Tarea3_DSepulveda_MNIST.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "import matplotlib.pyplot as plt\n",
        "import cv2\n",
        "import numpy as np\n",
        "#------------Descarga de dataset------------#\n",
        "(x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data()\n",
        "\n"
      ],
      "metadata": {
        "id": "_rGbXRCV8LM0"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "x_train = x_train/255.0\n",
        "x_test = x_test/255.0"
      ],
      "metadata": {
        "id": "U8_7uWH8853I"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#------------Modelo autoencoder------------#\n",
        "\n",
        "encoder_input = keras.Input(shape=(28,28,1), name = \"ae_input\")\n",
        "flat = keras.layers.Flatten()(encoder_input) #Aplana los valores\n",
        "h_layer1 = keras.layers.Dense(588, activation=\"relu\")(flat)\n",
        "h_layer2 = keras.layers.Dense(392, activation=\"relu\")(h_layer1)\n",
        "h_layer3 = keras.layers.Dense(196, activation=\"relu\")(h_layer2)\n",
        "h_layer4 = keras.layers.Dense(5, activation=\"relu\")(h_layer3)\n",
        "\n",
        "encoder = keras.Model(encoder_input, h_layer4, name=\"encoder\") #Modelo encoder para poder visualizar imagen de 5 dimensiones\n",
        "\n",
        "h_layer5 = keras.layers.Dense(196, activation=\"relu\")(h_layer4)\n",
        "h_layer6 = keras.layers.Dense(392, activation=\"relu\")(h_layer5)\n",
        "h_layer7 = keras.layers.Dense(588, activation=\"relu\")(h_layer6)\n",
        "\n",
        "h_layer8 = keras.layers.Dense(784, activation=\"relu\")(h_layer7)\n",
        "\n",
        "#En este paso de vuelve a tener un vector de valores de 28x28x1\n",
        "\n",
        "decoder_output = keras.layers.Reshape((28,28,1))(h_layer8)\n",
        "\n",
        "opt = keras.optimizers.Adam(learning_rate = 0.001, decay = 1e-6)\n",
        "\n",
        "autoencoder = keras.Model(encoder_input, decoder_output, name = \"autoencoder\") #Creación modelo autoencoder\n",
        "autoencoder.summary()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "U4PwxJ3i92wV",
        "outputId": "1f1cea64-70bb-460e-fbf5-04dce17ef733"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"autoencoder\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " img (InputLayer)            [(None, 28, 28, 1)]       0         \n",
            "                                                                 \n",
            " flatten_1 (Flatten)         (None, 784)               0         \n",
            "                                                                 \n",
            " dense_8 (Dense)             (None, 588)               461580    \n",
            "                                                                 \n",
            " dense_9 (Dense)             (None, 392)               230888    \n",
            "                                                                 \n",
            " dense_10 (Dense)            (None, 196)               77028     \n",
            "                                                                 \n",
            " dense_11 (Dense)            (None, 5)                 985       \n",
            "                                                                 \n",
            " dense_12 (Dense)            (None, 196)               1176      \n",
            "                                                                 \n",
            " dense_13 (Dense)            (None, 392)               77224     \n",
            "                                                                 \n",
            " dense_14 (Dense)            (None, 588)               231084    \n",
            "                                                                 \n",
            " dense_15 (Dense)            (None, 784)               461776    \n",
            "                                                                 \n",
            " reshape_1 (Reshape)         (None, 28, 28, 1)         0         \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 1,541,741\n",
            "Trainable params: 1,541,741\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "autoencoder.compile(opt, loss=\"mse\")"
      ],
      "metadata": {
        "id": "1GQndJZ8AS6w"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "autoencoder.fit(x_train, x_train, epochs=50, batch_size=32, validation_split=0.1)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FaAoAmbE_9K2",
        "outputId": "b40bd6af-b999-4347-a4dc-52d0e235aefb"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/50\n",
            "1688/1688 [==============================] - 44s 25ms/step - loss: 0.0354 - val_loss: 0.0287\n",
            "Epoch 2/50\n",
