cifar_soft-boundary SVDD .ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 272
    },
    "colab_type": "code",
    "id": "CRXgJFHTEOPI",
    "outputId": "b7a52b5a-88aa-4dd1-eab7-cbfd92e75147"
   },
   "outputs": [],
   "source": [
    "#!pip uninstall tensorflow"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 853
    },
    "colab_type": "code",
    "id": "0PW6ikG-ERZV",
    "outputId": "60475b27-a9d9-4247-b079-4f7e68135ff3"
   },
   "outputs": [],
   "source": [
    "#!pip install tensorflow-gpu==2.0.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "a80IPFtiYlm-",
    "outputId": "7e8cac94-a7ba-4019-a287-9e5a7d56e29e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.0.0\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.compat.v1 import enable_eager_execution\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import os\n",
    "import tensorflow_probability as tfp\n",
    "from tensorflow.keras import layers\n",
    "import time\n",
    "import numpy as np\n",
    "from pathlib import Path\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib import gridspec\n",
    "import matplotlib.patches as mpatches\n",
    "from sklearn.metrics import roc_auc_score\n",
    "import random\n",
    "from sklearn.metrics import confusion_matrix\n",
    "print(tf.__version__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.client.session.Session at 0x27244bb07f0>"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "config = tf.compat.v1.ConfigProto()\n",
    "config.gpu_options.allow_growth = True\n",
    "sess = tf.compat.v1.Session(config=config)\n",
    "config = tf.compat.v1.ConfigProto()\n",
    "config.gpu_options.per_process_gpu_memory_fraction = 0.95\n",
    "tf.compat.v1.Session(config=config)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "B1hkHlXiQ1io"
   },
   "source": [
    "## 參數設定"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "VW-G91MqQ1ip"
   },
   "outputs": [],
   "source": [
    "flatten_size = 32 * 32 * 3\n",
    "img_size = 32\n",
    "img_channel = 3\n",
    "regular = 1e-5\n",
    "\n",
    "z_dim = 128\n",
    "h_dim = [32, 64, 128]\n",
    "objective = 'deep-SVDD'     #'  'soft-boundary'\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "mLtweCA9Q1if"
   },
   "source": [
    "## 匯入資料（MNIST）"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "names = np.array(['airplane',\n",
    "'automobile',\n",
    "'bird',\n",
    "'cat',\n",
    "'deer',\n",
    "'dog',\n",
    "'frog',\n",
    "'horse',\n",
    "'ship',\n",
    "'truck'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "location  = 'E:/咪聽/SVDD/deep SVDD/checkpoints/'\n",
    "location  = 'D:/7107018014/checkpoint/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 85
    },
    "colab_type": "code",
    "id": "F-rxo8dLYlnB",
    "outputId": "a90ed8ac-dbc9-4372-96e3-2269366f9f7c"
   },
   "outputs": [],
   "source": [
    "def load_data(label_num, batch_size = 200, outlier_nu = 0.1):\n",
    "    # 0: normal, 1: outlier\n",
    "    \n",
    "    x_train = np.load('D:/7107018014/cifar10/{}/{}_data{}_batch{}.npy'.format(label_num,'train',label_num, 0))\n",
    "    y_train = np.load('D:/7107018014/cifar10/{}/{}_label{}_batch{}.npy'.format(label_num,'train',label_num, 0))\n",
    "    x_train = np.transpose(x_train, (0, 2, 3, 1))\n",
    "    \n",
    "    \n",
    "    for i in np.arange(2):\n",
    "        data = np.load('D:/7107018014/cifar10/{}/{}_data{}_batch{}.npy'.format(label_num,'test',label_num, i))\n",
    "        label = np.load('D:/7107018014/cifar10/{}/{}_label{}_batch{}.npy'.format(label_num,'test',label_num, i))\n",
    "        \n",
    "        data = np.transpose(data, (0, 2, 3, 1))\n",
    "        try:\n",
    "            x_test = np.vstack((x_test, data))\n",
    "            y_test = np.hstack((y_test, label))\n",
    "        except:\n",
    "            x_test = data\n",
    "            y_test = label\n",
    "    ## outlier data\n",
    "    for index in np.delete(np.arange(10), label_num):\n",
    "        \n",
    "        out_data = np.load('D:/7107018014/cifar10/{}/{}_data{}_batch{}.npy'.format(index,'train',index, 0))\n",
    "        out_label = np.load('D:/7107018014/cifar10/{}/{}_label{}_batch{}.npy'.format(index,'train',index, 0))\n",
    "\n",
    "        out_data = np.transpose(out_data, (0, 2, 3, 1))\n",
    "        try:\n",
    "            out_datas_ = np.vstack((out_datas_, out_data))\n",
    "            out_labels_ = np.hstack((out_labels_, out_label))\n",
    "        except:\n",
    "            out_datas_ = out_data\n",
    "            out_labels_ = out_label\n",
    "        \n",
    "    ### 製造outlier\n",
    "    random.seed (1120)\n",
    "    normal_n = len(x_train)\n",
    "    outlier_n = len(out_datas_) ####\n",
    "    indices = np.random.choice(outlier_n, int(normal_n * outlier_nu), replace=False)\n",
    "    x_train = np.vstack( (x_train[y_train == label_num], out_datas_[indices]))\n",
    "    y_train = np.hstack( (y_train[y_train == label_num], out_labels_[indices]))\n",
    "    \n",
    "    \n",
    "    # Flatten the dataset\n",
    "    x_train = x_train.reshape((-1, flatten_size))\n",
    "    x_test = x_test.reshape((-1, flatten_size))\n",
    "\n",
    "    train_buf = x_train.shape[0]\n",
    "    train_dataset = tf.data.Dataset.from_tensor_slices(x_train)\n",
    "    train_dataset = train_dataset.shuffle(buffer_size=train_buf)\n",
    "    train_dataset = train_dataset.batch(batch_size)\n",
    "    \n",
    "    return train_dataset, x_train, x_test, y_train, y_test"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "uwrm-Sw7Q1ir"
   },
   "source": [
    "## 建構模型"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "7fNOIyn8Q1iv"
   },
   "source": [
    "## Prtrain"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Autoencoder():\n",
    "    \n",
    "    def __init__(self, label_num, img_size = img_size, z_dim = z_dim, h_dim = h_dim, img_channel = img_channel):\n",
    "        \n",
    "        self.label_num = label_num\n",
    "        self.flatten_size = img_size * img_size * img_channel\n",
    "        self.z_dim = z_dim\n",
    "        self.h_dim = h_dim\n",
    "        self.encoder = self.make_encoder()\n",
    "        self.decoder = self.make_decoder()\n",
    "        self.ae_optimizer = tf.keras.optimizers.Adam(lr = 0.001, epsilon = 1e-6)\n",
    "        \n",
    "\n",
    "    def make_encoder(self):\n",
    "\n",
    "        inputs = tf.keras.Input(shape=(self.flatten_size,), name='inputs')\n",
    "        x = tf.reshape(inputs, shape = (-1, 32, 32, 3))\n",
    "        x = tf.keras.layers.Conv2D(self.h_dim[0], (5, 5), activation='linear', padding='same',  use_bias = False)(x)\n",
    "        x = tf.keras.layers.BatchNormalization(epsilon = 1e-04)(x)\n",
    "        x = tf.keras.layers.ELU()(x)\n",
    "        x = tf.keras.layers.MaxPool2D((2, 2), padding='same')(x)\n",
    "\n",
    "        x = tf.keras.layers.Conv2D(self.h_dim[1], (5, 5), activation='linear', padding='same',  use_bias = False)(x)\n",
    "        x = tf.keras.layers.BatchNormalization(epsilon = 1e-04)(x)\n",
    "        x = tf.keras.layers.ELU()(x)\n",
    "        x = tf.keras.layers.MaxPool2D((2, 2), padding='same')(x)\n",
    "        \n",
    "        x = tf.keras.layers.Conv2D(self.h_dim[2], (5, 5), activation='linear', padding='same',  use_bias = False)(x)\n",
    "        x = tf.keras.layers.BatchNormalization(epsilon = 1e-04)(x)\n",
    "        x = tf.keras.layers.ELU()(x)\n",
    "        x = tf.keras.layers.MaxPool2D((2, 2), padding='same')(x)\n",
    "\n",
    "        x = tf.keras.layers.Flatten(name = 'Flatten')(x)\n",
    "        x = tf.keras.layers.BatchNormalization(epsilon = 1e-04)(x)\n",
    "        x = tf.keras.layers.Dense(self.z_dim, use_bias = False, name = 'dense')(x)\n",
    "        \n",
    "        \n",
    "        model = tf.keras.Model(inputs=inputs, outputs=x)\n",
