auto_encoders.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "gpuType": "T4",
      "authorship_tag": "ABX9TyO+gZhpNX7KWrmP/yifye47",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/miller00315/ia_studies/blob/main/auto_encoders.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 37,
      "metadata": {
        "id": "PvOSP2LC9e8b"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from keras.datasets import mnist\n",
        "from keras.models import Sequential, Model\n",
        "from keras.layers import Dense, Input"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "(X_train, _, ), (X_test, _) = mnist.load_data()"
      ],
      "metadata": {
        "id": "NmzHT7lYADP-"
      },
      "execution_count": 38,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "X_train.shape"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cWYCrKJbAkp9",
        "outputId": "c17ebaca-30c4-4f9f-c0bf-70d5cf7c0345"
      },
      "execution_count": 39,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(60000, 28, 28)"
            ]
          },
          "metadata": {},
          "execution_count": 39
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "X_train[0]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 66
        },
        "id": "0ExVAZlZApRY",
        "outputId": "b94b5aa8-0678-4d10-9fa4-3ef7896c7230"
      },
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   3,\n",
              "         18,  18,  18, 126, 136, 175,  26, 166, 255, 247, 127,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,  30,  36,  94, 154, 170,\n",
              "        253, 253, 253, 253, 253, 225, 172, 253, 242, 195,  64,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,  49, 238, 253, 253, 253, 253,\n",
              "        253, 253, 253, 253, 251,  93,  82,  82,  56,  39,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,  18, 219, 253, 253, 253, 253,\n",
              "        253, 198, 182, 247, 241,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,  80, 156, 107, 253, 253,\n",
              "        205,  11,   0,  43, 154,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,  14,   1, 154, 253,\n",
              "         90,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0, 139, 253,\n",
              "        190,   2,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,  11, 190,\n",
              "        253,  70,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,  35,\n",
              "        241, 225, 160, 108,   1,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "         81, 240, 253, 253, 119,  25,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,  45, 186, 253, 253, 150,  27,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,  16,  93, 252, 253, 187,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0, 249, 253, 249,  64,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,  46, 130, 183, 253, 253, 207,   2,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,  39,\n",
              "        148, 229, 253, 253, 253, 250, 182,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,  24, 114, 221,\n",
              "        253, 253, 253, 253, 201,  78,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,  23,  66, 213, 253, 253,\n",
              "        253, 253, 198,  81,   2,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,  18, 171, 219, 253, 253, 253, 253,\n",
              "        195,  80,   9,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,  55, 172, 226, 253, 253, 253, 253, 244, 133,\n",
              "         11,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0, 136, 253, 253, 253, 212, 135, 132,  16,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0]], dtype=uint8)"
            ],
            "text/html": [
              "<style>\n",
              "      .ndarray_repr .ndarray_raw_data {\n",
              "        display: none;\n",
              "      }\n",
              "      .ndarray_repr.show_array .ndarray_raw_data {\n",
              "        display: block;\n",
              "      }\n",
              "      .ndarray_repr.show_array .ndarray_image_preview {\n",
              "        display: none;\n",
              "      }\n",
              "      </style>\n",
              "      <div id=\"id-9edf3ad2-b815-46d0-b9a0-66974c99be18\" class=\"ndarray_repr\"><pre>ndarray (28, 28) <button style=\"padding: 0 2px;\">show data</button></pre><img src=\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABwAAAAcCAAAAABXZoBIAAABAElEQVR4nGNgGMyAWUhIqK5jvdSy/9/rGRgYGFhgEnJsVjYCwQwMDAxPJgV+vniQgYGBgREqZ7iXH8r6l/SV4dn7m8gmCt3++/fv37/Htn3/iMW+gDnZf/+e5WbQnoXNNXyMs/5GoQoxwVmf/n9kSGFiwAW49/11wynJoPzx4YIcRlyygR/+/i2XxCWru+vv32nSuGQFYv/83Y3b4p9/fzpAmSyoMnohpiwM1w5h06Q+5enfv39/bcMiJVF09+/fv39P+mFKiTtd/fv3799jgZiBJLT69t+/f/8eDuDEkDJf8+jv379/v7Ryo4qzMDAwMAQGMjBc3/y35wM2V1IfAABFF16Aa0wAOwAAAABJRU5ErkJggg==\" class=\"ndarray_image_preview\" /><pre class=\"ndarray_raw_data\">array([[  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   3,\n",
