code/ch15/ch15.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 15장. 심층 합성곱 신경망으로 이미지 분류"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**아래 링크를 통해 이 노트북을 주피터 노트북 뷰어(nbviewer.jupyter.org)로 보거나 구글 코랩(colab.research.google.com)에서 실행할 수 있습니다.**\n",
    "\n",
    "<table class=\"tfo-notebook-buttons\" align=\"left\">\n",
    "  <td>\n",
    "    <a target=\"_blank\" href=\"https://nbviewer.jupyter.org/github/rickiepark/python-machine-learning-book-2nd-edition/blob/master/code/ch15/ch15.ipynb\"><img src=\"https://jupyter.org/assets/main-logo.svg\" width=\"28\" />주피터 노트북 뷰어로 보기</a>\n",
    "  </td>\n",
    "  <td>\n",
    "    <a target=\"_blank\" href=\"https://colab.research.google.com/github/rickiepark/python-machine-learning-book-2nd-edition/blob/master/code/ch15/ch15.ipynb\"><img src=\"https://www.tensorflow.org/images/colab_logo_32px.png\" />구글 코랩(Colab)에서 실행하기</a>\n",
    "  </td>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`watermark`는 주피터 노트북에 사용하는 파이썬 패키지를 출력하기 위한 유틸리티입니다. `watermark` 패키지를 설치하려면 다음 셀의 주석을 제거한 뒤 실행하세요."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#!pip install watermark"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2019-04-26 \n",
      "\n",
      "numpy 1.16.3\n",
      "scipy 1.2.1\n",
      "tensorflow 2.0.0-alpha0\n"
     ]
    }
   ],
   "source": [
    "%load_ext watermark\n",
    "%watermark -d -p numpy,scipy,tensorflow"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**이 노트북을 실행하려면 텐서플로 2.0.0-alpha0 버전 이상이 필요합니다. 이전 버전의 텐서플로가 설치되어 있다면 다음 셀의 주석을 제거한 뒤 실행하세요.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#!pip install tensorflow==2.0.0-alpha0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**코랩을 사용할 때는 다음 셀의 주석을 제거하고 GPU 버전의 텐서플로 2.0.0-alpha0 버전을 설치하세요.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#!pip install tensorflow-gpu==2.0.0-alpha0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 합성곱 신경망의 구성 요소"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 이산 합성곱 수행하기"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 합성곱 출력 크기 계산하기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Conv1d 구현:  [ 5. 14. 16. 26. 24. 34. 19. 22.]\n",
      "넘파이 결과:  [ 5 14 16 26 24 34 19 22]\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "\n",
    "def conv1d(x, w, p=0, s=1):\n",
    "    w_rot = np.array(w[::-1])\n",
    "    x_padded = np.array(x)\n",
    "    if p > 0:\n",
    "        zero_pad = np.zeros(shape=p)\n",
    "        x_padded = np.concatenate([zero_pad, x_padded, zero_pad])\n",
    "    res = []\n",
    "    for i in range(0, int(len(x)/s),s):\n",
    "        res.append(np.sum(x_padded[i:i+w_rot.shape[0]] * w_rot))\n",
    "    return np.array(res)\n",
    "\n",
    "## 테스트:\n",
    "x = [1, 3, 2, 4, 5, 6, 1, 3]\n",
    "w = [1, 0, 3, 1, 2]\n",
    "print('Conv1d 구현: ', \n",
    "      conv1d(x, w, p=2, s=1))\n",
    "print('넘파이 결과: ', \n",
    "      np.convolve(x, w, mode='same'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2D 이산 합성곱 수행하기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Conv2d 구현: \n",
      " [[11. 25. 32. 13.]\n",
      " [19. 25. 24. 13.]\n",
      " [13. 28. 25. 17.]\n",
      " [11. 17. 14.  9.]]\n",
      "싸이파이 결과: \n",
      " [[11 25 32 13]\n",
      " [19 25 24 13]\n",
      " [13 28 25 17]\n",
      " [11 17 14  9]]\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import scipy.signal\n",
    "\n",
    "\n",
    "def conv2d(X, W, p=(0,0), s=(1,1)):\n",
    "    W_rot = np.array(W)[::-1,::-1]\n",
    "    X_orig = np.array(X)\n",
    "    n1 = X_orig.shape[0] + 2*p[0]\n",
    "    n2 = X_orig.shape[1] + 2*p[1]\n",
    "    X_padded = np.zeros(shape=(n1,n2))\n",
    "    X_padded[p[0]:p[0] + X_orig.shape[0], \n",
    "             p[1]:p[1] + X_orig.shape[1]] = X_orig\n",
    "\n",
    "    res = []\n",
    "    for i in range(0, int((X_padded.shape[0] - \n",
