5.3-using-a-pretrained-convnet.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'2.2.2'"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import keras\n",
    "keras.__version__"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 사전 훈련된 컨브넷 사용하기\n",
    "\n",
    "이 노트북은 [케라스 창시자에게 배우는 딥러닝](https://tensorflow.blog/%EC%BC%80%EB%9D%BC%EC%8A%A4-%EB%94%A5%EB%9F%AC%EB%8B%9D/) 책의 5장 3절의 코드 예제입니다. 책에는 더 많은 내용과 그림이 있습니다. 이 노트북에는 소스 코드에 관련된 설명만 포함합니다.\n",
    "\n",
    "----\n",
    "\n",
    "작은 이미지 데이터셋에 딥러닝을 적용하는 일반적이고 매우 효과적인 방법은 사전 훈련된 네트워크를 사용하는 것입니다. 사전 훈련된 네트워크는 일반적으로 대규모 이미지 분류 문제를 위해 대량의 데이터셋에서 미리 훈련되어 저장된 네트워크입니다. 원본 데이터셋이 충분히 크고 일반적이라면 사전 훈련된 네트워크에 의해 학습된 특성의 계층 구조는 실제 세상에 대한 일반적인 모델로 효율적인 역할을 할 수 있습니다. 새로운 문제가 원래 작업과 완전히 다른 클래스에 대한 것이더라도 이런 특성은 많은 컴퓨터 비전 문제에 유용합니다. 예를 들어 (대부분 동물이나 생활 용품으로 이루어진) ImageNet 데이터셋에 네트워크를 훈련합니다. 그다음 이 네트워크를 이미지에서 가구 아이템을 식별하는 것 같은 다른 용도로 사용할 수 있습니다. 학습된 특성을 다른 문제에 적용할 수 있는 이런 유연성은 이전의 많은 얕은 학습 방법과 비교했을 때 딥러닝의 핵심 장점입니다. 이런 방식으로 작은 데이터셋을 가진 문제에도 딥러닝이 효율적으로 작동할 수 있습니다.\n",
    "\n",
    "여기에서는 (1.4백만 개의 레이블된 이미지와 1,000개의 클래스로 이루어진) ImageNet 데이터셋에서 훈련된 대규모 컨브넷을 사용해 보겠습니다. ImageNet 데이터셋은 다양한 종의 강아지와 고양이를 포함해 많은 동물들을 포함하고 있습니다. 그래서 강아지 vs. 고양이 분류 문제에 좋은 성능을 낼 것 같습니다.\n",
    "\n",
    "캐런 시몬연과 앤드류 지서먼이 2014년에 개발한 VGG16 구조를 사용하겠습니다. VGG16은 간단하고 ImageNet 데이터셋에 널리 사용되는 컨브넷 구조입니다. VGG16은 조금 오래되었고 최고 수준의 성능에는 못미치며 최근의 다른 모델보다는 조금 무겁습니다. 하지만 이 모델의 구조가 이전에 보았던 것과 비슷해서 새로운 개념을 도입하지 않고 이해하기 쉽기 때문에 선택했습니다. 아마 VGG가 처음 보는 모델 애칭일지 모르겠습니다. 이런 이름에는 VGG, ResNet, Inception, Inception-ResNet, Xception 등이 있습니다. 컴퓨터 비전을 위해 딥러닝을 계속 공부하다보면 이런 이름을 자주 만나게 될 것입니다.\n",
    "\n",
    "사전 훈련된 네트워크를 사용하는 두 가지 방법이 있습니다. 특성 추출과 미세 조정입니다. 이 두 가지를 모두 다루어 보겠습니다. 먼저 특성 추출부터 시작하죠."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 특성 추출\n",
    "\n",
    "특성 추출은 사전에 학습된 네트워크의 표현을 사용해 새로운 샘플에서 흥미로운 특성을 뽑아내는 것입니다. 이런 특성을 사용하여 새로운 분류기를 처음부터 훈련합니다.\n",
    "\n",
    "앞서 보았듯이 컨브넷은 이미지 분류를 위해 두 부분으로 구성됩니다. 먼저 연속된 합성곱과 풀링 층으로 시작해서 완전 연결 분류기로 끝납니다. 첫 번째 부분을 모델의 합성곱 기반층(convolutional base)이라고 부르겠습니다. 컨브넷의 경우 특성 추출은 사전에 훈련된 네트워크의 합성곱 기반층을 선택해 새로운 데이터를 통과시키고 그 출력으로 새로운 분류기를 훈련합니다.\n",
    "\n",
    "![swapping FC classifiers](https://s3.amazonaws.com/book.keras.io/img/ch5/swapping_fc_classifier.png)\n",
    "\n",
    "왜 합성곱 층만 재사용할까요? 완전 연결 분류기도 재사용할 수 있을까요? 일반적으로 권장하지 않습니다. 합성곱 층에 의해 학습된 표현이 더 일반적이어서 재사용 가능하기 때문입니다. 컨브넷의 특성 맵은 사진에 대한 일반적인 컨셉의 존재 여부를 기록한 맵입니다. 그래서 주어진 컴퓨터 비전 문제에 상관없이 유용하게 사용할 수 있습니다. 하지만 분류기에서 학습한 표현은 모델이 훈련된 클래스 집합에 특화되어 있습니다. 분류기는 전체 사진에 어떤 클래스가 존재할 확률에 관한 정보만을 담고 있습니다. 더군다나 완전 연결 층에서 찾은 표현은 더 이상 입력 이미지에 있는 객체의 위치 정보를 가지고 있지 않습니다. 완전 연결 층들은 공간 개념을 제거하지만 합성곱의 특성 맵은 객체의 위치를 고려합니다. 객체의 위치가 중요한 문제라면 완전 연결 층에서 만든 특성은 크게 쓸모가 없습니다.\n",
    "\n",
    "특정 합성곱 층에서 추출한 표현의 일반성(그리고 재사용성)의 수준은 모델에 있는 층의 깊이에 달려 있습니다. 모델의 하위 층은 (에지, 색깔, 질감 등과 같이) 지역적이고 매우 일반적인 특성 맵을 추출합니다. 반면 상위 층은 ('강아지 눈'이나 '고양이 귀'와 같이) 좀 더 추상적인 개념을 추출합니다. 만약 새로운 데이터셋이 원본 모델이 훈련한 데이터셋과 많이 다르다면 전체 합성곱 기반층을 사용하는 것보다는 모델의 하위 층 몇 개만 특성 추출에 사용하는 것이 좋습니다.\n",
    "\n",
    "ImageNet의 클래스 집합에는 여러 종류의 강아지와 고양이를 포함하고 있습니다. 이런 경우 원본 모델의 완전 연결 층에 있는 정보를 재사용하는 것이 도움이 될 것 같습니다. 하지만 새로운 문제의 클래스가 원본 모델의 클래스 집합과 겹치지 않는 좀 더 일반적인 경우를 다루기 위해서 여기서는 완전 연결 층을 사용하지 않겠습니다."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "ImageNet 데이터셋에 훈련된 VGG16 네트워크의 합성곱 기반층을 사용하여 강아지와 고양이 이미지에서 유용한 특성을 추출해 보겠습니다. 그런 다음 이 특성으로 강아지 vs. 고양이 분류기를 훈련합니다.\n",
    "\n",
    "VGG16 모델은 케라스에 패키지로 포함되어 있습니다. `keras.applications` 모듈에서 임포트할 수 있습니다. `keras.applications` 모듈에서 사용 가능한 이미지 분류 모델은 다음과 같습니다(모두 ImageNet 데이터셋에서 훈련되었습니다):\n",
    "\n",
    "* Xception\n",
    "* InceptionV3\n",
    "* ResNet50\n",
    "* VGG16\n",
    "* VGG19\n",
    "* MobileNet\n",
    "\n",
    "VGG16 모델을 만들어 보죠:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.applications import VGG16\n",
    "\n",
    "conv_base = VGG16(weights='imagenet',\n",
    "                  include_top=False,\n",
    "                  input_shape=(150, 150, 3))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "VGG16 함수에 세 개의 매개변수를 전달합니다:\n",
    "\n",
    "* `weights`는 모델을 초기화할 가중치 체크포인트를 지정합니다.\n",
    "* `include_top`은 네트워크의 최상위 완전 연결 분류기를 포함할지 안할지를 지정합니다. 기본값은 ImageNet의 1,000개의 클래스에 대응되는 완전 연결 분류기를 포함합니다. 별도의 (강아지와 고양이 두 개의 클래스를 구분하는) 완전 연결 층을 추가하려고 하므로 이를 포함시키지 않습니다.\n",
    "* `input_shape`은 네트워크에 주입할 이미지 텐서의 크기입니다. 이 매개변수는 선택사항입니다. 이 값을 지정하지 않으면 네트워크가 어떤 크기의 입력도 처리할 수 있습니다.\n",
    "\n",
    "다음은 VGG16 합성곱 기반층의 자세한 구조입니다. 이 구조는 앞에서 보았던 간단한 컨브넷과 비슷합니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 150, 150, 3)       0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, 150, 150, 64)      1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, 150, 150, 64)      36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, 75, 75, 64)        0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, 75, 75, 128)       73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, 75, 75, 128)       147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, 37, 37, 128)       0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, 37, 37, 256)       295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, 37, 37, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, 37, 37, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, 18, 18, 256)       0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, 18, 18, 512)       1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, 18, 18, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, 18, 18, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, 9, 9, 512)         0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, 9, 9, 512)         2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, 9, 9, 512)         2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, 9, 9, 512)         2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, 4, 4, 512)         0         \n",