            "1688/1688 [==============================] - 42s 25ms/step - loss: 0.0278 - val_loss: 0.0266\n",
            "Epoch 3/50\n",
            "1688/1688 [==============================] - 43s 25ms/step - loss: 0.0256 - val_loss: 0.0247\n",
            "Epoch 4/50\n",
            "1688/1688 [==============================] - 43s 26ms/step - loss: 0.0242 - val_loss: 0.0237\n",
            "Epoch 5/50\n",
            "1688/1688 [==============================] - 41s 24ms/step - loss: 0.0234 - val_loss: 0.0231\n",
            "Epoch 6/50\n",
            "1688/1688 [==============================] - 42s 25ms/step - loss: 0.0228 - val_loss: 0.0229\n",
            "Epoch 7/50\n",
            "1688/1688 [==============================] - 44s 26ms/step - loss: 0.0224 - val_loss: 0.0226\n",
            "Epoch 8/50\n",
            "1688/1688 [==============================] - 45s 27ms/step - loss: 0.0220 - val_loss: 0.0225\n",
            "Epoch 9/50\n",
            "1688/1688 [==============================] - 45s 26ms/step - loss: 0.0217 - val_loss: 0.0223\n",
            "Epoch 10/50\n",
            "1688/1688 [==============================] - 44s 26ms/step - loss: 0.0214 - val_loss: 0.0218\n",
            "Epoch 11/50\n",
            "1688/1688 [==============================] - 45s 26ms/step - loss: 0.0212 - val_loss: 0.0217\n",
            "Epoch 12/50\n",
            "1688/1688 [==============================] - 45s 26ms/step - loss: 0.0209 - val_loss: 0.0219\n",
            "Epoch 13/50\n",
            "1688/1688 [==============================] - 44s 26ms/step - loss: 0.0208 - val_loss: 0.0215\n",
            "Epoch 14/50\n",
            "1688/1688 [==============================] - 43s 26ms/step - loss: 0.0206 - val_loss: 0.0219\n",
            "Epoch 15/50\n",
            "1688/1688 [==============================] - 47s 28ms/step - loss: 0.0205 - val_loss: 0.0213\n",
            "Epoch 16/50\n",
            "1688/1688 [==============================] - 46s 27ms/step - loss: 0.0203 - val_loss: 0.0215\n",
            "Epoch 17/50\n",
            "1688/1688 [==============================] - 45s 27ms/step - loss: 0.0201 - val_loss: 0.0213\n",
            "Epoch 18/50\n",
            "1688/1688 [==============================] - 45s 27ms/step - loss: 0.0200 - val_loss: 0.0209\n",
            "Epoch 19/50\n",
            "1688/1688 [==============================] - 44s 26ms/step - loss: 0.0199 - val_loss: 0.0212\n",
            "Epoch 20/50\n",
            "1688/1688 [==============================] - 46s 27ms/step - loss: 0.0198 - val_loss: 0.0210\n",
            "Epoch 21/50\n",
            "1688/1688 [==============================] - 44s 26ms/step - loss: 0.0197 - val_loss: 0.0210\n",
            "Epoch 22/50\n",
            "1688/1688 [==============================] - 45s 27ms/step - loss: 0.0197 - val_loss: 0.0211\n",
            "Epoch 23/50\n",
            "1688/1688 [==============================] - 44s 26ms/step - loss: 0.0196 - val_loss: 0.0210\n",
            "Epoch 24/50\n",
            "1688/1688 [==============================] - 43s 25ms/step - loss: 0.0195 - val_loss: 0.0212\n",
            "Epoch 25/50\n",
            "1688/1688 [==============================] - 45s 27ms/step - loss: 0.0194 - val_loss: 0.0208\n",
            "Epoch 26/50\n",
            "1688/1688 [==============================] - 45s 27ms/step - loss: 0.0193 - val_loss: 0.0211\n",
            "Epoch 27/50\n",
            "1688/1688 [==============================] - 42s 25ms/step - loss: 0.0193 - val_loss: 0.0210\n",
            "Epoch 28/50\n",
            "1688/1688 [==============================] - 41s 24ms/step - loss: 0.0192 - val_loss: 0.0209\n",
            "Epoch 29/50\n",
            "1688/1688 [==============================] - 44s 26ms/step - loss: 0.0191 - val_loss: 0.0209\n",
            "Epoch 30/50\n",
            "1688/1688 [==============================] - 46s 27ms/step - loss: 0.0191 - val_loss: 0.0209\n",
            "Epoch 31/50\n",
            "1688/1688 [==============================] - 45s 27ms/step - loss: 0.0190 - val_loss: 0.0208\n",
            "Epoch 32/50\n",
            "1688/1688 [==============================] - 44s 26ms/step - loss: 0.0190 - val_loss: 0.0210\n",
            "Epoch 33/50\n",
            "1688/1688 [==============================] - 44s 26ms/step - loss: 0.0189 - val_loss: 0.0210\n",
            "Epoch 34/50\n",
            "1688/1688 [==============================] - 45s 27ms/step - loss: 0.0189 - val_loss: 0.0209\n",
            "Epoch 35/50\n",
            "1688/1688 [==============================] - 45s 27ms/step - loss: 0.0188 - val_loss: 0.0208\n",
            "Epoch 36/50\n",
            "1688/1688 [==============================] - 43s 26ms/step - loss: 0.0188 - val_loss: 0.0208\n",
            "Epoch 37/50\n",
