7-3_Neural_Network_Model.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pqxnslml3tu6"
      },
      "source": [
        "# 신경망 모델 훈련"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fN7zF6Zy3tvE"
      },
      "source": [
        "<table align=\"left\">\n",
        "  <td>\n",
        "    <a target=\"_blank\" href=\"https://colab.research.google.com/github/rickiepark/hg-mldl/blob/master/7-3.ipynb\"><img src=\"https://www.tensorflow.org/images/colab_logo_32px.png\" />구글 코랩에서 실행하기</a>\n",
        "  </td>\n",
        "</table>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XGP-X65EmJBg"
      },
      "source": [
        "## 손실 곡선"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "hZsGl9udlqZk",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5d626818-831f-4811-ff86-d318970a9f08"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-labels-idx1-ubyte.gz\n",
            "32768/29515 [=================================] - 0s 0us/step\n",
            "40960/29515 [=========================================] - 0s 0us/step\n",
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-images-idx3-ubyte.gz\n",
            "26427392/26421880 [==============================] - 0s 0us/step\n",
            "26435584/26421880 [==============================] - 0s 0us/step\n",
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-labels-idx1-ubyte.gz\n",
            "16384/5148 [===============================================================================================] - 0s 0us/step\n",
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-images-idx3-ubyte.gz\n",
            "4423680/4422102 [==============================] - 0s 0us/step\n",
            "4431872/4422102 [==============================] - 0s 0us/step\n"
          ]
        }
      ],
      "source": [
        "from tensorflow import keras\n",
        "from sklearn.model_selection import train_test_split\n",
        "(train_input, train_target), (test_unput, test_target) = keras.datasets.fashion_mnist.load_data()\n",
        "train_scaled = train_input / 255.0\n",
        "train_scaled, val_scaled, train_target, val_target = train_test_split(train_scaled, train_target, test_size=0.2, random_state=42)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "iogH7o0Ll6uL"
      },
      "outputs": [],
      "source": [
        "def model_fn(a_layer=None):\n",
        "  model = keras.Sequential()\n",
        "  model.add(keras.layers.Flatten(input_shape=(28,28)))\n",
        "  model.add(keras.layers.Dense(100,activation='relu'))\n",
        "  if a_layer:\n",
        "    model.add(a_layer)\n",
        "  model.add(keras.layers.Dense(10, activation='softmax'))\n",
        "  return model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5Eh6hM4DNdzu",
        "outputId": "31e1a741-401e-4945-c008-c1ebdf0080f6"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " flatten (Flatten)           (None, 784)               0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 100)               78500     \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 10)                1010      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 79,510\n",
            "Trainable params: 79,510\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "model = model_fn()\n",
        "model.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "P-UK21N_mCM0"
      },
      "outputs": [],
      "source": [
        "model.compile(loss='sparse_categorical_crossentropy', metrics='accuracy')\n",
        "history = model.fit(train_scaled, train_target, epochs=5, verbose=0)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1ix_EA2LpaR0",
        "outputId": "45c2ea4c-77a3-460c-dc0c-47ea9ef9f8cc"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "dict_keys(['loss', 'accuracy'])\n"
          ]
        }
      ],
      "source": [
        "print(history.history.keys())"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "uz_TCdfPmG6e",
        "outputId": "05f03c6c-5b3b-400e-f725-6dcc4a327228"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.plot(history.history['loss'])\n",
        "plt.xlabel('epoch')\n",
        "plt.ylabel('loss')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 283
        },
        "id": "CpmK9lXQcBe9",
        "outputId": "db23db7b-1605-4c45-f812-8d790c71c45c"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plt.plot(history.history['accuracy'])\n",
        "plt.xlabel('epoch')\n",
        "plt.ylabel('accuracy')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "5cJlWITXqJWr"
      },
      "outputs": [],
      "source": [
        "model = model_fn()\n",
        "model.compile(loss='sparse_categorical_crossentropy', metrics='accuracy')\n",
        "\n",
        "history = model.fit(train_scaled, train_target, epochs=20, verbose=0)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        },
        "id": "YT87Fjo2qKPC",
        "outputId": "c25228a4-4b80-48c2-fda5-e7108942477d"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plt.plot(history.history['loss'])\n",
