IA_NN_classif_multiclasse_MNIST.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "IA_NN_classif_multiclasse_MNIST.ipynb",
      "provenance": [],
      "authorship_tag": "ABX9TyNLi43P8KEn80DwN5T+s1o3"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "-vyJarzwhPQh",
        "outputId": "d46d72b4-ef78-48f0-d82e-9712ae4d41eb"
      },
      "source": [
        "#https://www.tensorflow.org/tutorials/keras/classification?hl=pt-br\n",
        "# importa as bibliotecas\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import tensorflow as tf\n",
        "\n",
        "#importa o dataset (as imagens da base \"mnist\")\n",
        "number_mnist = tf.keras.datasets.mnist\n",
        "(train_images, train_labels), (test_images, test_labels) = number_mnist.load_data()\n",
        "\n",
        "#Inspecionando o dataset\n",
        "print('train shape: ',train_images.shape) \n",
        "print('test shape: ',test_images.shape)\n",
        "\n",
        "plt.rcParams.update({'font.size': 14})\n",
        "plt.figure(figsize=(8,4))\n",
        "for i in range(2*4):\n",
        "    plt.subplot(2,4,i+1)\n",
        "    plt.xticks([]);plt.yticks([])\n",
        "    plt.imshow(train_images[i], cmap=plt.cm.binary)\n",
        "    plt.xlabel(str(train_labels[i]))\n",
        "plt.show()\n",
        "\n",
        "# criando a rede neural, já com as camadas\n",
        "N_class=10  #define o numero de classes (saida da rede)\n",
        "model = tf.keras.Sequential([\n",
        "    tf.keras.layers.Flatten(input_shape=(28, 28)),\n",
        "    tf.keras.layers.Dense(128, activation='relu'),\n",
        "    tf.keras.layers.Dense(N_class, activation='softmax')\n",
        "],name='rede_IF_MultClass_MNIST')\n",
        "\n",
        "#compilando a rede\n",
        "opt=tf.keras.optimizers.Adam(learning_rate=0.002)\n",
        "model.compile(optimizer=opt, loss='sparse_categorical_crossentropy',\n",
        "              metrics=['accuracy'])\n",
        "print(model.summary())\n",
        "\n",
        "# treinando a rede\n",
        "history=model.fit(train_images, train_labels, epochs=16,verbose=1)\n",
        "\n",
        "#mostra a performace do treinamento da rede\n",
        "plt.figure()\n",
        "plt.subplot(2,1,1);plt.semilogy(history.history['loss'],'k')\n",
        "plt.legend(['loss'])\n",
        "plt.subplot(2,1,2);plt.plot(history.history['accuracy'],'k')\n",
        "plt.legend(['acuracia'])\n",
        "plt.tight_layout()\n",
        "\n",
        "#testando a rede com os dados de teste\n",
        "pred = model.predict(test_images)\n",
        "test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=0)\n",