              "         18,  18,  18, 126, 136, 175,  26, 166, 255, 247, 127,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,  30,  36,  94, 154, 170,\n",
              "        253, 253, 253, 253, 253, 225, 172, 253, 242, 195,  64,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,  49, 238, 253, 253, 253, 253,\n",
              "        253, 253, 253, 253, 251,  93,  82,  82,  56,  39,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,  18, 219, 253, 253, 253, 253,\n",
              "        253, 198, 182, 247, 241,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,  80, 156, 107, 253, 253,\n",
              "        205,  11,   0,  43, 154,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,  14,   1, 154, 253,\n",
              "         90,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0, 139, 253,\n",
              "        190,   2,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,  11, 190,\n",
              "        253,  70,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,  35,\n",
              "        241, 225, 160, 108,   1,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "         81, 240, 253, 253, 119,  25,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,  45, 186, 253, 253, 150,  27,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,  16,  93, 252, 253, 187,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0, 249, 253, 249,  64,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,  46, 130, 183, 253, 253, 207,   2,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,  39,\n",
              "        148, 229, 253, 253, 253, 250, 182,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,  24, 114, 221,\n",
              "        253, 253, 253, 253, 201,  78,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,  23,  66, 213, 253, 253,\n",
              "        253, 253, 198,  81,   2,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,  18, 171, 219, 253, 253, 253, 253,\n",
              "        195,  80,   9,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,  55, 172, 226, 253, 253, 253, 253, 244, 133,\n",
              "         11,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0, 136, 253, 253, 253, 212, 135, 132,  16,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0],\n",
              "       [  0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,   0,\n",
              "          0,   0]], dtype=uint8)</pre></div><script>\n",
              "      (() => {\n",
              "      const titles = ['show data', 'hide data'];\n",
              "      let index = 0\n",
              "      document.querySelector('#id-9edf3ad2-b815-46d0-b9a0-66974c99be18 button').onclick = (e) => {\n",
              "        document.querySelector('#id-9edf3ad2-b815-46d0-b9a0-66974c99be18').classList.toggle('show_array');\n",
              "        index = (++index) % 2;\n",
              "        document.querySelector('#id-9edf3ad2-b815-46d0-b9a0-66974c99be18 button').textContent = titles[index];\n",
              "        e.preventDefault();\n",
              "        e.stopPropagation();\n",
              "      }\n",
              "      })();\n",
              "    </script>"
            ]
          },
          "metadata": {},
          "execution_count": 40
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.imshow(X_train[4], cmap='gray')\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 430
        },
        "id": "FQlzTqvvA_RJ",
        "outputId": "830af7da-dc7d-4e46-9f51-63d0b4d2fadd"
      },
      "execution_count": 41,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "X_train = X_train.astype('float32') / 255.0\n",
        "X_test = X_test.astype('float32') / 255.0"
      ],
      "metadata": {
        "id": "LNjFs5J2BUM1"
      },
      "execution_count": 42,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.imshow(X_train[4], cmap='gray')\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 430
        },
        "id": "wQzZY8OkBq8R",
        "outputId": "49378f5f-e610-4234-ab9a-a03367914d3c"
      },
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "X_train = X_train.reshape((len(X_train), np.prod(X_train.shape[1:])))\n",
        "X_test = X_test.reshape((len(X_test), np.prod(X_test.shape[1:])))"
      ],
      "metadata": {
        "id": "h1ZYUpvGBtO2"
      },
      "execution_count": 44,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_train_noisy = X_train + 0.5 + np.random.normal(loc=0.0, scale=1.0, size=X_train.shape)\n",
        "x_test_noisy = X_test + 0.5 + np.random.normal(loc=0.0, scale=1.0, size=X_test.shape)"
      ],
      "metadata": {
        "id": "_rMOxgecC-B1"
