[Project_100]_Image_Classification_with_Tensorflow.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "[Project 100] Image Classification with Tensorflow.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "authorship_tag": "ABX9TyOatmU2U0l6jSflymWzCpRB",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/id-shiv/project_notebooks/blob/master/%5BProject_100%5D_Image_Classification_with_Tensorflow.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IfqA6165Vtfb",
        "colab_type": "text"
      },
      "source": [
        "# Neural Networks\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IQVt84VOgqW3",
        "colab_type": "text"
      },
      "source": [
        "## Introduction\n",
        "\n",
        "* Input Layer : is the first layer.\n",
        "> Consists of recommendation for a certain input.  \n",
        "> Flatten : Convert multi-dimentional (array within an array) data into one-dimentional data.  \n",
        "* Hidden Layer : include hidden layers for computation to improve the accuracy of the model.  \n",
        "> Number of neurons in hiddden layer may be percentage of number of input features\n",
        "* Output Layer : is the last layer.\n",
        "> Outputs probability of recommendation being correct.\n",
        "\n",
        "* Each connection from one neuron to another neuron is associated with weights."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "43vZfrrQgg5e",
        "colab_type": "text"
      },
      "source": [
        "## Model\n",
        "\n",
        "(known as Weighted Sum)\n",
        "```\n",
        "Output = Sum ( Input Value X Weights) + Sum (Biases)\n",
        "```\n",
        "e.g.:\n",
        "```\n",
        "Model = v1 * w1 + v2 * w2 + v3 * w3 + v4 * w4 + b1 + b2 + b3 + b4\n",
        "```\n",
        "\n",
        "## Working\n",
        "\n",
        "* Model starts off with random weights and biases and checks for correctness of output in training data.\n",
        "* Model adjuts weights and biases to predict the training data against its output with multiple repeatations. \n",
        "* All of the output neurons will be fired to certain value which indicates the percentage \\ probabibility of that output being accurate"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RRY0I63_hGF5",
        "colab_type": "text"
      },
      "source": [
        "### Activation Function\n",
        "\n",
        "Context : Above model is similar to a linear function (mx + b) limits the degree of complexity.  \n",
        "Allows you to create the above model with more complex non-linear higher degree function.\n",
        "\n",
        "Output : \n",
        "```\n",
        "f(Sum ( Input Value X Weights) + Sum (Biases))\n",
        "```\n",
        "\n",
