(KDT)_숫자이미지_오토인코더.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "(KDT) 숫자이미지 오토인코더",
      "provenance": [],
      "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/learn-programmers/programmers_kdt_II/blob/main/(KDT)_%EC%88%AB%EC%9E%90%EC%9D%B4%EB%AF%B8%EC%A7%80_%EC%98%A4%ED%86%A0%EC%9D%B8%EC%BD%94%EB%8D%94.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cv2rcOr48imA"
      },
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from keras.layers import Input, Dense\n",
        "from keras.models import Model\n",
        "from keras.datasets import mnist"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KUr2MERZA5XS",
        "outputId": "36951348-f9bb-42dc-fabe-bea453e8e6c7"
      },
      "source": [
        "(x_train, y_train), (x_test, y_test) = mnist.load_data()\n",
        "\n",
        "x_train = x_train.astype('float32') / 256\n",
        "x_test = x_test.astype('float32') / 256\n",
        "\n",
        "print(x_train.shape)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz\n",
            "11493376/11490434 [==============================] - 0s 0us/step\n",
            "(60000, 28, 28)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5t7KNCvPoIz1",
        "outputId": "0edf2a88-0252-4604-bf78-73aacf87aa53"
      },
      "source": [
        "x_train = x_train.reshape(len(x_train), 28*28)\n",
        "x_test = x_test.reshape(len(x_test), 28*28)\n",
        "\n",
        "print(x_train.shape)\n",
        "print(x_test.shape)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(60000, 784)\n",
            "(10000, 784)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MCtTpQRjA9X4"
      },
      "source": [
        "input_layer = Input(shape=(784,))\n",
        "encoded = Dense(64, activation='relu')(input_layer)\n",
        "decoded = Dense(784, activation='sigmoid')(encoded)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LOMYwzl0BAtE"
      },
      "source": [
        "autoencoder = Model(input_layer, decoded)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wuo7vQ2MBHZW"
      },
      "source": [
        "encoder = Model(input_layer, encoded)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xSvPdJDLBJrS"
      },
      "source": [
        "input_layer_decoder = Input(shape=(64,))\n",
        "decoder_layer = autoencoder.layers[-1](input_layer_decoder)\n",
        "decoder = Model(input_layer_decoder,decoder_layer)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_4vqBbPjBPh3",
        "outputId": "a901a452-eb08-4179-ea78-ff74a60a9498"
      },
      "source": [
        "autoencoder.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_1 (InputLayer)         [(None, 784)]             0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 64)                50240     \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 784)               50960     \n",
            "=================================================================\n",
            "Total params: 101,200\n",
            "Trainable params: 101,200\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Fn_BHg3YBgo6",
        "outputId": "4f9aa3ea-d449-4f20-813b-c5188b7e2002"
      },
      "source": [
        "autoencoder.compile(optimizer='adam', loss='binary_crossentropy')\n",