    "\n",
    "        return model    \n",
    "   \n",
    "    def make_decoder(self):\n",
    "\n",
    "        inputs = tf.keras.Input(shape=(self.z_dim,), name='inputs')\n",
    "        x = tf.reshape(inputs, shape = (-1, 4, 4, 8))\n",
    "        x = tf.keras.layers.LeakyReLU()(x)\n",
    "        #x = tf.keras.layers.UpSampling2D((2, 2))(x)    \n",
    "\n",
    "        x = tf.keras.layers.Conv2DTranspose(self.h_dim[2], (5, 5), padding='same',  use_bias = False)(x)\n",
    "        x = tf.keras.layers.BatchNormalization()(x)\n",
    "        x = tf.keras.layers.ELU()(x)\n",
    "        x = tf.keras.layers.UpSampling2D((2, 2))(x)   \n",
    "\n",
    "        x = tf.keras.layers.Conv2DTranspose(self.h_dim[1], (5, 5), padding='same',  use_bias = False)(x)\n",
    "        x = tf.keras.layers.BatchNormalization()(x)\n",
    "        x = tf.keras.layers.ELU()(x)\n",
    "        x = tf.keras.layers.UpSampling2D((2, 2))(x)\n",
    "        \n",
    "        x = tf.keras.layers.Conv2DTranspose(self.h_dim[0], (5, 5), padding='same',  use_bias = False)(x)\n",
    "        x = tf.keras.layers.BatchNormalization()(x)\n",
    "        x = tf.keras.layers.ELU()(x)\n",
    "        x = tf.keras.layers.UpSampling2D((2, 2))(x) \n",
    "        \n",
    "        x = tf.keras.layers.Conv2DTranspose(3, (5, 5), activation='sigmoid', padding='same',  use_bias = False)(x)\n",
    "        #x = tf.reshape(x, shape = (-1, self.flatten_size))\n",
    "        x = tf.keras.layers.Flatten(name = 'Flatten')(x)\n",
    "        model = tf.keras.Model(inputs=inputs, outputs=x)\n",
    "\n",
    "        return model   \n",
    "    \n",
    "    @tf.function\n",
    "    def train_ae_model(self, batch_x):\n",
    "        with tf.GradientTape() as ae_tape:\n",
    "            ae_output = self.decoder(self.encoder(batch_x, training=True), training=True)\n",
    "            loss = tf.reduce_mean((ae_output - batch_x) ** 2)\n",
    "\n",
    "        grads = ae_tape.gradient(loss, self.encoder.trainable_variables + self.decoder.trainable_variables)\n",
    "        self.ae_optimizer.apply_gradients(zip(grads, self.encoder.trainable_variables + self.decoder.trainable_variables))\n",
    "\n",
    "        return loss\n",
    "    \n",
    "    def train_ae( self, n_ae_epochs, train_dataset, ae_lr_milestone=250 ):\n",
    "        #loss_list = []\n",
    "        for epoch in range(n_ae_epochs):\n",
    "            epoch_ae_loss_avg = tf.metrics.Mean()\n",
    "            \n",
    "            if (epoch+1) % ae_lr_milestone == 0:\n",
    "                self.ae_optimizer.lr = 0.1 * self.ae_optimizer.lr\n",
    "\n",
    "            for batch, (batch_x) in enumerate(train_dataset):\n",
    "\n",
    "                ae_loss = self.train_ae_model(batch_x)\n",
    "                #epoch_ae_loss_avg(ae_loss)\n",
    "            #loss_list.append(epoch_ae_loss_avg.result())\n",
    "        \n",
    "        self.encoder.save_weights('D:/7107018014/checkpoint/oneclassAE_label_{}' .format(self.label_num))\n",
    "        #print('LOSS: {:.4f} '.format(epoch_ae_loss_avg.result()))\n",
    "        #plt.title('label = {}'.format(self.label_num))\n",
    "        #plt.plot(loss_list)\n",
    "        #plt.show()\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "ae = Autoencoder('airplane')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "inputs (InputLayer)          [(None, 3072)]            0         \n",
      "_________________________________________________________________\n",
      "tf_op_layer_Reshape (TensorF [(None, 32, 32, 3)]       0         \n",
      "_________________________________________________________________\n",
      "conv2d (Conv2D)              (None, 32, 32, 32)        2400      \n",
      "_________________________________________________________________\n",
      "batch_normalization (BatchNo (None, 32, 32, 32)        128       \n",
      "_________________________________________________________________\n",
      "elu (ELU)                    (None, 32, 32, 32)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 16, 16, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 16, 16, 64)        51200     \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 16, 16, 64)        256       \n",
      "_________________________________________________________________\n",
      "elu_1 (ELU)                  (None, 16, 16, 64)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 8, 8, 64)          0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 8, 8, 128)         204800    \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 8, 8, 128)         512       \n",
      "_________________________________________________________________\n",
      "elu_2 (ELU)                  (None, 8, 8, 128)         0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 4, 4, 128)         0         \n",
      "_________________________________________________________________\n",
      "Flatten (Flatten)            (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_3 (Batch (None, 2048)              8192      \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 128)               262144    \n",
      "=================================================================\n",
      "Total params: 529,632\n",
      "Trainable params: 525,088\n",
      "Non-trainable params: 4,544\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "encoder = ae.encoder\n",
    "encoder.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "inputs (InputLayer)          [(None, 128)]             0         \n",
      "_________________________________________________________________\n",
      "tf_op_layer_Reshape_1 (Tenso [(None, 4, 4, 8)]         0         \n",
      "_________________________________________________________________\n",
      "leaky_re_lu (LeakyReLU)      (None, 4, 4, 8)           0         \n",
      "_________________________________________________________________\n",
      "conv2d_transpose (Conv2DTran (None, 4, 4, 128)         25600     \n",
      "_________________________________________________________________\n",
      "batch_normalization_4 (Batch (None, 4, 4, 128)         512       \n",
      "_________________________________________________________________\n",
      "elu_3 (ELU)                  (None, 4, 4, 128)         0         \n",
      "_________________________________________________________________\n",
      "up_sampling2d (UpSampling2D) (None, 8, 8, 128)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_1 (Conv2DTr (None, 8, 8, 64)          204800    \n",
      "_________________________________________________________________\n",
      "batch_normalization_5 (Batch (None, 8, 8, 64)          256       \n",
      "_________________________________________________________________\n",
      "elu_4 (ELU)                  (None, 8, 8, 64)          0         \n",
      "_________________________________________________________________\n",
      "up_sampling2d_1 (UpSampling2 (None, 16, 16, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_2 (Conv2DTr (None, 16, 16, 32)        51200     \n",
      "_________________________________________________________________\n",
      "batch_normalization_6 (Batch (None, 16, 16, 32)        128       \n",
      "_________________________________________________________________\n",
      "elu_5 (ELU)                  (None, 16, 16, 32)        0         \n",
      "_________________________________________________________________\n",
      "up_sampling2d_2 (UpSampling2 (None, 32, 32, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_3 (Conv2DTr (None, 32, 32, 3)         2400      \n",
      "_________________________________________________________________\n",
      "Flatten (Flatten)            (None, 3072)              0         \n",
      "=================================================================\n",