        "print('\\n accuracia dos dados de teste: ', test_acc)\n",
        "\n",
        "# mostra alguns resultados\n",
        "q_ex=[2, 8] #quais exemplos de teste queremos olhar\n",
        "class_names = ['0', '1', '2', '3', '4','5', '6', '7', '8', '9']\n",
        "plt.rcParams.update({'font.size': 8})\n",
        "plt.figure(),\n",
        "plt.subplot(3,2,1),plt.imshow(test_images[q_ex[0]],cmap=plt.cm.binary)\n",
        "plt.subplot(3,2,2),plt.imshow(test_images[q_ex[1]],cmap=plt.cm.binary)\n",
        "plt.subplot(3,1,2),plt.bar(class_names,pred[q_ex[0]])\n",
        "plt.legend(['resposta esperada:  '+class_names[test_labels[q_ex[0]]]])\n",
        "plt.subplot(3,1,3),plt.bar(class_names,pred[q_ex[1]])\n",
        "plt.legend(['resposta esperada:  '+class_names[test_labels[q_ex[1]]]])\n",
        "plt.tight_layout()\n",
        "\n",
        "#encontra a classe de maior probabilidade\n",
        "labels_pred=np.argmax(pred,axis=1)\n",
        "#mostra 15 resultados esperados e os alcançados lado a lado\n",
        "print('data and pred = \\n',np.concatenate(\n",
        "    (test_labels[None].T[0:15], labels_pred[None].T[0:15]),axis=1))\n"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "train shape:  (60000, 28, 28)\n",
            "test shape:  (10000, 28, 28)\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x288 with 8 Axes>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"rede_IF_MultClass_MNIST\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "flatten (Flatten)            (None, 784)               0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 128)               100480    \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 10)                1290      \n",
            "=================================================================\n",
            "Total params: 101,770\n",
            "Trainable params: 101,770\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n",
            "Epoch 1/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 1.8871 - accuracy: 0.8408\n",
            "Epoch 2/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.4117 - accuracy: 0.9019\n",
            "Epoch 3/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3760 - accuracy: 0.9109\n",
            "Epoch 4/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3747 - accuracy: 0.9155\n",
            "Epoch 5/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3714 - accuracy: 0.9130\n",
            "Epoch 6/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3638 - accuracy: 0.9175\n",
