Convolution_AutoEncoder.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Convolution_AutoEncoder.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "7XmXo3zqq4cW"
      },
      "source": [
        "from tensorflow.keras.layers import Input, Dense, LeakyReLU, Conv2D, Conv2DTranspose\n",
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.utils import plot_model\n",
        "from tensorflow.keras.optimizers import Adam #since we want to modify our leaning rate for our optimizer so functional API call"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jCr_8lSsq6g3"
      },
      "source": [
        "#ENCODER MODEL"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Y-vRe1p-rWH9"
      },
      "source": [
        "# The input is (28,28) instead of (784) because we are using Conv layers \n",
        "encoder_input = Input(shape=(28,28,1), name=\"encoder_input\")"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "evDbQl9DrliA"
      },
      "source": [
        "#The encoder\n",
        "encoder_conv_layer1 = Conv2D(32,(3,3),name=\"encoder_conv\")(encoder_input)\n",
        "encoder_activ_layer1 = LeakyReLU(name=\"encoder_leakyrelu_1\")(encoder_conv_layer1)\n",
        "\n",
        "encoder_dense_layer2 = Dense(units=10, name=\"encoder_dense_2\")(encoder_activ_layer1)\n",
        "encoder_output = LeakyReLU(name=\"encoder_output\")(encoder_dense_layer2)"
      ],
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XMar6iq5sB5f"
      },
      "source": [
        "encoder = Model(encoder_input, encoder_output, name=\"encoder_model\")"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tranvrZ1sVQl",
        "outputId": "08cc83bf-104a-4723-bfb9-737b219acfa5"
      },
      "source": [
        "#The parameters in the encoder model\n",
        "encoder.summary()"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"encoder_model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "encoder_input (InputLayer)   [(None, 28, 28, 1)]       0         \n",
            "_________________________________________________________________\n",
            "encoder_conv (Conv2D)        (None, 26, 26, 32)        320       \n",
            "_________________________________________________________________\n",
            "encoder_leakyrelu_1 (LeakyRe (None, 26, 26, 32)        0         \n",
            "_________________________________________________________________\n",
            "encoder_dense_2 (Dense)      (None, 26, 26, 10)        330       \n",
            "_________________________________________________________________\n",
            "encoder_output (LeakyReLU)   (None, 26, 26, 10)        0         \n",
            "=================================================================\n",
            "Total params: 650\n",
            "Trainable params: 650\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 465
        },
        "id": "MC2TVQLrskuj",
        "outputId": "e47d8047-0ff6-46ce-86ea-6b35811fd6c7"
      },
      "source": [
        "#Model structure\n",
        "plot_model(encoder)"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "12V9lDEfs68r"
      },
      "source": [
        "#DECODER MODEL"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SGIbZHt-tJzx"
      },
      "source": [
        "#Decoder Input\n",
        "decoder_input = Input(shape=(26,26,10),name=\"decoder_input\")"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bY8gPXJos6oh"
      },
      "source": [
        "#The decoder\n",
        "decoder_conv_layer1 = Conv2DTranspose(32,(3,3), name=\"decoder_convtranspose\")(decoder_input)\n",
        "decoder_activ_layer1 = LeakyReLU(name=\"decoder_leakyrelu_1\")(decoder_conv_layer1)\n",
        "\n",
        "decoder_dense_layer2 = Dense(units=1, name=\"decoder_dense_2\")(decoder_activ_layer1) #We want to get the orginal vector back\n",
        "decoder_output = LeakyReLU(name=\"decoder_output\")(decoder_dense_layer2)"
      ],
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sCllREI8uCDV"
      },
      "source": [
        "decoder = Model(decoder_input,decoder_output,name=\"Decoder\")"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mG4_JTzysoE7",
        "outputId": "37d1203b-b9a8-4809-8d3e-36245fabeea9"
      },
      "source": [
        "#The decoder model parameters\n",
        "decoder.summary()"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"Decoder\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "decoder_input (InputLayer)   [(None, 26, 26, 10)]      0         \n",