            "1688/1688 [==============================] - 43s 26ms/step - loss: 0.0187 - val_loss: 0.0207\n",
            "Epoch 38/50\n",
            "1688/1688 [==============================] - 42s 25ms/step - loss: 0.0187 - val_loss: 0.0210\n",
            "Epoch 39/50\n",
            "1688/1688 [==============================] - 45s 26ms/step - loss: 0.0187 - val_loss: 0.0210\n",
            "Epoch 40/50\n",
            "1688/1688 [==============================] - 44s 26ms/step - loss: 0.0186 - val_loss: 0.0209\n",
            "Epoch 41/50\n",
            "1688/1688 [==============================] - 43s 25ms/step - loss: 0.0186 - val_loss: 0.0208\n",
            "Epoch 42/50\n",
            "1688/1688 [==============================] - 43s 26ms/step - loss: 0.0185 - val_loss: 0.0207\n",
            "Epoch 43/50\n",
            "1688/1688 [==============================] - 43s 25ms/step - loss: 0.0185 - val_loss: 0.0209\n",
            "Epoch 44/50\n",
            "1688/1688 [==============================] - 45s 26ms/step - loss: 0.0185 - val_loss: 0.0209\n",
            "Epoch 45/50\n",
            "1688/1688 [==============================] - 45s 27ms/step - loss: 0.0184 - val_loss: 0.0210\n",
            "Epoch 46/50\n",
            "1688/1688 [==============================] - 43s 25ms/step - loss: 0.0184 - val_loss: 0.0207\n",
            "Epoch 47/50\n",
            "1688/1688 [==============================] - 43s 25ms/step - loss: 0.0183 - val_loss: 0.0211\n",
            "Epoch 48/50\n",
            "1688/1688 [==============================] - 42s 25ms/step - loss: 0.0183 - val_loss: 0.0208\n",
            "Epoch 49/50\n",
            "1688/1688 [==============================] - 46s 27ms/step - loss: 0.0183 - val_loss: 0.0207\n",
            "Epoch 50/50\n",
            "1688/1688 [==============================] - 46s 27ms/step - loss: 0.0182 - val_loss: 0.0210\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f105fafac50>"
            ]
          },
          "metadata": {},
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "predict_autoencoder = encoder.predict([x_test[0].reshape(-1,28,28,1)])[0]\n",
        "print(predict_autoencoder)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ujdkA0KrAe7l",
        "outputId": "4a7d4ace-4341-4ab7-a394-f8e5cc70505c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1/1 [==============================] - 0s 21ms/step\n",
            "[ 646.6637   388.8617    89.8376  1005.5409   125.09917]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.title(\"Componente principal de un parámetro\")\n",
        "plt.imshow(predict_autoencoder.reshape((5,1)), cmap=\"gray\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 298
        },
        "id": "NiY2hoQEA41W",
        "outputId": "6397ede2-5448-4580-a2ff-f9d7cf525ad9"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f105cf7a910>"
            ]
          },
          "metadata": {},
          "execution_count": 227
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "ae_out = autoencoder.predict([x_test[100].reshape(-1,28,28,1)])[0]\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "nJAAGgbPBYy_",
        "outputId": "3365d2b9-b3db-49dd-c61a-8853fc8387cd"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "1/1 [==============================] - 0s 19ms/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.subplot(1,2,1)\n",
        "plt.title(\"Imagen original\")\n",
        "plt.imshow(x_test[100],cmap=\"gray\")\n",
        "plt.subplot(1,2,2)\n",
        "plt.title(\"Reconstrucción autoencoder\")\n",
        "plt.imshow(ae_out.reshape(28,28),cmap=\"gray\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 233
        },
        "id": "I_PcjksEBlhY",
        "outputId": "ec8dff04-0124-4abe-979c-750b8b62719b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f10476a5a50>"
            ]
          },
          "metadata": {},
          "execution_count": 228
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#----------------Creación PCA-------------------#\n",
        "\n",
        "from sklearn.decomposition import PCA #Descargamos PCA de lib sklearn\n",
        "pca = PCA(n_components = 5) #Cantidad de componentes principales a la cual se disminuye imagen\n",
        "\n",
        "x_train_flat = x_train.reshape((60000, 28*28))\n",
        "x_test_flat = x_test.reshape((10000, 28*28))\n",
        "PC_train = pca.fit_transform(x_train_flat)\n",
        "PC_test = pca.fit_transform(x_test_flat)\n",
        "\n",
        "\n",
        "\n",
        "plt.title(\"Componente principal de un parámetro\")\n",
        "plt.imshow(PC_train[0].reshape(5,1), cmap=\"gray\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 298
        },
        "id": "-54hZ1-0L883",
