nb_ex6_2_ae_conv_mnist.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 6장. 케라스2로 구현하는 AE(오토인코더)\n",
    "## 6.3 합성곱 계층을 이용한 AE 구현 \n",
    "이미지 처리에 효과적인 합성곱 계층을 이용한 AE 방식"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.3.1 합성곱 AE 모델링\n",
    "1. 합성곱 계층을 이용하는 AE 신경망의 모델링"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import layers, models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def Conv2D(filters, kernel_size, padding='same', activation='relu'):\n",
    "    return layers.Conv2D(filters, kernel_size, padding=padding, activation=activation)\n",
    "\n",
    "class AE(models.Model):\n",
    "    def __init__(self, org_shape=(1, 28, 28)):\n",
    "        # Input\n",
    "        original = layers.Input(shape=org_shape)\n",
    "\n",
    "        # encoding-1\n",
    "        x = Conv2D(4, (3, 3))(original)\n",
    "        x = layers.MaxPooling2D((2, 2), padding='same')(x)\n",
    "\n",
    "        # encoding-2\n",
    "        x = Conv2D(8, (3, 3))(x)\n",
    "        x = layers.MaxPooling2D((2, 2), padding='same')(x)\n",
    "\n",
    "        # encoding-3: encoding output: 7x7 pixels\n",
    "        z = Conv2D(1, (7, 7))(x)\n",
    "\n",
    "        # decoding-1\n",
    "        y = Conv2D(16, (3, 3))(z)\n",
    "        y = layers.UpSampling2D((2, 2))(y)\n",
    "\n",
    "        # decoding-2\n",
    "        y = Conv2D(8, (3, 3))(y)\n",
    "        y = layers.UpSampling2D((2, 2))(y)\n",
    "\n",
    "        # decoding-3\n",
    "        y = Conv2D(4, (3, 3))(y)\n",
    "\n",
    "        # decoding & Output\n",
    "        decoded = Conv2D(1, (3, 3), activation='sigmoid')(y)\n",
    "\n",
    "        super().__init__(original, decoded)\n",
    "        self.compile(optimizer='adadelta', loss='binary_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.3.2 데이터 준비 및 학습 효과 분석 \n",
    "2. 데이터 준비와 학습 효과 분석"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "###########################\n",
    "# 데이타 불러오기\n",
    "###########################\n",
    "from ex4_1_cnn_mnist_cl import DATA\n",
    "\n",
    "\n",
    "###########################\n",
    "# 학습 효과 분석\n",
    "###########################\n",
    "from keraspp.skeras import plot_loss, plot_acc\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.3.3 합성곱 AE 결과 시각화\n",
    "3. 결과 데이터의 차원을 조정해 합성곱 AE 결과를 시각화"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import backend"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def show_ae(autoencoder, data):\n",
    "    x_test = data.x_test\n",
    "    decoded_imgs = autoencoder.predict(x_test)\n",
    "    print(decoded_imgs.shape, data.x_test.shape)\n",
    "\n",
    "    if backend.image_data_format() == 'channels_first':\n",
    "        N, n_ch, n_i, n_j = x_test.shape\n",
    "    else:\n",
    "        N, n_i, n_j, n_ch = x_test.shape\n",
    "\n",
    "    x_test = x_test.reshape(N, n_i, n_j)\n",
    "    decoded_imgs = decoded_imgs.reshape(decoded_imgs.shape[0], n_i, n_j)\n",
    "    \n",
    "    n = 10\n",
    "    plt.figure(figsize=(20, 4))\n",
    "    for i in range(n):\n",
    "\n",
    "        ax = plt.subplot(2, n, i + 1)\n",
    "        plt.imshow(x_test[i], cmap='gray')\n",
    "        # plt.gray()\n",
    "        ax.get_xaxis().set_visible(False)\n",
    "        ax.get_yaxis().set_visible(False)\n",
    "\n",
    "        ax = plt.subplot(2, n, i + 1 + n)\n",
    "        plt.imshow(decoded_imgs[i], cmap='gray')\n",
    "        # plt.gray()\n",
    "        ax.get_xaxis().set_visible(False)\n",
    "        ax.get_yaxis().set_visible(False)\n",
    "\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.3.4 합성곱 AE 학습 및 성능 평가 \n",
    "4. 합성곱 AE 모델링, 데이터 준비, 학습, 결과 시각화 코드를 준비했으니 이제 실제 학습과 성능 평가를 수행"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def main(epochs=20, batch_size=128):\n",
    "    data = DATA()\n",
    "    autoencoder = AE(data.input_shape)\n",
    "\n",
    "    history = autoencoder.fit(data.x_train, data.x_train,\n",
    "                              epochs=epochs,\n",
    "                              batch_size=batch_size,\n",
    "                              shuffle=True,\n",
    "                              validation_split=0.2)\n",
    "\n",
    "    plot_acc(history, '(a) 정확도 학습 곡선')\n",
    "    plt.show()\n",
    "    plot_loss(history, '(b) 손실 학습 곡선')\n",
    "    plt.show()\n",
    "\n",