        "e.g. : \n",
        "* Sigmoid function (Sets the output to be in between -1 and 1).\n",
        "* Rectified linear unit (relu) - takes all the values that are negative and makes to 0 and all the values that are positive makes them higher positive.\n",
        "* Softmax function"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "V5QADLaRhLkb",
        "colab_type": "text"
      },
      "source": [
        "### Loss function\n",
        "\n",
        "Loss functions are the ones responsible to adjust the weights and biases to predict outputs with better accuracy.  \n",
        "Loss functions is way of calculating error while predicting outputs (in training).  \n",
        "\n",
        "e.g.: Mean Squared Error.\n",
        "\n",
        "* Loss functions are non-linears - hence handles high degree of complexity."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "A92k11GDfHFW",
        "colab_type": "text"
      },
      "source": [
        "# Code\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bvjGVoyioSSq",
        "colab_type": "text"
      },
      "source": [
        "## Libraries"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zP95f71hF4Jn",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import tensorflow as tf\n",
        "\n",
        "# Keras is an API from tensorflow to write less code\n",
        "from tensorflow import keras\n",
        "\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import pandas"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OFDTi2G4n_E1",
        "colab_type": "text"
      },
      "source": [
        "## Data"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bSry26NmitPb",
        "colab_type": "text"
      },
      "source": [
        "### Import"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_i1ee4dKi2H8",
        "colab_type": "text"
      },
      "source": [
        "Data to be imported is 70k+ labelled images of fashion accessories"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zLu5ZoV7h9ui",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "data = keras.datasets.fashion_mnist\n",
        "(train_images, train_labels), (test_images, test_labels) = data.load_data()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8fXfRvaljRNm",
        "colab_type": "text"
      },
      "source": [
        "Map the labels (integers) to actual text"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MaTcrcDrjOy9",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "labels = ['T-shirt/ Top', 'Trouser', 'Pullover', 'Dress', 'Coat',\n",
        "          'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle Boot']"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VEqnJzD7kgO8",
        "colab_type": "text"
      },
      "source": [