            "_________________________________________________________________\n",
            "decoder_convtranspose (Conv2 (None, 28, 28, 32)        2912      \n",
            "_________________________________________________________________\n",
            "decoder_leakyrelu_1 (LeakyRe (None, 28, 28, 32)        0         \n",
            "_________________________________________________________________\n",
            "decoder_dense_2 (Dense)      (None, 28, 28, 1)         33        \n",
            "_________________________________________________________________\n",
            "decoder_output (LeakyReLU)   (None, 28, 28, 1)         0         \n",
            "=================================================================\n",
            "Total params: 2,945\n",
            "Trainable params: 2,945\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 465
        },
        "id": "XEClhaN1uNd7",
        "outputId": "65efa5a0-1195-4f5b-98cc-cb244afe4350"
      },
      "source": [
        "#Decoder model\n",
        "plot_model(decoder)"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAW0AAAHBCAIAAAAtvX3gAAAABmJLR0QA/wD/AP+gvaeTAAAgAElEQVR4nO3deVxTV/4//nMDSW72gIKAgGwqWrGj4lKrM1LHsX5srQoIivLBFj+odZtaZQYsY63WKipOFdtabfsQPw8F0VFLXTq1VTutpWpVLAruC1KMskOQQLi/P+6n+eUbQoAcIAFez7/MXc55n5vk5V3CvQzHcQQAgILA1gUAQKeHHAEAWsgRAKCFHAEAWo62LqAtnTt3bsuWLbauAqB5L7zwwltvvWXrKtpMl9ofefjwYWZmpq2rAGjGTz/9dO7cOVtX0Za61P4I78CBA7YuAcCS8PBwW5fQxrrU/ggA2ARyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABodfcciY2NVSgUDMNcvny58/Z17NgxlUr15Zdftm2zNH766acBAwYIBAKGYXr16rV27doO6/rgwYN+fn4MwzAM4+bmNnv27A7rutvq7jmya9euTz/9tLP3ZYcPDxk1atT169f/8pe/EELy8/NXrVrVYV2HhobeuXPH399fpVIVFRXt3bu3w7rutrp7jnQNkydPLi8vf/XVV9u7o5qamtGjR7d3L1aw28K6CeQIYRimS/bVHnbv3q3RaGxdhRl2W1g30R1zhOO45OTk/v37i8VilUq1YsUK47l6vT4pKcnb21sikQwePDg9Pd0wKy0tLTg4mGVZmUzm4+Pz3nvv8a1t2bJlwIABYrHYyclp6tSpeXl5NH1t3LhRKpUqFAqNRrN8+fLevXvn5+dbGM5//vMfb29vhmG2b99OCNmxY4dMJpNKpUeOHJk0aZJSqfT09Ny3bx+/8IcffsiyrKur6/z5893d3VmWHT16dHZ2Nj93yZIlIpHIzc2Nf/nmm2/KZDKGYZ4+fUoIWbZs2fLly2/fvs0wTEBAACHkxIkTSqVy3bp1LdnsHVlYS3z//fcDBw5UqVQsywYFBZ08eZIQEhsby59Y8ff3v3TpEiFk7ty5UqlUpVIdPXq0rd6yLojrQvg3tdnFEhMTGYbZvHlzaWmpVqtNTU0lhFy6dImf+/bbb4vF4szMzNLS0oSEBIFAcP78eY7jUlJSCCHr168vLi4uKSn55JNPoqKiOI5LSkoSiURpaWllZWU5OTlDhw7t2bNnUVERTV+JiYmEkKVLl27btm369OnXr1+3PKKHDx8SQrZt22bolBBy6tSp8vJyjUYzduxYmUym0+n4uXFxcTKZ7Nq1a8+ePcvNzR0+fLhCoXjw4AE/NyoqqlevXoaWk5OTCSFPnjzhX4aGhvr7+xvmZmVlKRSKNWvWNFXYxIkTCSGlpaUdXBjHcfz5EQsb7cCBA6tXry4pKSkuLh41alSPHj0MTTk4ODx69Miw5KxZs44ePcr/u03esrCwsLCwMAsLdDrdLke0Wq1UKp0wYYJhCv9fIv/drqmpkUqlkZGRhoXFYvHChQt1Op1arQ4JCTGsVV9fv3XrVq1WK5fLDctzHPfzzz8TQvivlnV9cb9/KGtqalo4cLM5YlidD69bt27xL+Pi4oy/YOfPnyeEvPvuu/zL1n5dLTObIx1TWLM5Yuz9998nhGg0Go7jvvnmG0LI2rVr+Vnl5eV9+/atr6/n2u4t63o50u2Oa27duqXVasePH292bn5+vlarHTRoEP9SIpG4ubnl5eXl5OSUlZXx3wqeg4PD0qVLc3Nzq6qqgoODDdOHDx8uEon4HXLr+mqTYRoTiUSEkLq6OrNzg4ODpVJpe/TbLPspTCgUEkL0ej0h5KWXXurXr99nn33GcRwhZP/+/ZGRkQ4ODqQD37JOp9vlSEFBASHExcXF7Nzq6mpCyKpVq5jf3b9/X6vVVlRUEELUarXJ8mVlZYQQuVxuPFGtVldWVlrdF934rCEWi588edLx/TarXQv76quvxo0b5+LiIhaLV65caZjOMMz8+fPv3Llz6tQpQsiePXveeOMNfpb9vGX2ptvlCMuyhJDa2lqzc/nvfEpKivE+27lz5zw8PAgh/Fk9Y3yy8KlhUFZW5unpaXVfdONrtbq6OkPBdqU9Cjt79ix/nuvBgwfTpk1zc3PLzs4uLy/fsGGD8WIxMTEsy+7atSs/P1+pVPbp04efbidvmR3qdjkyaNAggUBw5swZs3O9vLxYlm38e1MfHx9nZ+evv/66cWtyufzChQuGKdnZ2TqdbtiwYVb31cFOnz7NcdyoUaP4l46Ojk0daHSw9ijs4sWLMpmMEHL16tW6urqFCxf6+fmxLGtyPd7JySkiIuLw4cObNm2aN2+eYbqdvGV2qNvliIuLS2hoaGZm5u7duysqKnJycnbu3GmYy7Ls3Llz9+3bt2PHjoqKCr1eX1BQ8Ntvv4nF4oSEhLNnzy5ZsuTRo0cNDQ2VlZXXrl1jWXb58uWHDh3au3dvRUXF1atXFyxY4O7uHhcXZ3VfHbARGhoaSktL6+vrc3Jyli1b5u3tHRMTw88KCAgoKSk5fPhwXV3dkydP7t+/b7yis7NzYWHhvXv3Kisr6+rqjh8/3vLrvh1ZWOOW6+rqHj9+fPr0aT5HvL29CSHffPPNs2fPbt68abjAbLBgwYLa2tqsrCzjX/fZ8C2zd+1+JrcDtfC6b2VlZWxsbI8ePeRy+ZgxY5KSkgghnp6eV65c4TiutrY2Pj7e29vb0dGRD4Lc3Fx+xe3btwcFBbEsy7LskCFDUlNTOY5raGhITk7u27evUCh0cnKaNm1afn4+TV8bNmyQSCSEEC8vr7S0tGaHs23bNv6HFVKpdMqUKampqVKplBDSt2/f27dv79y5U6lUEkL69Olz48YNjuPi4uKEQmHv3r0dHR2VSuXUqVNv375taK24uDgkJIRlWV9f38WLF/M/eAkICOCvv/7yyy99+vSRSCRjxowpKio6duyYQqEwXNow9tNPPz333HMCgYAQ4ubmtm7dug4r7KOPPvL392/qA3/o0CG+wfj4eGdnZ7VaHR4ezv/0xt/f33CZmeO4IUOG/P3vfzcZV5u8ZV3vek13zJFuLi4uztnZ2dZVmGFvhf3Xf/3XnTt32qPlrpcj3e64BsjvFzjtkM0LMxwT5eTk8Ps+tq2ns0CO2Lu8vDymaZGRkbYusEuJj4+/efPmjRs35s6dy//dA7QEcsTeBQYGWtif3L9/f6taS0hI+Pzzz8vLy319fTMzM9upZivYSWFSqTQwMPDPf/7z6tWrBw4caKsyOh2Gs79bV1gtIyMjIiKiK40IuqTw8HBCyIEDB2xdSJvB/ggA0EKOAAAt5AgA0EKOAAAt5AgA0EKOAAAt5AgA0EKOAAAt5AgA0EKOAAAt5AgA0EKOAAAt5AgA0HK0dQFtj/9jSgC79dNPPxnuX901dKn9ES8vr7CwMFtX0dUcPXq0sLDQ1lV