    "    show_ae(autoencoder, data)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "375/375 [==============================] - 23s 63ms/step - loss: 0.6921 - accuracy: 0.5884 - val_loss: 0.6918 - val_accuracy: 0.6391\n",
      "Epoch 2/20\n",
      "375/375 [==============================] - 22s 59ms/step - loss: 0.6913 - accuracy: 0.6770 - val_loss: 0.6908 - val_accuracy: 0.7095\n",
      "Epoch 3/20\n",
      "375/375 [==============================] - 24s 63ms/step - loss: 0.6902 - accuracy: 0.7329 - val_loss: 0.6895 - val_accuracy: 0.7524\n",
      "Epoch 4/20\n",
      "375/375 [==============================] - 21s 55ms/step - loss: 0.6886 - accuracy: 0.7663 - val_loss: 0.6877 - val_accuracy: 0.7778\n",
      "Epoch 5/20\n",
      "375/375 [==============================] - 20s 52ms/step - loss: 0.6866 - accuracy: 0.7857 - val_loss: 0.6852 - val_accuracy: 0.7925\n",
      "Epoch 6/20\n",
      "375/375 [==============================] - 19s 52ms/step - loss: 0.6836 - accuracy: 0.7969 - val_loss: 0.6817 - val_accuracy: 0.8012\n",
      "Epoch 7/20\n",
      "375/375 [==============================] - 19s 51ms/step - loss: 0.6792 - accuracy: 0.8034 - val_loss: 0.6761 - val_accuracy: 0.8059\n",
      "Epoch 8/20\n",
      "375/375 [==============================] - 19s 51ms/step - loss: 0.6718 - accuracy: 0.8065 - val_loss: 0.6661 - val_accuracy: 0.8078\n",
      "Epoch 9/20\n",
      "375/375 [==============================] - 20s 52ms/step - loss: 0.6573 - accuracy: 0.8078 - val_loss: 0.6451 - val_accuracy: 0.8089\n",
      "Epoch 10/20\n",
      "375/375 [==============================] - 19s 52ms/step - loss: 0.6230 - accuracy: 0.8085 - val_loss: 0.5910 - val_accuracy: 0.8092\n",
      "Epoch 11/20\n",
      "375/375 [==============================] - 20s 54ms/step - loss: 0.5432 - accuracy: 0.8087 - val_loss: 0.4960 - val_accuracy: 0.8093\n",
      "Epoch 12/20\n",
      "375/375 [==============================] - 19s 52ms/step - loss: 0.4805 - accuracy: 0.8087 - val_loss: 0.4688 - val_accuracy: 0.8093\n",
      "Epoch 13/20\n",
      "375/375 [==============================] - 19s 52ms/step - loss: 0.4640 - accuracy: 0.8087 - val_loss: 0.4558 - val_accuracy: 0.8093\n",
      "Epoch 14/20\n",
      "375/375 [==============================] - 19s 52ms/step - loss: 0.4514 - accuracy: 0.8087 - val_loss: 0.4434 - val_accuracy: 0.8093\n",
      "Epoch 15/20\n",
      "375/375 [==============================] - 19s 52ms/step - loss: 0.4389 - accuracy: 0.8087 - val_loss: 0.4310 - val_accuracy: 0.8093\n",
      "Epoch 16/20\n",
      "375/375 [==============================] - 20s 52ms/step - loss: 0.4263 - accuracy: 0.8087 - val_loss: 0.4183 - val_accuracy: 0.8093\n",
      "Epoch 17/20\n",
      "375/375 [==============================] - 21s 56ms/step - loss: 0.4137 - accuracy: 0.8087 - val_loss: 0.4059 - val_accuracy: 0.8093\n",
      "Epoch 18/20\n",
      "375/375 [==============================] - 21s 56ms/step - loss: 0.4016 - accuracy: 0.8087 - val_loss: 0.3944 - val_accuracy: 0.8093\n",
      "Epoch 19/20\n",
      "375/375 [==============================] - 21s 56ms/step - loss: 0.3907 - accuracy: 0.8087 - val_loss: 0.3843 - val_accuracy: 0.8093\n",
      "Epoch 20/20\n",
      "375/375 [==============================] - 21s 57ms/step - loss: 0.3813 - accuracy: 0.8087 - val_loss: 0.3756 - val_accuracy: 0.8093\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 51221 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 54869 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 46020 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 54617 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 49845 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 44257 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 49440 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 51221 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 54869 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 46020 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 54617 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 49845 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 44257 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 49440 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 49552 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 49892 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 49552 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 49892 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10000, 28, 28, 1) (10000, 28, 28, 1)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1440x288 with 20 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "main()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "### 6.3.5 전체 코드"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(60000, 784)\n",