        "### View Data (Images)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EYP7pqRUlHHd",
        "colab_type": "text"
      },
      "source": [
        "Plot the image\n",
        "\n",
        "cmap binary - for black and white"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_GjU4EAhkwVZ",
        "colab_type": "code",
        "outputId": "dff8492f-a2b9-4c36-f23e-c1b52149b0ac",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        }
      },
      "source": [
        "print(labels[train_labels[3]])\n",
        "\n",
        "plt.imshow(train_images[3], cmap=plt.cm.binary)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Dress\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD4CAYAAAAq5pAIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/D/Y7avbI09FywAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "o1arH2i1nUSD",
        "colab_type": "text"
      },
      "source": [
        "### Prepare Data"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PpveYfhhnhcA",
        "colab_type": "text"
      },
      "source": [
        "Reduce the values (ranging from 0 to 255) to decimal"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4Y2y2mbunXkK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "test_images = test_images / 255.0\n",
        "train_images = train_images / 255.0"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "x3X8xbKroLa1",
        "colab_type": "text"
      },
      "source": [
        "## Model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LbGqITpcpLqI",
        "colab_type": "text"
      },
      "source": [
        "* Flatten Data :  \n",
        "Current Data has 28 rows X 28 pixels in each row = 784 pixels\n",
        "---\n",
        "* Dense Layer :  \n",
        "Fully connected layer - each neuron connected to each layer\n",
        "---\n",
        "* Output Layer :  \n",
        "Current problem requires 10 neurons to represent 10 categories of prediction (i.e. image labels).  \n",
        "Require 10 classes of labels for recommendation, set activation function as softmax"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uUNqYXPGp4rO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model = keras.Sequential([keras.layers.Flatten(input_shape=(28, 28)),\n",
        "                          keras.layers.Dense(128, activation='relu'),\n",
        "                          keras.layers.Dense(10, activation='softmax'),\n",
        "])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "F_1jD6fJp--k",
        "colab_type": "text"
      },
      "source": [
        "### Compile"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "z_UbCN-5zG5n",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8_nnIi8fDon4",
        "colab_type": "text"
      },