      },
      "execution_count": 45,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_train_noise = np.clip(x_train_noisy, 0.0, 1.0)\n",
        "x_test_noisy = np.clip(x_test_noisy, 0.0, 1.0)"
      ],
      "metadata": {
        "id": "kPMDvABtEKGD"
      },
      "execution_count": 46,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "inputs = Input(shape=(784,))\n",
        "\n",
        "encoder = Dense(32, activation='relu')(inputs)\n",
        "\n",
        "decoder = Dense(784, activation='sigmoid')(encoder)\n",
        "\n",
        "autoencoder = Model(inputs, decoder)\n",
        "\n",
        "autoencoder.compile(optimizer='adam', loss='binary_crossentropy')\n",
        "\n",
        "autoencoder.fit(x_train_noise, X_train, epochs=100, batch_size=256, shuffle=True, validation_data=(x_test_noisy, X_test))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KNQoWb2VEo0j",
        "outputId": "b18ec658-c0ff-4f71-8906-94061a9307f5"
      },
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "235/235 [==============================] - 3s 10ms/step - loss: 0.2924 - val_loss: 0.2608\n",
            "Epoch 2/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.2518 - val_loss: 0.2407\n",
            "Epoch 3/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.2314 - val_loss: 0.2210\n",
            "Epoch 4/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.2155 - val_loss: 0.2096\n",
            "Epoch 5/100\n",
            "235/235 [==============================] - 3s 13ms/step - loss: 0.2079 - val_loss: 0.2044\n",
            "Epoch 6/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.2030 - val_loss: 0.2001\n",
            "Epoch 7/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1987 - val_loss: 0.1953\n",
            "Epoch 8/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1948 - val_loss: 0.1932\n",
            "Epoch 9/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1927 - val_loss: 0.1913\n",
            "Epoch 10/100\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.1912 - val_loss: 0.1902\n",
            "Epoch 11/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1902 - val_loss: 0.1892\n",
            "Epoch 12/100\n",
            "235/235 [==============================] - 3s 11ms/step - loss: 0.1894 - val_loss: 0.1887\n",
            "Epoch 13/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1889 - val_loss: 0.1884\n",
            "Epoch 14/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1885 - val_loss: 0.1881\n",
            "Epoch 15/100\n",
            "235/235 [==============================] - 3s 11ms/step - loss: 0.1882 - val_loss: 0.1878\n",
            "Epoch 16/100\n",
            "235/235 [==============================] - 3s 13ms/step - loss: 0.1879 - val_loss: 0.1875\n",
            "Epoch 17/100\n",
            "235/235 [==============================] - 2s 11ms/step - loss: 0.1877 - val_loss: 0.1872\n",
            "Epoch 18/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1876 - val_loss: 0.1875\n",
            "Epoch 19/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1874 - val_loss: 0.1872\n",
            "Epoch 20/100\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.1873 - val_loss: 0.1870\n",
            "Epoch 21/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1872 - val_loss: 0.1870\n",
            "Epoch 22/100\n",
            "235/235 [==============================] - 3s 13ms/step - loss: 0.1871 - val_loss: 0.1871\n",
            "Epoch 23/100\n",
            "235/235 [==============================] - 3s 11ms/step - loss: 0.1871 - val_loss: 0.1869\n",
            "Epoch 24/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1870 - val_loss: 0.1867\n",
            "Epoch 25/100\n",
            "235/235 [==============================] - 3s 15ms/step - loss: 0.1870 - val_loss: 0.1870\n",
            "Epoch 26/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1869 - val_loss: 0.1868\n",
            "Epoch 27/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1868 - val_loss: 0.1868\n",
            "Epoch 28/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1868 - val_loss: 0.1866\n",
            "Epoch 29/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1867 - val_loss: 0.1867\n",
            "Epoch 30/100\n",
            "235/235 [==============================] - 3s 13ms/step - loss: 0.1867 - val_loss: 0.1866\n",
            "Epoch 31/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1866 - val_loss: 0.1865\n",
            "Epoch 32/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1866 - val_loss: 0.1864\n",
            "Epoch 33/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1865 - val_loss: 0.1866\n",
            "Epoch 34/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1865 - val_loss: 0.1866\n",