      "=================================================================\n",
      "Total params: 14,714,688\n",
      "Trainable params: 14,714,688\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "conv_base.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "최종 특성 맵의 크기는 `(4, 4, 512)`입니다. 이 특성 위에 완전 연결 층을 놓을 것입니다.\n",
    "이 지점에서 두 가지 방식이 가능합니다.\n",
    "\n",
    "* 새로운 데이터셋에서 합성곱 기반층을 실행하고 출력을 넘파이 배열로 디스크에 저장합니다. 그다음 이 데이터를 이 책의 1부에서 보았던 것과 비슷한 독립된 완전 연결 분류기에 입력으로 사용합니다. 합성곱 연산은 전체 과정 중에서 가장 비싼 부분입니다. 이 방식은 모든 입력 이미지에 대해 합성곱 기반층을 한 번만 실행하면 되기 때문에 빠르고 비용이 적게 듭니다. 하지만 이런 이유 때문에 이 기법에는 데이터 증식을 사용할 수 없습니다.\n",
    "* 준비한 모델(`conv_base`) 위에 `Dense` 층을 쌓아 확장합니다. 그다음 입력 데이터에서 엔드 투 엔드로 전체 모델을 실행합니다. 모델에 노출된 모든 입력 이미지가 매번 합성곱 기반층을 통과하기 때문에 데이터 증식을 사용할 수 있습니다. 하지만 이런 이유로 이 방식은 첫 번째 방식보다 훨씬 비용이 많이 듭니다.\n",
    "\n",
    "두 가지 방식을 모두 다루어 보겠습니다. 첫 번째 방식을 구현하는 코드를 살펴봅니다. `conv_base`에 데이터를 주입하고 출력을 기록합니다. 이 출력을 새로운 모델의 입력으로 사용하겠습니다.\n",
    "\n",
    "먼저 앞서 소개한 `ImageDataGenerator`를 사용해 이미지와 레이블을 넘파이 배열로 추출하겠습니다. `conv_base` 모델의 `predict` 메서드를 호출하여 이 이미지에서 특성을 추출합니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2000 images belonging to 2 classes.\n",
      "Found 1000 images belonging to 2 classes.\n",
      "Found 1000 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "base_dir = './datasets/cats_and_dogs_small'\n",
    "\n",
    "train_dir = os.path.join(base_dir, 'train')\n",
    "validation_dir = os.path.join(base_dir, 'validation')\n",
    "test_dir = os.path.join(base_dir, 'test')\n",
    "\n",
    "datagen = ImageDataGenerator(rescale=1./255)\n",
    "batch_size = 20\n",
    "\n",
    "def extract_features(directory, sample_count):\n",
    "    features = np.zeros(shape=(sample_count, 4, 4, 512))\n",
    "    labels = np.zeros(shape=(sample_count))\n",
    "    generator = datagen.flow_from_directory(\n",
    "        directory,\n",
    "        target_size=(150, 150),\n",
    "        batch_size=batch_size,\n",
    "        class_mode='binary')\n",
    "    i = 0\n",
    "    for inputs_batch, labels_batch in generator:\n",
    "        features_batch = conv_base.predict(inputs_batch)\n",
    "        features[i * batch_size : (i + 1) * batch_size] = features_batch\n",
    "        labels[i * batch_size : (i + 1) * batch_size] = labels_batch\n",
    "        i += 1\n",
    "        if i * batch_size >= sample_count:\n",
    "            # 제너레이터는 루프 안에서 무한하게 데이터를 만들어내므로 모든 이미지를 한 번씩 처리하고 나면 중지합니다\n",
    "            break\n",
    "    return features, labels\n",
    "\n",
    "train_features, train_labels = extract_features(train_dir, 2000)\n",
    "validation_features, validation_labels = extract_features(validation_dir, 1000)\n",
    "test_features, test_labels = extract_features(test_dir, 1000)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "추출된 특성의 크기는 `(samples, 4, 4, 512)`입니다. 완전 연결 분류기에 주입하기 위해서 먼저 `(samples, 8192)` 크기로 펼칩니다:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_features = np.reshape(train_features, (2000, 4 * 4 * 512))\n",
    "validation_features = np.reshape(validation_features, (1000, 4 * 4 * 512))\n",
    "test_features = np.reshape(test_features, (1000, 4 * 4 * 512))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "그러고 나서 완전 연결 분류기를 정의하고(규제를 위해 드롭아웃을 사용합니다) 저장된 데이터와 레이블을 사용해 훈련합니다:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 2000 samples, validate on 1000 samples\n",
      "Epoch 1/30\n",
      "2000/2000 [==============================] - 1s 361us/step - loss: 0.6234 - acc: 0.6500 - val_loss: 0.4730 - val_acc: 0.8210\n",
      "Epoch 2/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.4368 - acc: 0.8035 - val_loss: 0.3854 - val_acc: 0.8350\n",
      "Epoch 3/30\n",
      "2000/2000 [==============================] - 0s 193us/step - loss: 0.3694 - acc: 0.8425 - val_loss: 0.3403 - val_acc: 0.8640\n",
      "Epoch 4/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.3275 - acc: 0.8655 - val_loss: 0.3082 - val_acc: 0.8800\n",
      "Epoch 5/30\n",
      "2000/2000 [==============================] - 0s 193us/step - loss: 0.2981 - acc: 0.8805 - val_loss: 0.2907 - val_acc: 0.8850\n",
      "Epoch 6/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.2724 - acc: 0.8900 - val_loss: 0.2764 - val_acc: 0.8900\n",
      "Epoch 7/30\n",
      "2000/2000 [==============================] - 0s 193us/step - loss: 0.2522 - acc: 0.9010 - val_loss: 0.2678 - val_acc: 0.8970\n",
      "Epoch 8/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.2420 - acc: 0.9025 - val_loss: 0.2665 - val_acc: 0.8920\n",
      "Epoch 9/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.2303 - acc: 0.9065 - val_loss: 0.2565 - val_acc: 0.8990\n",