    "                           W_rot.shape[0])/s[0])+1, s[0]):\n",
    "        res.append([])\n",
    "        for j in range(0, int((X_padded.shape[1] - \n",
    "                               W_rot.shape[1])/s[1])+1, s[1]):\n",
    "            X_sub = X_padded[i:i+W_rot.shape[0], j:j+W_rot.shape[1]]\n",
    "            res[-1].append(np.sum(X_sub * W_rot))\n",
    "    return(np.array(res))\n",
    "    \n",
    "X = [[1, 3, 2, 4], [5, 6, 1, 3], [1 , 2,0, 2], [3, 4, 3, 2]]\n",
    "W = [[1, 0, 3], [1, 2, 1], [0, 1, 1]]\n",
    "print('Conv2d 구현: \\n', \n",
    "      conv2d(X, W, p=(1,1), s=(1,1)))\n",
    "\n",
    "print('싸이파이 결과: \\n', \n",
    "      scipy.signal.convolve2d(X, W, mode='same'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 기본 구성 요소를 사용하여 CNN 구성하기"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 여러 개의 입력 또는 컬러 채널을 다루기"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**코랩을 사용할 때는 다음 셀의 주석을 제거하고 실행하세요.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#!wget https://github.com/rickiepark/python-machine-learning-book-2nd-edition/raw/master/code/ch15/example-image.png"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "이미지 크기: (252, 221, 3)\n",
      "채널 수: 3\n",
      "이미지 데이터 타입: uint8\n",
      "[[[179 134 110]\n",
      "  [182 136 112]]\n",
      "\n",
      " [[180 135 111]\n",
      "  [182 137 113]]]\n"
     ]
    }
   ],
   "source": [
    "import imageio\n",
    "\n",
    "img = imageio.imread('./example-image.png', pilmode='RGB')\n",
    "    \n",
    "print('이미지 크기:', img.shape)\n",
    "print('채널 수:', img.shape[2])\n",
    "print('이미지 데이터 타입:', img.dtype)\n",
    "\n",
    "print(img[100:102, 100:102, :])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 텐서플로를 사용해 심층 합성곱 신경망 구현하기"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 데이터 적재와 전처리"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**코랩을 사용할 때는 다음 셀의 주석을 제거하고 실행하세요.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#!wget https://github.com/rickiepark/python-machine-learning-book-2nd-edition/raw/master/code/ch12/t10k-images-idx3-ubyte.gz\n",
    "#!wget https://github.com/rickiepark/python-machine-learning-book-2nd-edition/raw/master/code/ch12/t10k-labels-idx1-ubyte.gz\n",
    "#!wget https://github.com/rickiepark/python-machine-learning-book-2nd-edition/raw/master/code/ch12/train-images-idx3-ubyte.gz\n",
    "#!wget https://github.com/rickiepark/python-machine-learning-book-2nd-edition/raw/master/code/ch12/train-labels-idx1-ubyte.gz"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "## MNIST 데이터셋 압축을 풉니다.\n",
    "\n",
    "import sys\n",
    "import gzip\n",
    "import shutil\n",
    "import os\n",
    "\n",
    "\n",
    "if (sys.version_info > (3, 0)):\n",
    "    writemode = 'wb'\n",
    "else:\n",
    "    writemode = 'w'\n",
    "\n",
    "zipped_mnist = [f for f in os.listdir('./')\n",
    "                if f.endswith('ubyte.gz')]\n",
    "for z in zipped_mnist:\n",
    "    with gzip.GzipFile(z, mode='rb') as decompressed, open(z[:-3], writemode) as outfile:\n",
    "        outfile.write(decompressed.read())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "행: 60000,  열: 784\n",
      "행: 10000,  열: 784\n",
      "훈련 세트:  (50000, 784) (50000,)\n",
      "검증 세트:  (10000, 784) (10000,)\n",
      "테스트 세트:  (10000, 784) (10000,)\n"
     ]
    }
   ],
   "source": [
    "import struct\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "def load_mnist(path, kind='train'):\n",
    "    \"\"\"`path`에서 MNIST 데이터셋을 적재합니다.\"\"\"\n",
    "    labels_path = os.path.join(path,\n",
    "                               '%s-labels-idx1-ubyte'\n",
    "                                % kind)\n",
    "    images_path = os.path.join(path,\n",
    "                               '%s-images-idx3-ubyte'\n",