      "Total params: 284,896\n",
      "Trainable params: 284,448\n",
      "Non-trainable params: 448\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "decoder = ae.decoder\n",
    "decoder.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## result "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "j-iGsWCOQ1i8"
   },
   "outputs": [],
   "source": [
    "def threshold_img(data, dists):\n",
    "    dict_score = {}\n",
    "    for i in range(len(dists)):\n",
    "        dict_score[float(dists[i])] = i\n",
    "    sort_score = [(k, dict_score[k]) for k in sorted(dict_score.keys())]\n",
    "    \n",
    "    n_digits = 20  # how many digits we will display\n",
    "    fig = plt.figure(figsize=(20, 4))\n",
    "    for i in range(1, n_digits):\n",
    "        # display normal\n",
    "        ax = plt.subplot(2, n_digits, i + 1)\n",
    "        plt.imshow(data[sort_score[i][1]].reshape(img_size, img_size, img_channel), cmap = plt.cm.gray)\n",
    "        ax.get_xaxis().set_visible(False)\n",
    "        ax.get_yaxis().set_visible(False)\n",
    "\n",
    "        # display anomaly\n",
    "        ax = plt.subplot(2, n_digits, i + 1 + n_digits)\n",
    "        plt.imshow(data[sort_score[-i][1]].reshape(img_size, img_size, img_channel), cmap = plt.cm.gray)\n",
    "        ax.get_xaxis().set_visible(False)\n",
    "        ax.get_yaxis().set_visible(False)\n",
    "\n",
    "    plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def threshold_uncertain_img(data, dists, r):\n",
    "    dist2radius = (dists - r**2)\n",
    "    normal = dist2radius[dist2radius<0]\n",
    "    anomaly = dist2radius[dist2radius>0]\n",
    "    normal_score = {}\n",
    "    anomaly_score = {}\n",
    "    for i in range(len(normal)):\n",
    "        normal_score[float(normal[i])] = i\n",
    "    for i in range(len(anomaly)):\n",
    "        anomaly_score[float(anomaly[i])] = i\n",
    "    normalsort_score = [(k, normal_score[k]) for k in sorted(normal_score.keys())]\n",
    "    anomalysort_score = [(k, anomaly_score[k]) for k in sorted(anomaly_score.keys())]\n",
    "    \n",
    "    n_digits = min([len(normal),len(anomaly),20])#20  # how many digits we will display\n",
    "    fig = plt.figure(figsize=(20, 4))\n",
    "    #print('normal:', len(normal), 'anomaly', len(anomaly))\n",
    "    for i in range(1, n_digits):\n",
    "        # display normal\n",
    "        \n",
    "        ax = plt.subplot(2, n_digits, i + 1)\n",
    "        plt.imshow(data[normalsort_score[-i][1]].reshape(img_size, img_size, img_channel), cmap = plt.cm.gray)\n",
    "        ax.get_xaxis().set_visible(False)\n",
    "        ax.get_yaxis().set_visible(False)\n",
    "\n",
    "        # display anomaly\n",
    "        ax = plt.subplot(2, n_digits, i + 1 + n_digits)\n",
    "        plt.imshow(data[anomalysort_score[i][1]].reshape(img_size, img_size, img_channel), cmap = plt.cm.gray)\n",
    "        ax.get_xaxis().set_visible(False)\n",
    "        ax.get_yaxis().set_visible(False)\n",
    "\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "UKX2VlDnQ1i5"
   },
   "outputs": [],
   "source": [
    "def distance_plot(label_num, dist, data, radius_end, objective, train = True, label = None,  nu = 0.1):\n",
    "    treshold = radius_end ** 2\n",
    "    if train:\n",
    "        plt.title(str(objective) + 'label = {} train data'.format(label_num))\n",
    "        if objective == 'soft-boundary':\n",
    "            treshold = 0\n",
    "            plt.hist(dis, rwidth=0.3, color='b' ,bins = 100)\n",
    "            plt.axvline(treshold,  color='r')\n",
    "        else:\n",
    "            score = dist\n",
    "            plt.hist(score, rwidth=0.3, color='b' ,bins = 100)\n",
    "            plt.axvline(float(treshold),  color='g')\n",
    "        plt.show()       \n",
    "    \n",
    "    else:\n",
    "        plt.title(str(objective) + 'label = {} test data'.format(label_num))\n",
    "        if objective == 'soft-boundary':\n",
    "            plt.hist(dis[label!=label_num], rwidth=0.3, color='r' ,bins = 100,label = 'anomaly')\n",
    "            plt.hist(dis[label==label_num], rwidth = 0.3, color = 'b',label = 'normal', bins = 100)\n",
    "            plt.axvline(treshold,  color='g')\n",
    "            plt.legend()\n",
    "            plt.show()\n",
    "            plt.hist(dist[label==label_num], rwidth=0.3, color='b' ,bins = 100,label = 'normal')\n",
    "            plt.axvline(treshold, color='g')\n",
    "            plt.legend()\n",
    "            plt.show()\n",
    "            plt.hist(dist[label!=label_num], rwidth=0.3, color='r' ,bins = 100,label = 'anomaly')\n",
    "            plt.axvline(treshold, color='g')\n",
    "        else:\n",
    "            score = dist\n",
    "            plt.hist(score[label!=label_num], rwidth=0.3, color='r' ,bins = 100,label = 'anomaly')\n",
    "            plt.hist(score[label==label_num], rwidth=0.3, color='b' ,bins = 100,label = 'normal')\n",
    "            plt.axvline(float(treshold), color='g')\n",
    "            plt.legend()\n",
    "            plt.show()\n",
    "            plt.hist(score[label==label_num], rwidth=0.3, color='b' ,bins = 100,label = 'normal')\n",
    "            plt.axvline(float(treshold), color='g')\n",
    "            plt.legend()\n",
    "            plt.show()\n",
    "            plt.hist(score[label!=label_num], rwidth=0.3, color='r' ,bins = 100,label = 'anomaly')\n",
    "            plt.axvline(float(treshold), color='g')\n",
    "        plt.legend()\n",
    "        plt.show()\n",
    "    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def false_img(data, labels_,  dists, r, label_num):\n",
    "    dist2radius = (dists - r**2)\n",
    "    dist_n = dist2radius[labels_ != label_num]\n",
    "    dist_p = dist2radius[labels_ == label_num]\n",
    "    data_n = data[labels_ != label_num]\n",
    "    data_p = data[labels_ == label_num]\n",
    "    fn = dist_n[dist_n<0]\n",
    "    fp = dist_p[dist_p>0]\n",
    "    fn_score = {}\n",
    "    fp_score = {}\n",
    "    for i in range(len(fn)):\n",
    "        fn_score[float(fn[i])] = i\n",
    "    for i in range(len(fp)):\n",
    "        fp_score[float(fp[i])] = i\n",
    "        \n",
    "    fn_score_sort = [(k, fn_score[k]) for k in sorted(fn_score.keys())]\n",
    "    fp_score_sort = [(k, fp_score[k]) for k in sorted(fp_score.keys())]\n",
    "    \n",
    "    n_digits = min([len(fn),len(fp),20])#20  # how many digits we will display\n",
    "    fig = plt.figure(figsize=(20, 4))\n",
    "    #print('normal:', len(normal), 'anomaly', len(anomaly))\n",
    "    for i in range(1, n_digits):\n",
    "        # display fn\n",
    "        ax = plt.subplot(2, n_digits, i + 1)\n",
    "        plt.imshow(data_n[fn_score_sort[-i][1]].reshape(img_size, img_size, img_channel), cmap = plt.cm.gray)\n",
    "        ax.get_xaxis().set_visible(False)\n",
    "        ax.get_yaxis().set_visible(False)\n",
    "\n",
    "        # display fp\n",
    "        ax = plt.subplot(2, n_digits, i + 1 + n_digits)\n",
    "        plt.imshow(data_p[fp_score_sort[i][1]].reshape(img_size, img_size, img_channel), cmap = plt.cm.gray)\n",
    "        ax.get_xaxis().set_visible(False)\n",
    "        ax.get_yaxis().set_visible(False)\n",
    "\n",
    "    plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "GbTlov0uQ1it"
   },
   "outputs": [],
   "source": [
    "class train_SVDD():\n",
    "    \n",
    "    def __init__(self, label_num , objective , encoder, train_dataset, nu = 0.1, z_dim = z_dim):\n",
    "        names = np.array(['airplane',\n",
    "        'automobile',\n",
    "        'bird',\n",
    "        'cat',\n",
    "        'deer',\n",
    "        'dog',\n",
    "        'frog',\n",
    "        'horse',\n",
    "        'ship',\n",
    "        'truck'])\n",
    "        \n",
    "        self.label_num = label_num\n",