            "Epoch 7/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3604 - accuracy: 0.9203\n",
            "Epoch 8/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3376 - accuracy: 0.9229\n",
            "Epoch 9/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3429 - accuracy: 0.9236\n",
            "Epoch 10/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3307 - accuracy: 0.9257\n",
            "Epoch 11/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3369 - accuracy: 0.9245\n",
            "Epoch 12/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3249 - accuracy: 0.9262\n",
            "Epoch 13/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3279 - accuracy: 0.9258\n",
            "Epoch 14/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3287 - accuracy: 0.9248\n",
            "Epoch 15/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3311 - accuracy: 0.9241\n",
            "Epoch 16/16\n",
            "1875/1875 [==============================] - 3s 2ms/step - loss: 0.3175 - accuracy: 0.9263\n",
            "\n",
            " accuracia dos dados de teste:  0.9218000173568726\n",
            "data and pred = \n",
            " [[7 7]\n",
            " [2 2]\n",
            " [1 1]\n",
            " [0 0]\n",
            " [4 4]\n",
            " [1 1]\n",
            " [4 4]\n",
            " [9 9]\n",
            " [5 6]\n",
            " [9 9]\n",
            " [0 0]\n",
            " [6 6]\n",
            " [9 9]\n",
            " [0 0]\n",
            " [1 1]]\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 4 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 757
        },
        "id": "NBukrq251Dzm",
        "outputId": "c23b2349-c569-492e-ba11-452b763ba841"
      },
      "source": [
        "#código para mostrar como a camada Flatten funciona\n",
        "import tensorflow as tf\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "#importa o dataset (as imagens da base \"mnist\")\n",
        "number_mnist = tf.keras.datasets.mnist\n",
        "(train_images, train_labels), (test_images, test_labels) = number_mnist.load_data()\n",
        "\n",
        "# criando uma rede neural, só com a camada Flatten\n",
        "rede_teste_flatten = tf.keras.Sequential([\n",
        "    tf.keras.layers.Flatten(input_shape=(28, 28)),\n",
        "],name='rede_IF_MultClass_MNIST')\n",
        "\n",
        "#compilando a rede e mostrando o sumário\n",
        "rede_teste_flatten.compile()\n",
        "print(rede_teste_flatten.summary())\n",
        "#separa a primeira imagen, para testar a rede \n",
        "imagem_entrada=test_images[8:9]\n",
        "#passa a imagem pela rede\n",