        "outputId": "95185f79-c8d6-4e3b-a775-81288ce48e2f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f104be257d0>"
            ]
          },
          "metadata": {},
          "execution_count": 215
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#--------------Modelo para reconstruir PCA---------------------#\n",
        "\n",
        "decoder_input = keras.Input(shape=(5,), name = \"principal_component\")\n",
        "h_layer1 = keras.layers.Dense(196, activation=\"relu\")(decoder_input)\n",
        "h_layer2 = keras.layers.Dense(392, activation=\"relu\")(h_layer1)\n",
        "h_layer3 = keras.layers.Dense(588, activation=\"relu\")(h_layer2)\n",
        "h_layer4 = keras.layers.Dense(784, activation=\"relu\")(h_layer3)\n",
        "decoder_output = keras.layers.Reshape((28,28,1))(h_layer4)\n",
        "\n",
        "opt = keras.optimizers.Adam(learning_rate = 0.001, decay = 1e-6)\n",
        "\n",
        "decoder = keras.Model(decoder_input, decoder_output, name = \"decoder\")\n",
        "\n",
        "decoder.compile(opt, loss=\"mse\")\n",
        "decoder.summary()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YNOjvZu0V33x",
        "outputId": "c61fcead-d131-4af9-90a3-f761343afcb5"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"decoder\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " principal_component (InputL  [(None, 5)]              0         \n",
            " ayer)                                                           \n",
            "                                                                 \n",
            " dense_104 (Dense)           (None, 196)               1176      \n",
            "                                                                 \n",
            " dense_105 (Dense)           (None, 392)               77224     \n",
            "                                                                 \n",
            " dense_106 (Dense)           (None, 588)               231084    \n",
            "                                                                 \n",
            " dense_107 (Dense)           (None, 784)               461776    \n",
            "                                                                 \n",
            " reshape_9 (Reshape)         (None, 28, 28, 1)         0         \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 771,260\n",
            "Trainable params: 771,260\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "decoder.fit(PC_train, x_train, epochs=5, batch_size=32, validation_split=0.1)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KQ-tSpR0Xn9O",
        "outputId": "c510141f-6190-4bb1-8b15-34f5083dc4f1"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/5\n",
            "1688/1688 [==============================] - 26s 15ms/step - loss: 0.0379 - val_loss: 0.0345\n",
            "Epoch 2/5\n",
            "1688/1688 [==============================] - 25s 15ms/step - loss: 0.0345 - val_loss: 0.0335\n",
            "Epoch 3/5\n",
            "1688/1688 [==============================] - 24s 14ms/step - loss: 0.0337 - val_loss: 0.0329\n",
            "Epoch 4/5\n",
            "1688/1688 [==============================] - 24s 14ms/step - loss: 0.0333 - val_loss: 0.0328\n",
            "Epoch 5/5\n",
            "1688/1688 [==============================] - 27s 16ms/step - loss: 0.0330 - val_loss: 0.0326\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f1047ba30d0>"
            ]
          },
          "metadata": {},
          "execution_count": 190
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "predict = decoder.predict(PC_test)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "eScZAlGCZR9o",
        "outputId": "a6ec731e-b87a-40f8-a5b6-ea94866eec4a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 [==============================] - 1s 5ms/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.subplot(1,2,1)\n",
        "plt.title(\"Imagen original\")\n",
        "plt.imshow(x_test[3], cmap=\"gray\")\n",
        "plt.subplot(1,2,2)\n",
        "plt.title(\"Reconstrucción PCA\")\n",
        "plt.imshow(predict[3].reshape(28,28),cmap=\"gray\")\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 233
        },
        "id": "lPMQNvl8hP5S",
        "outputId": "8024da8f-2ea1-4a6c-9081-75fd1be264e4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f105c9dafd0>"
            ]
          },
          "metadata": {},
          "execution_count": 235
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Para ambas implementaciones se encuentra que el autoencoder predice las imágenes del dataset de manera efectiva. Además se tiene que la implementación de PCA no es muy eficiente, pudiendo reconstruir solo algunas de las imágenes.\n"
      ],
      "metadata": {
        "id": "SrEDts1BsMbd"
      }
    }
  ]
}