0KaNGjXrhhRdsXUVb6lL3H4H2wDBMenr6jBkzbF0I2K8utT8CADaBHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKDFcBxn6xrAvsyZM+fy5cuGl/fu3XNxcZHJZPxLoVD45Zdf9u7d20bVgT1ytHUBYHf69++/d+9e4ylVVVWGfwcGBiJEwASOa8DUzJkzGYYxO0soFMbExHRsOdAJ4LgGzBg2bNjly5cbGhpMpjMMc+fOHR8fH1sUBfYL+yNgRnR0tEBg+tlgGGbEiBEIEWgMOQJmRERENN4ZEQgE0dHRNqkH7BxyBMxwc3MbO3asg4ODyfTQ0FCb1AN2DjkC5s2ZM8f4pUAgCAkJ6dWrl63qAXuGHAHzwsPDTU6RmCQLgAFyBMxTKpUvv/yyo+P//cLIwcHhtddes21JYLeQI9Ck2bNn6/V6Qoijo+OUKVNUKpWtKwI7hRyBJk2ZMkUikRBC9Hp9VFSUrcsB+4UcgSaxLDt9+nRCiFQqnTRpkq3LAfv1//x9TUFBwY8//mirUsAOeXl5EUKGDx9+9OhRW9cCdsTLy+uFF174/19zRtLT021XGAB0GmFhYcbRYebvffEXN2Bs9erVq1atMly4AQgPDzeZgvMj0AyECDQLOQLNQIhAs5AjAEALOQIAtJAjAEALOQIAtJAjAEALOQIAtJAjAEALOQIAtJAjAEALOQIAtJAjAEALOQIAtGhzJDY2VqFQMAxj/IT6dtKRfUE7yc/PX7x48XPPPadQKBwdHVUqVb9+/SZPnnzu3Ln27nrNmjUDBw5UKpVisTggIGDlypWG558fPHjQz8+PMSISiVxdXceNG5ecnFxaWmpoJDIykrEoKyurvQdih2hzZNeuXZ9++mmblGJXfUF72L17d1BQUE5OzpYtWx4+fFhdXX3p0qX33nuvrKzs6tWr7d37t99+u2jRonv37j19+vT999/funWr4T4aoaGhd+7c8ff3V6lUHMc1NDRoNJqMjAxfX9/4+PjnnnvuwoULhna+/vrrsrKyurq63377jRAyZcoUnU5XXV2t0WjmzZvX3qOwT/iTcNuoqakZP358t7qL5U8//RQXF/enP/3p5MmThnsR+Pn5+fn5qdXqmzdvtncBcrk8Li6Of0jgjBkzDh48mJGR8fDhQ/7ekcYYhlGr1ePGjRs3btzkyZMjIiImT55848YNlUrFMMyLL74olUqNFxYKhUKhUCqVDhs2rL1HYZ/a4PwIwzD0jdhhX+1q9+7dGo3G1lV0qLVr1+r1+vXr1ze+ocnEiRMXLVrU3gVkZWUZP2m0Z8+ehBCtVmt5rbCwsJiYGI1G8/HHHxNC9u3bZxwiJuLi4l555ZU2qrczsSZHOI5LTk7u37+/WCxWqVQrVqwwnqvX65OSkry9vSUSyeDBg43v+ZqWlhYcHMyyrEwm8/Hxee+99/jWtmzZMmDAALFY7OTkNHXq1Ly8PJq+Nm7cKJVKFQqFRqNZvnx579698/Pzmx1Ua2vbsWOHTCaTSqVHjhyZNGmSUqn09PTct28fP3fAgAEMwwgEgmHDhvGf1JUrV6pUKpZlv/jii2XLli1fvvz27dsMwwQEBJgt+Pvvvx84cCC/SlBQ0MmTJ5vtlBBy5syZESNGSKVSpVIZFBRUUVHx4Ycfsizr6uo6f/58d3d3lmVHjx6dnZ1tvIUtbP/GDVp4i0+cOKFUKtetW9d48+p0ulOnTvXo0WPEiBGW34h22uaNO3r06JFEIvH19bVcDyEkJiaGEHL8+PFmlyRNfPw647vZOo3v88w1JzExkWGYzZs3l5aWarXa1NRUQsilS5f4uW+//bZYLM7MzCwtLU1ISBAIBOfPn+c4LiUlhRCyfv364uLikpKSTz75JCoqiuO4pKQkkUiUlpZWVlaWk5MzdOjQnj17FhUV0fSVmJhICFm6dOm2bdumT59+/fp1yyOyujZCyKlTp8rLyzUazdixY2UymU6n4ziuvr7ex8fH29u7vr7e0Mtf//rXlJQU/t+hoaH+/v7Gm9Sk4AMHDqxevbqkpKS4uHjUqFE9evRottOqqiqlUrlhw4aampqioqLp06c/efKE47i4uDiZTHbt2rVnz57l5uYOHz5coVA8ePCAb9DCGJtqsKnNnpWVpVAo1qxZ03gL37hxgxAyatSoZj9d7bfNjVVXVysUiiVLlhhPNJwfMcF/37y8vEym8+dHXnvtNZPpXePdtCAsLMzkPs+tzhGtViuVSidMmGCYwico/92uqamRSqWRkZGGhcVi8cKFC3U6nVqtDgkJMaxVX1+/detWrVYrl8sNy3Mc9/PPPxNC+M+idX0Z3p6amhrLY+FZV1vjXviMu3XrFv+Sz6aMjAz+ZXV1tbe3d3l5Of/SbI40VfD7779PCNFoNJY7/fXXXwkhWVlZJqvHxcUZfz3Onz9PCHn33Xf5jWZhjGYbtLDZLeDPU/75z3+2vFi7bnNjiYmJ/fr1q6ioMJ7YVI5wHMefMTGZaDlHuvC72ThHWn1cc+vWLa1WO378eLNz8/PztVrtoEGD+JcSicTNzS0vLy8nJ6esrGzixImGJR0cHJYuXZqbm1tVVRUcHGyYPnz4cJFIxO+qWddXa0dkXW2NiUQiQkhdXR3/MjY2VqVSbd26lX+5d+/eqVOnKpXK1pZHCBEKhYQQ/hGZFjr18/NzdXWdPXv26tWr792711RrwcHBUqmU31CWx2i2Qes2u1wuJy04GdEx2/zQoUMZGRknT55UKBSW6+FVV1dzHGfde9dYF3g3G2t1jhQUFBBCXFxczM6trq4mhKxatcpwOf3+/ftarZbfM1Sr1SbLl5WVkd8/ZAZqtbqystLqvlo7Iutqa5ZcLv+f//mfH3/8kf8/4aOPPlqyZEnLq/rqq6/GjRvn4uIiFotXrlzZklUkEsm33347ZsyYdevW+fn5RUZG1tTUmF1SLBY/efKENDdGsw1at9l9fHxYluWPbizogG2+f//+Dz744PTp0z4+Pi1pkxDClx0YGNjC5RvrYu9mY63OEZZlCSG1tbVm5/LfeZMj0nPnznl4eBBCnj59arI8/+01+ZSUlZV5enpa3VdrR2RdbS2xZMkSoVCYkpJy9uxZLy8vf3//Fq744MGDadOmubm5ZWdnl5eXb9iwoYUrPvfcc19++WVhYWF8fHx6evqmTZsaL1NXV2cYRbNjbNygdZtdLBZPnDjx6dOnP/zwQ+O5JSUlsbGxLamnWZa3+bZt2/bu3fvtt9/yb3oLnThxghBi9ZNJu9672Virc2TQoEECgeDMmTNm53p5ebEs2/j3pj4+Ps7Ozl9//XXj1uRyufGPfLKzs3U6HX8d3rq+Wsu62lrC09NzxowZmZmZ77zzzrJly1pe0tWrV+vq6hYuXOjn58eybAuvdhcWFl67do0Q4uLisn79+qFDh/IvTZw+fZrjuFGjRpHmxmi2Qas3++rVq8Vi8VtvvdX4/9Vff/2Vvxjcftuc47j4+PirV68ePnzY5H9sy4qKilJSUjw9PV9//fWWr2WsS76bJlqdIy4uLqGhoZmZmbt3766oqMjJydm5c6dhLsuyc+fO3bdv344dOyoqKvR6fUFBwW+//SYWixMSEs6ePbtkyZJHjx41NDRUVlZeu3aNZdnly5cfOnRo7969FRUVV69eXbBggbu7e1xcnNV9tXZE1tXWQsuXL6+vry8tLX3ppZeMpzs7OxcWFt67d6+ystJweG/g7e1NCPnmm2+ePXt28+bNps4OmCgsLJw/f35eXp5Op7t06dL9+/f5jxchpKGhobS0tL6+PicnZ9myZd7e3vy1TMtjNNughc1+/Pjxpq77EkL+8Ic//O///u+vv/46duzYY8eOlZeX19XV3b1799NPP33jjTf4swbtt82vXbu2cePGTz/9VCgUGv+M3eQ/eY7jqqqqGhoaOI578uRJenr6iy++6ODgcPjwYavPj3TSd7N1jPdnWnjdt7KyMjY2tkePHnK5fMyYMUlJSYQQT0/PK1eucBxXW1sbHx/v7e3t6OjIB0Fubi6/4vbt24OCgliWZVl2yJAhqampHMc1NDQkJyf37dtXKBQ6OTlNmzYtPz+fpq8NGzZIJBJCiJeXV1paWrPDsa621NRU/vdIffv2vX379s6dO/nPWZ8+fW7cuGHcckhIyK5du0y6++WXX/r06SORSMaMGfPWW281Ljg+Pt7Z2VmtVoeHh2/fvp0Q4u/v/7e//c1Cp/fu3Rs9erSTk5ODg4OHh0diYiJ/BTQuLk4oFPbu3dvR0VGpVE6dOvX27duGSiyMsakGm3qLjx07plAo1q5da2E7P3jw4O233w4KCpLL5Q4ODmq1esiQIW+88cYPP/zQbD0027yp390nJydzHHf06NHBgwdLpVKRSCQQCMjvP2kdMWLEmjVriouLTUZRUVHxxz/+0dnZmRAiEAgCAgLWrVvHzzL78euM76YFja/XMJzR03wzMjIiIiI4PN+3a5k/f/6BAweKi4ttXQi0AXt4N/m/Szpw4