    "        self.label_name = names[label_num]\n",
    "        self.objective = objective\n",
    "        self.nu = nu\n",
    "        self.z_dim = z_dim\n",
    "        self.encoder = encoder\n",
    "        self.label_num = label_num\n",
    "        self.optimizer = tf.keras.optimizers.Adam(lr = 0.0001, epsilon = 1e-6)\n",
    "        self.train_dataset = train_dataset\n",
    "        self.center = self.init_center_c( eps = 0.1)\n",
    "        \n",
    "        \n",
    "    def init_center_c(self, eps = 0.1):\n",
    "        \n",
    "        self.encoder.load_weights('D:/7107018014/checkpoint/oneclassAE_label_{}' .format(self.label_num))\n",
    "        positive = tf.fill([self.z_dim], eps)\n",
    "        negative = tf.fill([self.z_dim], -eps)\n",
    "        c = tf.zeros(shape = [self.z_dim], dtype = tf.float32, name = 'c')\n",
    "        \n",
    "        for batch, (batch_x) in enumerate(self.train_dataset):\n",
    "            output = self.encoder(batch_x, training=False)            \n",
    "            if batch ==0:\n",
    "                modeloutput = output\n",
    "            else:\n",
    "                modeloutput = tf.concat([modeloutput, output], 0)\n",
    "        \n",
    "        c = tf.reduce_mean(modeloutput, axis = 0, name = 'outputs')\n",
    "        #c = tf.reduce_mean( self.encoder(data, training = False), axis = 0, name = 'outputs') \n",
    "\n",
    "        c = tf.where(tf.logical_and(tf.less(c, 0), tf.greater(c, -eps)), negative, c)\n",
    "        c = tf.where(tf.logical_and(tf.greater(c, 0), tf.less(c, eps)), positive, c)\n",
    "        \n",
    "        return c\n",
    "        \n",
    "    def get_radius(self,dist):\n",
    "        \"\"\"Optimally solve for radius R via the (1-nu)-quantile of distances.\"\"\"\n",
    "        return tfp.stats.percentile(tf.sqrt(dist), (1-self.nu)*100)\n",
    "\n",
    "\n",
    "    def loss(self, objective, dist, R = None):\n",
    "\n",
    "        if objective == 'soft-boundary':\n",
    "            scores = dist - R**2\n",
    "            avg_loss = R**2 + (1/nu) * tf.reduce_mean(tf.maximum(tf.zeros_like(scores), scores))\n",
    "\n",
    "        if objective=='deep-SVDD':\n",
    "            avg_loss = tf.reduce_mean(dist)\n",
    "        return avg_loss\n",
    "\n",
    "    @tf.function\n",
    "    def train_model(self, batch_x, center, objective, R = None):\n",
    "        with tf.GradientTape() as tape:\n",
    "            output = self.encoder(batch_x, training=True)\n",
    "            dist = tf.reduce_sum(tf.math.square(output - self.center) , axis=1)\n",
    "            output_loss = self.loss(self.objective, dist, R)\n",
    "\n",
    "        grads = tape.gradient(output_loss, self.encoder.trainable_variables)\n",
    "        self.optimizer.apply_gradients(zip(grads, self.encoder.trainable_variables))\n",
    "\n",
    "        return output_loss, dist\n",
    "    def train(self, x_train, n_epochs = 150, warm_up_n_epochs = 10, lr_milestone = 50):\n",
    "        \n",
    "        if self.objective == 'soft-boundary':\n",
    "            self.R = tf.zeros(shape = [1], dtype = tf.float32, name = 'R')\n",
    "        else:\n",
    "            self.R = None\n",
    "        loss_list = []\n",
    "        \n",
    "        for epoch in range(n_epochs):\n",
    "            epoch_loss_avg = tf.metrics.Mean()\n",
    "            \n",
    "            if (epoch+1) % lr_milestone == 0:\n",
    "                self.optimizer.lr = 0.1 * self.optimizer.lr\n",
    "            \n",
    "            for batch, (batch_x) in enumerate(self.train_dataset):\n",
    "                loss_, dist = self.train_model(batch_x, self.center, self.objective, self.R )\n",
    "                epoch_loss_avg(loss_)\n",
    "                if (self.objective == 'soft-boundary') and (epoch >= warm_up_n_epochs):\n",
    "                    self.R = self.get_radius(dist)\n",
    "            loss_list.append(loss_)\n",
    "        self.encoder.save_weights('D:/7107018014/checkpoint/{}_label_{}_outlier_{}' .format\n",
    "                                  (self.objective, self.label_name, self.nu))\n",
    "        np.save('D:/7107018014/checkpoint/radius_{}_label_{}_outlier_{}' .format\n",
    "                                  (self.objective, self.label_name,self.nu), self.R.numpy() )\n",
    "        plt.title('label = {}'.format(self.label_name))\n",
    "        plt.plot(loss_list)\n",
    "        plt.show()\n",
    "        \n",
    "    def distance(self, data, label, load_model = True):\n",
    "        if load_model:\n",
    "            self.encoder.load_weights('D:/7107018014/checkpoint/{}_label_{}_outlier_{}' .format\n",
    "                                  (self.objective, self.label_name, self.nu))\n",
    "        batch_size = 200\n",
    "        train_dataset = tf.data.Dataset.from_tensor_slices((data, label))\n",
    "        train_dataset = train_dataset.batch(batch_size)\n",
    "\n",
    "        for batch, (batch_x,batch_y) in enumerate(train_dataset):\n",
    "            output = self.encoder(batch_x, training=False)\n",
    "            dist = tf.reduce_sum(tf.math.square(output - self.center) , axis=1) ## dist**2\n",
    "            \n",
    "            if batch ==0:\n",
    "                modeloutput = dist\n",
    "                labels_ = batch_y\n",
    "                datas = batch_x\n",
    "            else:\n",
    "                modeloutput = tf.concat([modeloutput, dist], 0)\n",
    "                labels_ = tf.concat([labels_, batch_y], 0)\n",
    "                datas = tf.concat([datas, batch_x], 0)\n",
    "                \n",
    "        labels_ = labels_.numpy()\n",
    "        datas = datas.numpy()\n",
    "        \n",
    "        return modeloutput, labels_, datas\n",
    "    \n",
    "    def decide_treshold(self, x_train, y_train):\n",
    "        dis = 0\n",
    "        if self.objective == 'soft-boundary':\n",
    "            radius_end = np.load('D:/7107018014/checkpoint/radius_{}_label_{}_outlier_{}.npy' .format\n",
    "                                  (self.objective, self.label_name, self.nu))\n",
    "        else:\n",
    "            dis, labels_, datas = self.distance(x_train, x_train)\n",
    "            radius_end = self.get_radius(dis).numpy()\n",
    "            \n",
    "        return float(radius_end), dis\n",
    "    \n",
    "    def train_result(self, x_train, y_train, show_img = False, load_model = True):\n",
    "        \n",
    "        radius_end, dis = self.decide_treshold(x_train, x_train)\n",
    "        \n",
    "        if show_img:\n",
    "            #if self.objective == 'soft-boundary':\n",
    "            dis, labels_, datas = self.distance(x_train, y_train)\n",
    "            dis = dis.numpy()\n",
    "            distance_plot(self.label_num, dis, datas, radius_end, self.objective, self.nu, label = labels_)\n",
    "            print(' train label = {} uncertain_img'.format(self.label_name))\n",
    "            threshold_uncertain_img(datas, dis, radius_end)\n",
    "            print('train label = {} distance oder'.format(self.label_name))\n",
    "            threshold_img(datas, dis)\n",
    "            print('train label = {} false_img'.format(self.label_name))\n",
    "            false_img(datas, labels_,  dist, radius_end, self.label_num)\n",
    "            \n",
    "        return radius_end\n",
    "    \n",
    "    def result(self, dist, y_test, radius_end, show_img = False):\n",
    "        \n",
    "        y_true = y_test == self.label_num\n",
    "        scores = dist - radius_end ** 2\n",
    "        y_prob = 1/dist\n",
    "        auc = roc_auc_score(y_true, y_prob)\n",
    "        confusionmatrix = confusion_matrix(y_true, scores<0)\n",
    "        if show_img:\n",
    "            print('AUC:', auc)\n",
