1mAatcWpUK4LSxrngxs5OOfSujuNNtr8xXlpDD58bYzjBx\n9RyKemlXV0R/aCxrED9zzC+78zijUZoYEwpVUQ7SB3b0rx7XrCHTLtPJcS2lygmjOQSpPUH8aXSB\nBGdgSAAEnEg9fQ1sNa2F4hjlkjGR9yJev1NZ11YrZyR/Z1VQwwMDFf/Z\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "H_TWbsTLcGbV",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "X = np.expand_dims(X, axis=-1)\n",
        "np.save('X.npy', X)\n",
        "np.save('y.npy', y)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vbP9HQKYcGbb",
        "colab_type": "code",
        "outputId": "d854003f-9c1c-4d39-9170-94514bacd7fd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 69
        }
      },
      "source": [
        "import numpy as np\n",
        "from keras.models import Sequential\n",
        "from keras.layers.core import Dense, Dropout, Flatten, Reshape\n",
        "from keras.layers.convolutional import Conv2D, MaxPooling2D\n",
        "from keras.utils import np_utils\n",
        "\n",
        "# load the data\n",
        "X = np.load('X.npy')\n",
        "y = np.load('y.npy')\n",
        "\n",
        "# convert classes to vector\n",
        "nb_classes = 2\n",
        "y = np_utils.to_categorical(y, nb_classes).astype(np.float32)\n",
        "\n",
        "# shuffle all the data\n",
        "indices = np.arange(len(X))\n",
        "np.random.shuffle(indices)\n",
        "X = X[indices]\n",
        "y = y[indices]\n",
        "\n",
        "# prepare weighting for classes since they're unbalanced\n",
        "class_totals = y.sum(axis=0)\n",
        "class_weight = class_totals.max() / class_totals\n",
        "\n",
        "print(X.dtype, X.min(), X.max(), X.shape)\n",
        "print(y.dtype, y.min(), y.max(), y.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "float32 0.0 1.0 (13165, 32, 32, 1)\n",
            "float32 0.0 1.0 (13165, 2)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vkUa9CA4cGbj",
        "colab_type": "code",
        "outputId": "1e39e5b2-e2f9-4080-8228-e5cab8ae48e8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 416
        }
      },
      "source": [
        "nb_filters = 32\n",
        "nb_pool = 2\n",
        "nb_conv = 3\n",
        "\n",
        "model = Sequential()\n",
        "\n",
        "model.add(Conv2D(nb_filters, (nb_conv, nb_conv), activation='relu', input_shape=X.shape[1:]))\n",
        "model.add(Conv2D(nb_filters, (nb_conv, nb_conv), activation='relu'))\n",
        "model.add(MaxPooling2D(pool_size=(nb_pool, nb_pool)))\n",
        "model.add(Dropout(0.25))\n",
        "model.add(Flatten())\n",
        "model.add(Dense(128, activation='relu'))\n",
        "model.add(Dropout(0.5))\n",
        "model.add(Dense(nb_classes, activation='softmax'))\n",
        "\n",
        "model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_1 (Conv2D)            (None, 30, 30, 32)        320       \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 28, 28, 32)        9248      \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 14, 14, 32)        0         \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 14, 14, 32)        0         \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 6272)              0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 128)               802944    \n",
            "_________________________________________________________________\n",
            "dropout_2 (Dropout)          (None, 128)               0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 2)                 258       \n",
            "=================================================================\n",
            "Total params: 812,770\n",
            "Trainable params: 812,770\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TmEzI7LYcGbr",
        "colab_type": "code",
        "outputId": "c658dc00-0d30-45be-ab0a-52727136a822",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        }
      },
      "source": [
        "validation_split = 0.10\n",
        "model.fit(X, y, batch_size=128, class_weight=class_weight, epochs=5, verbose=1, validation_split=validation_split)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.cast instead.\n",
            "Train on 11848 samples, validate on 1317 samples\n",
            "Epoch 1/5\n",
            "11848/11848 [==============================] - 5s 422us/step - loss: 0.4918 - acc: 0.7567 - val_loss: 0.3680 - val_acc: 0.8345\n",
            "Epoch 2/5\n",
            "11848/11848 [==============================] - 1s 121us/step - loss: 0.3498 - acc: 0.8498 - val_loss: 0.2942 - val_acc: 0.8853\n",
            "Epoch 3/5\n",
            "11848/11848 [==============================] - 1s 121us/step - loss: 0.3156 - acc: 0.8710 - val_loss: 0.2636 - val_acc: 0.8975\n",
            "Epoch 4/5\n",
            "11848/11848 [==============================] - 1s 120us/step - loss: 0.2829 - acc: 0.8868 - val_loss: 0.2501 - val_acc: 0.9028\n",
            "Epoch 5/5\n",
            "11848/11848 [==============================] - 1s 120us/step - loss: 0.2702 - acc: 0.8937 - val_loss: 0.2392 - val_acc: 0.9081\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f0231ed8470>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 114
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YOgt-Z0ScGby",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "open('model.json', 'w').write(model.to_json())\n",
        "model.save_weights('weights.h5')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GAYDOJY8cGcG",
        "colab_type": "code",
        "outputId": "e752cb90-61e1-4b6f-c8b9-dcfe40acf136",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 320
        }
      },
      "source": [
        "%matplotlib inline\n",
        "from matplotlib import pyplot as plt\n",
        "plt.plot(model.model.history.history['loss'])\n",
        "plt.plot(model.model.history.history['acc'])\n",
        "plt.plot(model.model.history.history['val_loss'])\n",
        "plt.plot(model.model.history.history['val_acc'])\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/keras/engine/sequential.py:110: UserWarning: `Sequential.model` is deprecated. `Sequential` is a subclass of `Model`, you can just use your `Sequential` instance directly.\n",
            "  warnings.warn('`Sequential.model` is deprecated. '\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXIAAAD4CAYAAADxeG0DAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIVzu/wP+/Fcr0l++JgAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jND3ZLLtcGcL",
        "colab_type": "code",
        "outputId": "f5ef6da3-20ac-41dd-c017-bd4e45fa66b1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "from sklearn.metrics import roc_auc_score\n",
        "n_validation = int(len(X) * validation_split)\n",
        "y_predicted = model.predict(X[-n_validation:])\n",
        "print(roc_auc_score(y[-n_validation:], y_predicted))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.9585530065545739\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}