IBhCu4b0C2YvcoInZQdvptdP0fy8vIs/JV3ZGSkrQsE6PS6fo4EBgZaONLbv3+/rQtsXwkJCZ9//nl5ebmvr29mZqatywEqdvtu4vwIALQOzo8AQNtDjgAALeQIANBCjgAALeQIANBCjgAALeQIANBCjgAALeQIANBCjgAALeQIANBCjgAALeQIANAy85zwjIyMjq8DADqLgoICkzv4m8mRiIiIjqoHADqlsLAw45cM7jYCljEMk56ePmPGDFsXAvYL50cAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgBZyBABoIUcAgJajrQsAu7Nz587S0lLjKUeOHLl7967hZUxMTK9evTq8LrBfDMdxtq4B7EtcXNzOnTvFYjH/kuM4hmH4f9fX16tUqqKiIqFQaLsCwe7guAZMzZw5kxBS+zudTmf4t0AgmDlzJkIETGB/BEw1NDS4u7trNBqzc//zn/+8+OKLHVwS2Dnsj4ApgUAwe/ZskUjUeJa7u/vo0aM7viSwc8gRMGPmzJk6nc5kolAojI6ONpwrATDAcQ2Y5+fnZ3yNhnf58uXnn3/eJvWAPcP+CJgXHR1tcj7Vz88PIQJmIUfAvNmzZ9fV1RleCoXCuXPn2rAesGc4roEmDR48+NdffzV8Qm7cuNG3b1/blgT2Cfsj0KTo6GgHBwdCCMMwQ4YMQYhAU5Aj0KRZs2bp9XpCiIODw3//93/buhywX8gRaJKHh8fo0aMZhmloaAgPD7d1OWC/kCNgyZw5cziO++Mf/+jh4WHrWsCOcR0uPT3d1oMG6LLCwsI6/ktts/sGIE06i82bN8fFxcnlclsXAs1LSUmxSb82y5EZM2bYqmtoldGjR3t6etq6CmiRAwcO2KRfnB+BZiBEoFnIEQCghRwBAFrIEQCghRwBAFrIEQCghRwBAFrIEQCghRwBAFrIEQCghRwBAFrIEQCghRwBAFrIEQCg1TlyJDY2VqFQMAxz+fLlTtrXpk2bXF1dGYb5+OOPrVh9/fr1KpWqvbcAZZFmNTQ0pKSktOppngcPHvTz82MYhmEYNze32bNn05dBswGN6+GxLOvr6/v66683flSY5dWbGs4///lPDw8PhmEEAkG/fv2++eYbw6xXXnlFqVQKBILAwMAffvihtcV3kI6/dRJ/B6PWrrVv3z5CyKVLl9qjpI7p6+bNm4SQjz76yK6qMkFZpIkbN27wDxV//vnnW7uuv7+/SqVqkzJ4lBvQUI9er3/8+PGePXukUqmrq+vTp09btboFhJCRI0c2nv7dd9+NHz++Jb2EhYXZ5H5onWN/BDqjK1eu/O1vf1uwYMEf/vAHW9fSlgQCgaur65w5cxYtWqTRaIz3HbqtTpMjHfl4ajwKu008//zzBw8ejIqKEovFtq6lXQQEBBBCioqKbF2I7dlvjnAcl5yc3L9/f7FYrFKpVqxYYTxXr9cnJSV5e3tLJJLBgwcb3+01LS0tODiYZVmZTObj4/Pee+/xrW3ZsmXAgAFisdjJyWnq1Kl5eXk0fW3cuFEqlSoUCo1Gs3z58t69e+fn57dqgE0N4fvvvx84cKBKpWJZNigo6OTJk43Xffz4sY+Pj6Oj48svv0wIiY2N5Q+//f39L126RAiZO3euVCpVqVRHjx41W6qFDWiwZMkSkUjk5ubGv3zzzTdlMhnDME+fPm3VSM06ceKEUqlct26d1S102AZsqgD+GND4mcct2apdU8cfSrXw/EhiYiLDMJs3by4tLdVqtampqcTo4Pbtt98Wi8WZmZmlpaUJCQkCgeD8+fMcx/H3uV2/fn1xcXFJScknn3wSFRXFcVxSUpJIJEpLSysrK8vJyRk6dGjPnj2Liopo+kpMTCSELF26dNu2bdOnT79+/brlEZmcemiq2QMHDqxevbqkpKS4uHjUqFE9evTglzc+vNfpdKGhoUeOHDE0Hhoa6uDg8OjRI8OUWbNmHT161DBAk1Kb6t2kyKioqF69ehnaTE5OJoQ8efKk2bfP2MiRIxufH8nKylIoFGvWrGlqrWZPKHTkBjSpp7S09IsvvpBKpZMnT25JSS0ZDteZz4/YaY5otVqpVDphwgTDFOMPQU1NjVQqjYyMNCwsFosXLlyo0+nUanVISIhhrfr6+q1bt2q1Wrlcblie47iff/6ZEMJ/iK3ri/v9y1lTU9PCgRt/RS00a+z9998nhGg0GuOq6urqZs6cefz4ceMl+aP0tWvX8i/Ly8v79u1bX1/PvzQp1ULvHZYjzbL8xevgDcjXY/wfMMMwa9eu1el0LSypa+eInR7X3Lp1S6vVjh8/3uzc/Px8rVY7aNAg/qVEInFzc8vLy8vJySkrK5s4caJhSQcHh6VLl+bm5lZVVQUHBxumDx8+XCQSZWdnW90X5QBb2KxQKCSE8A/H5On1+lmzZrm6uvI75AYvvfRSv379PvvsM47jCCH79++PjIzkn87bYYPqSDbZgIYgWLFiBcdxKpWKb79VJXVJdpojBQUFhBAXFxezc6urqwkhq1atMlzPv3//vlarraioIISo1WqT5cvKygghJk9gUavVlZWVVvdFNz5LzX711Vfjxo1zcXERi8UrV640WXHRokU3b978+OOPr127ZjydYZj58+ffuXPn1KlThJA9e/a88cYbVvTeWdh2A77zzjtubm4JCQkPHz5sSUkt19DQ0HiiXq83Diw7ZKc5wrIsIaS2ttbsXP47n5KSYrxnde7cOf7ZkY3PAvLJwkVo9YkAABskSURBVKeGQVlZGf9EBev6ohtfk80+ePBg2rRpbm5u2dnZ5eXlGzZsMFlxxowZ//73v9VqdXR0dH19vfGsmJgYlmV37dqVn5+vVCr79OnT2t4pB9UBzp49y58Cs+0GVCgUH3zwQWVl5cKFCw0TrdiqhuHwnJ2dCwsLGy929+5dLy8vC+3YnJ3myKBBgwQCwZkzZ8zO9fLyYlm28Q8TfXx8nJ2dv/7668atyeXyCxcuGKZkZ2frdLphw4ZZ3Relppq9evVqXV3dwoUL/fz8WJZtfAU6JCSkZ8+eO3fuvHjx4tq1a41nOTk5RUREHD58eNOmTfPmzbOi98YcHR3r6upaNqaOcPHiRZlMRmy9AQkh0dHRI0eOzMrKysjI4KdY8VExDIf30ksvPXr06McffzRehuO4L774YuTIkS1vtuPZaY64uLiEhoZmZmbu3r27oqIiJydn586dhrksy86dO3ffvn07duyoqKjQ6/UFBQW//fabWCxOSEg4e/bskiVLHj161NDQUFlZee3aNZZlly9ffujQob1791ZUVFy9enXBggXu7u5xcXFW90U5wKaa9fb2JoR88803z549u3nzJn8Gp7EpU6bExMSsW7fu4sWLxtMXLFhQW1ublZX16quvWtF74yUDAgJKSkoOHz5cV1f35MmT+/fvUwz6/3H8+PFWXfetq6t7/Pjx6dOn+S+ebTcgIYRhmA8//JBhmCVLlpSWllooqSXD4a1du1atVoeHh//rX/+qrq6ura29cuXKrFmz6uvr58yZ08INZRt0p2mt0cLrvpWVlbGxsT169JDL5WPGjElKSiKEeHp6XrlyheO42tra+Ph4b29vR0dHPghyc3P5Fbdv3x4UFMSyLMuyQ4YMSU1N5TiuoaEhOTm5b9++QqHQyclp2rRp+fn5NH1t2LBBIpEQQry8vNLS0podzubNm3v16kUIkclk06dPtzCE+Ph4Z2dn/vO0fft2Qoi/v//WrVudnJwIIT4+PhqNpqKigt/Rlcvle/bsMe5oyJAhf//7342nmC3VbO+NiywuLg4JCeH/nGTx4sX8L2sCAgIePHjQ7JDPnTv34osvuru78580Nze30aNHnzlzhp977NgxhUJhuD5i7NChQyYXR4wdOnTIwhDaYwP+8MMP/fr143v38PCYP3++YVZMTAwhRK1Wr1+/vqmSWjgc3t27d+fNm+fr6ysSiSQSycCBA5OSkqqqqprd2jxbXa9hOI6jC6JWy8jIiIiI6Ph+u4nJkydv377d19fX1oV0Vp16A4aHhxNbPOXXTo9roFUMpzBycnL4fQfb1tPpYANSQo60jby8PKZpkZGR7dp7fHz8zZs3b9y4MXfuXP7vANqbbcfb5jp+A3YxjrYuoIsIDAy04ZGaVCoNDAzs3bt3amrqwIEDO6BH2463zXX8BuxicH4EoOvA+REA6KyQIwBACzkCALSQIwBACzkCALSQIwBACzkCALSQIwBACzkCALSQIwBACzkCALSQIwBACzkCALRsdt8APEMXoD2EhYV1fKc2uG9AQUGByR2xwZ5FREQsW7bshRdesHUh0CJeXl4d/2bZIEegc2EYJj09fcaMGbYuBOwXzo8AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQQo4AAC3kCADQcrR1AWB37t+/r9frjac8fvz4zp07hpfu7u4SiaTD6wL7xXAcZ+sawL5MmjTpxIkTTc11dHQsKirq0aNHR5YEdg7HNWAqMjKSYRizswQCwYQJExAiYAI5AqamT58uFAqbmjtnzpyOLAY6BeQImFIoFK+88orZKBEKha+++mrHlwR2DjkCZkRFRdXX15tMdHR0nDZtmlwut0lJYM+QI2DG5MmTZTKZyUS9Xh8VFWWTesDOIUfADLFYHBYWJhKJjCfK5fK//OUvtioJ7BlyBMybNWuWTqczvBQKhZGRkSbJAsDD70fAvIaGhl69ej19+tQw5bvvvhs3bpztKgL7hf0RME8gEMyaNcuwA+Li4jJ27FjblgR2CzkCTZo5cyZ/aCMSiaKjox0cHGxdEdgpHNdAkziO69Onz8OHDwkh58+fDw4OtnVFYKewPwJNYhgmOjqaENKnTx+ECFjQLf7eNzw83NYldFYVFRWEEJlMhm1otbfeeuuFF16wdRXtq1vsj2RmZhYUFNi6ik5JqVSqVCpPT09bF9JZZWZm8geGXVu32B8hhPz1r3+dMWOGravolE6ePDlx4kRbV9FZNfWX011Mt9gfARoIEWgWcgQAaCFHAIAWcgQAaCFHAIAWcgQAaCFHAIAWcgQAaCFHAIAWcgQAaCFHAIAWcgQAaCFHAIAWcgQAaCFHzIiNjVUoFAzDXL58uWv01ZEjaqE1a9YMHDhQqVSKxeKAgICVK1dWVVW1ZMWDBw/6+fkxRkQikaur67hx45KTk0tLS9u7cjCD6wYIIenp6a1aZd++fYSQS5cutVNJHd9XR46oJf70pz+lpqYWFxdXVFSkp6cLhcKXX3655av7+/urVCqO4xoaGkpLS7/77ruYmBiGYdzd3c+fP99uVbeaFZ+9zgj7I2Abcrk8Li7O2dlZoVDMmDFj2rRpJ06csOLWYQzDqNXqcePGff755xkZGY8fP548eXJ5eXl71AxNQY6Y15G3seqYvuztxlxZWVnGD7Lo2bMnIUSr1dK0GRYWFhMTo9FoPv74Y9r6oDWQI/+H47jk5OT+/fuLxWKVSrVixQrjuXq9PikpydvbWyKRDB48OD093TArLS0tODiYZVmZTObj4/Pee+/xrW3ZsmXAgAFisdjJyWnq1Kl5eXk0fW3cuFEqlSoUCo1Gs3z58t69e+fn57f5iHbs2CGTyaRS6ZEjRyZNmqRUKj09PfkDIt6ZM2dGjBghlUqVSmVQUBB/F2gLG6flHj16JJFIfH19+ZcnTpxQKpXr1q1rbTsxMTGEkOPHj9vnMLss2x5WdQzSgmPUxMREhmE2b95cWlqq1WpTU1OJ0dmEt99+WywWZ2ZmlpaWJiQkCAQC/iA8JSWFELJ+/fri4uKSkpJPPvkkKiqK47ikpCSRSJSWllZWVpaTkzN06NCePXsWFRXR9JWYmEgIWbp06bZt26ZPn379+vX2GBHfy6lTp8rLyzUazdixY2UymU6n4ziuqqpKqVRu2LChpqamqKho+vTpT548sdBUy1VXVysUiiVLlhimZGVlKRSKNWvWNLWK4fyICf477+XlZSfDbMlnrwtAjnAcx2m1WqlUOmHCBMMU47OSNTU1Uqk0MjLSsLBYLF64cKFOp1Or1SEhIYa16uvrt27dqtVq5XK5YXmO437++WdCCP+tsK4v7vePfk1NTUuG3Fa98Olz69YtjuN+/fVXQkhWVpZxRxaaarnExMR+/fpVVFS0fJWmcoTjOP6MiZ0Ms5vkCI5rCCHk1q1bWq12/PjxZufm5+drtdpBgwbxLyUSiZubW15eXk5OTllZmfFtkB0cHJYuXZqbm1tVVWX84Kjhw4eLRKLs7Gyr++qYETVekn++b11dHSHEz8/P1dV19uzZq1evvnfvXlsVfOjQoYyMjJMnTyoUipav1ZTq6mqO45RKZatq64Bhdm3IEUII4Z9u4+LiYnZudXU1IWTVqlWGHyzcv39fq9Xyu9Bqtdpk+bKyMkKIXC43nqhWqysrK63uq2NGZLlNiUTy7bffjhkzZt26dX5+fpGRkTU1NZQF79+//4MPPjh9+rSPj0/LR2fBjRs3CCGBgYHEnobZ5SFHCCGEZVlCSG1trdm5/LcxJSXFeEfu3LlzHh4ehJCnT5+aLM8nC58aBmVlZfzTpKzrq2NG1Gyzzz333JdffllYWBgfH5+enr5p0yaagrdt27Z3795vv/2W35Jt4sSJE4SQSZMmEbsZZneAHCGEkEGDBgkEgjNnzpid6+XlxbJs41+C+vj4ODs7f/31141bk8vlFy5cMEzJzs7W6XTDhg2zuq/Wao9eCgsLr127RghxcXFZv3790KFDr127Zl1THMfFx8dfvXr18OHDJjtuNIqKilJSUjw9PV9//XViB8PsRtriJIu9Iy041xUeHu7g4LBr167y8vIrV66EhIQQo6sbCxYsEIlEqamp5eXl9fX1Dx8+LCws5Dhu06ZNhJDFixcXFBTo9fqKiorc3FyO4/7xj38IhcK0tLTy8vKcnJwhQ4a4u7tXVVXR9NWq86xt1cunn35KCOGvDZ0/f37s2LHXr1+vra395ZdfevTosXXrVgtNWcCfy2wsOTmZX+DYsWMKhWLt2rVNteDv769UKisrK/V6fUNDg0aj2b9/v5+fn5ub24ULFwyL2XaYXLc5z4oc+T+VlZWxsbE9evSQy+VjxoxJSkoihHh6el65coXjuNra2vj4eG9vb0dHRxcXl9DQUD4vOI7bvn17UFAQy7Isyw4ZMiQ1NZXjuIaGhuTk5L59+wqFQicnp2nTpuXn59P0tWHDBolEQgjx8vJKS0tryait6CU1NVUqlRJC+vbte/v27Z07d/InLPv06XPjxo179+6NHj3aycnJwcHBw8MjMTGxvr7e8sZpytWrV63OkaNHjw4ePFgqlYpEIoFAQH7/SeuIESPWrFlTXFxsvLBth8l1mxxhOI5rmx0bO8YwTHp6Op7vCx2vm3z2cH4EAGghRzqlvLw8pmmRkZEoDzqSo60LAGsEBgba8wGpnZcHbQ77IwBACzkCALSQIwBACzkCALSQIwBACzkCALSQIwBACzkCALSQIwBACzkCALSQIwBACzkCALSQIwBACzkCALS6y30DUlJSDhw4YOsqALqmbrE/EhYWxj/zAaxw9OjRwsJCW1fRWYWFhXl5edm6inbXLe7PCjS6yR1GgUa32B8BgHaFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKCFHAEAWsgRAKDFcBxn6xrAvsyZM+fy5cuGl/fu3XNxcZHJZPxLoVD45Zdf9u7d20bVgT1ytHUBYHf69++/d+9e4ylVVVWGfwcGBiJEwASOa8DUzJkzGYYxO0soFMbExHRsOdAJ4LgGzBg2bNjly5cbGhpMpjMMc+fOHR8fH1sUBfYL+yNgRnR0tEBg+tlgGGbEiBEIEWgMOQJmRERENN4ZEQgE0dHRNqkH7BxyBMxwc3MbO3asg4ODyfTQ0FCb1AN2DjkC5s2ZM8f4pUAgCAkJ6dWrl63qAXuGHAHzwsPDTU6RmCQLgAFyBMxTKpUvv/yyo+P//cLIwcHhtddes21JYLeQI9Ck2bNn6/V6Qoijo+OUKVNUKpWtKwI7hRyBJk2ZMkUikRBC9Hp9VFSUrcsB+4UcgSaxLDt9+nRCiFQqnTRpkq3LAfvVyf6+pqCg4Mcff7R1Fd2Il5cXIWT48OFHjx61dS3diJeX1wsvvGDrKlqD61TS09NtvcEA2l1YWJitv2qt08n2R3gc/iaoA61evXrVqlWGCzfQ3sLDw21dQqvh/Ag0AyECzUKOQDMQItAs5AgA0EKOAAAt5AgA0EKOAAAt5AgA0EKOAAAt5AgA0EKOAAAt5AgA0EKOAAAt5AgA0EKOAACtrp8jsbGxCoWCYZjLly93pb7s3MGDB/38/BiGYRjGzc1t9uzZ9G2uX79epVJZt3mN6+GxLOvr6/v666/fvXu3Vas3NZx//vOfHh4eDMMIBIJ+/fp98803hlmvvPKKUqkUCASBgYE//PBDa4vvBGx9A5TW4e9j1Nq19u3bRwi5dOlSe5Rkw77sn7+/v0qlasMGKTevoR69Xv/48eM9e/ZIpVJXV9enT5+2anULCCEjR45sPP27774bP358S3oJCwvrdPcx6vr7I1BTUzN69Gh7a8q2BAKBq6vrnDlzFi1apNFojPcdwArdIkcYhumSfbXQ7t27NRqNvTVlJwICAgghRUVFti6kc+uaOcJxXHJycv/+/cVisUqlWrFihfFcvV6flJTk7e0tkUgGDx5sfM/XtLS04OBglmVlMpmPj897773Ht7Zly5YBAwaIxWInJ6epU6fm5eXR9LVx40apVKpQKDQazfLly3v37p2fn9/siJqqYcmSJSKRyM3NjX/55ptvymQyhmGePn1KCFm2bNny5ctv377NMExAQMCHH37Isqyrq+v8+fPd3d1Zlh09enR2drYVTRFCTpw4oVQq161b18r3p/n34vvvvx84cKBKpWJZNigo6OTJk43Xffz4sY+Pj6Oj48svv0wIiY2N5c9f+Pv7X7p0iRAyd+5cqVSqUqks3KT65s2bhJDnn3++2ZLAEtseVrVWC8+PJCYmMgyzefPm0tJSrVabmppKjA6q3377bbFYnJmZWVpampCQIBAIzp8/z3FcSkoKIWT9+vXFxcUlJSWffPJJVFQUx3FJSUkikSgtLa2srCwnJ2fo0KE9e/YsKiqi6SsxMZEQsnTp0m3btk2fPv369euWR2S5hqioqF69ehkWTk5OJoQ8efKEfxkaGurv72+YGxcXJ5PJrl279uzZs9zc3OHDhysUigcPHljRVFZWlkKhWLNmTVNlN3tCoantc+DAgdWrV5eUlBQXF48aNapHjx788sbnR3Q6XWho6JEjRwythYaGOjg4PHr0yDBl1qxZR48eNVtPaWnpF198IZVKJ0+e3JKSWjIcrrueH+mCOaLVaqVS6YQJEwxTjD98NTU1Uqk0MjLSsLBYLF64cKFOp1Or1SEhIYa16uvrt27dqtVq5XK5YXmO437++WdCCP/lsa4v7vccqampacmoLdfAtT5HjL8M58+fJ4S8++67VjTVLMtfPAvbx9j7779PCNFoNJzR5q2rq5s5c+bx48eNl+RPc6xdu5Z/WV5e3rdv3/r6euN6jP8TZRhm7dq1Op2uhSUhR5rSBY9rbt26pdVqx48fb3Zufn6+VqsdNGgQ/1Iikbi5ueXl5eXk5JSVlU2cONGwpIODw9KlS3Nzc6uqqoKDgw3Thw8fLhKJ+GMB6/pq7Ygs10ApODhYKpVaURW9Fm4foVBICOGfEMrT6/WzZs1ydXXlj2gMXnrppX79+n322WccxxFC9u/fHxkZ6eDgYLyMIQhWrFjBcZxKpeLbb1VJYKIL5khBQQEhxMXFxezc6upqQsiqVasMvyO4f/++VqutqKgghKjVapPly8rKCCFyudx4olqtrqystLqv1o7Icg30xGLxkydP2qSpVrGwfb766qtx48a5uLiIxeKVK1earLho0aKbN29+/PHH165dM57OMMz8+fPv3Llz6tQpQsiePXveeOONpnp/55133NzcEhISHj582JKSWq6hoaHxRL1ebxxYXUwXzBGWZQkhtbW1Zufy3/mUlBTjvbJz5855eHgQQvgTisb4ZDH5xpaVlXl6elrdV2tHZLkGSnV1dW3VVAudPXuWPxXV1PZ58ODBtGnT3NzcsrOzy8vLN2zYYNLCjBkz/v3vf6vV6ujo6Pr6euNZMTExLMvu2rUrPz9fqVT26dOnqTIUCsUHH3xQWVm5cOFCw0Qr3jLDcHjOzs6FhYWNF7t79y7/cMIuqQvmyKBBgwQCwZkzZ8zO9fLyYlm28Q8ifXx8nJ2dv/7668atyeXyCxcuGKZkZ2frdLphw4ZZ3VdrWa6BEOLo6FhXV2dd46dPn+Y4btSoUfRNtdDFixdlMhlpevtcvXq1rq5u4cKFfn5+LMs2vpQeEhLSs2fPnTt3Xrx4ce3atcaznJycIiIiDh8+vGnTpnnz5lmuJDo6euTIkVlZWRkZGfwUK94yw3B4L7300qNHj0yeHstx3BdffDFy5MiWN9vJtP0pl/bUwus14eHhDg4Ou3btKi8vv3LlSkhICDG6hrJgwQKRSJSamlpeXl5fX//w4cPCwkKO4zZt2kQIWbx4cUFBgV6vr6ioyM3N5TjuH//4h1AoTEtLKy8vz8nJGTJkiLu7e1VVFU1frTrP2mwN/PXpf/3rXzqdTqPRLFq0iBidHJ03b55EIrl7925FRYVOp4uLi1MoFCUlJXV1dVeuXBk4cKC3t/ezZ8+saOrYsWMKhcJwXrMxkxOTOp2uqKhoypQpe/futbB9cnJyCCHvvPNOTU3NjRs3wsLCCCG//fYb1+j3rDExMY6OjhcuXDDu9KeffiKEuLm51dXVWa6H47js7GyGYXr16lVSUmKhpBYOh+O4vLw8tVrt4eFx6NChqqqqZ8+eXb58OTIycuTIkbW1tU1tKGOd8Txr18yRysrK2NjYHj16yOXyMWPGJCUlEUI8PT2vXLnCcVxtbW18fLy3t7ejo6OLi0toaCifFxzHbd++PSgoiGVZlmWHDBmSmprKcVxDQ0NycnLfvn2FQqGTk9O0adPy8/Np+tqwYYNEIiGEeHl5paWltWTglmsoLi4OCQnh/2Bk8eLF/G9YAgIC+Ku5v/zyS58+fSQSyZgxY4qKiuLi4oRCYe/evR0dHZVK5dSpU2/fvm1dUxZy5NChQyYXR4wdOnSIX6yp9yI+Pt7Z2VmtVoeHh2/fvp0Q4u/vv3XrVicnJ0KIj4+PRqOpqKjgjxTkcvmePXuMex8yZMjf//534yk//PBDv379+N49PDzmz59vmBUTE0MIUavV69evb6qkFg6Hd/fu3Xnz5vn6+opEIolEMnDgwKSkJEPoN6sz5gjDdapn5WZkZERERHSumu3N/PnzDxw4UFxcbOtC2tHkyZO3b9/u6+tr60KswT/f98CBA7YupBW64PkRaJbxNdQuw3BaJycnh9+fsm093QpyxPby8vKYpkVGRtq6wM4hPj7+5s2bN27cmDt3Ln+WBzoMHgFte4GBgR12pJaQkPD555/rdDpfX9/k5GT+FGbXIJVKAwMDe/funZqaOnDgQFuX073g/AiAfcH5EQDojpAjAEALOQIAtJAjAEALOQIAtJAjAEALOQIAtJAjAEALOQIAtJAjAEALOQIAtJAjAEALOQIAtDrlfQMMd+UF6HoKCgo68vb9baJT5khERIStSwBoR53uvjCd7P4jAGCHcH4EAGghRwCAFnIEAGghRwCA1v8HQZxVSwD/fLUAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lEbAkzSguep6"
      },
      "source": [
        "#The Complete AutoEncoder Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TFK-t0kfuSuE"
      },
      "source": [
        "AutoEncoder_output = decoder(encoder(encoder_input))"
      ],
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tDICIqJFvSm3"
      },
      "source": [
        "#The autoencoder model\n",
        "AutoEncoder = Model(encoder_input,AutoEncoder_output,name=\"Autoencoder\")"
      ],
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TumywvdDvevU",
        "outputId": "45316518-4e06-4c26-b18d-dac377cca9f5"
      },
      "source": [
        "#Autoencoder parameters\n",
        "AutoEncoder.summary()"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"Autoencoder\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "encoder_input (InputLayer)   [(None, 28, 28, 1)]       0         \n",
            "_________________________________________________________________\n",
            "encoder_model (Functional)   (None, 26, 26, 10)        650       \n",
            "_________________________________________________________________\n",
            "Decoder (Functional)         (None, 28, 28, 1)         2945      \n",
            "=================================================================\n",
            "Total params: 3,595\n",
            "Trainable params: 3,595\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 272
        },
        "id": "1u8A8Zn9vji9",
        "outputId": "f1e8338a-b69c-4599-a20c-700879b3d191"
      },
      "source": [
        "#Autoencoder layers\n",
        "plot_model(AutoEncoder)"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wsMj21vRwB8Z"
      },
      "source": [