        "plt.xlabel('epoch')\n",
        "plt.ylabel('loss')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bGqf6ceRr3zO"
      },
      "source": [
        "## 검증 손실"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "4_iHvMxwu2D2"
      },
      "outputs": [],
      "source": [
        "model = model_fn()\n",
        "model.compile(loss='sparse_categorical_crossentropy', metrics='accuracy')\n",
        "\n",
        "history = model.fit(train_scaled, train_target, epochs=20, verbose=0,\n",
        "                    validation_data=(val_scaled, val_target))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "nJ5RGEmLu5KI",
        "outputId": "dc2fafbb-7f16-48a8-b678-d15352c613c3"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "dict_keys(['loss', 'accuracy', 'val_loss', 'val_accuracy'])\n"
          ]
        }
      ],
      "source": [
        "print(history.history.keys())"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "zcpm7CpXu5vC",
        "outputId": "84a85864-c7a3-48e6-d416-d1b8ab5b9faa"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.xlabel('epoch')\n",
        "plt.ylabel('loss')\n",
        "plt.legend(['train', 'val'])\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "qC8gDwo3qcJv"
      },
      "outputs": [],
      "source": [
        "model = model_fn()\n",
        "model.compile(optimizer='adam', loss='sparse_categorical_crossentropy',\n",
        "              metrics='accuracy')\n",
        "\n",
        "history = model.fit(train_scaled, train_target, epochs=20, verbose=0,\n",
        "                    validation_data=(val_scaled, val_target))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "k8wWnyFzsLKb",
        "outputId": "4e804e9b-9be1-4d6d-99ee-fe1a03e026ca"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.xlabel('epoch')\n",
        "plt.ylabel('loss')\n",
        "plt.legend(['train', 'val'])\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "F-dFY8lrYXm3"
      },
      "source": [
        "## 드롭아웃"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AppFtFKgsk--",
        "outputId": "23264d3f-1b2a-44ba-cc81-14598007fba5"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_5\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " flatten_5 (Flatten)         (None, 784)               0         \n",
            "                                                                 \n",
            " dense_10 (Dense)            (None, 100)               78500     \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 100)               0         \n",
            "                                                                 \n",
            " dense_11 (Dense)            (None, 10)                1010      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 79,510\n",
            "Trainable params: 79,510\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "model = model_fn(keras.layers.Dropout(0.3))\n",
        "model.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "_TSe7oM9v1lW"
      },
      "outputs": [],
      "source": [
        "model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics='accuracy')\n",
        "\n",
        "history = model.fit(train_scaled, train_target, epochs=20, verbose=0,\n",
        "                    validation_data=(val_scaled, val_target))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "Rj_syB_iv30l",
        "outputId": "3d5605f0-68ce-4efa-e6ae-6ac8fe400d39"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.xlabel('epoch')\n",
        "plt.ylabel('loss')\n",
        "plt.legend(['train', 'val'])\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hQi91PCQorHr"
      },
      "source": [
        "## 모델 저장과 복원"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "npsYdX3rv6Oa"
      },
      "outputs": [],
      "source": [
        "model = model_fn(keras.layers.Dropout(0.3))\n",
        "model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics='accuracy')\n",
        "\n",
        "history = model.fit(train_scaled, train_target, epochs=10, verbose=0,\n",
        "                    validation_data=(val_scaled, val_target))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "hWVYzt0Y2FPm"
      },
      "outputs": [],
      "source": [
        "model.save_weights('model-weights.h5')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "NissHzcq3xbN"
      },
      "outputs": [],
      "source": [
        "model.save('model-whole.h5')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2vljkGGu3AUl",
        "outputId": "81ce3331-40e2-4b04-833b-431a99edbbb7",
        "scrolled": true
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "-rw-r--r-- 1 root root 333448 Jan 23 13:29 model-weights.h5\n",
            "-rw-r--r-- 1 root root 982664 Jan 23 13:29 model-whole.h5\n"
          ]
        }
      ],
      "source": [
        "!ls -al *.h5"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "7-t6gC5Z3GCM"
      },
      "outputs": [],