        "vetor_saida = rede_teste_flatten.predict(imagem_entrada)\n",
        "\n",
        "#mostra o resultado da rede flatten\n",
        "print('imagem_entrada shape: ',imagem_entrada.shape) \n",
        "print('vetor_saida shape: ',vetor_saida.shape)\n",
        "\n",
        "plt.figure()\n",
        "plt.imshow(imagem_entrada[0], cmap=plt.cm.binary)\n",
        "plt.colorbar()\n",
        "plt.figure()\n",
        "plt.plot(vetor_saida[0],'k-')"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"rede_IF_MultClass_MNIST\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "flatten_1 (Flatten)          (None, 784)               0         \n",
            "=================================================================\n",
            "Total params: 0\n",
            "Trainable params: 0\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "None\n",
            "imagem_entrada shape:  (1, 28, 28)\n",
            "vetor_saida shape:  (1, 784)\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<matplotlib.lines.Line2D at 0x7f4f4fa7b750>]"
            ]
          },
          "metadata": {},
          "execution_count": 3
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAASoAAAD3CAYAAACq7RlqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAATKElEQVR4nO3df4xU5b3H8c8HWXKtiNawNGktbNKW1kixmCVIBAGh1qSp8R9tom2TxpTrf4038EdNk2tzmwZMmsVE/tn7p1raaHI1Eo2CccMSUNm1VtPeVK/3VmojuATESrqI+L1/zGG6rDvPzM6cmXkG3q9kkjnP9/z4crL58pxnnnOOI0IAkLM53U4AAOqhUAHIHoUKQPYoVACyR6ECkL257T7AwoULY2BgoN2HAS5q4+PjxyKiv5V92J7NFIDnIuLWVo43G20vVAMDAxobG2v3YYCLmu13OnzIhZ08WNOXfraHbI/afqjMhAB0j+2GPnX2scr2Adv7bQ8VbSdtjxSfq4q2u4v1dttekNpnU4XK9vWS5kfEWknzbK9sZj8A8jJnzpyGPnW8I+nmiFgjaZHtb0p6IyLWF5/jtvsk3SvpJkmPSPrXZF5N/ntukLSn+L5X0uqpQdubbY/ZHpuYmGjyEAA6rYweVUQciYjJYvGMpLOSrimuwLa5soOvqVK8PtEMNWS6ZgvVlZI+LL6fLJanJjocEYMRMdjf39L4HoAOabRIFYVq4bnOSPHZPMP+lkvqj4g/qVKYbpL0eUnfU50aMl2zg+knJZ27plwg6YMm9wMgI/V6S1Mci4jBxH6ukvSwpDslKSKOF+1PSloh6SnNooY026M6KGlj8X2TpJea3A+AjJQ0mD5X0qOStkTEEduX2b6kCN8o6W1Jb0paVrTXrSFN9agi4lXbk7ZHJb0WEa80sx8AeZlFjyrlDkkrJT1Y7O9nknba/kjS/0n694g4a/s/JY1KOiHprtQOm55HFRE/bXZbAPmx3cgvenVFxC5Ju6Y1Xz/Deo+o8otfXW2f8Amgd5TUoyodhQpAFYUKQPYoVACyR6ECkLWyBtPbgUIFoIoeFYDsUagAZI9CBSBrjdwe0y0UKgBVFCoA2eNXPwDZo0cFIGuMUQHoCRQqANmjUAHIHoPpALLGGBWAnkChApA9ChWA7FGoAGSPQgVIOnHiRDJ++PDhth17yZIlyfjQ0FAyvmzZsmR86dKlyfh1112XjHcbD84D0BPoUQHIHoUKQPYoVACyxoRPAD2BQgUge/zqByB79Khwwdi9e3cy/vTTT9eMjYyMJLd96623mkmpIV//+teT8b/85S/J+OnTp1s6/qefftrS9u12wY1R2R6Q9LKk/5b0cUTcUmJOALrkgipUhT0R8YPSMgHQdbkWqlZGzjbYHrV9X2nZAOiqOXPmNPTpeF5NbveepKWSNkjaZHv51KDtzbbHbI9NTEy0miOADjg3RtXIp9OaKlQRcToiTkXEJ5J2S1o2LT4cEYMRMdjf319GngA64