    "            print('confusion_matrix:')\n",
    "            print(confusionmatrix)\n",
    "\n",
    "        return auc, confusionmatrix \n",
    "    \n",
    "    def test(self, x_test, y_test, radius_end, show_img = False, load_model = True):\n",
    "        \n",
    "        dist, labels_, datas = self.distance(x_test, y_test, load_model = load_model)\n",
    "        dist = dist.numpy()\n",
    "        test_auc, confusionmatrix = self.result(dist, labels_, radius_end, show_img = show_img)\n",
    "        if show_img:\n",
    "            distance_plot(self.label_num, dist, datas, radius_end, self.objective, train = False, label = labels_,\n",
    "                          nu = self.nu )\n",
    "            print('test label = {} uncertain_img'.format(self.label_name))\n",
    "            threshold_uncertain_img(datas, dist, radius_end)\n",
    "            print('test label = {} distance oder'.format(self.label_name))\n",
    "            threshold_img(datas, dist)\n",
    "            print('test label = {} false_img'.format(self.label_name))\n",
    "            false_img(datas, labels_,  dist, radius_end, self.label_num)\n",
    "            \n",
    "        return test_auc, confusionmatrix \n",
    "    "
   ]
  },
  {
   "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": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5935175555555555\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6275768888888887\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.47063599999999994\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6101826666666667\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5227634999999999\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6447898888888889\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6663095555555556\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAEICAYAAACzliQjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAbuUlEQVR4nO3df5RcZZ3n8fenqjppIIkJEGJMgkEMjsDRyGb4sczOcUX5ta7BOeNumFnNKDNxd2AXPc7sgJwRVw87M44jI2cAJy4ZcRdBVkGzbJTJRHZnR0ckOBgIMdLyKw0xBPkNIXR3ffeP+1T6Vlf1jyTdXZV7P69z6vSt5z5V9dzbXZ/79HN/KSIwM7NyqHS6AWZmNn0c+mZmJeLQNzMrEYe+mVmJOPTNzErEoW9mViIOfetqkh6V9O4J1g1Jbz7Azzng147xnhNuu9l0ceibmZWIQ9+sy0mqdboNVhwOfTtkSDpV0j9Kek7STkl/JWnGiGrnS3pY0tOS/lxSJff6j0jaJulZSXdKeuM0NHu5pC2Snpf0dUm9ufb8nqQ+Sc9IWi/pDbl5IeliSQ8BDylztaSn0nttkXRyqjtT0uclPS5pl6QvSTpsGpbNDkEOfTuUDAEfB44GzgDOAn5/RJ33AyuAU4CVwEcAJF0AfBL4DWA+8P+AmyfyoZKuSxuado8t47z83wDnAscBbwN+J73nu4A/SfMXAo8Bt4x47QXAacCJwNnArwMnAHOBfwv8MtX7s1S+HHgzsAj41ESWzUooIvzwo2sfwKPAu0eZ9zHg9tzzAM7NPf99YFOa/g5wUW5eBXgFeGPutW+egrb/u9zzzwFfStM3AJ/LzZsFDABLc+15V27+u4CfAacDlVy5gJeB43NlZwCPdPp350d3PtzTt0OGpBMk3SHpF5JeAP4rWa8/b0du+jGgMWTyRuCLjR468AxZYC6a4mb/Ijf9Clm4k9r1WGNGRLxE1nPPt2dHbv73gL8CrgV2SVoraQ7Zfy2HA/fmlu27qdyshUPfDiXXAz8FlkXEHLLhGo2osyQ3fSzwZJreAXw0IubmHodFxA/G+9A0Rv7SKI+tB7gsT5JtiBqfcQRwFPBErk7TJXAj4pqI+GfASWTDOX8IPA3sAU7KLdfrImIWZm049O1QMht4AXhJ0q8A/6FNnT+UNE/SEuBS4Oup/EvA5ZJOApD0OkkfmMiHRsS/j4hZozxOOsBl+RrwYUnLJc0k+6/l7oh4tF1lSb8q6TRJPWTDOa8CQxFRB74MXC3pmFR3kaRzDrBdVnAOfTuU/AHwW8CLZEH39TZ1vg3cC9wH/G+ysXMi4nayHZ63pKGhB4DzpqHNbUXEJuCPgW8CO4HjgVVjvGQO2TI/SzYs9Evg82neHwF9wA/Tsv0d8Japabkd6hThm6iYmZWFe/pmZiXi0DczKxGHvplZiTj0zcxKpKsv5HT00UfH0qVLO90MM7NDyr333vt0RLQ9Qa+rQ3/p0qVs3ry5080wMzukSHpstHke3jEzKxGHvplZiTj0zcxKxKFvZlYiDn0zsxJx6JuZlYhD38ysRAoZ+i/vHeQLf7udf3r82U43xcysqxQy9PcO1rnme31s6X++000xM+sqhQz9aiW7g95g3fcKMDPLK2To11LoD9XrHW6JmVl3KWTou6dvZtZeIUN/X09/yKFvZpZXyNB3T9/MrL1Chr4kqhUx5NA3M2tSyNCHrLfvnr6ZWbPChn6tIgaHfPSOmVleYUPfPX0zs1aFDf2easVj+mZmIxQ29N3TNzNrVdjQr1XkM3LNzEYobOi7p29m1qqwoV/zcfpmZi0KG/ru6ZuZtSps6NcqFV97x8xshMKGvnv6ZmatChv6taoY9NE7ZmZNChv6vuCamVmrcUNf0hJJd0naJmmrpEtT+aclPSHpvvQ4P/eayyX1Sdou6Zxc+bmprE/SZVOzSJns2jsOfTOzvNoE6gwCn4iIH0uaDdwraWOad3VEfD5fWdKJwCrgJOANwN9JOiHNvhZ4D9AP3CNpfUQ8OBkLMlKt4sswmJmNNG7oR8ROYGeaflHSNmDRGC9ZCdwSEXuBRyT1AaemeX0R8TCApFtS3akJ/arYOzg0FW9tZnbI2q8xfUlLgXcAd6eiSyRtkbRO0rxUtgjYkXtZfyobrXzkZ6yRtFnS5t27d+9P85p4TN/MrNWEQ1/SLOCbwMci4gXgeuB4YDnZfwJ/0aja5uUxRnlzQcTaiFgRESvmz58/0ea1qPmQTTOzFhMZ00dSD1ng3xQRtwFExK7c/C8Dd6Sn/cCS3MsXA0+m6dHKJ517+mZmrSZy9I6AG4BtEfGFXPnCXLX3Aw+k6fXAKkkzJR0HLAN+BNwDLJN0nKQZZDt710/OYrSqVSru6ZuZjTCRnv6ZwAeB+yXdl8o+CVwoaTnZEM2jwEcBImKrpFvJdtAOAhdHxBCApEuAO4EqsC4itk7isjRxT9/MrNVEjt75B9qPx28Y4zVXAVe1Kd8w1usmU60iBnyPXDOzJj4j18ysRAob+tm1dxz6ZmZ5xQ19n5FrZtaisKFfrYhBj+mbmTUpbOj7dolmZq0KG/pVj+mbmbUobOi7p29m1qqwoV9NZ+RGOPjNzBoKG/q1SnY+mTv7ZmbDChv61RT6vk+umdmwwoZ+o6fvWyaamQ0rbOgP9/Qd+mZmDYUN/Z5qtmg+gsfMbFhhQ99j+mZmrQob+o0xfff0zcyGFTb0q96Ra2bWorChX6u6p29mNlJhQ79ayRbNR++YmQ0rbOh7TN/MrFVhQ99H75iZtSps6PuMXDOzVoUNfZ+Ra2bWqrCh7zNyzcxaFTb0PaZvZtaqsKHvo3fMzFoVNvQ9pm9m1mrc0Je0RNJdkrZJ2irp0lR+pKSNkh5KP+elckm6RlKfpC2STsm91+pU/yFJq6dusaCWTs4a8tE7Zmb7TKSnPwh8IiLeCpwOXCzpROAyYFNELAM2pecA5wHL0mMNcD1kGwngSuA04FTgysaGYiq4p29m1mrc0I+InRHx4zT9IrANWASsBG5M1W4ELkjTK4GvRuaHwFxJC4FzgI0R8UxEPAtsBM6d1KXJ8bV3zMxa7deYvqSlwDuAu4EFEbETsg0DcEyqtgjYkXtZfyobrXzkZ6yRtFnS5t27d+9P85r46B0zs1YTDn1Js4BvAh+LiBfGqtqmLMYoby6IWBsRKyJixfz58yfavBY+I9fMrNWEQl9SD1ng3xQRt6XiXWnYhvTzqVTeDyzJvXwx8OQY5VOi6kM2zcxaTOToHQE3ANsi4gu5WeuBxhE4q4Fv58o/lI7iOR14Pg3/3AmcLWle2oF7diqbEo0zcr0j18xsWG0Cdc4EPgjcL+m+VPZJ4E+BWyVdBDwOfCDN2wCcD/QBrwAfBoiIZyR9Frgn1ftMRDwzKUvRxnBP32P6ZmYN44Z+RPwD7cfjAc5qUz+Ai0d5r3XAuv1p4IGq+ZBNM7MWhT8j12P6ZmbDChv6Nd8u0cysRWFD3z19M7NWhQ19H6dvZtaqsKFfqQjJR++YmeUVNvQh6+0PeHjHzGyfQod+tSKP6ZuZ5RQ69GuVisf0zcxyih36VXlM38wsp9ihX5GP0zczyyl06HtM38ysWaFDv1apuKdvZpZT6NB3T9/MrFmhQ99j+mZmzQod+llP30fvmJk1FD70B3ycvpnZPoUO/ew4fYe+mVlDoUO/6qN3zMyaFDr0ezymb2bWpNChX63I194xM8spdOh7TN/MrFmhQ99j+mZmzQod+jWfkWtm1qTQoV/1GblmZk0KHfo1H71jZtak0KHvo3fMzJqNG/qS1kl6StIDubJPS3pC0n3pcX5u3uWS+iRtl3ROrvzcVNYn6bLJX5RWvuCamVmzifT0vwKc26b86ohYnh4bACSdCKwCTkqvuU5SVVIVuBY4DzgRuDDVnVLVSsU7cs3McmrjVYiIv5e0dILvtxK4JSL2Ao9I6gNOTfP6IuJhAEm3pLoP7neL90NPVQx6TN/MbJ+DGdO/RNKWNPwzL5UtAnbk6vSnstHKW0haI2mzpM27d+8+iOb5JipmZiMdaOhfDxwPLAd2An+RytWmboxR3loYsTYiVkTEivnz5x9g8zIe0zczazbu8E47EbGrMS3py8Ad6Wk/sCRXdTHwZJoerXzKVCsVhnz0jpnZPgfU05e0MPf0/UDjyJ71wCpJMyUdBywDfgTcAyyTdJykGWQ7e9cfeLMnplZ1T9/MLG/cnr6km4F3AkdL6geuBN4paTnZEM2jwEcBImKrpFvJdtAOAhdHxFB6n0uAO4EqsC4itk760ozgMX0zs2YTOXrnwjbFN4xR/yrgqjblG4AN+9W6g5SN6fvoHTOzhsKfkVsPqLu3b2YGFDz0a5XsoCGP65uZZQod+tVKtnge1zczyxQ69HuqjZ6+x/XNzKDgoV9Nwzvu6ZuZZQod+h7TNzNrVujQ95i+mVmzQoe+e/pmZs0KHfr7xvR9/R0zM6DgoV/z0TtmZk0KHfpVD++YmTUpdOjvG9P38I6ZGVDw0PfRO2ZmzQod+h7TNzNrVuzQ9xm5ZmZNCh363pFrZtas0KFf85i+mVmTQoe+e/pmZs0KHfrDY/rekWtmBgUP/aqP0zcza1Lo0B8+ZNOhb2YGRQ99j+mbmTUpdOgPn5HrMX0zMyh46PvaO2ZmzYod+lWfkWtmllfo0Pdx+mZmzcYNfUnrJD0l6YFc2ZGSNkp6KP2cl8ol6RpJfZK2SDol95rVqf5DklZPzeI08xm5ZmbNJtLT/wpw7oiyy4BNEbEM2JSeA5wHLEuPNcD1kG0kgCuB04BTgSsbG4qp5J6+mVmzcUM/Iv4eeGZE8UrgxjR9I3BBrvyrkfkhMFfSQuAcYGNEPBMRzwIbad2QTLrhHbk+esfMDA58TH9BROwESD+PSeWLgB25ev2pbLTyFpLWSNosafPu3bsPsHmZ3p4qAK+8NnRQ72NmVhSTvSNXbcpijPLWwoi1EbEiIlbMnz//oBpTrYgjZlR58dXBg3ofM7OiONDQ35WGbUg/n0rl/cCSXL3FwJNjlE+52b09vLR3YDo+ysys6x1o6K8HGkfgrAa+nSv/UDqK53Tg+TT8cydwtqR5aQfu2alsys3qrbmnb2aW1MarIOlm4J3A0ZL6yY7C+VPgVkkXAY8DH0jVNwDnA33AK8CHASLiGUmfBe5J9T4TESN3Dk+J2b01Xtrr0DczgwmEfkRcOMqss9rUDeDiUd5nHbBuv1o3CWb39vD8Hg/vmJlBwc/IBZg9s8aLrzr0zcygDKHfW+Mlj+mbmQElCX3vyDUzyxQ+9GfN7GHPwBADPivXzKz4oT+7N9tX/bKP4DEzK0/oe4jHzMyhb2ZWKiUI/R4AH7ZpZkYpQt89fTOzhsKH/qyZWej7UgxmZiUIfQ/vmJkNK0HoZz39Fzy8Y2ZW/NCfWavQU5WHd8zMKEHoS2J2b4+Hd8zMKEHoQ7Yz1xddMzMrSej7omtmZhmHvplZiZQi9GfN7OFF78g1MytH6M/p9d2zzMygJKHv4R0zs0wpQn9Wb42X9g6S3bfdzKy8ShH6s3t7GKoHewaGOt0UM7OOKknop4uueYjHzEquFKHfuNKmr79jZmVXitCf4yttmpkBJQn9fcM7PlbfzEruoEJf0qOS7pd0n6TNqexISRslPZR+zkvlknSNpD5JWySdMhkLMBGzfPcsMzNgcnr6/zIilkfEivT8MmBTRCwDNqXnAOcBy9JjDXD9JHz2hPhGKmZmmakY3lkJ3JimbwQuyJV/NTI/BOZKWjgFn9/C98k1M8scbOgH8LeS7pW0JpUtiIidAOnnMal8EbAj99r+VNZE0hpJmyVt3r1790E2L3PEjBoSvLDHPX0zK7faQb7+zIh4UtIxwEZJPx2jrtqUtZwiGxFrgbUAK1asmJRTaKsV8fo5vTzx3KuT8XZmZoesg+rpR8ST6edTwO3AqcCuxrBN+vlUqt4PLMm9fDHw5MF8/v5YMu9wdjz7ynR9nJlZVzrg0Jd0hKTZjWngbOABYD2wOlVbDXw7Ta8HPpSO4jkdeL4xDDQdFs87jCee3TNdH2dm1pUOZnhnAXC7pMb7fC0ivivpHuBWSRcBjwMfSPU3AOcDfcArwIcP4rP32+J5h/Gt+/YwMFSnp1qK0xPMzFoccOhHxMPA29uU/xI4q015ABcf6OcdrMVHHk49YOdzr3LsUYd3qhlmZh1Vmi7v4nmHAdDvcX0zK7HShP6SeVnv3jtzzazMShP6C1/XS7Ui+r0z18xKrDShX6tWeP2cXnY8456+mZVXaUIfYMmRh7mnb2alVqrQX+wTtMys5EoW+oex64W97B30vXLNrJxKFfqNI3h8Zq6ZlVWpQn/4WH2HvpmVU6lCf8mRPlbfzMqtVKG/YE4vPVXx8O6XO90UM7OOKFXoVyvijOOP5s6tvyC7FJCZWbmUKvQBVr79DfQ/u4cfP/5cp5tiZjbtShf6Z5+0gJm1Cv/rJ9N2/xYzs65RutCf3dvDWW89hju2PMngUL3TzTEzm1alC32A9719EU+/9Bo/+PkvO90UM7NpVcrQf+db5jOnt8bffP+RTjfFzGxalTL0e3uq/Md3LeOu7bvZ+OCuTjfHzGzalDL0AX7nzKWcsGAWn16/lT2v+Vo8ZlYOpQ39nmqFz648mSee28NVGx70cftmVgqlDX2A0950FL/7a8fxP374OJ+5w8FvZsVX63QDOu2Kf/VW6gHrvv8IL+wZ5DMrT+KImaVfLWZWUKVPN0n88XvfyuzeGtd87yF+/PizfHHVct62eG6nm2ZmNulKPbzTIImPv+cEvva7p/PqwBC/cd0PuO7/9DFU93CPmRWLQz/njOOP4juX/gvOPmkBn/vuds6++v/yJxu28bNdL3a6aWZmk8KhP8Lcw2dw7W+dwhdXLWfBnF7Wff8RfvP6H7Dzed94xcwOfdMe+pLOlbRdUp+ky6b78ydCEiuXL+Jrv3c6d37s1xkYCj5x60+oe7jHzA5x07ojV1IVuBZ4D9AP3CNpfUQ8OJ3t2B9vmj+LT/3rE7n8tvu54lv385YFs6lWK9QqoloRtYqoVSv01ir0VCsMDNUJYNbMGjNrFV5+bYg9rw1RUXY9/30PDU9X0vtUNPyelVQHoB5BPSAiqFbEjFqFGbUKM6tVqlVREQiRqreVPxo1aN54VZS9VmTvte/5WG/YJSKtm2ql+9tq1g2m++idU4G+iHgYQNItwEqga0MfYNWvLuEff/5Lbv7Rjk43ZVplG4Is/JUrg2wDwb7p5nI1lWvfdL4+jfcdpe7w9qa5znC7xMBQnRdeHWBgKNsYzqxVmJk2vsC+TdvwBi+ano/8v234c5Vb9uFlym9W8q8da4PaPK99eX7OaPWBfRvkRudgIsbbbk9kuy7GrjTRvsHwem/+PYycHi7LrZcx3qflvdrUHW8dj/dZ+Tqj/OpGvG74WW9PlcNmVKlWRETr38hoTlw4h7/+4IoJ1d0f0x36i4B8cvYDp+UrSFoDrAE49thjp69lY5DEF1ct57MXnMxQPRis16nXYbBeZ6geDAzVeXWgzmtDdWakwHl57yB7B+scMbNKb0+VCBiqB0MR2c96UK8Hg6msMV0fUUf7et7ZV2+oHrw2WGfvUJ29A0PUI4iAemT/EYz8Aua/sPl5jcmA9PogRrxXoxc98ksajPjiNZUP1x0ZukGM+iVqVydfng/sRlm1Kl53WA+9tSoDQ3X2Dg6xd7DOa4P1po1Gftnbbrxynzv8mbHvs/Lz8qu3eX22X8+tz8f+fYz2vsHwf3vZ3wbjGjdcJpA941WZyAmN+fXW1Alomhhe1v1ZL01l46zn/XqvprqtW7Xxfvf579fewSFe3pt9V0d2osZy7FGHT6DW/pvu0G+3rM0bzoi1wFqAFStWdM0gupQFjJnZoWy6d+T2A0tyzxcDvoWVmdk0me7QvwdYJuk4STOAVcD6aW6DmVlpTevwTkQMSroEuBOoAusiYut0tsHMrMym/do7EbEB2DDdn2tmZj4j18ysVBz6ZmYl4tA3MysRh76ZWYmom28RKGk38NhBvMXRwNOT1Jyp0u1t7Pb2gds4WdzGydENbXxjRMxvN6OrQ/9gSdocEZN/8YpJ1O1t7Pb2gds4WdzGydHtbfTwjplZiTj0zcxKpOihv7bTDZiAbm9jt7cP3MbJ4jZOjq5uY6HH9M3MrFnRe/pmZpbj0DczK5FChn433nxd0hJJd0naJmmrpEtT+ZGSNkp6KP2c1wVtrUr6J0l3pOfHSbo7tfHr6bLYnWzfXEnfkPTTtD7P6Kb1KOnj6Xf8gKSbJfV2wzqUtE7SU5IeyJW1XW/KXJO+Q1skndKh9v15+j1vkXS7pLm5eZen9m2XdM5Ut2+0Nubm/YGkkHR0ej7t63AiChf6uZuvnwecCFwo6cTOtgqAQeATEfFW4HTg4tSuy4BNEbEM2JSed9qlwLbc8z8Drk5tfBa4qCOtGvZF4LsR8SvA28na2hXrUdIi4D8BKyLiZLJLiK+iO9bhV4BzR5SNtt7OA5alxxrg+g61byNwckS8DfgZcDlA+u6sAk5Kr7kuffc70UYkLQHeAzyeK+7EOhxf7Ls3ajEewBnAnbnnlwOXd7pdbdr5bbI/ku3AwlS2ENje4XYtJvvyvwu4g+wWl08DtXbrtwPtmwM8QjoIIVfeFeuR4ftAH0l26fI7gHO6ZR0CS4EHxltvwF8DF7arN53tGzHv/cBNabrpe012j44zOrEOU9k3yDogjwJHd3IdjvcoXE+f9jdfX9ShtrQlaSnwDuBuYEFE7ARIP4/pXMsA+EvgPwONW28fBTwXEYPpeafX55uA3cDfpCGo/ybpCLpkPUbEE8DnyXp8O4HngXvprnWYN9p668bv0UeA76TprmmfpPcBT0TET0bM6po25hUx9Me9+XonSZoFfBP4WES80On25El6L/BURNybL25TtZPrswacAlwfEe8AXqY7hsQASGPiK4HjgDcAR5D9mz9S1/xNjqKrfu+SriAbIr2pUdSm2rS3T9LhwBXAp9rNblPW8d97EUO/a2++LqmHLPBviojbUvEuSQvT/IXAU51qH3Am8D5JjwK3kA3x/CUwV1LjLmudXp/9QH9E3J2ef4NsI9At6/HdwCMRsTsiBoDbgH9Od63DvNHWW9d8jyStBt4L/HakcRK6p33Hk23gf5K+N4uBH0t6Pd3TxiZFDP2uvPm6JAE3ANsi4gu5WeuB1Wl6NdlYf0dExOURsTgilpKtt+9FxG8DdwG/map1uo2/AHZIeksqOgt4kO5Zj48Dp0s6PP3OG+3rmnU4wmjrbT3woXQEyunA841hoOkk6Vzgj4D3RcQruVnrgVWSZko6jmxn6Y+mu30RcX9EHBMRS9P3ph84Jf2ddsU6bNHpnQpTtKPlfLI9/T8Hruh0e1Kbfo3sX7stwH3pcT7ZmPkm4KH088hOtzW1953AHWn6TWRfqD7gfwIzO9y25cDmtC6/BczrpvUI/Bfgp8ADwH8HZnbDOgRuJtvPMEAWTheNtt7IhiauTd+h+8mORupE+/rIxsUb35kv5epfkdq3HTivU+twxPxHGd6RO+3rcCIPX4bBzKxEiji8Y2Zmo3Dom5mViEPfzKxEHPpmZiXi0DczKxGHvplZiTj0zcxK5P8DiMGkF0THoC4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6527662777777778\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.7711713888888889\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6995227222222222\n"
     ]
    }
   ],
   "source": [
    "objective = 'soft-boundary'\n",
    "i = 0\n",
    "nu = 0.1\n",
    "outlier_nu = 0\n",
    "label_list = []\n",
    "auc_list = []\n",
    "time_list = []\n",
    "confusionmatrix_list = []\n",
    "\n",
    "for label_num in np.arange(10):\n",
    "    ## 資料設定\n",
    "    train_dataset, x_train, x_test, y_train, y_test = load_data(label_num, batch_size = 200,\n",
    "                                                                outlier_nu = outlier_nu)\n",
    "    ## 訓練 autoencoder\n",
    "    pre_time = time.time()\n",
    "    ae = Autoencoder(label_num)\n",
    "    ae.train_ae(350, train_dataset)\n",
    "    encoder = ae.encoder\n",
    "    ## 訓練 SVDD\n",
    "    SVDD = train_SVDD(train_dataset = train_dataset, label_num = label_num , objective = objective, encoder = encoder)\n",
    "    SVDD.train(x_train, n_epochs = 150)\n",
    "    radius_end =SVDD.train_result(x_train, y_train)\n",
    "    test_auc, confusionmatrix  = SVDD.test(x_test, y_test, radius_end)\n",
    "    train_time = time.time() - pre_time\n",
    "    print('auc = ', test_auc)\n",
    "    label_list.append(label_num)\n",
    "    auc_list.append(test_auc)\n",
    "    time_list.append(train_time)\n",
    "    confusionmatrix_list.append(confusionmatrix)\n",
    "    i += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "label =  airplane\n",
      "auc =  0.5935175555555555\n",
      "time =  246.15763521194458\n",
      "[[1487 7513]\n",
      " [ 118  882]]\n",
      "label =  automobile\n",
      "auc =  0.6275768888888887\n",
      "time =  248.14074611663818\n",
      "[[2788 6212]\n",
      " [ 155  845]]\n",
      "label =  bird\n",
      "auc =  0.47063599999999994\n",
      "time =  246.39579486846924\n",
      "[[ 832 8168]\n",
      " [ 109  891]]\n",
      "label =  cat\n",
      "auc =  0.6101826666666667\n",
      "time =  244.83347082138062\n",
      "[[2161 6839]\n",
      " [ 121  879]]\n",
      "label =  deer\n",
      "auc =  0.5227634999999999\n",
      "time =  244.29768633842468\n",
      "[[1755 7245]\n",
      " [ 154  846]]\n",
      "label =  dog\n",
      "auc =  0.6447898888888889\n",
      "time =  249.2629246711731\n",
      "[[2972 6028]\n",
      " [ 158  842]]\n",
      "label =  frog\n",
      "auc =  0.6663095555555556\n",
      "time =  247.43330478668213\n",
      "[[2678 6322]\n",
      " [  92  908]]\n",
      "label =  horse\n",
      "auc =  0.6527662777777778\n",
      "time =  248.0103781223297\n",
      "[[2937 6063]\n",
      " [ 120  880]]\n",
      "label =  ship\n",
      "auc =  0.7711713888888889\n",
      "time =  246.05769515037537\n",
      "[[3187 5813]\n",
      " [  89  911]]\n",
      "label =  truck\n",
      "auc =  0.6995227222222222\n",
      "time =  247.93882060050964\n",
      "[[3980 5020]\n",
      " [ 196  804]]\n"
     ]
    }
   ],
   "source": [
    "for i in range(len(names)):\n",
    "    print('label = ', names[i])\n",
    "    print('auc = ', auc_list[i])\n",
    "    print('time = ', time_list[i])\n",
    "    print(confusionmatrix_list[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "objective = 'soft-boundary'\n",
    "i = 0\n",
    "nu = 0.1\n",
    "outlier_nu = 0\n",
    "\n",
    "\n",
    "for label_num in np.arange(10):\n",
    "    ## 資料設定\n",
    "    train_dataset, x_train, x_test, y_train, y_test = load_data(label_num, batch_size = 200,\n",
    "                                                                outlier_nu = outlier_nu)\n",
    "    ## 訓練 autoencoder\n",
    "    pre_time = time.time()\n",
    "    ae = Autoencoder(label_num)\n",
    "    encoder = ae.encoder\n",
    "    ## 訓練 SVDD\n",
    "    SVDD = train_SVDD(train_dataset = train_dataset, label_num = label_num , objective = objective, encoder = encoder)\n",
    "    radius_end =SVDD.train_result(x_train, y_train)\n",
    "    test_auc, confusionmatrix  = SVDD.test(x_test, y_test, radius_end)\n",
    "    train_time = time.time() - pre_time\n",