            "Epoch 35/100\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.1865 - val_loss: 0.1863\n",
            "Epoch 36/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1864 - val_loss: 0.1862\n",
            "Epoch 37/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1864 - val_loss: 0.1865\n",
            "Epoch 38/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1863 - val_loss: 0.1861\n",
            "Epoch 39/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1863 - val_loss: 0.1862\n",
            "Epoch 40/100\n",
            "235/235 [==============================] - 3s 14ms/step - loss: 0.1863 - val_loss: 0.1861\n",
            "Epoch 41/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1862 - val_loss: 0.1862\n",
            "Epoch 42/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1862 - val_loss: 0.1863\n",
            "Epoch 43/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1862 - val_loss: 0.1859\n",
            "Epoch 44/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1862 - val_loss: 0.1860\n",
            "Epoch 45/100\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.1862 - val_loss: 0.1861\n",
            "Epoch 46/100\n",
            "235/235 [==============================] - 3s 11ms/step - loss: 0.1861 - val_loss: 0.1860\n",
            "Epoch 47/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1861 - val_loss: 0.1860\n",
            "Epoch 48/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1861 - val_loss: 0.1859\n",
            "Epoch 49/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1861 - val_loss: 0.1859\n",
            "Epoch 50/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1860 - val_loss: 0.1863\n",
            "Epoch 51/100\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.1860 - val_loss: 0.1859\n",
            "Epoch 52/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1860 - val_loss: 0.1862\n",
            "Epoch 53/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1860 - val_loss: 0.1857\n",
            "Epoch 54/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1860 - val_loss: 0.1858\n",
            "Epoch 55/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1859 - val_loss: 0.1861\n",
            "Epoch 56/100\n",
            "235/235 [==============================] - 3s 13ms/step - loss: 0.1859 - val_loss: 0.1858\n",
            "Epoch 57/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1859 - val_loss: 0.1858\n",
            "Epoch 58/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1859 - val_loss: 0.1858\n",
            "Epoch 59/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1858 - val_loss: 0.1858\n",
            "Epoch 60/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1858 - val_loss: 0.1857\n",
            "Epoch 61/100\n",
            "235/235 [==============================] - 3s 13ms/step - loss: 0.1857 - val_loss: 0.1855\n",
            "Epoch 62/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1857 - val_loss: 0.1856\n",
            "Epoch 63/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1857 - val_loss: 0.1854\n",
            "Epoch 64/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1858 - val_loss: 0.1857\n",
            "Epoch 65/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1857 - val_loss: 0.1855\n",
            "Epoch 66/100\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.1856 - val_loss: 0.1855\n",
            "Epoch 67/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1856 - val_loss: 0.1855\n",
            "Epoch 68/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1856 - val_loss: 0.1854\n",
            "Epoch 69/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1856 - val_loss: 0.1855\n",
            "Epoch 70/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1855 - val_loss: 0.1858\n",
            "Epoch 71/100\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.1855 - val_loss: 0.1854\n",
            "Epoch 72/100\n",
            "235/235 [==============================] - 3s 11ms/step - loss: 0.1855 - val_loss: 0.1854\n",
            "Epoch 73/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1855 - val_loss: 0.1857\n",
            "Epoch 74/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1854 - val_loss: 0.1852\n",
            "Epoch 75/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1854 - val_loss: 0.1853\n",
            "Epoch 76/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1854 - val_loss: 0.1856\n",
            "Epoch 77/100\n",
            "235/235 [==============================] - 3s 13ms/step - loss: 0.1853 - val_loss: 0.1852\n",
            "Epoch 78/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1853 - val_loss: 0.1850\n",
            "Epoch 79/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1853 - val_loss: 0.1854\n",