      "source": [
        "model = model_fn(keras.layers.Dropout(0.3))\n",
        "\n",
        "model.load_weights('model-weights.h5')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pHV9tBnzi8St",
        "outputId": "3a7d04f4-430c-4866-de90-7c90c6d64f23"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0.8731666666666666\n"
          ]
        }
      ],
      "source": [
        "import numpy as np\n",
        "\n",
        "val_labels = np.argmax(model.predict(val_scaled), axis=-1)\n",
        "print(np.mean(val_labels == val_target))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sRM3Vpki4QyH",
        "outputId": "241588f0-7ede-4075-d39e-30ed545c4892"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "375/375 [==============================] - 1s 3ms/step - loss: 0.3492 - accuracy: 0.8732\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.34922581911087036, 0.8731666803359985]"
            ]
          },
          "metadata": {},
          "execution_count": 37
        }
      ],
      "source": [
        "model = keras.models.load_model('model-whole.h5')\n",
        "\n",
        "model.evaluate(val_scaled, val_target)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4NTCF3YD3EyA"
      },
      "source": [
        "## 콜백"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "L2lKN_934VB4",
        "outputId": "864c1e31-3a5c-4b1e-87fc-58420cce7a49"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f464edca390>"
            ]
          },
          "metadata": {},
          "execution_count": 38
        }
      ],
      "source": [
        "model = model_fn(keras.layers.Dropout(0.3))\n",
        "model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics='accuracy')\n",
        "\n",
        "checkpoint_cb = keras.callbacks.ModelCheckpoint('best-model.h5', save_best_only=True)\n",
        "\n",
        "model.fit(train_scaled, train_target, epochs=20, verbose=0,\n",
        "          validation_data=(val_scaled, val_target),\n",
        "          callbacks=[checkpoint_cb])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qYISeH6U5oh9",
        "outputId": "64c9ada4-ba3c-4343-9ed9-ba0edd5b4a7d"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "375/375 [==============================] - 1s 3ms/step - loss: 0.3213 - accuracy: 0.8839\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.3212670087814331, 0.8839166760444641]"
            ]
          },
          "metadata": {},
          "execution_count": 39
        }
      ],
      "source": [
        "model = keras.models.load_model('best-model.h5')\n",
        "\n",
        "model.evaluate(val_scaled, val_target)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "HLLlkR0s5Nd8",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "e1c89ec0-9e18-4cb1-c63a-bd09df3f3e59"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f46513ad990>"
            ]
          },
          "metadata": {},
          "execution_count": 41
        }
      ],
      "source": [
        "# early stopping\n",
        "\n",
        "model = model_fn(keras.layers.Dropout(0.3))\n",
        "model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics='accuracy')\n",
        "\n",
        "checkpoint_cb = keras.callbacks.ModelCheckpoint('best-model.h5', save_best_only=True)\n",
        "early_stopping_cb = keras.callbacks.EarlyStopping(patience=2, restore_best_weights=True)\n",
        "\n",
        "model.fit(train_scaled, train_target, epochs=20, verbose=0,\n",
        "          validation_data=(val_scaled, val_target),\n",
        "          callbacks=[checkpoint_cb])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "b6TazMcDxqXD",
        "outputId": "65da73d4-e418-4fad-b4fc-10fe8657b1f3"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0\n"
          ]
        }
      ],
      "source": [
        "print(early_stopping_cb.stopped_epoch)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 280
        },
        "id": "QrUNYGPB6Kq7",
        "outputId": "fa9dbfac-fd1b-45ba-fb8c-b1e4e771547d"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.xlabel('epoch')\n",
        "plt.ylabel('loss')\n",
        "plt.legend(['train', 'val'])\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "H-7y1qlg5yqO",
        "outputId": "5db4a6e9-464f-409e-f70f-fd3d90a8feb7"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "375/375 [==============================] - 1s 3ms/step - loss: 0.3252 - accuracy: 0.8883\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.32518965005874634, 0.8883333206176758]"
            ]
          },
          "metadata": {},
          "execution_count": 44
        }
      ],
      "source": [
        "model.evaluate(val_scaled, val_target)"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [],
      "name": "7-3_Neural_Network_Model",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3 (ipykernel)",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.3"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}