      "Epoch 10/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.2143 - acc: 0.9170 - val_loss: 0.2571 - val_acc: 0.8930\n",
      "Epoch 11/30\n",
      "2000/2000 [==============================] - 0s 193us/step - loss: 0.1988 - acc: 0.9265 - val_loss: 0.2472 - val_acc: 0.8990\n",
      "Epoch 12/30\n",
      "2000/2000 [==============================] - 0s 195us/step - loss: 0.1956 - acc: 0.9280 - val_loss: 0.2453 - val_acc: 0.8990\n",
      "Epoch 13/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.1807 - acc: 0.9335 - val_loss: 0.2537 - val_acc: 0.8920\n",
      "Epoch 14/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.1741 - acc: 0.9380 - val_loss: 0.2400 - val_acc: 0.9010\n",
      "Epoch 15/30\n",
      "2000/2000 [==============================] - 0s 193us/step - loss: 0.1683 - acc: 0.9395 - val_loss: 0.2386 - val_acc: 0.9010\n",
      "Epoch 16/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.1620 - acc: 0.9400 - val_loss: 0.2481 - val_acc: 0.8990\n",
      "Epoch 17/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.1480 - acc: 0.9465 - val_loss: 0.2372 - val_acc: 0.9040\n",
      "Epoch 18/30\n",
      "2000/2000 [==============================] - 0s 193us/step - loss: 0.1480 - acc: 0.9480 - val_loss: 0.2360 - val_acc: 0.9010\n",
      "Epoch 19/30\n",
      "2000/2000 [==============================] - 0s 193us/step - loss: 0.1376 - acc: 0.9575 - val_loss: 0.2469 - val_acc: 0.8990\n",
      "Epoch 20/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.1358 - acc: 0.9510 - val_loss: 0.2469 - val_acc: 0.8950\n",
      "Epoch 21/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.1322 - acc: 0.9560 - val_loss: 0.2342 - val_acc: 0.9020\n",
      "Epoch 22/30\n",
      "2000/2000 [==============================] - 0s 193us/step - loss: 0.1287 - acc: 0.9550 - val_loss: 0.2350 - val_acc: 0.9030\n",
      "Epoch 23/30\n",
      "2000/2000 [==============================] - 0s 193us/step - loss: 0.1165 - acc: 0.9650 - val_loss: 0.2488 - val_acc: 0.8960\n",
      "Epoch 24/30\n",
      "2000/2000 [==============================] - 0s 195us/step - loss: 0.1172 - acc: 0.9560 - val_loss: 0.2338 - val_acc: 0.9050\n",
      "Epoch 25/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.1115 - acc: 0.9645 - val_loss: 0.2350 - val_acc: 0.9050\n",
      "Epoch 26/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.1134 - acc: 0.9620 - val_loss: 0.2338 - val_acc: 0.9060\n",
      "Epoch 27/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.0998 - acc: 0.9755 - val_loss: 0.2343 - val_acc: 0.9010\n",
      "Epoch 28/30\n",
      "2000/2000 [==============================] - 0s 193us/step - loss: 0.1000 - acc: 0.9665 - val_loss: 0.2369 - val_acc: 0.9030\n",
      "Epoch 29/30\n",
      "2000/2000 [==============================] - 0s 198us/step - loss: 0.0966 - acc: 0.9710 - val_loss: 0.2489 - val_acc: 0.9000\n",
      "Epoch 30/30\n",
      "2000/2000 [==============================] - 0s 194us/step - loss: 0.0917 - acc: 0.9715 - val_loss: 0.2395 - val_acc: 0.9010\n"
     ]
    }
   ],
   "source": [
    "from keras import models\n",
    "from keras import layers\n",
    "from keras import optimizers\n",
    "\n",
    "model = models.Sequential()\n",
    "model.add(layers.Dense(256, activation='relu', input_dim=4 * 4 * 512))\n",
    "model.add(layers.Dropout(0.5))\n",
    "model.add(layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "model.compile(optimizer=optimizers.RMSprop(lr=2e-5),\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['acc'])\n",
    "\n",
    "history = model.fit(train_features, train_labels,\n",
    "                    epochs=30,\n",
    "                    batch_size=20,\n",
    "                    validation_data=(validation_features, validation_labels))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "두 개의 `Dense` 층만 처리하면 되기 때문에 훈련이 매우 빠릅니다. CPU를 사용하더라도 한 에포크에 걸리는 시간이 1초 미만입니다.\n",
    "\n",
    "훈련 손실과 정확도 곡선을 살펴보죠:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "약 90%의 검증 정확도에 도달했습니다. 이전 절에서 처음부터 훈련시킨 작은 모델에서 얻은 것보다 훨씬 좋습니다. 하지만 이 그래프는 많은 비율로 드롭아웃을 사용했음에도 불구하고 훈련이 시작하면서 거의 바로 과대적합되고 있다는 것을 보여줍니다. 작은 이미지 데이터셋에서는 과대적합을 막기 위해 필수적인 데이터 증식을 사용하지 않았기 때문입니다.\n",
    "\n",
    "이제 특성 추출을 위해 두 번째로 언급한 방법을 살펴보겠습니다. 이 방법은 훨씬 느리고 비용이 많이 들지만 훈련하는 동안 데이터 증식 기법을 사용할 수 있습니다. `conv_base` 모델을 확장하고 입력 데이터를 사용해 엔드 투 엔드로 실행합니다.\n",
    "\n",
    "이 기법은 연산 비용이 크기 때문에 GPU를 사용할 수 있을 때 시도해야 합니다. CPU에서는 적용하기 매우 힘듭니다. GPU를 사용할 수 없다면 첫 번째 방법을 사용하세요.\n",
    "\n",
    "모델은 층과 동일하게 작동하므로 층을 추가하듯이 `Sequential` 모델에 (`conv_base` 같은) 다른 모델을 추가할 수 있습니다:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import models\n",
    "from keras import layers\n",
    "\n",
    "model = models.Sequential()\n",
    "model.add(conv_base)\n",
    "model.add(layers.Flatten())\n",
    "model.add(layers.Dense(256, activation='relu'))\n",
    "model.add(layers.Dense(1, activation='sigmoid'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "이 모델의 구조는 다음과 같습니다:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "vgg16 (Model)                (None, 4, 4, 512)         14714688  \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 8192)              0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 256)               2097408   \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 1)                 257       \n",