            "Epoch 80/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1853 - val_loss: 0.1853\n",
            "Epoch 81/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1853 - val_loss: 0.1850\n",
            "Epoch 82/100\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.1852 - val_loss: 0.1851\n",
            "Epoch 83/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1852 - val_loss: 0.1853\n",
            "Epoch 84/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1852 - val_loss: 0.1851\n",
            "Epoch 85/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1851 - val_loss: 0.1855\n",
            "Epoch 86/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1851 - val_loss: 0.1852\n",
            "Epoch 87/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1851 - val_loss: 0.1858\n",
            "Epoch 88/100\n",
            "235/235 [==============================] - 3s 12ms/step - loss: 0.1851 - val_loss: 0.1850\n",
            "Epoch 89/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1850 - val_loss: 0.1849\n",
            "Epoch 90/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1849 - val_loss: 0.1848\n",
            "Epoch 91/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1849 - val_loss: 0.1848\n",
            "Epoch 92/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1849 - val_loss: 0.1847\n",
            "Epoch 93/100\n",
            "235/235 [==============================] - 3s 14ms/step - loss: 0.1849 - val_loss: 0.1849\n",
            "Epoch 94/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1848 - val_loss: 0.1848\n",
            "Epoch 95/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1848 - val_loss: 0.1848\n",
            "Epoch 96/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1848 - val_loss: 0.1846\n",
            "Epoch 97/100\n",
            "235/235 [==============================] - 2s 10ms/step - loss: 0.1847 - val_loss: 0.1846\n",
            "Epoch 98/100\n",
            "235/235 [==============================] - 3s 13ms/step - loss: 0.1847 - val_loss: 0.1846\n",
            "Epoch 99/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1847 - val_loss: 0.1845\n",
            "Epoch 100/100\n",
            "235/235 [==============================] - 2s 9ms/step - loss: 0.1847 - val_loss: 0.1845\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.src.callbacks.History at 0x7b7c4df8f3d0>"
            ]
          },
          "metadata": {},
          "execution_count": 48
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "idx = np.random.randint(X_test.shape[1])"
      ],
      "metadata": {
        "id": "eKkXP19oISI4"
      },
      "execution_count": 55,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "original_image = X_test[idx]"
      ],
      "metadata": {
        "id": "fYSXf0MCKKKB"
      },
      "execution_count": 56,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "original_shape = X_test[idx]"
      ],
      "metadata": {
        "id": "dTgB4AR_If-S"
      },
      "execution_count": 57,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "noisy_image = original_shape + 0.5 + np.random.normal(loc=0.0, scale=1.0, size=original_shape.shape)\n",
        "\n",
        "noisy_image = np.clip(noisy_image, 0.0, 1.0)\n",
        "\n",
        "denoised_image = autoencoder.predict(np.expand_dims(noisy_image, axis=0))\n",
        "\n",
        "denoised_image = denoised_image.reshape((28, 28))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0FDG164NIlKt",
        "outputId": "2e3aaa84-5a8c-43b2-c49f-9a0875da7053"
      },
      "execution_count": 58,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1/1 [==============================] - 0s 23ms/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize=(10, 4))\n",
        "plt.subplot(1,3,1)\n",
        "plt.imshow(original_image.reshape(28,28), cmap=\"gray\")\n",
        "plt.title(\"Imagem original\")\n",
        "plt.axis('off')\n",
        "\n",
        "plt.subplot(1,3,2)\n",
        "plt.imshow(noisy_image.reshape(28,28), cmap=\"gray\")\n",
        "plt.title(\"Imagem ruido\")\n",
        "plt.axis('off')\n",
        "\n",
        "plt.subplot(1,3,3)\n",
        "plt.imshow(denoised_image, cmap=\"gray\")\n",
        "plt.title(\"Removido o ruido\")\n",
        "plt.axis('off')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 304
        },
        "id": "9dOVhl-DJUSe",
        "outputId": "c3a931c8-daa4-475c-d9ce-675f9e5f04d7"
      },
      "execution_count": 59,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(-0.5, 27.5, 27.5, -0.5)"
            ]
          },
          "metadata": {},
          "execution_count": 59
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1000x400 with 3 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    }
  ]
}