        "### Compiling the autoencoder model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZWWxf8x4vs_6"
      },
      "source": [
        "AutoEncoder.compile(optimizer=Adam(learning_rate=0.0005),loss='mse')"
      ],
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tShj_K1awrDO"
      },
      "source": [
        "# Loading the data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "41q2OIZywi9w"
      },
      "source": [
        "from tensorflow.keras.datasets.mnist import load_data #we used the mnist dataset\n",
        "import numpy as np"
      ],
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5DjMI9XMxFQc",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d11708a0-531f-46aa-b921-098ee21c5f89"
      },
      "source": [
        "(x_train_orig,y_train),(x_test_orig,y_test) = load_data() #downloading the data"
      ],
      "execution_count": 24,
      "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"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pLuJGUDJxNO0",
        "outputId": "2d1461ad-da9b-478f-8070-2281818aebb1"
      },
      "source": [
        "x_train_orig.shape, x_test_orig.shape"
      ],
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((60000, 28, 28), (10000, 28, 28))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wJVR2q7wxTLL"
      },
      "source": [
        "#Normalising the pixels of the images\n",
        "x_train_orig = x_train_orig.astype('float32')/255.0\n",
        "x_test_orig = x_test_orig.astype('float32')/255.0"
      ],
      "execution_count": 26,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OoJL850-xiaK"
      },
      "source": [
        "#Since the images are in a matrix so converting them into a vector\n",
        "x_train = x_train_orig.reshape((x_train_orig.shape[0],28,28,1))\n",
        "x_test = x_test_orig.reshape((x_test_orig.shape[0],28,28,1))"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SBQjyEVZx0f7",
        "outputId": "021efd58-8516-4f1f-acfe-6fb4b498c97a"
      },
      "source": [
        "#new shapes\n",
        "x_train.shape, x_test.shape"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((60000, 28, 28, 1), (10000, 28, 28, 1))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6Y0-8pyByHBe",
        "outputId": "6fb396e5-ce4c-4021-89cb-a2beabd427ba"
      },
      "source": [
        "#Now that our data is in vector order, we provide it as an input and train our model\n",
        "\n",
        "#The labels y have not been given because we don't need that\n",
        "AutoEncoder.fit(x_train,x_train,batch_size=256,epochs=20,shuffle=True,validation_data=(x_test,x_test),verbose = 1)"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "235/235 [==============================] - 35s 25ms/step - loss: 0.0106 - val_loss: 8.9230e-04\n",
            "Epoch 2/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 3.9247e-04 - val_loss: 1.6438e-04\n",
            "Epoch 3/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 1.2240e-04 - val_loss: 9.3285e-05\n",
            "Epoch 4/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 8.1611e-05 - val_loss: 6.8493e-05\n",
            "Epoch 5/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 6.2239e-05 - val_loss: 5.3838e-05\n",
            "Epoch 6/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 4.9864e-05 - val_loss: 4.3878e-05\n",
            "Epoch 7/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 4.0988e-05 - val_loss: 3.6405e-05\n",
            "Epoch 8/20\n",
            "235/235 [==============================] - 6s 24ms/step - loss: 3.4284e-05 - val_loss: 3.0833e-05\n",
            "Epoch 9/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 2.9309e-05 - val_loss: 2.6725e-05\n",
            "Epoch 10/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 2.5546e-05 - val_loss: 2.3580e-05\n",
            "Epoch 11/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 2.2589e-05 - val_loss: 2.1018e-05\n",
            "Epoch 12/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 2.0223e-05 - val_loss: 1.8999e-05\n",
            "Epoch 13/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 1.8364e-05 - val_loss: 1.7824e-05\n",
            "Epoch 14/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 1.7006e-05 - val_loss: 1.6427e-05\n",
            "Epoch 15/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 1.5801e-05 - val_loss: 1.5744e-05\n",
            "Epoch 16/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 1.4812e-05 - val_loss: 1.9012e-05\n",