      "=================================================================\n",
      "Total params: 16,812,353\n",
      "Trainable params: 16,812,353\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "여기서 볼 수 있듯이 VGG16의 합성곱 기반층은 14,714,688개의 매우 많은 파라미터를 가지고 있습니다. 합성곱 기반층 위에 추가한 분류기는 2백만 개의 파라미터를 가집니다.\n",
    "\n",
    "모델을 컴파일하고 훈련하기 전에 합성곱 기반층을 동결하는 것이 아주 중요합니다. 하나 이상의 층을 동결한다는 것은 훈련하는 동안 가중치가 업데이트되지 않도록 막는다는 뜻입니다. 이렇게 하지 않으면 합성곱 기반층에 의해 사전에 학습된 표현이 훈련하는 동안 수정될 것입니다. 맨 위의 `Dense` 층은 랜덤하게 초기화되었기 때문에 매우 큰 가중치 업데이트 값이 네트워크에 전파될 것입니다. 이는 사전에 학습된 표현을 크게 훼손하게 됩니다.\n",
    "\n",
    "케라스에서는 `trainable` 속성을 `False`로 설정하여 네트워크를 동결할 수 있습니다:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "conv_base를 동결하기 전 훈련되는 가중치의 수: 30\n"
     ]
    }
   ],
   "source": [
    "print('conv_base를 동결하기 전 훈련되는 가중치의 수:', \n",
    "      len(model.trainable_weights))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "conv_base.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "conv_base를 동결한 후 훈련되는 가중치의 수: 4\n"
     ]
    }
   ],
   "source": [
    "print('conv_base를 동결한 후 훈련되는 가중치의 수:', \n",
    "      len(model.trainable_weights))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "이렇게 설정하면 추가한 두 개의 `Dense` 층의 가중치만 훈련될 것입니다. 층마다 두 개씩(가중치 행렬과 편향 벡터) 총 네 개의 텐서가 훈련됩니다. 변경 사항을 적용하려면 먼저 모델을 컴파일해야 합니다. 컴파일 단계 후에 `trainable` 속성을 변경하면 반드시 모델을 다시 컴파일해야 합니다. 그렇지 않으면 변경 사항이 적용되지 않습니다.\n",
    "\n",
    "이제 앞의 예제에서 사용했던 데이터 증식을 사용하여 모델 훈련을 시작할 수 있습니다:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2000 images belonging to 2 classes.\n",
      "Found 1000 images belonging to 2 classes.\n",
      "Epoch 1/30\n",
      " - 13s - loss: 0.5522 - acc: 0.7325 - val_loss: 0.4184 - val_acc: 0.8390\n",
      "Epoch 2/30\n",
      " - 12s - loss: 0.4120 - acc: 0.8335 - val_loss: 0.3403 - val_acc: 0.8780\n",
      "Epoch 3/30\n",
      " - 12s - loss: 0.3549 - acc: 0.8540 - val_loss: 0.3026 - val_acc: 0.8900\n",
      "Epoch 4/30\n",
      " - 12s - loss: 0.3191 - acc: 0.8715 - val_loss: 0.2988 - val_acc: 0.8750\n",
      "Epoch 5/30\n",
      " - 12s - loss: 0.3084 - acc: 0.8700 - val_loss: 0.2750 - val_acc: 0.8890\n",
      "Epoch 6/30\n",
      " - 12s - loss: 0.2868 - acc: 0.8905 - val_loss: 0.2729 - val_acc: 0.8850\n",
      "Epoch 7/30\n",
      " - 12s - loss: 0.2722 - acc: 0.8875 - val_loss: 0.2571 - val_acc: 0.8940\n",
      "Epoch 8/30\n",
      " - 12s - loss: 0.2691 - acc: 0.8960 - val_loss: 0.2538 - val_acc: 0.8920\n",
      "Epoch 9/30\n",
      " - 12s - loss: 0.2641 - acc: 0.9025 - val_loss: 0.2477 - val_acc: 0.8980\n",
      "Epoch 10/30\n",
      " - 12s - loss: 0.2506 - acc: 0.8985 - val_loss: 0.2618 - val_acc: 0.8880\n",
      "Epoch 11/30\n",
      " - 12s - loss: 0.2542 - acc: 0.8885 - val_loss: 0.2415 - val_acc: 0.9060\n",
      "Epoch 12/30\n",
      " - 12s - loss: 0.2378 - acc: 0.9095 - val_loss: 0.2478 - val_acc: 0.9020\n",
      "Epoch 13/30\n",
      " - 12s - loss: 0.2369 - acc: 0.8975 - val_loss: 0.2630 - val_acc: 0.8840\n",
      "Epoch 14/30\n",
      " - 12s - loss: 0.2477 - acc: 0.8990 - val_loss: 0.2376 - val_acc: 0.9090\n",
      "Epoch 15/30\n",
      " - 12s - loss: 0.2178 - acc: 0.9105 - val_loss: 0.2559 - val_acc: 0.8930\n",
      "Epoch 16/30\n",
      " - 12s - loss: 0.2263 - acc: 0.9095 - val_loss: 0.2356 - val_acc: 0.9070\n",
      "Epoch 17/30\n",
      " - 12s - loss: 0.2223 - acc: 0.9110 - val_loss: 0.2360 - val_acc: 0.9090\n",
      "Epoch 18/30\n",
      " - 12s - loss: 0.2178 - acc: 0.9155 - val_loss: 0.2400 - val_acc: 0.9050\n",
      "Epoch 19/30\n",
      " - 12s - loss: 0.2109 - acc: 0.9180 - val_loss: 0.2345 - val_acc: 0.9090\n",
      "Epoch 20/30\n",
      " - 12s - loss: 0.2161 - acc: 0.9100 - val_loss: 0.2349 - val_acc: 0.9050\n",
      "Epoch 21/30\n",
      " - 12s - loss: 0.2104 - acc: 0.9200 - val_loss: 0.2357 - val_acc: 0.9100\n",
      "Epoch 22/30\n",
      " - 12s - loss: 0.2005 - acc: 0.9235 - val_loss: 0.2342 - val_acc: 0.9090\n",
      "Epoch 23/30\n",
      " - 12s - loss: 0.1952 - acc: 0.9235 - val_loss: 0.2415 - val_acc: 0.9040\n",
      "Epoch 24/30\n",
      " - 12s - loss: 0.1992 - acc: 0.9185 - val_loss: 0.2349 - val_acc: 0.9030\n",
      "Epoch 25/30\n",
      " - 12s - loss: 0.1939 - acc: 0.9180 - val_loss: 0.2339 - val_acc: 0.9030\n",
      "Epoch 26/30\n",
      " - 12s - loss: 0.1915 - acc: 0.9220 - val_loss: 0.2379 - val_acc: 0.9090\n",
      "Epoch 27/30\n",
      " - 12s - loss: 0.1920 - acc: 0.9205 - val_loss: 0.2346 - val_acc: 0.9050\n",
      "Epoch 28/30\n",
      " - 12s - loss: 0.1855 - acc: 0.9310 - val_loss: 0.2330 - val_acc: 0.9010\n",
      "Epoch 29/30\n",
      " - 12s - loss: 0.1958 - acc: 0.9210 - val_loss: 0.2378 - val_acc: 0.9030\n",
      "Epoch 30/30\n",
      " - 12s - loss: 0.1758 - acc: 0.9285 - val_loss: 0.2504 - val_acc: 0.8960\n"
     ]
    }
   ],
   "source": [
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "train_datagen = ImageDataGenerator(\n",
    "      rescale=1./255,\n",
    "      rotation_range=20,\n",
    "      width_shift_range=0.1,\n",
    "      height_shift_range=0.1,\n",
    "      shear_range=0.1,\n",
    "      zoom_range=0.1,\n",
    "      horizontal_flip=True,\n",
    "      fill_mode='nearest')\n",
    "\n",