    "                               % kind)\n",
    "\n",
    "    with open(labels_path, 'rb') as lbpath:\n",
    "        magic, n = struct.unpack('>II',\n",
    "                                 lbpath.read(8))\n",
    "        labels = np.fromfile(lbpath,\n",
    "                             dtype=np.uint8)\n",
    "\n",
    "    with open(images_path, 'rb') as imgpath:\n",
    "        magic, num, rows, cols = struct.unpack(\">IIII\",\n",
    "                                               imgpath.read(16))\n",
    "        images = np.fromfile(imgpath,\n",
    "                             dtype=np.uint8).reshape(len(labels), 784)\n",
    "\n",
    "    return images, labels\n",
    "\n",
    "\n",
    "X_data, y_data = load_mnist('./', kind='train')\n",
    "print('행: %d,  열: %d' % (X_data.shape[0], X_data.shape[1]))\n",
    "X_test, y_test = load_mnist('./', kind='t10k')\n",
    "print('행: %d,  열: %d' % (X_test.shape[0], X_test.shape[1]))\n",
    "\n",
    "X_train, y_train = X_data[:50000,:], y_data[:50000]\n",
    "X_valid, y_valid = X_data[50000:,:], y_data[50000:]\n",
    "\n",
    "print('훈련 세트: ', X_train.shape, y_train.shape)\n",
    "print('검증 세트: ', X_valid.shape, y_valid.shape)\n",
    "print('테스트 세트: ', X_test.shape, y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "mean_vals = np.mean(X_train, axis=0)\n",
    "std_val = np.std(X_train)\n",
    "\n",
    "X_train_centered = (X_train - mean_vals)/std_val\n",
    "X_valid_centered = (X_valid - mean_vals)/std_val\n",
    "X_test_centered = (X_test - mean_vals)/std_val"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train_centered = X_train_centered.reshape((-1, 28, 28, 1))\n",
    "X_valid_centered = X_valid_centered.reshape((-1, 28, 28, 1))\n",
    "X_test_centered = X_test_centered.reshape((-1, 28, 28, 1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras.utils import to_categorical\n",
    "\n",
    "y_train_onehot = to_categorical(y_train)\n",
    "y_valid_onehot = to_categorical(y_valid)\n",
    "y_test_onehot = to_categorical(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5 [0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n"
     ]
    }
   ],
   "source": [
    "print(y_train[0], y_train_onehot[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "훈련 세트:  (50000, 28, 28, 1) (50000, 10)\n",
      "검증 세트:  (10000, 28, 28, 1) (10000, 10)\n",
      "테스트 세트:  (10000, 28, 28, 1) (10000, 10)\n"
     ]
    }
   ],
   "source": [
    "print('훈련 세트: ', X_train_centered.shape, y_train_onehot.shape)\n",
    "print('검증 세트: ', X_valid_centered.shape, y_valid_onehot.shape)\n",
    "print('테스트 세트: ', X_test_centered.shape, y_test_onehot.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 텐서플로 Keras API로 CNN 구성하기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras import layers, models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = models.Sequential()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.add(layers.Conv2D(32, (5, 5), padding='valid', \n",
    "                        activation='relu', input_shape=(28, 28, 1)))\n",
    "model.add(layers.MaxPool2D((2, 2)))\n",
    "model.add(layers.Conv2D(64, (5, 5), padding='valid', \n",
    "                        activation='relu'))\n",
    "model.add(layers.MaxPool2D((2, 2)))\n",
    "model.add(layers.Flatten())\n",
    "model.add(layers.Dense(1024, activation='relu'))\n",
    "model.add(layers.Dropout(0.5))\n",
    "model.add(layers.Dense(10, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 24, 24, 32)        832       \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 12, 12, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 8, 8, 64)          51264     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 4, 4, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 1024)              1049600   \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 10)                10250     \n",
      "=================================================================\n",
      "Total params: 1,111,946\n",
      "Trainable params: 1,111,946\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 합성곱 신경망 모델 훈련하기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss='categorical_crossentropy', \n",
    "              optimizer='adam', metrics=['acc'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "import time\n",
    "from tensorflow.keras.callbacks import ModelCheckpoint, TensorBoard\n",
    "\n",
    "callback_list = [ModelCheckpoint(filepath='cnn_checkpoint.h5', \n",
    "                                 monitor='val_loss', \n",
    "                                 save_best_only=True), \n",
    "                 TensorBoard(log_dir=\"logs/{}\".format(time.asctime()))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 50000 samples, validate on 10000 samples\n",
      "Epoch 1/20\n",
      "50000/50000 [==============================] - 4s 87us/sample - loss: 0.1377 - acc: 0.9564 - val_loss: 0.0577 - val_acc: 0.9838\n",
      "Epoch 2/20\n",
      "50000/50000 [==============================] - 3s 69us/sample - loss: 0.0481 - acc: 0.9852 - val_loss: 0.0508 - val_acc: 0.9852\n",
      "Epoch 3/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0363 - acc: 0.9891 - val_loss: 0.0431 - val_acc: 0.9873\n",
      "Epoch 4/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0273 - acc: 0.9915 - val_loss: 0.0329 - val_acc: 0.9922\n",
      "Epoch 5/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0255 - acc: 0.9918 - val_loss: 0.0371 - val_acc: 0.9907\n",
      "Epoch 6/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0195 - acc: 0.9937 - val_loss: 0.0370 - val_acc: 0.9924\n",
      "Epoch 7/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0184 - acc: 0.9941 - val_loss: 0.0393 - val_acc: 0.9911\n",
      "Epoch 8/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0183 - acc: 0.9945 - val_loss: 0.0322 - val_acc: 0.9929\n",
      "Epoch 9/20\n",
      "50000/50000 [==============================] - 4s 71us/sample - loss: 0.0147 - acc: 0.9954 - val_loss: 0.0427 - val_acc: 0.9909\n",
      "Epoch 10/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0138 - acc: 0.9956 - val_loss: 0.0430 - val_acc: 0.9911\n",
      "Epoch 11/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0145 - acc: 0.9953 - val_loss: 0.0421 - val_acc: 0.9901\n",
      "Epoch 12/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0100 - acc: 0.9968 - val_loss: 0.0502 - val_acc: 0.9908\n",
      "Epoch 13/20\n",
      "50000/50000 [==============================] - 3s 68us/sample - loss: 0.0130 - acc: 0.9962 - val_loss: 0.0553 - val_acc: 0.9899\n",
      "Epoch 14/20\n",
      "50000/50000 [==============================] - 3s 68us/sample - loss: 0.0110 - acc: 0.9967 - val_loss: 0.0506 - val_acc: 0.9910\n",
      "Epoch 15/20\n",
      "50000/50000 [==============================] - 3s 69us/sample - loss: 0.0113 - acc: 0.9969 - val_loss: 0.0569 - val_acc: 0.9908\n",
      "Epoch 16/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0111 - acc: 0.9965 - val_loss: 0.0686 - val_acc: 0.9903\n",
      "Epoch 17/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0124 - acc: 0.9967 - val_loss: 0.0794 - val_acc: 0.9893\n",
      "Epoch 18/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0083 - acc: 0.9977 - val_loss: 0.0573 - val_acc: 0.9909\n",
      "Epoch 19/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0098 - acc: 0.9974 - val_loss: 0.0779 - val_acc: 0.9900\n",
      "Epoch 20/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0114 - acc: 0.9970 - val_loss: 0.0795 - val_acc: 0.9902\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(X_train_centered, y_train_onehot, \n",
    "                    batch_size=64, epochs=20, \n",
    "                    validation_data=(X_valid_centered, y_valid_onehot),\n",
    "                    callbacks=callback_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "epochs = np.arange(1, 21)\n",
    "plt.plot(epochs, history.history['loss'])\n",
    "plt.plot(epochs, history.history['val_loss'])\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('loss')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(epochs, history.history['acc'])\n",
    "plt.plot(epochs, history.history['val_acc'])\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('accuracy')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "합계 16\r\n",
      "drwxr-xr-x 4 haesun haesun 4096  3월 15 11:24  .\r\n",
      "drwxrwxr-x 7 haesun haesun 4096  3월 15 11:24  ..\r\n",
      "drwxr-xr-x 2 haesun haesun 4096  3월 15 11:23  .ipynb_checkpoints\r\n",
      "drwxr-xr-x 5 haesun haesun 4096  3월 15 11:24 'Fri Mar 15 11:24:15 2019'\r\n"
     ]
    }
   ],
   "source": [
    "!ls -al logs/"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "모델 저장"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save('cnn_model.h5')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "모델 복원"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras.models import load_model\n",
    "\n",
    "restored_model = load_model('cnn_model.h5')\n",
    "restored_model.load_weights('cnn_checkpoint.h5')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "테스트 정확도 확인"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 0s 41us/sample - loss: 0.0279 - acc: 0.9923\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.02787502632992919, 0.9923]"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "restored_model.evaluate(X_test_centered, y_test_onehot)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "테스트 샘플 예측 확인 + 이미지 출력"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([7, 2, 1, 0, 4, 1, 4, 9, 5, 9])"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.argmax(restored_model.predict(X_test_centered[:10]), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([7, 2, 1, 0, 4, 1, 4, 9, 5, 9], dtype=uint8)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 10 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(10, 5))\n",
    "for i in range(10):\n",
    "    fig.add_subplot(2, 5, i+1)\n",
    "    plt.imshow(X_test_centered[i].reshape(28, 28))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 활성화 출력과 필터 시각화하기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<tensorflow.python.keras.layers.convolutional.Conv2D object at 0x7fed375caf60>\n"
     ]
    }
   ],
   "source": [
    "first_layer = model.layers[0]\n",
    "print(first_layer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor(\"conv2d_input:0\", shape=(None, 28, 28, 1), dtype=float32)\n"
     ]
    }
   ],
   "source": [
    "print(model.input)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "first_activation = models.Model(inputs=model.input, \n",
    "                                outputs=first_layer.output)\n",
    "activation = first_activation.predict(X_test_centered[:10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10, 24, 24, 32)\n"
     ]
    }
   ],
   "source": [
    "print(activation.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x1080 with 32 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(10, 15))\n",
    "for i in range(32):\n",
    "    fig.add_subplot(7, 5, i+1)\n",
    "    plt.imshow(activation[0, :, :, i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x1080 with 32 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(10, 15))\n",
    "for i in range(32):\n",
    "    fig.add_subplot(7, 5, i+1)\n",
    "    plt.imshow(activation[3, :, :, i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x1080 with 32 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(10, 15))\n",