        "autoencoder.fit(x_train,x_train, epochs=50, batch_size=256, shuffle=True, validation_data=(x_test,x_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/50\n",
            "235/235 [==============================] - 4s 13ms/step - loss: 0.3471 - val_loss: 0.1650\n",
            "Epoch 2/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.1553 - val_loss: 0.1285\n",
            "Epoch 3/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.1241 - val_loss: 0.1099\n",
            "Epoch 4/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.1077 - val_loss: 0.0988\n",
            "Epoch 5/50\n",
            "235/235 [==============================] - 3s 11ms/step - loss: 0.0979 - val_loss: 0.0916\n",
            "Epoch 6/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0911 - val_loss: 0.0865\n",
            "Epoch 7/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0865 - val_loss: 0.0831\n",
            "Epoch 8/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0833 - val_loss: 0.0807\n",
            "Epoch 9/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0811 - val_loss: 0.0790\n",
            "Epoch 10/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0794 - val_loss: 0.0779\n",
            "Epoch 11/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0783 - val_loss: 0.0770\n",
            "Epoch 12/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0775 - val_loss: 0.0765\n",
            "Epoch 13/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0770 - val_loss: 0.0760\n",
            "Epoch 14/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0766 - val_loss: 0.0757\n",
            "Epoch 15/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0762 - val_loss: 0.0755\n",
            "Epoch 16/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0759 - val_loss: 0.0752\n",
            "Epoch 17/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0757 - val_loss: 0.0750\n",
            "Epoch 18/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0755 - val_loss: 0.0749\n",
            "Epoch 19/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0753 - val_loss: 0.0747\n",
            "Epoch 20/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0752 - val_loss: 0.0746\n",
            "Epoch 21/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0751 - val_loss: 0.0745\n",
            "Epoch 22/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0749 - val_loss: 0.0745\n",
            "Epoch 23/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0748 - val_loss: 0.0743\n",
            "Epoch 24/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0747 - val_loss: 0.0743\n",
            "Epoch 25/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0748 - val_loss: 0.0743\n",
            "Epoch 26/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0747 - val_loss: 0.0742\n",
            "Epoch 27/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0746 - val_loss: 0.0741\n",
            "Epoch 28/50\n",
            "235/235 [==============================] - 3s 11ms/step - loss: 0.0746 - val_loss: 0.0741\n",
            "Epoch 29/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0745 - val_loss: 0.0741\n",