      "source": [
        "### Fit"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yFruB1MgEIAY",
        "colab_type": "text"
      },
      "source": [
        "Epochs : # of times the model views the data for processing, required since the accuracy of the model also depends on sequence in which the model view the data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7mQrpcIhDo0c",
        "colab_type": "code",
        "outputId": "f47c6785-d0d9-4924-9abb-650238ec1d3c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "model.fit(train_images, train_labels, epochs=2)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 60000 samples\n",
            "Epoch 1/2\n",
            "60000/60000 [==============================] - 4s 72us/sample - loss: 0.4957 - acc: 0.8255\n",
            "Epoch 2/2\n",
            "60000/60000 [==============================] - 4s 70us/sample - loss: 0.3752 - acc: 0.8637\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7fe5d7727b70>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 40
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lJ0sgvioEk_T",
        "colab_type": "text"
      },
      "source": [
        "### Evaluate"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KTGefwqqElKG",
        "colab_type": "code",
        "outputId": "ece21e32-bd5f-4f23-81c5-fe8630e88588",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "source": [
        "test_loss, test_accuracy = model.evaluate(test_images, test_labels)\n",
        "\n",
        "print(f'Test Loss : {test_loss}')\n",
        "print(f'Test Accuracy : {test_accuracy * 100} percentage')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "10000/10000 [==============================] - 0s 34us/sample - loss: 0.3727 - acc: 0.8667\n",
            "Test Loss : 0.3726923031926155\n",
            "Test Accuracy : 86.66999936103821 percentage\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MeQx_8yoqA5u",
        "colab_type": "text"
      },
      "source": [
        "## Predict\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Gew-UfltEAG0",
        "colab_type": "code",
        "outputId": "59c7d54e-e914-4f92-b5c8-92700aeb28e8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "print(f'To be predicted : {labels[test_labels[3]]}')\n",
        "predictions = model.predict(test_images)\n",
        "\n",
        "recommendations = predictions[3]  # Get probabilities to all labels for image #3\n",
        "predicted_image = np.argmax(recommendations)  # Get the label that has max probability\n",
        "print(f'Predicted : {labels[predicted_image]}')  # Map the label to label name"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "To be predicted : Trouser\n",
            "Predicted : Trouser\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qhJQW1iAIFD8",