    "# 검증 데이터는 증식되어서는 안 됩니다!\n",
    "test_datagen = ImageDataGenerator(rescale=1./255)\n",
    "\n",
    "train_generator = train_datagen.flow_from_directory(\n",
    "        # 타깃 디렉터리\n",
    "        train_dir,\n",
    "        # 모든 이미지의 크기를 150 × 150로 변경합니다\n",
    "        target_size=(150, 150),\n",
    "        batch_size=20,\n",
    "        # binary_crossentropy 손실을 사용하므로 이진 레이블이 필요합니다\n",
    "        class_mode='binary')\n",
    "\n",
    "validation_generator = test_datagen.flow_from_directory(\n",
    "        validation_dir,\n",
    "        target_size=(150, 150),\n",
    "        batch_size=20,\n",
    "        class_mode='binary')\n",
    "\n",
    "model.compile(loss='binary_crossentropy',\n",
    "              optimizer=optimizers.RMSprop(lr=2e-5),\n",
    "              metrics=['acc'])\n",
    "\n",
    "history = model.fit_generator(\n",
    "      train_generator,\n",
    "      steps_per_epoch=100,\n",
    "      epochs=30,\n",
    "      validation_data=validation_generator,\n",
    "      validation_steps=50,\n",
    "      verbose=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save('cats_and_dogs_small_3.h5')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "결과 그래프를 다시 그려 봅시다:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "여기서 볼 수 있듯이 검증 정확도가 이전과 비슷하지만 처음부터 훈련시킨 소규모 컨브넷보다 과대적합이 줄었습니다."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 미세 조정\n",
    "\n",
    "모델을 재사용하는 데 널리 사용되는 또 하나의 기법은 특성 추출을 보완하는 미세 조정입니다. 미세 조정은 특성 추출에 사용했던 동결 모델의 상위 층 몇 개를 동결에서 해제하고 모델에 새로 추가한 층(여기서는 완전 연결 분류기)과 함께 훈련하는 것입니다. 주어진 문제에 조금 더 밀접하게 재사용 모델의 표현을 일부 조정하기 때문에 미세 조정이라고 부릅니다.\n",
    "\n",
    "![fine-tuning VGG16](https://s3.amazonaws.com/book.keras.io/img/ch5/vgg16_fine_tuning.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "앞서 랜덤하게 초기화된 상단 분류기를 훈련하기 위해 VGG16의 합성곱 기반층을 동결해야 한다고 말했습니다. 같은 이유로 맨 위에 있는 분류기가 훈련된 후에 합성곱 기반의 상위 층을 미세 조정할 수 있습니다. 분류기가 미리 훈련되지 않으면 훈련되는 동안 너무 큰 오차 신호가 네트워크에 전파됩니다. 이는 미세 조정될 층들이 사전에 학습한 표현들을 망가뜨리게 될 것입니다. 네트워크를 미세 조정하는 단계는 다음과 같습니다:\n",
    "\n",
    "1. 사전에 훈련된 기반 네트워크 위에 새로운 네트워크를 추가합니다.\n",
    "2. 기반 네트워크를 동결합니다.\n",
    "3. 새로 추가한 네트워크를 훈련합니다.\n",
    "4. 기반 네트워크에서 일부 층의 동결을 해제합니다.\n",
    "5. 동결을 해제한 층과 새로 추가한 층을 함께 훈련합니다.\n",
    "\n",
    "처음 세 단계는 특성 추출을 할 때 이미 완료했습니다. 네 번째 단계를 진행해 보죠. `conv_base`의 동결을 해제하고 개별 층을 동결하겠습니다.\n",
    "\n",
    "기억을 되살리기 위해 합성곱 기반층의 구조를 다시 확인해 보겠습니다:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 150, 150, 3)       0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, 150, 150, 64)      1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, 150, 150, 64)      36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, 75, 75, 64)        0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, 75, 75, 128)       73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, 75, 75, 128)       147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, 37, 37, 128)       0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, 37, 37, 256)       295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, 37, 37, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, 37, 37, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, 18, 18, 256)       0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, 18, 18, 512)       1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, 18, 18, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, 18, 18, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, 9, 9, 512)         0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, 9, 9, 512)         2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, 9, 9, 512)         2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, 9, 9, 512)         2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, 4, 4, 512)         0         \n",
      "=================================================================\n",
      "Total params: 14,714,688\n",
      "Trainable params: 0\n",
      "Non-trainable params: 14,714,688\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "conv_base.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "마지막 세 개의 합성곱 층을 미세 조정하겠습니다. 즉, `block4_pool`까지 모든 층은 동결되고 `block5_conv1`, `block5_conv2`, `block5_conv3` 층은 학습 대상이 됩니다.\n",
    "\n",
    "왜 더 많은 층을 미세 조정하지 않을까요? 왜 전체 합성곱 기반층을 미세 조정하지 않을까요? 그렇게 할 수도 있지만 다음 사항을 고려해야 합니다:\n",
    "\n",
    "* 합성곱 기반층에 있는 하위 층들은 좀 더 일반적이고 재사용 가능한 특성들을 인코딩합니다. 반면 상위 층은 좀 더 특화된 특성을 인코딩합니다. 새로운 문제에 재활용하도록 수정이 필요한 것은 구체적인 특성이므로 이들을 미세 조정하는 것이 유리합니다. 하위 층으로 갈수록 미세 조정에 대한 효과가 감소합니다.\n",
    "* 훈련해야 할 파라미터가 많을수록 과대적합의 위험이 커집니다. 합성곱 기반층은 1천 5백만 개의 파라미터를 가지고 있습니다. 작은 데이터셋으로 전부 훈련하려고 하면 매우 위험합니다.\n",
    "\n",
    "그러므로 이런 상황에서는 합성곱 기반층에서 최상위 두 세개의 층만 미세 조정하는 것이 좋습니다.\n",
    "\n",
    "앞선 예제 코드에 이어서 미세 조정을 설정해보죠:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "conv_base.trainable = True\n",
    "\n",
    "set_trainable = False\n",
    "for layer in conv_base.layers:\n",
    "    if layer.name == 'block5_conv1':\n",
    "        set_trainable = True\n",
    "    if set_trainable:\n",
    "        layer.trainable = True\n",
    "    else:\n",
    "        layer.trainable = False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "이제 네트워크의 미세 조정을 시작하겠습니다. 학습률을 낮춘 RMSProp 옵티마이저를 사용합니다. 학습률을 낮추는 이유는 미세 조정하는 세 개의 층에서 학습된 표현을 조금씩 수정하기 위해서입니다. 변경량이 너무 크면 학습된 표현에 나쁜 영향을 끼칠 수 있습니다.\n",