    "for i in range(32):\n",
    "    fig.add_subplot(7, 5, i+1)\n",
    "    plt.imshow(first_layer.kernel[:, :, 0, i])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "20번 더 훈련"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 50000 samples, validate on 10000 samples\n",
      "Epoch 1/20\n",
      "50000/50000 [==============================] - 4s 80us/sample - loss: 0.0139 - acc: 0.9957 - val_loss: 0.0430 - val_acc: 0.9909\n",
      "Epoch 2/20\n",
      "50000/50000 [==============================] - 4s 78us/sample - loss: 0.0129 - acc: 0.9961 - val_loss: 0.0424 - val_acc: 0.9907\n",
      "Epoch 3/20\n",
      "50000/50000 [==============================] - 4s 78us/sample - loss: 0.0148 - acc: 0.9953 - val_loss: 0.0419 - val_acc: 0.9911\n",
      "Epoch 4/20\n",
      "50000/50000 [==============================] - 4s 72us/sample - loss: 0.0129 - acc: 0.9964 - val_loss: 0.0429 - val_acc: 0.9911\n",
      "Epoch 5/20\n",
      "50000/50000 [==============================] - 4s 71us/sample - loss: 0.0105 - acc: 0.9969 - val_loss: 0.0517 - val_acc: 0.9904\n",
      "Epoch 6/20\n",
      "50000/50000 [==============================] - 3s 70us/sample - loss: 0.0130 - acc: 0.9960 - val_loss: 0.0727 - val_acc: 0.9880\n",
      "Epoch 7/20\n",
      "50000/50000 [==============================] - 4s 70us/sample - loss: 0.0099 - acc: 0.9968 - val_loss: 0.0636 - val_acc: 0.9906\n",
      "Epoch 8/20\n",
      "50000/50000 [==============================] - 4s 70us/sample - loss: 0.0139 - acc: 0.9961 - val_loss: 0.0592 - val_acc: 0.9904\n",
      "Epoch 9/20\n",
      "50000/50000 [==============================] - 4s 70us/sample - loss: 0.0100 - acc: 0.9973 - val_loss: 0.0651 - val_acc: 0.9906\n",
      "Epoch 10/20\n",
      "50000/50000 [==============================] - 4s 70us/sample - loss: 0.0117 - acc: 0.9970 - val_loss: 0.0652 - val_acc: 0.9913\n",
      "Epoch 11/20\n",
      "50000/50000 [==============================] - 4s 70us/sample - loss: 0.0103 - acc: 0.9973 - val_loss: 0.0710 - val_acc: 0.9913\n",
      "Epoch 12/20\n",
      "50000/50000 [==============================] - 4s 71us/sample - loss: 0.0081 - acc: 0.9979 - val_loss: 0.0725 - val_acc: 0.9908\n",
      "Epoch 13/20\n",
      "50000/50000 [==============================] - 4s 70us/sample - loss: 0.0122 - acc: 0.9969 - val_loss: 0.0974 - val_acc: 0.9891\n",
      "Epoch 14/20\n",
      "50000/50000 [==============================] - 4s 70us/sample - loss: 0.0116 - acc: 0.9974 - val_loss: 0.0730 - val_acc: 0.9925\n",
      "Epoch 15/20\n",
      "50000/50000 [==============================] - 4s 70us/sample - loss: 0.0090 - acc: 0.9979 - val_loss: 0.0719 - val_acc: 0.9911\n",
      "Epoch 16/20\n",
      "50000/50000 [==============================] - 4s 70us/sample - loss: 0.0118 - acc: 0.9974 - val_loss: 0.0802 - val_acc: 0.9894\n",
      "Epoch 17/20\n",
      "50000/50000 [==============================] - 4s 70us/sample - loss: 0.0094 - acc: 0.9978 - val_loss: 0.0798 - val_acc: 0.9899\n",
      "Epoch 18/20\n",
      "50000/50000 [==============================] - 4s 71us/sample - loss: 0.0077 - acc: 0.9980 - val_loss: 0.0967 - val_acc: 0.9911\n",
      "Epoch 19/20\n",
      "50000/50000 [==============================] - 4s 71us/sample - loss: 0.0116 - acc: 0.9973 - val_loss: 0.0901 - val_acc: 0.9912\n",
      "Epoch 20/20\n",
      "50000/50000 [==============================] - 4s 70us/sample - loss: 0.0098 - acc: 0.9979 - val_loss: 0.0972 - val_acc: 0.9914\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7fecac248eb8>"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "restored_model.fit(X_train_centered, y_train_onehot, batch_size=64, epochs=20, \n",
    "                   validation_data=(X_valid_centered, y_valid_onehot))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": 2
}