      "(10000, 784)\n",
      "Epoch 1/10\n",
      "235/235 [==============================] - 2s 8ms/step - loss: 0.6938 - accuracy: 0.0010 - val_loss: 0.6938 - val_accuracy: 0.0012\n",
      "Epoch 2/10\n",
      "235/235 [==============================] - 2s 7ms/step - loss: 0.6936 - accuracy: 0.0010 - val_loss: 0.6935 - val_accuracy: 0.0012\n",
      "Epoch 3/10\n",
      "235/235 [==============================] - 2s 7ms/step - loss: 0.6934 - accuracy: 0.0010 - val_loss: 0.6933 - val_accuracy: 0.0012\n",
      "Epoch 4/10\n",
      "235/235 [==============================] - 2s 7ms/step - loss: 0.6932 - accuracy: 0.0010 - val_loss: 0.6931 - val_accuracy: 0.0014\n",
      "Epoch 5/10\n",
      "235/235 [==============================] - 2s 7ms/step - loss: 0.6929 - accuracy: 0.0010 - val_loss: 0.6928 - val_accuracy: 0.0014\n",
      "Epoch 6/10\n",
      "235/235 [==============================] - 2s 7ms/step - loss: 0.6927 - accuracy: 0.0011 - val_loss: 0.6926 - val_accuracy: 0.0014\n",
      "Epoch 7/10\n",
      "235/235 [==============================] - 2s 7ms/step - loss: 0.6925 - accuracy: 0.0011 - val_loss: 0.6924 - val_accuracy: 0.0015\n",
      "Epoch 8/10\n",
      "235/235 [==============================] - 2s 7ms/step - loss: 0.6923 - accuracy: 0.0011 - val_loss: 0.6921 - val_accuracy: 0.0016\n",
      "Epoch 9/10\n",
      "235/235 [==============================] - 2s 7ms/step - loss: 0.6920 - accuracy: 0.0011 - val_loss: 0.6919 - val_accuracy: 0.0016\n",
      "Epoch 10/10\n",
      "235/235 [==============================] - 2s 7ms/step - loss: 0.6918 - accuracy: 0.0011 - val_loss: 0.6917 - val_accuracy: 0.0016\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 54617 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 49845 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 44221 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 44284 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 50640 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 46384 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 47480 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 51221 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 54869 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 46020 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 48320 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 54868 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 52628 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 51060 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 54617 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 49845 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 44221 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 44284 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 50640 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 46384 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 47480 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 51221 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 54869 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 46020 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 48320 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 54868 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 52628 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 51060 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 49552 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 49892 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:238: RuntimeWarning: Glyph 44050 missing from current font.\n",
      "  font.set_text(s, 0.0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 49552 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 49892 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n",
      "/home/sjkim/anaconda3/envs/keras-gpu/lib/python3.8/site-packages/matplotlib/backends/backend_agg.py:201: RuntimeWarning: Glyph 44050 missing from current font.\n",
      "  font.set_text(s, 0, flags=flags)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1440x432 with 30 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#########################################################\n",
    "# Convolutional layer based AE with MNIST, Models/Class\n",