    "\n",
    "미세 조정을 진행해 보죠:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "100/100 [==============================] - 14s 141ms/step - loss: 0.1923 - acc: 0.9200 - val_loss: 0.2230 - val_acc: 0.9160\n",
      "Epoch 2/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.1687 - acc: 0.9340 - val_loss: 0.2537 - val_acc: 0.9000\n",
      "Epoch 3/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.1434 - acc: 0.9445 - val_loss: 0.2158 - val_acc: 0.9150\n",
      "Epoch 4/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.1150 - acc: 0.9565 - val_loss: 0.2321 - val_acc: 0.9120\n",
      "Epoch 5/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0967 - acc: 0.9605 - val_loss: 0.2366 - val_acc: 0.9170\n",
      "Epoch 6/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.1012 - acc: 0.9570 - val_loss: 0.2014 - val_acc: 0.9250\n",
      "Epoch 7/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0898 - acc: 0.9625 - val_loss: 0.2320 - val_acc: 0.9130\n",
      "Epoch 8/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0793 - acc: 0.9720 - val_loss: 0.2295 - val_acc: 0.9120\n",
      "Epoch 9/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0641 - acc: 0.9740 - val_loss: 0.2033 - val_acc: 0.9250\n",
      "Epoch 10/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0567 - acc: 0.9790 - val_loss: 0.2200 - val_acc: 0.9230\n",
      "Epoch 11/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0466 - acc: 0.9855 - val_loss: 0.2137 - val_acc: 0.9310\n",
      "Epoch 12/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0418 - acc: 0.9870 - val_loss: 0.2831 - val_acc: 0.9140\n",
      "Epoch 13/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0378 - acc: 0.9865 - val_loss: 0.2696 - val_acc: 0.9110\n",
      "Epoch 14/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0392 - acc: 0.9880 - val_loss: 0.2511 - val_acc: 0.9290\n",
      "Epoch 15/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0340 - acc: 0.9865 - val_loss: 0.2213 - val_acc: 0.9310\n",
      "Epoch 16/100\n",
      "100/100 [==============================] - 14s 135ms/step - loss: 0.0304 - acc: 0.9925 - val_loss: 0.2513 - val_acc: 0.9270\n",
      "Epoch 17/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0297 - acc: 0.9925 - val_loss: 0.2646 - val_acc: 0.9240\n",
      "Epoch 18/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0302 - acc: 0.9910 - val_loss: 0.2299 - val_acc: 0.9300\n",
      "Epoch 19/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0256 - acc: 0.9910 - val_loss: 0.2423 - val_acc: 0.9330\n",
      "Epoch 20/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0280 - acc: 0.9895 - val_loss: 0.2521 - val_acc: 0.9300\n",
      "Epoch 21/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0198 - acc: 0.9940 - val_loss: 0.2582 - val_acc: 0.9280\n",
      "Epoch 22/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0197 - acc: 0.9935 - val_loss: 0.2573 - val_acc: 0.9290\n",
      "Epoch 23/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0209 - acc: 0.9940 - val_loss: 0.2322 - val_acc: 0.9350\n",
      "Epoch 24/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0160 - acc: 0.9930 - val_loss: 0.2737 - val_acc: 0.9250\n",
      "Epoch 25/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0168 - acc: 0.9940 - val_loss: 0.2716 - val_acc: 0.9330\n",
      "Epoch 26/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0157 - acc: 0.9950 - val_loss: 0.3792 - val_acc: 0.9050\n",
      "Epoch 27/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0188 - acc: 0.9935 - val_loss: 0.2382 - val_acc: 0.9300\n",
      "Epoch 28/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0126 - acc: 0.9960 - val_loss: 0.2699 - val_acc: 0.9270\n",
      "Epoch 29/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0115 - acc: 0.9975 - val_loss: 0.4240 - val_acc: 0.9050\n",
      "Epoch 30/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0212 - acc: 0.9935 - val_loss: 0.2435 - val_acc: 0.9340\n",
      "Epoch 31/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0121 - acc: 0.9975 - val_loss: 0.2652 - val_acc: 0.9370\n",
      "Epoch 32/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0120 - acc: 0.9970 - val_loss: 0.2749 - val_acc: 0.9370\n",
      "Epoch 33/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0101 - acc: 0.9965 - val_loss: 0.2410 - val_acc: 0.9330\n",
      "Epoch 34/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0081 - acc: 0.9975 - val_loss: 0.2904 - val_acc: 0.9350\n",
      "Epoch 35/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0137 - acc: 0.9940 - val_loss: 0.2813 - val_acc: 0.9370\n",
      "Epoch 36/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0121 - acc: 0.9965 - val_loss: 0.2607 - val_acc: 0.9430\n",
      "Epoch 37/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0069 - acc: 0.9985 - val_loss: 0.3894 - val_acc: 0.9160\n",
      "Epoch 38/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0118 - acc: 0.9950 - val_loss: 0.2814 - val_acc: 0.9300\n",
      "Epoch 39/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0117 - acc: 0.9960 - val_loss: 0.2617 - val_acc: 0.9340\n",
      "Epoch 40/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0063 - acc: 0.9990 - val_loss: 0.3204 - val_acc: 0.9300\n",