            "Epoch 17/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 1.4138e-05 - val_loss: 1.4356e-05\n",
            "Epoch 18/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 1.3426e-05 - val_loss: 1.4625e-05\n",
            "Epoch 19/20\n",
            "235/235 [==============================] - 5s 23ms/step - loss: 1.2871e-05 - val_loss: 1.2246e-05\n",
            "Epoch 20/20\n",
            "235/235 [==============================] - 6s 23ms/step - loss: 1.2083e-05 - val_loss: 1.1720e-05\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7f77501ff350>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wb4sNGFs29uY"
      },
      "source": [
        "## Predicting using Encoder and Decoder independently"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wlbsPXMTzjln"
      },
      "source": [
        "#Now we predict first using the encoder model to get a vector of 10 latent features\n",
        "encoder_prediction = encoder.predict(x_train)"
      ],
      "execution_count": 30,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nQAGZMWXz1X2"
      },
      "source": [
        "#Predciting the decoder output using the output of the encoder which generates an image along with the depth channel of 1  \n",
        "decoder_prediction = decoder.predict(encoder_prediction)"
      ],
      "execution_count": 31,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0AVxzErv0MzM"
      },
      "source": [
        "#Converting the image vector to a matrix\n",
        "decoder_images =  decoder_prediction.reshape((decoder_prediction.shape[0],28,28))"
      ],
      "execution_count": 32,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AyjS0HGM02ES"
      },
      "source": [
        "import matplotlib"
      ],
      "execution_count": 35,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 268
        },
        "id": "ynw37X0-0sGp",
        "outputId": "432d9708-645f-41de-db22-6d25e1c8beb1"
      },
      "source": [
        "#Plotting the output\n",
        "num_images_to_show = 5\n",
        "for im_ind in range(num_images_to_show):\n",
        "    plot_ind = im_ind*2 + 1\n",
        "    rand_ind = np.random.randint(low=0, high=x_train.shape[0])\n",
        "    matplotlib.pyplot.subplot(num_images_to_show, 2, plot_ind)\n",
        "    matplotlib.pyplot.imshow(x_train_orig[rand_ind, :, :], cmap=\"gray\")\n",
        "    matplotlib.pyplot.subplot(num_images_to_show, 2, plot_ind+1)\n",
        "    matplotlib.pyplot.imshow(decoder_images[rand_ind, :, :], cmap=\"gray\")"
      ],
      "execution_count": 36,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 10 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "P5bI-Mcj2c-F"
      },
      "source": [
        "### Note: The quality of images are much better than Vanilla AE because we used CNN and also used more latent features"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JHdM-Jr3243v"
      },
      "source": [
        "## Predicting using complete Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pJK6WT0UyxnH"
      },
      "source": [
        "#Predicitng using the Overall Model\n",
        "#Since we want to generate the images themselevs, so we predict on x_train again\n",
        "autoencoder_output_vector = AutoEncoder.predict(x_train)"
      ],
      "execution_count": 39,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vyRJm7wizbdw"
      },
      "source": [
        "#Converting vector to matrix\n",
        "autoencoder_images = autoencoder_output_vector.reshape((autoencoder_output_vector.shape[0],28,28))"
      ],
      "execution_count": 40,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 268
        },
        "id": "l8iEp1V32uDw",
        "outputId": "1eb86250-eb5e-44f9-c7fb-6f35d015cf54"
      },
      "source": [
        "#Plotting these images\n",
        "num_images_to_show = 5\n",
        "for im_ind in range(num_images_to_show):\n",
        "    plot_ind = im_ind*2 + 1\n",
        "    rand_ind = np.random.randint(low=0, high=x_train.shape[0])\n",
        "    matplotlib.pyplot.subplot(num_images_to_show, 2, plot_ind)\n",
        "    matplotlib.pyplot.imshow(x_train_orig[rand_ind, :, :], cmap=\"gray\")\n",
        "    matplotlib.pyplot.subplot(num_images_to_show, 2, plot_ind+1)\n",
        "    matplotlib.pyplot.imshow(autoencoder_images[rand_ind, :, :], cmap=\"gray\")"
      ],
      "execution_count": 41,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 10 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}