    "#########################################################\n",
    "\n",
    "###########################\n",
    "# AE 모델링\n",
    "###########################\n",
    "from keras import layers, models\n",
    "\n",
    "def Conv2D(filters, kernel_size, padding='same', activation='relu'):\n",
    "    return layers.Conv2D(filters, kernel_size, padding=padding, activation=activation)\n",
    "\n",
    "class AE(models.Model):\n",
    "    def __init__(self, org_shape=(1, 28, 28)):\n",
    "        # Input\n",
    "        original = layers.Input(shape=org_shape)\n",
    "\n",
    "        # encoding-1\n",
    "        x = Conv2D(4, (3, 3))(original)\n",
    "        x = layers.MaxPooling2D((2, 2), padding='same')(x)\n",
    "\n",
    "        # encoding-2\n",
    "        x = Conv2D(8, (3, 3))(x)\n",
    "        x = layers.MaxPooling2D((2, 2), padding='same')(x)\n",
    "\n",
    "        # encoding-3: encoding output: 7x7 pixels\n",
    "        z = Conv2D(1, (7, 7))(x)\n",
    "\n",
    "        # decoding-1\n",
    "        y = Conv2D(16, (3, 3))(z)\n",
    "        y = layers.UpSampling2D((2, 2))(y)\n",
    "\n",
    "        # decoding-2\n",
    "        y = Conv2D(8, (3, 3))(y)\n",
    "        y = layers.UpSampling2D((2, 2))(y)\n",
    "\n",
    "        # decoding-3\n",
    "        y = Conv2D(4, (3, 3))(y)\n",
    "\n",
    "        # decoding & Output\n",
    "        decoded = Conv2D(1, (3, 3), activation='sigmoid')(y)\n",
    "\n",
    "        super().__init__(original, decoded)\n",
    "        self.compile(optimizer='adadelta', loss='binary_crossentropy', metrics=['accuracy'])\n",
    "\n",
    "\n",
    "###########################\n",
    "# 데이타 불러오기\n",
    "###########################\n",
    "from ex4_1_cnn_mnist_cl import DATA\n",
    "\n",
    "\n",
    "###########################\n",
    "# 학습 효과 분석\n",
    "###########################\n",
    "from keraspp.skeras import plot_loss, plot_acc\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "###########################\n",
    "# AE 결과 시각화\n",
    "###########################\n",
    "from keras import backend \n",
    "\n",
    "\n",
    "def show_ae(autoencoder, data):\n",
    "    x_test = data.x_test\n",
    "    decoded_imgs = autoencoder.predict(x_test)\n",
    "    print(decoded_imgs.shape, data.x_test.shape)\n",
    "\n",
    "    if backend.image_data_format() == 'channels_first':\n",
    "        N, n_ch, n_i, n_j = x_test.shape\n",
    "    else:\n",
    "        N, n_i, n_j, n_ch = x_test.shape\n",
    "\n",
    "    x_test = x_test.reshape(N, n_i, n_j)\n",
    "    decoded_imgs = decoded_imgs.reshape(decoded_imgs.shape[0], n_i, n_j)\n",
    "    \n",
    "    n = 10\n",
    "    plt.figure(figsize=(20, 4))\n",
    "    for i in range(n):\n",
    "\n",
    "        ax = plt.subplot(2, n, i + 1)\n",
    "        plt.imshow(x_test[i], cmap='gray')\n",
    "        # plt.gray()\n",
    "        ax.get_xaxis().set_visible(False)\n",
    "        ax.get_yaxis().set_visible(False)\n",
    "\n",
    "        ax = plt.subplot(2, n, i + 1 + n)\n",
    "        plt.imshow(decoded_imgs[i], cmap='gray')\n",
    "        # plt.gray()\n",
    "        ax.get_xaxis().set_visible(False)\n",
    "        ax.get_yaxis().set_visible(False)\n",
    "\n",
    "    plt.show()\n",
    "    \n",
    "\n",
    "###########################\n",
    "# 학습 및 확인\n",
    "###########################\n",
    "def main(epochs=20, batch_size=128):\n",
    "    data = DATA()\n",
    "    autoencoder = AE(data.input_shape)\n",
    "\n",
    "    history = autoencoder.fit(data.x_train, data.x_train,\n",
    "                              epochs=epochs,\n",
    "                              batch_size=batch_size,\n",
    "                              shuffle=True,\n",
    "                              validation_split=0.2)\n",
    "\n",
    "    plot_acc(history, '(a) 정확도 학습 곡선')\n",
    "    plt.show()\n",
    "    plot_loss(history, '(b) 손실 학습 곡선')\n",
    "    plt.show()\n",
    "\n",
    "    show_ae(autoencoder, data)\n",
    "    plt.show()\n",
    "\n",
    "main()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "keras-gpu",
   "language": "python",
   "name": "keras-gpu"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  },
  "toc": {
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": true,
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": "block",
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}