      "Epoch 41/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0077 - acc: 0.9975 - val_loss: 0.2644 - val_acc: 0.9340\n",
      "Epoch 42/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0081 - acc: 0.9975 - val_loss: 0.3062 - val_acc: 0.9320\n",
      "Epoch 43/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0080 - acc: 0.9970 - val_loss: 0.3106 - val_acc: 0.9300\n",
      "Epoch 44/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0064 - acc: 0.9980 - val_loss: 0.3433 - val_acc: 0.9260\n",
      "Epoch 45/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0042 - acc: 0.9995 - val_loss: 0.3642 - val_acc: 0.9250\n",
      "Epoch 46/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0034 - acc: 0.9990 - val_loss: 0.3005 - val_acc: 0.9320\n",
      "Epoch 47/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0032 - acc: 0.9990 - val_loss: 0.2969 - val_acc: 0.9340\n",
      "Epoch 48/100\n",
      "100/100 [==============================] - 14s 135ms/step - loss: 0.0018 - acc: 0.9995 - val_loss: 0.3157 - val_acc: 0.9350\n",
      "Epoch 49/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0044 - acc: 0.9980 - val_loss: 0.3320 - val_acc: 0.9290\n",
      "Epoch 50/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0092 - acc: 0.9965 - val_loss: 0.2891 - val_acc: 0.9340\n",
      "Epoch 51/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0177 - acc: 0.9955 - val_loss: 0.3986 - val_acc: 0.9150\n",
      "Epoch 52/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0039 - acc: 0.9995 - val_loss: 0.3452 - val_acc: 0.9300\n",
      "Epoch 53/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0081 - acc: 0.9970 - val_loss: 0.3618 - val_acc: 0.9260\n",
      "Epoch 54/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0045 - acc: 0.9990 - val_loss: 0.3616 - val_acc: 0.9350\n",
      "Epoch 55/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0106 - acc: 0.9975 - val_loss: 0.3061 - val_acc: 0.9380\n",
      "Epoch 56/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0062 - acc: 0.9980 - val_loss: 0.3208 - val_acc: 0.9290\n",
      "Epoch 57/100\n",
      "100/100 [==============================] - 14s 135ms/step - loss: 0.0043 - acc: 0.9990 - val_loss: 0.3035 - val_acc: 0.9340\n",
      "Epoch 58/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0035 - acc: 0.9985 - val_loss: 0.3182 - val_acc: 0.9420\n",
      "Epoch 59/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0030 - acc: 0.9985 - val_loss: 0.3125 - val_acc: 0.9360\n",
      "Epoch 60/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0058 - acc: 0.9980 - val_loss: 0.3554 - val_acc: 0.9340\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 61/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0059 - acc: 0.9975 - val_loss: 0.3441 - val_acc: 0.9270\n",
      "Epoch 62/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0031 - acc: 0.9995 - val_loss: 0.3273 - val_acc: 0.9340\n",
      "Epoch 63/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0021 - acc: 0.9995 - val_loss: 0.7382 - val_acc: 0.8840\n",
      "Epoch 64/100\n",
      "100/100 [==============================] - 14s 135ms/step - loss: 0.0068 - acc: 0.9985 - val_loss: 0.3556 - val_acc: 0.9350\n",
      "Epoch 65/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0068 - acc: 0.9980 - val_loss: 0.3997 - val_acc: 0.9220\n",
      "Epoch 66/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 9.7492e-04 - acc: 1.0000 - val_loss: 0.3453 - val_acc: 0.9260\n",
      "Epoch 67/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0019 - acc: 0.9995 - val_loss: 0.3480 - val_acc: 0.9280\n",
      "Epoch 68/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0038 - acc: 0.9985 - val_loss: 0.3466 - val_acc: 0.9320\n",
      "Epoch 69/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0012 - acc: 0.9995 - val_loss: 0.3426 - val_acc: 0.9370\n",
      "Epoch 70/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0038 - acc: 0.9995 - val_loss: 0.3239 - val_acc: 0.9360\n",
      "Epoch 71/100\n",
      "100/100 [==============================] - 14s 135ms/step - loss: 0.0095 - acc: 0.9975 - val_loss: 0.3718 - val_acc: 0.9330\n",
      "Epoch 72/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 7.9535e-04 - acc: 1.0000 - val_loss: 0.3522 - val_acc: 0.9340\n",
      "Epoch 73/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0011 - acc: 1.0000 - val_loss: 0.3539 - val_acc: 0.9350\n",
      "Epoch 74/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0072 - acc: 0.9990 - val_loss: 0.3428 - val_acc: 0.9330\n",
      "Epoch 75/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0060 - acc: 0.9975 - val_loss: 0.3387 - val_acc: 0.9360\n",
      "Epoch 76/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0034 - acc: 0.9980 - val_loss: 0.3790 - val_acc: 0.9310\n",
      "Epoch 77/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0034 - acc: 0.9985 - val_loss: 0.3285 - val_acc: 0.9370\n",
      "Epoch 78/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0073 - acc: 0.9970 - val_loss: 0.3709 - val_acc: 0.9350\n",
      "Epoch 79/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0019 - acc: 0.9995 - val_loss: 0.3700 - val_acc: 0.9360\n",