        "colab_type": "text"
      },
      "source": [
        "### View Output"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CpqehmPLIHOz",
        "colab_type": "code",
        "outputId": "f9d7ba05-ffc7-4ebe-8ef4-8610293a3a32",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "for i in range(5):\n",
        "  plt.grid(False)\n",
        "  plt.imshow(test_images[i], cmap=plt.cm.binary)\n",
        "  plt.xlabel(f'Actual Image : {labels[test_labels[i]]}')\n",
        "\n",
        "  # Get probabilities to all labels for image\n",
        "  recommendations = predictions[i]  \n",
        "\n",
        "  # Get the label that has max probability\n",
        "  predicted_image = np.argmax(recommendations) \n",
        "   \n",
        "  # Map the label to label name\n",
        "  plt.title(f'Predicted : {labels[predicted_image]}')\n",
        "\n",
        "  plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAEWCAYAAACg3+FOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nGRsyIwAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAEWCAYAAACg3+FOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nF7tZIVzsZoVwsZsVwsVuVoj/D5G6kFuroz1jAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAEWCAYAAACg3+FOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAEWCAYAAACg3+FOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAEWCAYAAACg3+FOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0\ndHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAbGklEQVR4nO3de7SddX3n8feH3ElCIFcSJAQCmIVy\nqZ6CMyCmU8pIXFxsKUodig4aOpVpddWO6LAUp4OytEjtWuo0CAXx0toRCm1BpSgTKCgcQ8J1MsSQ\nkITcLyQkmEDynT/2c2DnePbvd7L32Wcf8/u81jrr7L2/+3me337O/p5n7+f7/H4/RQRmdvA7pNMN\nMLPB4WQ3K4ST3awQTnazQjjZzQrhZDcrhJP915ikWZJC0vDq/r2SLh+E7V4r6VttWvcKSec0iL1T\n0tJ2bLcETvY2q968r0h6WdJ6SbdKGteObUXEeRFxWz/b1GdCtZukkZJukLS62icrJP1Vf5aNiAcj\n4s2Z9XfstQ11TvbBcX5EjAPeBnQB1/R+gmpK+Ht8ito+OB0YD8wFFrW60p5PN9ZYCW+uISMi1gD3\nAm8FkPSApOsk/RuwCzhO0gRJN0taK2mNpP8paVj1/GGS/lLSJknLgffUr79a34fr7n9E0rOSdkh6\nRtLbJN0OzAT+qTqy/rfque+Q9LCkbZKWSJpbt55jJf2faj33AZNb2A2/CdwZES9GzYqI+Gav55wm\n6QlJL0n6e0mjq3bMlbS6rl0rJH1S0hPATknf7eu1WSUi/NPGH2AFcE51+2jgaeAvqvsPAC8AbwGG\nAyOAO4G/AcYCU4FHgSur5/8R8H+r9UwEfgIEMLxufR+ubv8+sIZacgk4Hjimd5uq+0cBm4F51A4A\nv1Pdn1LFHwG+DIwCzgZ2AN9KvOZtwFkNYtdUr/mPgZMB9bG/HgVmVK/xWeCPqthcYHWv5y6u9seY\nvl6bf9748UefwfGPkl4DXgL+Bfh8XezWiHgaQNI0agl3eES8Qu1odSMwn9o/gEuAv4qIVdXzv0At\nAfryYeCLEfFYdX9Zon3/CbgnIu6p7t8nqRuYJ+kn1P5hnBMRu4GFkv4p9WIj4vBE+AvAVuADwI3A\nZkmfiv3PNfx1RLwIUG3rtMT6/rpnf1iak31wXBQR/9ogVv9GPYba0X2tpJ7HDql7zoxez1+Z2ObR\nwC/62b5jgN+XdH7dYyOofXKYAWyNiJ29tnt0P9e9n4jYC3wV+KqkMcB/Bm6R9GhEPFs9bV3dIruq\nNjTiRO8nJ3vn1Xc7XAXsBiZHxGt9PHct+yfZzMR6VwGz+7HNnufeHhEf6f1ESccAR0gaW5fwM/tY\nxwGrPr18VdLngJOofWQ/4NVk7lvFJ+iGkIhYC/wIuEHSYZIOkTRb0ruqp3wP+BNJb5J0BHB1YnXf\nAD4h6e3Vmf7jq8QFWA8cV/fcbwHnS/qP1UnA0dXJsDdFxEqgG/hcVTY7CzifJkn6WLXuMZKGV9cF\njAceb3advfR+bVZxsg89fwiMBJ6h9t32fwPTq9hNwA+BJdTKVXc0WklE/ANwHfAdaifU/pHaCS+o\nfW++pjrz/onqO++FwKeBjdSO9H/OG++PPwDOALYAnwV6nz3fT3Um/J0NwruAG6h9VN8EfBT4vYhY\nnlrnAdjvtQ3QOg8Kqs5gmtlBzkd2s0I42c0K4WQ3K4ST3awQg1pnnzx5csyaNWswN1mEnTt3Nozt\n27cvuWwu3qrU+keMGJFcdty4tnQOPKitWLGCTZs2qa9YS8ku6d3AV4BhwDci4vrU82fNmkV3d3cr\nm2wo96Y95JD0h5hWqhJ1V7t1xCOPPNIwtmvXruSye/bsScb37t3bVJt67N69u2FsypQpyWXPPvvs\nlrZdoq6uroaxpj/GVz2xvgqcR+3qp0slndTs+sysvVr5zn46sCwilkfEHuDvqF2YYWZDUCvJfhT7\nd0JYXT22H0nzJXVL6t64cWMLmzOzVrT9bHxELIiIrojoyn1HM7P2aSXZ17B/D6w3VY+Z2RDUSrI/\nBpxQDVk0Eng/cPfANMvMBlrTpbeIeE3SVdR6YQ0DbukZcaUTcuWvVjv8tFJe27FjRzL+4x//OBlf\ntCg9HuO9997bMPbmNycHY82+rpdffjkZ37x5czI+adKkhrFf/vKXyWWvu+66ZPz889M9bS+44IKG\nsZkzU0MBHJxaqrNXwxjdk32imXWcL5c1K4ST3awQTnazQjjZzQrhZDcrhJPdrBAHzbjxrdbZW6mj\nL1iwIBlfujQ9y3Cue+6cOXOS8fe9730NY4sXL04uO2rUqGT8tdf6Gr7+Dbk6/vjx4xvGxo4dm1w2\n15di5crUHBnw8Y9/vOltX399src2M2ak5q0YmnxkNyuEk92sEE52s0I42c0K4WQ3K4ST3awQB03p\nrZ2lNYCvfe1rDWNbtmxJLnvssccm47khlXMjvE6dOrVh7F3velfDGMAddzScGxKAI488MhkfOXJk\nMp4a1ffkk09OLpvqugtwwgknJOMTJkxoGMuV7a655ppk/JZbbknGhyIf2c0K4WQ3K4ST3awQTnaz\nQjjZzQrhZDcrhJPdrBCus1dWrVrVdPy4445LLpsbjjkn1x1z/fr1DWOzZ89OLpuLP/fcc8n4xIkT\nk/EzzjijYWzhwoXJZXPdSHNDUadmsB0zZkxy2XXr1iXjt99+ezJ+2WWXJeOp92u7ZgX2kd2sEE52\ns0I42c0K4WQ3K4ST3awQTnazQjjZzQpx0NTZU/2m+2PZsmXJ+LBhwxrGcsMtjxs3LhnfvXt3Mp7r\nz55a/7Zt25LLnnfeecn4Qw89lIzn6tWpfZPbb7m+9Dt37kzGU1Nl79mzJ7lsbojtxx9/PBnP1dnb\nVUtPaSnZJa0AdgB7gdciomsgGmVmA28gjuy/FRGbBmA9ZtZG/s5uVohWkz2AH0n6uaT5fT1B0nxJ\n3ZK6c9P5mFn7tJrsZ0XE24DzgI9KOrv3EyJiQUR0RUTXlClTWtycmTWrpWSPiDXV7w3AncDpA9Eo\nMxt4TSe7pLGSxvfcBs4FnhqohpnZwGrlbPw04M6qXjgc+E5E/GBAWtUBTz/9dDI+evTohrFcnTzX\n1/7QQw9NxnNTOqeuAdi+fXty2enTpyfj5557btPbzsWPP/745LK5/Zbrc56q4+f6wuc8+uijLS3f\nCU0ne0QsB04dwLaYWRu59GZWCCe7WSGc7GaFcLKbFcLJblaIg6aLa6tWr16djB922GENY7nSW860\nadOS8dSQyJAuMeWmg86VHHPTKm/dujUZTw0H/eKLLyaXzXXPTQ2hDekusrnXnZtme9KkScl4rgtt\nbqrrdvCR3awQTnazQjjZzQrhZDcrhJPdrBBOdrNCONnNClFMnT1Xk81JTbucqwefcsopyXiuFp4b\nSjol1wU1N2Ry7rXl6smpbqqvvvpqctm1a9cm47m2p9qWe105uW7HTzzxRDLe1TX4AzH7yG5WCCe7\nWSGc7GaFcLKbFcLJblYIJ7tZIZzsZoUops6+fPnyZDw3rXJq6OHc1MG56Xm3bNmSjOfq0a0Mi9zK\nMNWQf20bNmxoetnc68rtl9Q4AK1Ooz18eDp1nn/++WTcdXYzaxsnu1khnOxmhXCymxXCyW5WCCe7\nWSGc7GaFKKbOvmrVqmQ8NSUz5OvRKStXrkzGZ82alYznxhhP9XdPjSkPMH78+GQ819c+99pS+y1X\nq879TXL9/FNTOo8dOza5bO515+JLly5Nxjshe2SXdIukDZKeqntsoqT7JD1X/T6ivc00s1b152P8\nrcC7ez12NXB/RJwA3F/dN7MhLJvsEbEQ6H0954XAbdXt24CLBrhdZjbAmj1BNy0iegYIWwc0nKxM\n0nxJ3ZK6N27c2OTmzKxVLZ+Nj9qIgg1HFYyIBRHRFRFdU6ZMaXVzZtakZpN9vaTpANXvxl2bzGxI\naDbZ7wYur25fDtw1MM0xs3bJ1tklfReYC0yWtBr4LHA98D1JVwArgUva2ciBkJsL/JBD0v/3Wpmf\nfceOHS1tO1crT/U5zy2b23buteWuP0jV8XNzu+fq7K+88koynvqb5c4fTZgwIRnPve4lS5Yk452Q\nTfaIuLRB6LcHuC1m1ka+XNasEE52s0I42c0K4WQ3K4ST3awQxXRxTU25DPlupEcc0bhjX66b54UX\nXpiM59qWG3I51d0yVzrLxXPDNee6qaaWzw0VnXvdudLcnDlzGsbuuit9aUiuJJnr4porC3aCj+xm\nhXCymxXCyW5WCCe7WSGc7GaFcLKbFcLJblaIYursuZrumDFjkvFcPTnlpJNOSsYffPDBZDw3vXBK\nrl68bdu2ZDx1fUF/1p+qR+f2aW0QpOadeOKJDWO5Onhu26NGjUrGX3rppWS8E3xkNyuEk92sEE52\ns0I42c0K4WQ3K4ST3awQTnazQhw0dfbckMm5ftm56X9Tdddc3+YZM2Y0ve7+2LVrV8NYrpa9c+fO\nZHzSpEnJeK7PeSqeG0MgJ/c3O/744xvGUvusP+vOvZ9yYxSk4q1cV5HiI7tZIZzsZoVwspsVwslu\nVggnu1khnOxmhXCymxXioKmzb9q0KRnP1bJz9eJU3TVXZ89dA5CL52rle/bsaRjL1WwPPfTQZDxX\nT871C586dWrDWGqqacj/zXLLp65vyPXDz8mNf5B7P61bt65hLHV9QCuyr1jSLZI2SHqq7rFrJa2R\ntLj6mdeW1pnZgOnPv7dbgXf38fiNEXFa9XPPwDbLzAZaNtkjYiGwZRDaYmZt1MoXl6skPVF9zG84\nUJmk+ZK6JXVv3Lixhc2ZWSuaTfavA7OB04C1wA2NnhgRCyKiKyK6pkyZ0uTmzKxVTSV7RKyPiL0R\nsQ+4CTh9YJtlZgOtqWSXNL3u7nuBpxo918yGhmydXdJ3gbnAZEmrgc8CcyWdBgSwAriyjW3sl9z4\n57ladm6u79T6Z86cmVx2/PjxyXiuT/m0adOS8dRr27dvX3LZXK06N95+rs6eWn8rc7sD7NixIxlP\n9RnP7dPcfmll/AOADRs2NIy1q86eTfaIuLSPh29uQ1vMrI18uaxZIZzsZoVwspsVwsluVggnu1kh\nDpourrkuhbmunrkpeJcuXdowNmfOnJa23cp00JAuA+XKV7n9ltsvua6eqdJdrnyVa9uWLekuG2PH\njm0YO/nkk5PL5sp6uamsc23PDTXdDj6ymxXCyW5WCCe7WSGc7GaFcLKbFcLJblYIJ7tZIQ6aOvvm\nzZuT8VwtO9dV86WXXmoYO/XUU5PL5obj2r59ezKeq9mmaum7d+9OLpsbSjq333JDMqf2a677bW6I\n7lw31BdeeKFhbPbs2cllH3744WQ8937JXXuR+5u3g4/sZoVwspsVwsluVggnu1khnOxmhXCymxXC\nyW5WiIOmzr5o0aJkPFcXzcXXr1/fMJbr29zd3Z2M52rduVp2Kp7rMz5y5MhkPDcEd279qXiur3wu\nnrsGYMmSJQ1jhx12WHLZXD/93Ptl165dyXjqPXHxxRcnl22Wj+xmhXCymxXCyW5WCCe7WSGc7GaF\ncLKbFcLJblaI/kzZfDTwTWAatSmaF0TEVyRNBP4emEVt2uZLImJr+5qalhojHPJTD69ZsyYZT40j\nnuvPnqr3Ahx++OHJeK5mm5KbOjjX3z1XR8/1KU+NmZ+r8efWnevnv2LFioaxCy64ILnsFVdckYxf\ncsklyXju2onp06cn4+3QnyP7a8CfRcRJwDuAj0o6CbgauD8iTgDur+6b2RCVTfaIWBsRi6rbO4Bn\ngaOAC4HbqqfdBlzUrkaaWesO6Du7pFnAbwA/A6ZFxNoqtI7ax3wzG6L6neySxgHfBz4WEfsNoBW1\nL4Z9fjmUNF9St6Tu3FhsZtY+/Up2SSOoJfq3I+KO6uH1kqZX8enAhr6WjYgFEdEVEV1TpkwZiDab\nWROyya7aKc+bgWcj4st1obuBy6vblwN3DXzzzGyg9KeL65nAZcCTkhZXj30auB74nqQrgJVAuhbR\nZh/60IdaWj43he7y5csbxnLDEt9xxx3JeK6LbK5tqSGZc2W9TZs2JeN79uxJxnMlzVQX2Vz32VzX\n3qlTpybjP/3pTxvGrrzyyuSyua+cuWm4R48enYx3QjbZI+IhoFFB87cHtjlm1i6+gs6sEE52s0I4\n2c0K4WQ3K4ST3awQTnazQhw0Q0m3Klc3PeWUUxrGUt1fIT+d9MSJE5Px1JTMAEceeWTDWK57bK5t\nuS6yuW6qqVp5rnttbt05qde+ePHihjGAefPmtbTtochHdrNCONnNCuFkNyuEk92sEE52s0I42c0K\n4WQ3K0QxdfZcvTjVJxzSwxo/9NBDyWVzUwvn5IYlTrV92bJlyWWPO+64ptrUY926dcl4qm25vvC5\n4cFz0yofddRRDWMLFy5MLpurs+feT7lhrjvBR3azQjjZzQrhZDcrhJPdrBBOdrNCONnNCuFkNytE\nMXX2XN0zNz1wytKlS5Px3NjtuX7duTp9avvHHntsctlcLTs3lXWu7ak6+yuvvJJcttW+9Kl47vqA\nnNz7aSjW4X1kNyuEk92sEE52s0I42c0K4WQ3K4ST3awQTnazQmTr7JKOBr4JTAMCWBARX5F0LfAR\noGci609HxD3tami77d27NxlP1eFXrlyZXDZXiz7xxBOb3jbAnDlzGsZyY9I/88wzyXiuHpwb0z51\njcD48eOTy06YMCEZz80dn9rvufH0c3+zUaNGJeNDsc7en4tqXgP+LCIWSRoP/FzSfVXsxoj4y/Y1\nz8wGSjbZI2ItsLa6vUPSs0DjIUDMbEg6oO/skmYBvwH8rHroKklPSLpF0hENlpkvqVtS98aNG/t6\nipkNgn4nu6RxwPeBj0XEduDrwGzgNGpH/hv6Wi4iFkREV0R0TZkyZQCabGbN6FeySxpBLdG/HRF3\nAETE+ojYGxH7gJuA09vXTDNrVTbZVTtteDPwbER8ue7x6XVPey/w1MA3z8wGSn/Oxp8JXAY8Kaln\nnttPA5dKOo1aOW4FcGVbWjhIWimFfP7zn0/Gv/SlLyXj9957bzK+bdu2ZDzVjTXXPTbXzXTq1KnJ\n+NatW5Px7du3N71srhtqruvw5MmTG8auuuqq5LK50lpOaqrqTunP2fiHgL4y4de2pm5WoqH378fM\n2sLJblYIJ7tZIZzsZoVwspsVwsluVohihpLOaaUumps6+DOf+UzT6wZ44YUXkvFUN9X169cnl03V\nwSE/lXXOiBEjmooBzJw5Mxk/88wzk/Fx48Yl46Xxkd2sEE52s0I42c0K4WQ3K4ST3awQTnazQjjZ\nzQqh3JC3A7oxaSNQP+7yZGDToDXgwAzVtg3VdoHb1qyBbNsxEdHn+G+Dmuy/snGpOyK6OtaAhKHa\ntqHaLnDbmjVYbfPHeLNCONnNCtHpZF/Q4e2nDNW2DdV2gdvWrEFpW0e/s5vZ4On0kd3MBomT3awQ\nHUl2Se+WtFTSMklXd6INjUhaIelJSYsldXe4LbdI2iDpqbrHJkq6T9Jz1e8+59jrUNuulbSm2neL\nJc3rUNuOlvQTSc9IelrSn1aPd3TfJdo1KPtt0L+zSxoG/D/gd4DVwGPApRGRnih8kEhaAXRFRMcv\nwJB0NvAy8M2IeGv12BeBLRFxffWP8oiI+OQQadu1wMudnsa7mq1oev0048BFwAfp4L5LtOsSBmG/\ndeLIfjqwLCKWR8Qe4O+ACzvQjiEvIhYCW3o9fCFwW3X7NmpvlkHXoG1DQkSsjYhF1e0dQM804x3d\nd4l2DYpOJPtRwKq6+6sZWvO9B/AjST+XNL/TjenDtIhYW91eB0zrZGP6kJ3GezD1mmZ8yOy7ZqY/\nb5VP0P2qsyLibcB5wEerj6tDUtS+gw2l2mm/pvEeLH1MM/66Tu67Zqc/b1Unkn0NcHTd/TdVjw0J\nEbGm+r0BuJOhNxX1+p4ZdKvfGzrcntcNpWm8+5pmnCGw7zo5/Xknkv0x4ARJx0oaCbwfuLsD7fgV\nksZWJ06QNBY4l6E3FfXdwOXV7cuBuzrYlv0MlWm8G00zTof3XcenP4+IQf8B5lE7I/8L4L93og0N\n2nUcsKT6ebrTbQO+S+1j3avUzm1cAUwC7geeA/4VmDiE2nY78CTwBLXEmt6htp1F7SP6E8Di6mde\np/ddol2Dst98uaxZIXyCzqwQTnazQjjZzQrhZDcrhJPdrBBO9jaTdJGkkDSnH8/9oKQZLWxrrqR/\n7u/jQ4mkd0j6WdXr69mqU01Pj7BPNFjm4QaPHy7pj9vY3F9LTvb2uxR4qPqd80Gg6WT/NXcbMD8i\nTgPeCnwvt0BE/Pvej0kaDhwOONl7cbK3UXUN9FnULjh5f6/YJ6t+80skXS/pYqAL+HZ1dBtT9a2f\nXD2/S9ID1e3TJT0i6XFJD0t68wG06VpJt0l6UNJKSb8r6YtVW35QXc6JpM9IekzSU5IWVFd/Iek3\nqw4biyV9qac/u6Rh1f3HqviVB7i7plK7SIeoXTpa3+X5JEkPSFou6U/qXsvL1e+51eu5G3gGuB6Y\n3dPGA2zHwasTVziV8gN8ALi5uv0w8Pbq9nnV/UOr+xOr3w9Q60vfs/wKYHJ1uwt4oLp9GDC8un0O\n8P3q9lzgn/tox+uPA9dS+6QxAjgV2AWcV8XuBC6qb1N1+3bg/Or2U8C/q25fDzxV3Z4PXFPdHgV0\nA8f20ZZ7gBl9PP4ZYGvVhiuB0XXtfbha52RgMzCiir1c9/p29mwPmNXTLv+88eMje3tdSq2/PtXv\nno/y5wB/GxG7ACLiQPuFTwD+oTqq3gi85QCXvzciXqV2ieYw4AfV409SSxSA36q+Qz8J/AfgLZIO\nB8ZHxCPVc75Tt85zgT+UtJhat81JwAm9NxwR8yLixT4e/x/U/qH9CPiDujYB/EtE7I7agCIb6Ltr\n6qMR8Xz+pZdreKcbcLCSNJFakpwsKaglVUj68wNYzWu88VVrdN3jfwH8JCLeW/WLfuAAm7cbICL2\nSXo1qsMhsA8YLmk08DVqnzJWVSfLRve9qtcJ+K8R8cMDbMvrIuIXwNcl3QRslDSpvr2VvfT9vt3Z\n7HZL4SN7+1wM3B4Rx0TErIg4GngeeCdwH/AhSYfC6/8YAHYA4+vWsQJ4e3X79+oen8Ab3YI/2Ia2\n9yT2puq8w8UAEbEN2CHpjCpefx7ih8B/qfvOf2LVc7BfJL2n57wAtU8Ee4FtTba/9340nOztdCm1\n75/1vk9tvL0fUOvd1F197O0pLd0K/K+eE3TA54CvqDbw5d669XwR+IKkx2nDp7MqqW+i9v38h9S6\nJfe4AripavdY4KXq8W9QOzm2qPp68Td9tU3SPQ3Ki5cBS6v13g58ICL29vG8/rR/M/Bv1clFn6Cr\nuNebHRBJ4yKi5yz41dS6Y/5ph5tl/eDv7Hag3iPpU9TeOytpz9cIawMf2c0K4e/sZoVwspsVwslu\nVggnu1khnOxmhfj/PKlVzbbaAN4AAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}