            "Epoch 30/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0746 - val_loss: 0.0740\n",
            "Epoch 31/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0744 - val_loss: 0.0740\n",
            "Epoch 32/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0745 - val_loss: 0.0740\n",
            "Epoch 33/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0744 - val_loss: 0.0739\n",
            "Epoch 34/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0744 - val_loss: 0.0739\n",
            "Epoch 35/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0742 - val_loss: 0.0739\n",
            "Epoch 36/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0742 - val_loss: 0.0739\n",
            "Epoch 37/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0742 - val_loss: 0.0738\n",
            "Epoch 38/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0743 - val_loss: 0.0739\n",
            "Epoch 39/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0743 - val_loss: 0.0738\n",
            "Epoch 40/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0742 - val_loss: 0.0738\n",
            "Epoch 41/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0741 - val_loss: 0.0738\n",
            "Epoch 42/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0742 - val_loss: 0.0738\n",
            "Epoch 43/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0741 - val_loss: 0.0737\n",
            "Epoch 44/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0742 - val_loss: 0.0738\n",
            "Epoch 45/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0742 - val_loss: 0.0737\n",
            "Epoch 46/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0741 - val_loss: 0.0738\n",
            "Epoch 47/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0741 - val_loss: 0.0737\n",
            "Epoch 48/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0741 - val_loss: 0.0737\n",
            "Epoch 49/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0740 - val_loss: 0.0737\n",
            "Epoch 50/50\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.0740 - val_loss: 0.0737\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f5cbd2946d0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b5NbVnagD53c"
      },
      "source": [
        "encoded_images = encoder.predict(x_test)\n",
        "decoded_images = decoder.predict(encoded_images)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UAmQi8DhD-Yf"
      },
      "source": [
        "def visualize(data, size):\n",
        "    n = 8\n",
        "    plt.figure(figsize=(20,4))\n",
        "    plt.gray()\n",
        "    for i in range(n):\n",
        "        ax = plt.subplot(2,n,i+1)\n",
        "        plt.imshow(data[i].reshape(size,size))\n",
        "        ax.get_xaxis().set_visible(False)\n",
        "        ax.get_yaxis().set_visible(False)\n",
        "    plt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 365
        },
        "id": "3OF0nVYuEB35",
        "outputId": "d52b5e32-78f5-47b8-feb5-bc80ef27a98d"
      },
      "source": [
        "visualize(x_test, 28)\n",
        "visualize(encoded_images, 8)\n",