      "Epoch 80/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0018 - acc: 0.9995 - val_loss: 0.3775 - val_acc: 0.9320\n",
      "Epoch 81/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0017 - acc: 0.9995 - val_loss: 0.4030 - val_acc: 0.9300\n",
      "Epoch 82/100\n",
      "100/100 [==============================] - 14s 135ms/step - loss: 5.7382e-04 - acc: 1.0000 - val_loss: 0.4193 - val_acc: 0.9280\n",
      "Epoch 83/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0049 - acc: 0.9975 - val_loss: 0.3853 - val_acc: 0.9290\n",
      "Epoch 84/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0044 - acc: 0.9980 - val_loss: 0.3513 - val_acc: 0.9320\n",
      "Epoch 85/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0026 - acc: 0.9990 - val_loss: 0.4970 - val_acc: 0.9150\n",
      "Epoch 86/100\n",
      "100/100 [==============================] - 13s 134ms/step - loss: 0.0022 - acc: 0.9995 - val_loss: 0.3699 - val_acc: 0.9350\n",
      "Epoch 87/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0013 - acc: 0.9995 - val_loss: 0.7092 - val_acc: 0.8970\n",
      "Epoch 88/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0014 - acc: 0.9995 - val_loss: 0.5854 - val_acc: 0.9150\n",
      "Epoch 89/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0017 - acc: 0.9990 - val_loss: 0.3538 - val_acc: 0.9360\n",
      "Epoch 90/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0034 - acc: 0.9985 - val_loss: 0.3927 - val_acc: 0.9310\n",
      "Epoch 91/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0018 - acc: 0.9990 - val_loss: 0.5833 - val_acc: 0.9080\n",
      "Epoch 100/100\n",
      "100/100 [==============================] - 13s 135ms/step - loss: 0.0021 - acc: 0.9995 - val_loss: 0.5752 - val_acc: 0.9210\n"
     ]
    }
   ],
   "source": [
    "model.compile(loss='binary_crossentropy',\n",
    "              optimizer=optimizers.RMSprop(lr=1e-5),\n",
    "              metrics=['acc'])\n",
    "\n",
    "history = model.fit_generator(\n",
    "      train_generator,\n",
    "      steps_per_epoch=100,\n",
    "      epochs=100,\n",
    "      validation_data=validation_generator,\n",
    "      validation_steps=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save('cats_and_dogs_small_4.h5')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "이전과 동일한 코드로 결과 그래프를 그려 보겠습니다:"
   ]
  },
  {
   "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": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "그래프가 불규칙하게 보입니다. 그래프를 보기 쉽게하기 위해 지수 이동 평균으로 정확도와 손실 값을 부드럽게 표현할 수 있습니다. 다음은 지수 이동 평균을 구하기 위한 간단한 함수입니다:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def smooth_curve(points, factor=0.8):\n",
    "  smoothed_points = []\n",
    "  for point in points:\n",
    "    if smoothed_points:\n",
    "      previous = smoothed_points[-1]\n",
    "      smoothed_points.append(previous * factor + point * (1 - factor))\n",
    "    else:\n",
    "      smoothed_points.append(point)\n",
    "  return smoothed_points\n",
    "\n",
    "plt.plot(epochs,\n",
    "         smooth_curve(acc), 'bo', label='Smoothed training acc')\n",
    "plt.plot(epochs,\n",
    "         smooth_curve(val_acc), 'b', label='Smoothed validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs,\n",
    "         smooth_curve(loss), 'bo', label='Smoothed training loss')\n",
    "plt.plot(epochs,\n",
    "         smooth_curve(val_loss), 'b', label='Smoothed validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "검증 정확도 곡선이 훨씬 깨끗하게 보입니다. 정확도가 확실히 1% 이상 향상되었습니다.\n",
    "\n",
    "손실 곡선은 실제 어떤 향상을 얻지 못했습니다(사실 악화되었습니다). 손실히 감소되지 않았는데 어떻게 정확도가 안정되거나 향상될 수 있을까요? 답은 간단합니다. 그래프는 개별적인 손실 값의 평균을 그린 것입니다. 하지만 정확도에 영향을 미치는 것은 손실 값의 분포이지 평균이 아닙니다. 정확도는 모델이 예측한 클래스 확률이 어떤 임계값을 넘었는지에 대한 결과이기 때문입니다. 모델이 더 향상더라도 평균 손실에 반영되지 않을 수 있습니다.\n",
    "\n",
    "이제 마지막으로 테스트 데이터에서 이 모델을 평가하겠습니다:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 1000 images belonging to 2 classes.\n",
      "test acc: 0.9169999933242798\n"
     ]
    }
   ],
   "source": [
    "test_generator = test_datagen.flow_from_directory(\n",
    "        test_dir,\n",
    "        target_size=(150, 150),\n",
    "        batch_size=20,\n",
    "        class_mode='binary')\n",
    "\n",
    "test_loss, test_acc = model.evaluate_generator(test_generator, steps=50)\n",
    "print('test acc:', test_acc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "92%의 테스트 정확도를 얻을 것입니다. 이 데이터셋을 사용한 원래 캐글 경연 대회에서 꽤 높은 순위입니다. 하지만 최신 딥러닝 기법으로 훈련 데이터의 일부분(약 10%)만을 사용해서 이런 결과를 달성했습니다. 20,000개의 샘플에서 훈련하는 것과 2,000개의 샘플에서 훈련하는 것 사이에는 아주 큰 차이점이 있습니다!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 정리\n",
    "\n",
    "다음은 앞의 두 절에 있는 예제로부터 배운 것들입니다.\n",
    "\n",
    "* 컨브넷은 컴퓨터 비전 작업에 가장 뛰어난 머신 러닝 모델입니다. 아주 작은 데이터셋에서도 처음부터 훈련해서 괜찮은 성능을 낼 수 있습니다.\n",
    "* 작은 데이터셋에서는 과대적합이 큰 문제입니다. 데이터 증식은 이미지 데이터를 다룰 때 과대적합을 막을 수 있는 강력한 방법입니다.\n",
    "* 특성 추출 방식으로 새로운 데이터셋에 기존의 컨브넷을 쉽게 재사용할 수 있습니다. 작은 이미지 데이터셋으로 작업할 때 효과적인 기법입니다.\n",
    "* 특성 추출을 보완하기 위해 미세 조정을 사용할 수 있습니다. 미세 조정은 기존 모델에서 사전에 학습한 표현의 일부를 새로운 문제에 적응시킵니다. 이 기법은 조금 더 성능을 끌어올립니다.\n",
    "\n",
    "지금까지 이미지 분류 문제에서 특히 작은 데이터셋을 다루기 위한 좋은 도구들을 배웠습니다."
   ]
  }
 ],
 "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": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}