IIqVLYvn7J4o6S3y0kHQDeVUahsr7J9wPZ+20NF29Zi+THbfbXaamn20m+t7XHbByT9LSJebnI/ADJSUo/qHUk3R8QaSYtsr5O0oVh+XdLtthdNb0vtsKlf/SLiGUnPNLMt2u/tt9Md3J07dybjw8PDyfg//vGPZDwikvFu+fOf/9ztFLI3i8u6hbbHpiwPR8SwJEXEkSntZyRdK2mkWN4r6W5Jp2Zoe7zWwZjwCUDSrB+cdywiBuvsb7mkfkkfSDo32/WkpCuLz4fT2mrK88YeAF1R1mC67askPSzpHlUK0YIitECVwjVTW00UKgBVJQ2mz5X0qKQtxWXgIUnrivAmSS/VaKuJQgWgqqQe1R2SVkp60PaIpK9I2md7v6RvSXoyIt6f3pbaIWNUACSVd1NyROyStGta80FJ26ett316Wy0UKgBVud7rR6G6AL377rvJ+I4dOzqUSed94xvfqBmr97or8OA8AD2AHhWArF1wD84DcGGiUAHIHoUKQPYYTAeQNcaoAPQECtVF5tixY8l4vblMa9asScZvvfXWmrF58+Ylt73iiiuS8fnz5yfjH330UTL+ne98p2as3lymVatWJeMrVqxIxi+99NKascsuuyy5LShUAHoAhQpA9ihUALI2ywfndRSFCkAVPSoA2aNQAcgehQpA1pjweQE6depUMv7tb387Gf/DH/6QjD/5ZPLJrEmrV69Oxn//+98n4wMDA8n44cOHk/Grr766ZizXwVpUUKgAZC/X/0goVAAkcekHoEdQqABkj0IFIHsUKgBZ4xYaAD2BHlUP+vjjj2vG7rrrruS29eZJ3X///cn4pk2bkvFW1JsnVc/ixYvLSQTZybVQ1e3n2f6i7VdtT9qeW7QN2R61/VD7UwTQKeemKNT7dFojF6THJW2U9JIk2b5e0vyIWCtpnu2VbcwPQAf1bKGKiMmIODGl6QZJe4rveyWl79cA0BMaLVJZFqoZXCnpw+L7yWL5PLY32x6zPTYxMdFKfgA6aM6cOQ19Op5XE9uclLSg+L5A0gfTV4iI4YgYjIjB/v7+VvID0EEXUo/qoCpjVpK0ScXYFYDe17OFynaf7b2SrpP0nKQ+SZO2RyWdjYhX2pwjgA7IeYyq7jyqiDijSs9pqpfbk05n1Xs/3a9+9auasaeffjq5bb1L3q1btybjn/vc55JxoB16dh4VgItHGYPp0+de2h6wfdT2iO3np6y31fZ+24/Z7kvmVdK/D8AFoKRLv/PmXhb2RMT6iLilOM4iSRsiYo2k1yXdntohhQqApPLGqGaYeylJG4q7We4rlgcljRTf687H5F4/AFWzGKNaaHtsyvJwRAzXWPc9SUslnZb0lO0X1MB8zKkoVACqZlGojkXEYCMrRsRpVYqUbO+WtEyV4nTuLSAzzseciks/AFXtmJ5g+/IpizdKelvSIUnrira68zEv6h5VvVdSbdu2rWZsyZIlyW1HR0eT8SuuuCIZBzrNJT04r/gF71n9c+7lPtu3qdKrGo2Il4v19tneL+mwpB2pfV7UhQrA+cqYR1Vj7uUvZlhvu6TtjeyTQgWgKtcJnxQqAFUUKgDZo1AByFq3bjhuBIUKQBWvywKQPXpUGTpw4EDT265YsSIZv/rqq5NxIEcUKgBZY4wKQE+gUAHIHoPpALJHjwpA1hijAtATKFQAskehytATTzzR9LbPPvtsMv6LX3zmqRbnue2225LxevO0gHagUAHIWlkPzmsHChWAKnpUALJHoQKQPQoVgOxRqABkjQmfAHoCv/plaGJiIhlP/e9y+vTp5Lb15lH98pe/TMbvvffeZHzVqlU1Y3/961+T2371q19Nxq+99tpkvJ4//vGPNWOrV69ObstzvLor1x5V3fJp+4u2X7U9aXuu7QHbR22P2H6+E0kC6Ix2vCm5DI30qI5L2ijpv6a07YmIH7QnJQDdkPMYVd0eVURMRsSJac0bbI/avq9NeQHogl7uUU33nqSlqrxH/inbL0TE61NXsL1Z0mZJWrx4cctJAuiMXAfTZ51VRJyOiFMR8Ymk3ZKWzbDOcEQMRsRgf39/GXkC6IBce1SzLlS2L5+yeKOkt8tLB0C3NFqksrz0s90n6VlJ10l6TtI+27epcuk3GhEvtzdFAJ2S62B63UIVEWckbZrWnJ4k1CO2bNmSjP/6179u27HPnj2bjO/cubOleK4WLVqUjK9fvz4Z/+1vf1tiNpiuZwsVgIsHhQpA1nhwHoCeQI8KQPZyLVR59vMAdEUZ0xOm3x9ctA0Vd7M8NGW9z7TVQqECUFXSPKpz9we/VOzzeknzI2KtpHm2V87UltrhRX3pt23btmT8zjvvrBm7++67k9ueOXMmGX/33XeT8XrTF3rV+++/n4w//vjjyfiyZZ+5EeI8P//5z2edEyrKmswZEZOSJqfs6wZJe4rveyWtlvTJDG2Hau2THhWAqjlz5jT0kbTQ9tiUz+bEbq+U9GHx/WSxPFNbTRd1jwrA+WbRozoWEYMNrntS0oLi+wJJH0g6O0NbTfSoAFS16V6/g6qMWUmVu1xeqtFWE4UKgKTybkq23Wd7r/55f3CfKmNWo5LORsQrEfHq9LbUPrn0A1BV0mD6TPcHf+bhBRHx00b3SaECUMUtNACylvMz0y/qQnXJJZck4ytX1p6D9uabb7Z07BdeeCEZrzcP64EHHqgZe+WV5OV+1iIiGR8fH+9QJhcnChWA7FGoAGSPQgUgexQqAFnjwXkAegI9KgDZo1AByB6FCufZuHFj/ZUSXnvttZqxevOo+vr6kvEf//jHyfhPfvKTZHxoaKhm7De/+U1yW3QPEz4B9AQG0wFkjx4VgOxRqABkjTEqAD2BQgUgexQqANnjVz+U6pZbbqkZu//++5Pb1nvW1fDwcDL+1ltvJeMjIyPJeCu+9KUvtW3fF7ucx6jqlk/bq2wfsL3f9lDRtrVYfsx2evYggJ7RprfQtKyRft47km6OiDWSFtleJ2lDsfy6pNvbmSCAzunZQhURR4pXNEvSGUnXShopls+9ihnABSDXQtXwGJXt5ZL6VXmj6adF84yvYi5e77xZkhYvXtx6lgA6omfHqCTJ9lWSHpZ0j2Z+PfN5ImI4IgYjYrC/v7+sXAG00bkH5zXy6bRGBtPnSnpU0paIOCLpkKR1Rbjuq5gB9I5evvS7Q9JKSQ8WCf5M0j7b+yUdlrSjfemhlmuuuaZm7Pvf/35y29/97nctHfvFF19setu5c9N/ct/97neT8e3btzd9bNSX66Vf3UIVEbsk7ZrWfFASfzHABaZnCxWAi0POEz4pVACquIUGQPboUQHIHoUKQNYYowLQEyhUKNWll15aM7ZjR3pq29///vdkfHx8PBk/evRoMj4wMFAz9qMf/Si57QMPPJCMo70oVACyl+uvfnlmBaDjGr19pl6vy/aA7aO2R2w/X7S19Aw7ChWAqhLv9dsTEesj4hbbi9TiM+woVACqZlGoFtoem/LZPG1XG2yP2r5P0qBafIYdY1QAqmYxmH4sIgZrxN6TtFTSaUlPSbpc0vtFbMZn2NVDoQJQVcavfhFxWpUiJdu7JX0o6dxbOWZ8hl09XPoBkFTeg/NsXz5l8UZJ/6MWn2FHj+oC9IUvfCEZ3717dzL+yCOPJOMHDx5MxlNzoRYtWpTcFt1V0jyqtbb/Q5Ve1WhEvGy7pWfYUagAVJV06feMpGemtW1XC8+wo1ABqGJmOoCscVMygJ6Q6y00FCoAVfSoAGSPQgUga4xRoaf88Ic/bCmO3kWhApA9ChWA7PGrH4CsMUYFoCdQqABkj0IFIHsUKgDZo1AByNq5B+flqG5WtlfZPlC86maoaDtZvApnxPZV7U8TQCeU+BaaUjXSo3pH0s0RMVm8k+ubkt6IiPXtTQ1Ap+V66Ve3RxURRyJislg8I+mspGuKV+Fs8wz/Mtubz71GZ2JiouSUAbRLrj2qhi9IbS+X1B8Rf5L0NUk3Sfq8pO9NXzcihiNiMCIG+/v7S0sWQPuU9abkdmioUBXjUA9LukeSIuJ4RISkJyUta196ADqpjLfQtEPdMSrbcyU9KmlLRByxfZmkyYg4q8qrcN5oc44AOqRnx6gk3SFppaQHbY9IWi7pkO19kr4s6Yn2pQegk3K99Kvbo4qIXZJ2TWu+vj3pAOgWbkoG0BMoVACyR6ECkLWcb6GhUAGookcFIHsUKgDZo1AByB6FCkDWmEcFoCfwqx+A7NGjApC9XAtVnv08AB1X5vOobA8VD9d8qIzcKFQAqsooVLavlzQ/ItZKmmd7Zat5cekHoKqkwfQbJO0pvu+VtFrSoVZ22PZCNT4+fsz2O1OaFko61u7jNoncmkNus1d2Xkta3cH4+Phzthc2uPq/2B6bsjwcEcPF9ysl/W/x/aSka1vNre2FKiLOe2i67bGIGGz3cZtBbs0ht9nLMa+IuLWkXZ2UtKD4vkDSB63ukDEqAGU7KGlj8X2TpJda3SGFCkCpIuJVSZO2RyWdjYhXWt1nNwbTh+uv0jXk1hxym71c8ypFRPy0zP258tYrAMgXl34AskehApC9jhaqsqfVl8X2gO2jtkdsP9/tfCTJ9hdtv2p7sngJbDbnb3puuZw/26tsH7C93/ZQ0ba1WH7Mdl9muZ0sztlI8TZy1NCxQtWOafUl2xMR6yPilm4nUjiuyk+8L0nZnb/zcivkcP7ekXRzRKyRtMj2OkkbiuXXJd2eUW7flPRGcc7WR8TxLuaWvU72qGaaVp+TDUVv5b5uJyJJETEZESemNGVz/mbITcrg/EXEkYiYLBbPqDIjeqRY7vY5m57bWUnXFOdsm3N9bEEmOlmorpT0YfH9ZLGci/ckLZW0QdIm28u7nM9MOH8NKo7fr8qM6KzO2bncIuJPkr4m6SZJn5f0va4mlrlOFqrSp9WXJSJOR8SpiPhE0m5Jy7qd0ww4fw0oxnoelnSPMjtn03JTRByPyvygJ5Xn31w2OlmoSp9WXxbbl09ZvFHS293KJYHzVz+PuZIelbQlIo6ocsf+uiLc1XM2PTfbl9m+pAjn+jeXjY4VqnZMqy/RWtvjtg9I+ltEvNzthGz32d4r6TpJz0nqUybnb4bc/i2T83eHpJWSHrQ9IukrkvbZ3i/pW6r0XLplem7LJR2yvU/SlyU90cXcssfMdADZY8IngOxRqABkj0IFIHsUKgDZo1AByB6FCkD2KFQAsvf/7UwjxV2pwhIAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}