        "visualize(decoded_images, 28)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x288 with 8 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAABFoAAAB0CAYAAAC111bGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAR50lEQVR4nO3df4zXBf3A8ddxxx3H8ePAg+CCaUhJoSHOapJkFFY4hzJz6SaOsIVpmC2wLYcuyjLZYrVc9gPFtoYKLaOWCSwxbGAKmNMJCckBejDhOKEDBI7P94/mn3z3ft337T5+vns8/n6+36/Pj/e97/N5cc66SqUSAAAAAPzf9av2AwAAAAD4/8KiBQAAAKAkFi0AAAAAJbFoAQAAACiJRQsAAABASRoy8dChQysjR44s3B84cCD9gOrq6lJ9b29vqj/rrLNSfUREd3d3qs8+h4iIAQMGpPo33njjQKVSGZEeVAX9+vWrNDQUv9Ta2trSM7KvX/a6eeutt1J9RMSJEydS/cCBA9MzzjnnnFS/efPmmrluIiJaWloqra2thfv+/funZ7zb18LRo0dTfUTE2LFjU31XV1d6Rnt7e+G2s7Mzuru78ze2Kmlubq4MGTKkcN+X9yj7f+xrbm5+V/uIiJ6enlSfvW/2RS39roqIqK+vr2TuI5n70zuy71NTU1Oq79cv/29o2fvg22+/nZ5x8uTJwu2pU6eit7e3Zu45LS0tlWHDhr2rMxobG1P9m2++meqzn1kiIkaNGpXqjx8/np4xYkTu9vHyyy/X1D2nqamp0tLSUrjvy8/3sWPHUn32fe3L957sd8S+/F9yjxw5kjp/pVKpmXvOgAEDKoMGDSrcjxkzJj3j8OHDqf61115L9ZMnT071ERGHDh1K9X35fJz5/BgRsXfv3jPec1KLlpEjR8bSpUsL98uWLcuc/r8PKPGFPCJ/EcyZMyfVR0SsXr061ffly9748eNT/d13392RHlIlDQ0NkVnQzZ07Nz3jIx/5SKrP/qA++eSTqT4ioqMj9xZNmjQpPWP58uWpvq6urmaum4j/fom59dZbC/ejR49Oz8guTrL3gy1btqT6iIjFixen+hUrVqRnLFq0qHDbl5/JahoyZEhcf/31hfvNmzenZ5w+fTrVX3DBBal+4sSJqT4iYuPGjan+/PPPT8/IPu9FixbV1D2nf//+cfbZZxfur7rqqvSM7Ps0bty4VN+Xpf1//vOfVL9jx470jDfeeKNwu2/fvvT5q2nYsGGp31X19fXpGdkF/C9+8YtUv2fPnlQfEbFw4cJU/8orr6Rn3HLLLal+woQJNXXPaWlpic997nOF+74syLOv+x133JHq+7Joeeihh1J9dhkcEfHXv/61cNuXRWM1DRo0KGbOnFm4X7JkSXrGE088kepnz56d6v/2t7+l+oiI3//+96n+t7/9bXrG9OnTU/3ChQvPeM/xnw4BAAAAlMSiBQAAAKAkFi0AAAAAJbFoAQAAACiJRQsAAABASSxaAAAAAEpi0QIAAABQEosWAAAAgJJYtAAAAACUxKIFAAAAoCQNmbizszN++MMfFu7b29vTD+jYsWOpftKkSan+8ccfT/URERs3bkz1EyZMSM+49NJL08fUivr6+hg8eHDhftq0aekZP/rRj1L9FVdckepHjRqV6iMiZs2aleofeeSR9IwZM2akj6kljY2NMXbs2ML91q1b0zMuuuiiVP/WW2+l+lWrVqX6iIh9+/al+iFDhqRn7Nq1q3D79ttvp89fTcePH4/t27cX7sePH5+e0dbWlup7enpS/W233ZbqIyLmzZuX6l966aX0jNbW1vQxtaS+vj7189SX12PQoEGp/uDBg6m+u7s71UdETJ06NdW//vrr6Rn79+8v3J48eTJ9/mpqaGiIkSNHFu778vyy96ktW7ak+h07dqT6iIh777031fflujnvvPPSx9SS48ePx7Zt2wr3q1evTs+47LLLUv2aNWve1T4i4ktf+lKq37x5c3rGXXfdVbi9//770+evprq6umhsbCzc33nnnekZzc3NqT77eXrDhg2pPiLiscceS/WZ1+gdZX4n9xctAAAAACWxaAEAAAAoiUULAAAAQEksWgAAAABKYtECAAAAUBKLFgAAAICSWLQAAAAAlMSiBQAAAKAkFi0AAAAAJbFoAQAAACiJRQsAAABASSxaAAAAAErSkImbm5vj/PPPL9wPHTo0/YBaW1tT/cqVK1N9Z2dnqo+IGDZsWKrPPoeIiJ///OfpY2pFa2trzJo1q3A/d+7c9Iyenp5U39bWluonTpyY6iMiuru7U/1LL72UnnHDDTek+r/85S/pGdW0d+/eWLBgQeH++9//fnrGZz/72VT/6KOPpvqbb7451UdEXH755am+vb09PWP//v2F25MnT6bPX039+vWLlpaWwn1dXV16RvZ3z0033ZTqv/nNb6b6iIimpqZUv3HjxvSM+fPnp4+pJaNHj47vfOc7hfsVK1akZ0yaNCnVb9u2LdU/8cQTqT4ior6+PtVfeeWV6Rnjxo0r3P7hD39In7+aOjo64itf+Urhft26dekZ2c+WU6ZMSfUXX3xxqo+I+NSnPpXqZ86cmZ7xta99LX1MLRk8eHBMmzatcL9p06b0jF/+8pepPvs7f+fOnak+In8f/PSnP52ekfk92tXVlT5/NdXX18egQYMK92vXrk3P+PWvf53qK5VKqu/o6Ej1ERFbt25N9dnPXhH5n5f/jb9oAQAAACiJRQsAAABASSxaAAAAAEpi0QIAAABQEosWAAAAgJJYtAAAAACUxKIFAAAAoCQWLQAAAAAlsWgBAAAAKIlFCwAAAEBJLFoAAAAAStKQiQcMGBAf/vCHC/c7d+5MP6BKpZLqf/CDH6T6DRs2pPqIiNOnT6f6AQMGpGe88sor6WNqxenTp6Onp6dw/9WvfjU946yzzkr169evT/WPPvpoqo+IuOCCC1L9ww8/nJ7x5S9/OX1MLWlra4ubbrqpcL9jx470jLVr16b6s88+O9V//OMfT/UREddcc02qv/7669MzFi1aVLhtbGxMn7+aGhsbo729vXD/7LPPpmdMnz491a9bty7V79q1K9VHRAwcODDVT5gwIT1j5cqV6WNqSVdXVzzyyCOF++7u7vSMzOeoiIjt27en+quvvjrVR0QMHTo01f/qV79Kz8jcp2rtnjN69OiYN29e4f6ZZ55Jz5g4cWKqz94PZs6cmeojIubPn5/qf/KTn6RnfOITn0j1DzzwQHpGNQ0fPjyuu+66wv3BgwfTM3bv3p3qL7zwwlR/2223pfqIiOeeey7V9/b2pmdcdtllhdsDBw6kz19NDQ0N0dbWVrg/fPhwekb2NR8+fHiq37RpU6qPyN+nrr322vSM5cuXp485E3/RAgAAAFASixYAAACAkli0AAAAAJTEogUAAACgJBYtAAAAACWxaAEAAAAoiUULAAAAQEksWgAAAABKYtECAAAAUBKLFgAAAICSWLQAAAAAlKQhEx85ciTWrVtXuD906FD6AY0ZMybVjxo1KtV3d3en+oiI/v37p/qf/vSn6Rnf+973Uv2CBQvSM6qltbU1rrrqqsL92rVr0zPGjx+f6h977LFUf8kll6T6iIhLL7001a9YsSI9Y8qUKal+9+7d6RnVVF9fH4MHDy7cz5gxIz3jZz/7Warv6upK9c8++2yqj4hYtGhRqn/xxRfTM7Zt21a4PX78ePr81davX/F/Rxg7dmz6/I2Njak++57+61//SvUREdu3b0/1vb296RnPPPNM+pha0r9//2hvby/cr1+/Pj2jrq4u1T/55JOpfunSpak+IuLBBx9M9T09PekZlUrlXWnfC+rq6lL3nOuuuy4947777kv1V199dapfsmRJqo+IuPfee1P9Aw88kJ5x9OjR9DG1pLOzM+65557Cfebz9Duy79Orr76a6pctW5bqIyKuvfbaVL958+b0jMx3vux9udoqlUqcOnWqcD916tT0jOx18/jjj6f67HffiIgdO3ak+r///e/pGeeee276mDPxFy0AAAAAJbFoAQAAACiJRQsAAABASSxaAAAAAEpi0QIAAABQEosWAAAAgJJYtAAAAACUxKIFAAAAoCQWLQAAAAAlsWgBAAAAKIlFCwAAAEBJLFoAAAAAStKQiQcOHBiTJ08u3I8YMSL9gDo7O1P9okWLUv1HP/rRVB/x3+edceONN6Zn9Pb2po+pFQcPHoyHH364cL9p06b0jMbGxlSffb3PO++8VB8RsWXLllTf0JD6cYyIiOeeey59TC05depUdHV1Fe6bm5vTM7q7u1P96NGjU/2uXbtSfUREW1tbql+wYEF6xj/+8Y/CbU9PT/r81XT06NH45z//Wbi/4oor0jNWrVqV6p966qlU35ffVdnnsWHDhvSMiy66KNVn3of3ghMnTqR+ZqdOnZqekX1NLrnkklR/7NixVB8RMXv27FSfvW9GRLzwwguF2748h2rq379/tLe3F+737NmTnvHqq6+m+tbW1lR/+PDhVB8RqeccEXHrrbemZ7S0tKSPqSWjR4+OO++8s3B/8cUXp2dkv1s1NTWl+t/85jepPiJi5cqVqf673/1uesa3vvWtwm2tfQ8bNWpU6rPfyy+/nJ5xzjnnpPof//jHqb6uri7VR0S8//3vT/Vf//rX0zPuu+++9DFn4i9aAAAAAEpi0QIAAABQEosWAAAAgJJYtAAAAACUxKIFAAAAoCQWLQAAAAAlsWgBAAAAKIlFCwAAAEBJLFoAAAAASmLRAgAAAFASixYAAACAkjRk4hMnTsSePXsK988//3z6ATU1NaX6P/3pT6l+zpw5qT4i4vXXX0/1e/fuTc9Yt25d+pha0dzcHJMnTy7cjxgxIj3j3//+d6r/85//nOqffvrpVB8RsXLlylS/atWq9IynnnoqfUwtGTlyZNxyyy2F+wMHDqRnzJw5M9UvW7Ys1c+dOzfVR0Ts2rUr1Xd1daVn3H777YXb7du3p89fTb29vXHo0KHC/Zo1a9IzGhpSvz7jk5/8ZKofM2ZMqo+IuPvuu1P98uXL0zNWr16dPqaWVCqVqFQqhfvXXnstPWPhwoWpPnt9Xnjhhak+IuKhhx5K9TNmzEjP2Lp1a+H26NGj6fNXU1NTU3zgAx8o3Hd0dKRn3HHHHal+8eLFqf7mm29O9RGRus9G9O15f/CDH0wfU0s6OjpSr33mOnvHDTfckOoPHz6c6u+///5UHxFx4403pvrp06enZwwfPrxwe+LEifT5q2nPnj2p3yWzZs1Kz1iyZEmq/93vfpfqL7/88lQfkf98vHHjxvSMI0eOpI85E3/RAgAAAFASixYAAACAkli0AAAAAJTEogUAAACgJBYtAAAAACWxaAEAAAAoiUULAAAAQEksWgAAAABKYtECAAAAUBKLFgAAAICSWLQAAAAAlMSiBQAAAKAkDZl42LBhcc011xTuly5dmn5Ac+bMSfVf+MIXUv24ceNSfUTE5z//+VS/cOHC9IwvfvGLqX7nzp3pGdVy+PDhWLNmTeH+M5/5THrGH//4x1Q/ZsyYVN+X1zv7nj7//PPpGW1tbal++/bt6RnVdPLkydi/f3/hfvfu3ekZu3btSvUrVqxI9fPnz0/1ERFXXnllqt+8eXN6Rmtra+G2vr4+ff5qet/73he333574b4vz2/27NmpfsaMGal+7969qT4i4vTp06l+wYIF6Rn79u1LH1NLKpVKnDhxonDf2NiYnvHggw+m+gMHDqT648ePp/qI/O/Evny+Gzx4cOH21KlT6fNXU0tLS3zsYx8r3Pf09KRndHZ2pvrFixen+m984xupPiLirrvuSvXNzc3pGdOmTUv199xzT3pGNZ08eTJ1X33hhRfSM55++ulU/6EPfSjVz5s3L9VHROp7QUTEt7/97fSMF198sXC7bdu29Pmrrbe3t3C7bt269PmHDx+e6s8999xUP2XKlFQfkf9stH79+vSM5cuXp485E3/RAgAAAFASixYAAACAkli0AAAAAJTEogUAAACgJBYtAAAAACWxaAEAAAAoiUULAAAAQEksWgAAAABKYtECAAAAUBKLFgAAAICSWLQAAAAAlKSuUqkUj+vq3oyIjnfv4ZBwdqVSGVHtB1GE6+Y9pWaumwjXznuI64a+cu3QF64b+sq1Q1+4buirM147qUULAAAAAGfmPx0CAAAAKIlFCwAAAEBJLFoAAAAASmLRAgAAAFASixYAAACAkli0AAAAAJTEogUAAACgJBYtAAAAACWxaAEAAAAoyf8AdaheU9l9SKEAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 1440x288 with 8 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x288 with 8 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sXlp1WvQEM0o"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}