    "    i += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5889552222222222\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6041779444444445\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.4839379999999999\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5217078333333334\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5485768333333334\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5697901666666666\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAEICAYAAACzliQjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nO3de5RcZZ3u8e+vqvqW7ty7E0ISCJfIdeRiuA0zZxSUAOMSZpbMieOSLGRO1EHFOeOo0XOGo8hR1hxFOTPi4CSKDorI4MAgI+aAjKLDpQMYSEJISEjSuXbIpdPp9KWqfueP/VZ3VVd1upN0uiq7ns9avWrvd79V9e7dXc9++91vVZm7IyIi1SFR7gaIiMjYUeiLiFQRhb6ISBVR6IuIVBGFvohIFVHoi4hUEYW+HFfM7E0ze/cI67qZnX6Ez3PE9z3EY55hZi+Z2X4z++RoPrbISKXK3QCRKvIZ4Gl3v6DcDZHqpZ6+yNg5GVg51EYzS45hW6RKKfTluGVmF5vZf5rZXjPbZmZ/b2a1g6pda2brzWyXmf2dmSXy7v9hM1ttZnvM7AkzO/kYtvUp4F3A35tZp5m9zcy+Z2b3mNnjZnYAeJeZTTSz75tZu5ltNLP/kWuzmSXN7GthXzaY2cfDMJT+Y5cRU+jL8SwD/BXQDFwGXAn85aA6fwLMAy4ErgM+DGBm1wOfB/4UaAF+DfxoJE9qZt8KJ5pSPytK3cfdrwjP8XF3b3L318OmPwfuAMYDzwD/F5gInAr8EXAjcFOo+9+Aa4Dzw/5cP5L2iuRT6Mtxy92Xu/uz7p529zeBfyQKynx3uvtud98EfAP4QCj/CPAVd1/t7mngfwPnj6S37+5/6e6Thvh5+2HuxiPu/ht3zwJ9wH8FFrv7/rBPXwM+FOr+GfBNd29z9z3AVw/zuUQU+nL8CkMkj5nZdjPrIAru5kHVNuctbwRODMsnA9/M9dCB3YABM491uw/RvmaglqidORsZaNOJg+rnL4uMiEJfjmf3AK8Bc919AtFwjQ2qMztv+SRga1jeDHxkUC+9wd1/O9yTmtm3w7h8qZ8hL9QOIf9jbncR9fbz/9s4CdgSlrcBs4bYN5ERUejL8Ww80AF0mtmZwMdK1PkbM5tsZrOBW4Efh/JvA4vN7ByAcAH1hpE8qbt/NIzLl/o550h3xt0zwIPAHWY2Pgw1/Xfgn0OVB4FbzWymmU0CPnukzyXVS6Evx7NPE10I3Q98h4FAz/cIsBx4GfgZsATA3X8K3Ak8EIaGXiW6SFpunwAOAOuJLuz+EFgatn0H+AWwAngJeBxIE13QFhkR05eoiByfzOwa4Nvufsymmkr8qKcvcpwwswYzu9bMUmY2E7gN+Gm52yXHF/X0RY4TZjYO+A/gTOAg0XDVre7eUdaGyXFFoS8iUkU0vCMiUkUq+jM7mpubfc6cOeVuhojIcWX58uW73L2l1LaKDv05c+bQ2tpa7maIiBxXzGzjUNs0vCMiUkUU+iIiVUShLyJSRRT6IiJVRKEvIlJFFPoiIlVEoS8iUkViGfrb9h3k679Yw/r2znI3RUSkosQy9Hd29HD3U+vYsOtAuZsiIlJRYhn6yUT0jXnprD5MTkQk34hC38zeNLNXzOxlM2sNZVPMbJmZrQ23k0O5mdndZrbOzFaY2YV5j7Mw1F9rZguPzS5BKhmFfkahLyJS4HB6+u9y9/PdfV5Y/xzwpLvPBZ4M6xB95dzc8LOI6MurMbMpRF/6cAlwMXBb7kQx2lLq6YuIlHQ0wzvXAfeF5fuA6/PKv++RZ4FJZjYDmA8sc/fd7r4HWAZcfRTPP6RkItqtTDZ7LB5eROS4NdLQd+AXZrbczBaFsunuvg0g3E4L5TOBzXn3bQtlQ5UXMLNFZtZqZq3t7e0j35M8/T39jHr6IiL5RvrRype7+1YzmwYsM7PXDlHXSpT5IcoLC9zvBe4FmDdv3hGltsb0RURKG1FP3923htudRF/EfDGwIwzbEG53huptwOy8u88Cth6ifNRp9o6ISGnDhr6ZNZrZ+NwycBXwKvAokJuBsxB4JCw/CtwYZvFcCuwLwz9PAFeZ2eRwAfeqUDbqUv1j+gp9EZF8IxnemQ781Mxy9X/o7j83sxeAB83sZmATcEOo/zhwLbAO6AJuAnD33WZ2O/BCqPcld989anuSRz19EZHShg19d18PnFei/C3gyhLlDtwyxGMtBZYefjMPT+5CrmbviIgU0jtyRUSqSCxDv7+nrymbIiIFYhn66umLiJQWy9A3M5IJ0+wdEZFBYhn6EPX21dMXESkU29BPJUyzd0REBolt6KunLyJSLLahn9KYvohIkdiGfjKRUE9fRGSQ2IZ+KmGapy8iMkhsQ19j+iIixWIb+qmkZu+IiAwW29BXT19EpFhsQ1+zd0REisU29DV7R0SkWGxDXz19EZFisQ19jemLiBSLbejrs3dERIrFNvSTCSOtN2eJiBSIbehH8/QV+iIi+WIb+pq9IyJSLLahr9k7IiLFYhv6mr0jIlIstqGv2TsiIsViG/rq6YuIFItt6GtMX0SkWGxDP5lIaJ6+iMggsQ199fRFRIrFNvSTSY3pi4gMFtvQ1+wdEZFiIw59M0ua2Utm9lhYP8XMnjOztWb2YzOrDeV1YX1d2D4n7zEWh/I1ZjZ/tHcmn2bviIgUO5ye/q3A6rz1O4G73H0usAe4OZTfDOxx99OBu0I9zOxsYAFwDnA18C0zSx5d84emMX0RkWIjCn0zmwX8MfBPYd2AK4CHQpX7gOvD8nVhnbD9ylD/OuABd+9x9w3AOuDi0diJUvTZOyIixUba0/8G8BkgN0g+Fdjr7umw3gbMDMszgc0AYfu+UL+/vMR9+pnZIjNrNbPW9vb2w9iVQurpi4gUGzb0zey9wE53X55fXKKqD7PtUPcZKHC/193nufu8lpaW4Zo3pGQIfXcFv4hITmoEdS4H3mdm1wL1wASinv8kM0uF3vwsYGuo3wbMBtrMLAVMBHbnlefk32fUpRLROSaTdVLJUucbEZHqM2xP390Xu/ssd59DdCH2KXf/IPBL4P2h2kLgkbD8aFgnbH/Ko+72o8CCMLvnFGAu8Pyo7ckgyRD0GtcXERkwkp7+UD4LPGBmXwZeApaE8iXAD8xsHVEPfwGAu680sweBVUAauMXdM0fx/IeU39MXEZHIYYW+uz8NPB2W11Ni9o27dwM3DHH/O4A7DreRRyKZiP6JUU9fRGRArN+RC+rpi4jki23oJxO5MX19FIOISE5sQ189fRGRYrEN/f6evj5TX0SkX2xDPzc3Xz19EZEBsQ19zd4RESkW29DXmL6ISLHYhr5m74iIFItt6KunLyJSLLahP9DTV+iLiOTENvRT4UKuevoiIgNiG/qapy8iUiy2oa95+iIixWIb+pq9IyJSLLahr9k7IiLFYhv6mr0jIlIstqGv2TsiIsViG/rq6YuIFItt6A+M6etCrohITmxDX/P0RUSKxTb0NU9fRKRYbENfY/oiIsViG/qavSMiUiy2oa+evohIsdiGvmbviIgUi23oq6cvIlIstqHf39PXlE0RkX6xDX319EVEisU29M2MZMI0e0dEJM+woW9m9Wb2vJn9zsxWmtkXQ/kpZvacma01sx+bWW0orwvr68L2OXmPtTiUrzGz+cdqp3KSCVNPX0Qkz0h6+j3AFe5+HnA+cLWZXQrcCdzl7nOBPcDNof7NwB53Px24K9TDzM4GFgDnAFcD3zKz5GjuzGCphGn2johInmFD3yOdYbUm/DhwBfBQKL8PuD4sXxfWCduvNDML5Q+4e4+7bwDWARePyl4MQT19EZFCIxrTN7Okmb0M7ASWAW8Ae909Haq0ATPD8kxgM0DYvg+Yml9e4j75z7XIzFrNrLW9vf3w9yhPSmP6IiIFRhT67p5x9/OBWUS987NKVQu3NsS2ocoHP9e97j7P3ee1tLSMpHlDSiYS6umLiOQ5rNk77r4XeBq4FJhkZqmwaRawNSy3AbMBwvaJwO788hL3OSZSCdM8fRGRPCOZvdNiZpPCcgPwbmA18Evg/aHaQuCRsPxoWCdsf8rdPZQvCLN7TgHmAs+P1o6UojF9EZFCqeGrMAO4L8y0SQAPuvtjZrYKeMDMvgy8BCwJ9ZcAPzCzdUQ9/AUA7r7SzB4EVgFp4BZ3z4zu7hRKJTV7R0Qk37Ch7+4rgAtKlK+nxOwbd+8Gbhjise4A7jj8Zh4Z9fRFRArF9h25oNk7IiKDxTr0NXtHRKRQrENfPX0RkUKxDn2N6YuIFIp16Ouzd0RECsU69JMJI603Z4mI9It16Efz9BX6IiI5sQ59zd4RESkU69DX7B0RkUKxDn3N3hERKRTr0NfsHRGRQrEOffX0RUQKxTr0NaYvIlIo1qGfTCQ0T19EJE+sQ189fRGRQrEO/WRSY/oiIvliHfqavSMiUijWoa/ZOyIihWId+hrTFxEpFOvQ12fviIgUinXoq6cvIlIo1qGfDKHvruAXEYGYh34qYQDq7YuIBLEO/WQyCn2N64uIRGId+urpi4gUinXoJxPR7qmnLyISiXXoq6cvIlIo1qGfTOTG9PVRDCIiEPPQV09fRKRQrEO/v6evz9QXEQFGEPpmNtvMfmlmq81spZndGsqnmNkyM1sbbieHcjOzu81snZmtMLML8x5rYai/1swWHrvdiqSS6umLiOQbSU8/Dfy1u58FXArcYmZnA58DnnT3ucCTYR3gGmBu+FkE3APRSQK4DbgEuBi4LXeiOFY0e0dEpNCwoe/u29z9xbC8H1gNzASuA+4L1e4Drg/L1wHf98izwCQzmwHMB5a5+2533wMsA64e1b0ZRGP6IiKFDmtM38zmABcAzwHT3X0bRCcGYFqoNhPYnHe3tlA2VPng51hkZq1m1tre3n44zSui2TsiIoVGHPpm1gT8C/Apd+84VNUSZX6I8sIC93vdfZ67z2tpaRlp80pST19EpNCIQt/MaogC/353fzgU7wjDNoTbnaG8DZidd/dZwNZDlB8zAz19hb6ICIxs9o4BS4DV7v71vE2PArkZOAuBR/LKbwyzeC4F9oXhnyeAq8xscriAe1UoO2ZS4UKuevoiIpHUCOpcDnwIeMXMXg5lnwe+CjxoZjcDm4AbwrbHgWuBdUAXcBOAu+82s9uBF0K9L7n77lHZiyFonr6ISKFhQ9/dn6H0eDzAlSXqO3DLEI+1FFh6OA08GpqnLyJSqDrekavZOyIiQMxDX7N3REQKxTr0NXtHRKRQrEM/N3tHF3JFRCKxDv2GmiQAXb3pMrdERKQyxDr0x9dHk5P2dyv0RUQg5qHfpNAXESkQ69CvSSZoqEmyv7uv3E0REakIsQ59iIZ41NMXEYlUR+j3qKcvIgJVEfo16umLiARVEPopOhT6IiJAFYT+hPoaXcgVEQliH/q6kCsiMqBKQl89fRERqIrQr6G7L0tfRh+vLCJSBaGvd+WKiORUQejXAGiIR0SEqgh99fRFRHKqJvQ71NMXEYl/6E/oH95RT19EJPahr+EdEZEBsQ/9CbqQKyLSL/ahry9SEREZEPvQ1xepiIgMiH3oQ/ikzYPq6YuIVE3o64tURESqJvT1RSoiIlA1oa8vUhERgRGEvpktNbOdZvZqXtkUM1tmZmvD7eRQbmZ2t5mtM7MVZnZh3n0WhvprzWzhsdmd0vRFKiIikZH09L8HXD2o7HPAk+4+F3gyrANcA8wNP4uAeyA6SQC3AZcAFwO35U4UY0FfpCIiEhk29N39V8DuQcXXAfeF5fuA6/PKv++RZ4FJZjYDmA8sc/fd7r4HWEbxieSY0RepiIhEjnRMf7q7bwMIt9NC+Uxgc169tlA2VPmY0BepiIhERvtCrpUo80OUFz+A2SIzazWz1vb29lFplD5/R0QkcqShvyMM2xBud4byNmB2Xr1ZwNZDlBdx93vdfZ67z2tpaTnC5hXSF6mIiESONPQfBXIzcBYCj+SV3xhm8VwK7AvDP08AV5nZ5HAB96pQNibU0xcRiaSGq2BmPwLeCTSbWRvRLJyvAg+a2c3AJuCGUP1x4FpgHdAF3ATg7rvN7HbghVDvS+4++OLwMaMvUhERiQwb+u7+gSE2XVmirgO3DPE4S4Glh9W6UaIvUhERiVTFO3InNkShv69LPX0RqW5VEfozJtZTm0rwRntnuZsiIlJWVRH6qWSC01uaeG37/nI3RUSkrKoi9AHOPGE8r23vKHczRETKqmpC/4wTxrOjo4e9Xb3lboqISNlUVegDGuIRkapWNaF/5gkTAFij0BeRKlY1oT99Qh0TG2qKevrb9h2kJ50pU6tERMZW1YS+mXHmCeNZk3cxd29XL1d+7T9Y8syGMrZMRGTsVE3oQzSD5/UdnWSz0Qd8PrZiG129Gdbt0Px9EakOVRX6Z5wwgc6eNFv2HgTgX1/aAkBbWBcRibthP3snTnIzeFZt68AdWjfuIZkwtuxR6ItIdaiqnv5ZM8YzpbGW2x5ZyTeefB2A6847ke0d3aT1rVoiUgWqKvTH1aa4/y8uIZ11Hn5xC5eeOoWLTplCJuvs2N9T7uaJiBxzVRX6AGfNmMBDH72Mi+ZM5mPvPJ2ZkxoAaNvdVeaWiYgce1U1pp8zp7mRn3z09wH6P3lziy7mikgVqLqe/mC5nr4u5opINaj60K+vSdLcVKuevohUhaoPfYCZk8cp9EWkKij0gVmTGjS8IyJVQaEPzJzcQNveg/0fzyAiElcKfaKLub3pLLsOaK6+iMSbQh+YNXnoGTz7uvro6k2PdZNERI4JhT7R8A4Uz9VPZ7Jc/63f8DcPrShHs0RERp1Cn4G5+i9u3EtveuAzeP5txVY27DrAr9a067N5RCQWFPrA+PoaTmtpZOlvNvCOLy/jO79aTzbr3PP0G9QmE+zvSbNiy75yN1NE5Kgp9IOfffIPWbJwHvNOnswdj69m4Xef5/UdnSy+9kwAfrtuV5lbKCJy9BT6QX1NkivPms6ShRdx42Un8+u1uzhpyjg+dOnJnHPiBJ5R6ItIDFTlB64dSiJhfPF953DuiRM5bVoTqWSCy09v5nu/eZODvRkaapPH5Hk7uvv4SWsbCy6aTWOdfi0icmyop1+CmfFnF83mHSdPBuDy05vpzWR54c3dAGSyzsHeDJkRvJnL3enuywxb747HVnP7Y6u4a9nrR9d4EZFDGPPQN7OrzWyNma0zs8+N9fMfiYvmTKYmaXz5Z6u47CtPctrnH+esv/05f3DnU/zwuU30DTGzZ317JwvufZZ33L6MZ9YOPTz03Pq3+HHrZpqb6vjub99kzfb9x2pX5Di05JkNzL/rV7y2vWNUHs/d+fXadtr1xUFVydzH7qMHzCwJvA68B2gDXgA+4O6rStWfN2+et7a2jln7DuVj/7ycZ9e/xWWnTeVt08dTl0qybNV2Xty0l+amOv7L3GZmTKpn1dYOtu3rxsx4o72TulSClqY62vYc5NPz38am3V280rYPM2NcbZJzZ07k/63eQW86y4MfuYxr7/41c6c18ce/N4NXt3bQUJNkalMtsyeP4+Sp4+hNZ9l3sI8JDTVMn1BPY12SZMLY351m1/4e0lknlTBSSSOZSPQvp8JywgyzaJ8Gbg0L68mEMb6uBjNo39/D3q4++rJZkmbMmFjPlMZaDvRmONATvWEtmTAmjauhLjUw7NWTzrCzo4fWjbtZuaWDt8+exBVnTqNp0LBVZ0+aN3Z2svdgH919GU6c2MAZJ4ynNjV6fRF3J+tRO3OyWedgX4Z0xpnQkMLMONibYef+bpqb6vqH17JZZ/OeLja+1cXsKeOYM3UcZlbw2Pnr+Q70pPld217c4aQp49iw6wA/X7md2mSCD15yEnOnjx9R+5c8s4HbH1tFKmE01qX47k0XccHsSUM+b6n9f237fp5Zu4szThjP22dN5H8+spJ/+91WpjTW8tU//T3ec/Z0+jKOGaQSVrSPwIifbzS4O1v3dbO+vZM5UxuZNbnhmD+/u3OgN0Mm72/iSHV097FxVxczJzcwpbF22PpvdfawfOMeJjbUcN7sSdTXHP0Qspktd/d5JbeNcehfBvwvd58f1hcDuPtXStWvpNCH4he5u/PUazt55OWtPLNuF/sO9nF6SxOzp4wDnGkT6vnUlXOpSyW56XvP8+KmvYyrTXL+7Emkkgn2dfWyett+ejNZvnvTRbzrjGn86PlNLH74FQCam2pJZ529XX1l2uORa6xNYmaks1m6+wb+80kmjEw4EdUkE/RlstQkE9SmEuw7WLxftakEE+prAMcdHMh6WHYn/681OlFFJ7H+ZXInMyOTzbK/O00664yrTVJfk+Rgb4aDecNtNUmjqS7FnrxjPL4+BQ7d6Qx9mYFnHF+XojaVoCedpSdsq00lGF+XIuNOV0+GVHi8XZ09DB79a6xN0pd1etPZ/veGwMCJKbd3CRs4ObftOcg1557AZ64+k4VLn2fT7i6SCaM+laChNkldKhntv4GRd0IPx+Ngb4btHd0Dxywcq4/80Wn8em07r27p6P8d9f8Owu8n97usSRoTGwpP7Ll2A/2/k/woye1Lrqy4zkDlwXV601k6ewbeBT+1sZbGuhQe/iaKnivv78K98Ll90HPmP1f+/br7Mv1/t+PrUrRMqMMKWjmwMvBcPngTPX3ZguPd3FRHQ22iZLsB0tksOzoG/uOqTSaY2lRLwowrz5rGl647lyNRSaH/fuBqd/+LsP4h4BJ3/3henUXAIoCTTjrpHRs3bhyz9h2NbNbpy2aLXhg53X0ZXt+xnzNOGF/UK97V2dsfAu7Of77xFic3N/aX9aazbN7TxabdXdSnkkxsqKGju48dHd0c7M3Ql3XG16VobqqjJhm9gNNZJ53Nks7klp10JhuFS354hhdJ7s+gL5OlsydD1p2W8XVMaqihJpUgnXG27TvI7gO9NNWlaKxLkTDoyzh7DvSyNwR4MhEFxJTGWs6bNYm505t4efNenl6zk75MFP59mShMTphYz+nTmmhuqqUulWTjW12saNtLR3e6P5wSg0IdouAaeBF5/wvbB72ok2aMr09Rk0xwoCdNdzpDQ02ShtoU42qTpBLGrs5eOrr7OHFiPdMn1LOrs5ft+w6SSBh1qSRzpo7jpKnj2PRWFyu3dpB1py6VpK4mQW0yQXc6w/7uNDUJo6E2RToTnWimTajjwpMnU5dKsOmtLqY21fGHc5s50JPmwdY2Xt+xv+CklQihDdFJLvd7mjm5gU9cMZfaVIL2/T08/GIbnT3p/pNXd1827xgMhHAu8JIGF58ylXee0cKqrR389o23mH/OdC45dSq96Sz3P7eRtzp7qa+J/rvqzUQnpd50llQyOrn0hY5H/hsX808uhetWXKe/yAbVpbguRjJhnDatiVObG1nf3smrWzroDUOolnfH4pNcifJwh0PVM6LORnNTHcmEsXl3F7s6e/MaN3hfbYh9h2QiwaktjZza3MiWvQdZu6NzYPh30DHK3e/UlkYunjOFPV19tL65mz1dvWQdzps1kQ9dNocjUUmhfwMwf1DoX+zunyhVv9J6+iIix4NDhf5YX8htA2bnrc8Cto5xG0REqtZYh/4LwFwzO8XMaoEFwKNj3AYRkao1pu8Ccve0mX0ceAJIAkvdfeVYtkFEpJqN+Vs/3f1x4PGxfl4REdE7ckVEqopCX0Skiij0RUSqiEJfRKSKjOmbsw6XmbUDR/OW3Gagkj8Iv9LbB2rjaFEbR4faODInu3tLqQ0VHfpHy8xah3pXWiWo9PaB2jha1MbRoTYePQ3viIhUEYW+iEgViXvo31vuBgyj0tsHauNoURtHh9p4lGI9pi8iIoXi3tMXEZE8Cn0RkSoSy9CvxC9fN7PZZvZLM1ttZivN7NZQPsXMlpnZ2nA7ucztTJrZS2b2WFg/xcyeC+37cfhI7LIys0lm9pCZvRaO52WVdBzN7K/C7/hVM/uRmdVXwnE0s6VmttPMXs0rK3ncLHJ3eA2tMLMLy9S+vwu/5xVm9lMzm5S3bXFo3xozm3+s2zdUG/O2fdrM3Myaw/qYH8ORiF3oW/Tl6/8AXAOcDXzAzM4ub6sASAN/7e5nAZcCt4R2fQ540t3nAk+G9XK6FVidt34ncFdo3x7g5rK0qtA3gZ+7+5nAeUTtrYjjaGYzgU8C89z9XKKPEF9AZRzH7wFXDyob6rhdA8wNP4uAe8rUvmXAue7+duB1YDFAeO0sAM4J9/lWeO2Xo42Y2WzgPcCmvOJyHMPhuXusfoDLgCfy1hcDi8vdrhLtfIToj2QNMCOUzQDWlLFNs4he+FcAjxF9q+cuIFXq2JapjROADYRJCHnlFXEcgZnAZmAK0UeXPwbMr5TjCMwBXh3uuAH/CHygVL2xbN+gbX8C3B+WC17XRN/RcVk5jmEoe4ioA/Im0FzOYzjcT+x6+gy86HLaQlnFMLM5wAXAc8B0d98GEG6nla9lfAP4DJD7BuypwF53T4f1SjiWpwLtwHfDMNQ/mVkjFXIc3X0L8H+IenzbgH3AcirvOOYMddwq8XX0YeDfw3LFtM/M3gdscfffDdpUMW3MF8fQtxJlFTMv1cyagH8BPuXuHeVuT46ZvRfY6e7L84tLVC33sUwBFwL3uPsFwAHKPyTWL4yJXwecApwINBL9mz9YuY/jcCrqd29mXyAaIr0/V1Si2pi3z8zGAV8A/rbU5hJlZf+9xzH0K/bL182shijw73f3h0PxDjObEbbPAHaWqXmXA+8zszeBB4iGeL4BTDKz3DesVcKxbAPa3P25sP4Q0UmgUo7ju4EN7t7u7n3Aw8DvU3nHMWeo41YxryMzWwi8F/igh3ESKqd9pxGd4H8XXjuzgBfN7AQqp40F4hj6Ffnl62ZmwBJgtbt/PW/To8DCsLyQaKx/zLn7Ynef5e5ziI7ZU+7+QeCXwPvL3b4cd98ObDazM0LRlcAqKuQ4Eg3rXGpm48LvPNe+ijqOeYY6bo8CN4YZKJcC+3LDQGPJzK4GPgu8z9278jY9CiwwszozO4XoYunzY90+d3/F3ae5+5zw2mkDLgx/pxVxDIuU+6LCMbrQci3Rlf43gC+Uuz2hTX9A9K/dCuDl8HMt0bj5k8DacDulAtr6TuCxsHwq0SxoCDsAAACSSURBVItpHfAToK4C2nc+0BqO5b8CkyvpOAJfBF4DXgV+ANRVwnEEfkR0naGPKJxuHuq4EQ1N/EN4Db1CNBupHO1bRzQunnvNfDuv/hdC+9YA15TrGA7a/iYDF3LH/BiO5EcfwyAiUkXiOLwjIiJDUOiLiFQRhb6ISBVR6IuIVBGFvohIFVHoi4hUEYW+iEgV+f8TQWnHaX4o5wAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5914675555555555\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5826312222222222\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.7246950000000001\n"
     ]
    }
   ],
   "source": [
    "objective = 'soft-boundary'\n",
    "i = 0\n",
    "nu = 0.1\n",
    "outlier_nu = 0.1\n",
    "label_list = []\n",
    "auc_list = []\n",
    "time_list = []\n",
    "confusionmatrix_list = []\n",
    "\n",
    "for label_num in np.arange(10):\n",
    "    ## 資料設定\n",
    "    train_dataset, x_train, x_test, y_train, y_test = load_data(label_num, batch_size = 200,\n",
    "                                                                outlier_nu = outlier_nu)\n",
    "    ## 訓練 autoencoder\n",
    "    pre_time = time.time()\n",
    "    ae = Autoencoder(label_num)\n",
    "    ae.train_ae(350, train_dataset)\n",
    "    encoder = ae.encoder\n",
    "    ## 訓練 SVDD\n",
    "    SVDD = train_SVDD(train_dataset = train_dataset, label_num = label_num , objective = objective, encoder = encoder)\n",
    "    SVDD.train(x_train, n_epochs = 150)\n",
    "    radius_end =SVDD.train_result(x_train, y_train)\n",
    "    test_auc, confusionmatrix  = SVDD.test(x_test, y_test, radius_end)\n",
    "    train_time = time.time() - pre_time\n",
    "    print('auc = ', test_auc)\n",
    "    label_list.append(label_num)\n",
    "    auc_list.append(test_auc)\n",
    "    time_list.append(train_time)\n",
    "    confusionmatrix_list.append(confusionmatrix)\n",
    "    i += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "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"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6174309444444445\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5945320000000001\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.4962631111111111\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.510377388888889\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAEICAYAAACzliQjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nO3de5QcZ33m8e9T3XPR1ZI8A8iSbMmsnMR2HNkIY0II7HLxBYINGxI7OeAEssIEAuS2wWFPINl4s0m4hQ0xsYMDJI4NsbloiRMivFlgAxhGYGT5hmVLtiQLe2TZumtmuvu3f1T1TPVMjzSeW4+qn885fab6raquX9dMP/3O21VdigjMzKw9JK0uwMzMZo9D38ysjTj0zczaiEPfzKyNOPTNzNqIQ9/MrI049O2kIWmHpFdOcNmQ9B8muZ1Jr/sstvEpSX88k9swa8ahb2bWRhz6ZgUiqdzqGmxuc+jbSUnShZK+JekZSXsk/aWkzlGLXSbpEUl7Jf25pCS3/lsk3S/paUlfkXTGDNd7vqTvSToo6bNA96j5r5V0d/Z8vinpvNy80yTdLqlf0nZJ78rN+4Ck2yT9vaQDwK/M5POwk59D305WVeA3gR7gxcArgF8ftczrgfXABcDlwFsAJF0B/D7wBqAX+AZwy0Q2KumvsmBudtsyzjqdwBeBvwOWAf8I/Ofc/AuAm4C3AacCfw1slNSVvVH9b+AHwIrseb5H0sW5TVwO3AYsAW6eyPOw9uXQt5NSRGyOiG9HRCUidpAG5ctGLfanEbEvIh4DPgpclbW/DfiTiLg/IirA/wDWTaS3HxG/HhFLxrmdN85qFwEdwEcjYigibgO+m5v/X4C/joi7IqIaEZ8GBrL1Xgj0RsQfRcRgRDwC3AhcmVv/WxHxxYioRcTREz0Ha28e/7OTkqSzgA+T9uTnk/4tbx612M7c9KPAadn0GcBfSPpQ/iFJe9KPzkC5pwG7o/HbDfPbOQO4WtJv5No6s/WqwGmSnsnNK5H+d1KXf55mx+Wevp2srgceANZGxGLS4RqNWmZVbvp04PFseifwtlG99HkR8c0TbVTSJyQdGud27zir7QFWSMrXd3pueidw3ah65kfELdm87aPmLYqIy3Lr+6tybcIc+nayWgQcAA5J+nHg7U2W+V1JSyWtAt4NfDZr/wRwraRzACSdIumNE9loRFwTEQvHuZ0zzmrfAirAuySVJb0BuDA3/0bgGkkvUmqBpNdIWgR8Bzgg6fckzZNUknSupBdOpF6z0Rz6drL6HeCXgIOkofnZJst8iXTI527gn4BPAkTEF4A/BW7NjnjZClw6U4VGxCDph8a/AjwN/CLw+dz8PtJx/b/M5m/LliUiqsDPAeuA7cBe4G+AU2aqXis2+SIqZmbtwz19M7M24tA3M2sjDn0zszbi0DczayNz/uSsnp6eWL16davLMDM7aWzevHlvRPQ2mzfnQ3/16tX09fW1ugwzs5OGpHHPLPfwjplZG3Hom5m1EYe+mVkbceibmbURh76ZWRtx6JuZtRGHvplZGyls6H/szof42g/7W12GmdmcUtjQv/7/Psy/b9vb6jLMzOaUwoZ+KRGVqq8VYGaWV+jQr/kCMWZmDQob+uVEVGq1VpdhZjanFDb0k0RUa+7pm5nlFTb0yw59M7MxChv6iUTFoW9m1qCwoV8uiZpD38ysQWFDv+SevpnZGCcMfUk3SXpS0tZc22cl3Z3ddki6O2tfLelobt4ncuu8QNI9krZJ+pgkzcxTSvmQTTOzsSZyucRPAX8JfKbeEBG/WJ+W9CFgf275hyNiXZPHuR7YAHwbuAO4BPjnZ1/yxPjkLDOzsU7Y04+IrwP7ms3Leuu/ANxyvMeQtBxYHBHfioggfQO54tmXO3ElH71jZjbGVMf0Xwo8EREP5drWSPq+pK9JemnWtgLYlVtmV9bWlKQNkvok9fX3T+5L08qJqHp4x8yswVRD/yoae/l7gNMj4nzgt4B/kLQYaDZ+P24iR8QNEbE+Itb39vZOqjCfnGVmNtZExvSbklQG3gC8oN4WEQPAQDa9WdLDwFmkPfuVudVXAo9PdtsT4ZOzzMzGmkpP/5XAAxExPGwjqVdSKZs+E1gLPBIRe4CDki7KPgd4M/ClKWz7hHxylpnZWBM5ZPMW4FvAj0naJemt2awrGfsB7s8CWyT9ALgNuCYi6h8Cvx34G2Ab8DAzeOQO+OQsM7NmTji8ExFXjdP+K03abgduH2f5PuDcZ1nfpJWShEqtOlubMzM7KRT4jFw8pm9mNkpxQz9JHPpmZqMUOPTd0zczG62woV9OEp+cZWY2SmFD3ydnmZmNVdjQ98lZZmZjFTb0Ezn0zcxGK2zolxNRqdVaXYaZ2ZxS2NAvlUTVmW9m1qC4oS9RdU/fzKxBcUPfH+SamY3h0DczayOFDX1fOcvMbKzChr5PzjIzG6uwoZ8esunQNzPLK2zolxIRgS+kYmaWU9zQV3otdo/rm5mNmMjlEm+S9KSkrbm2D0jaLenu7HZZbt61krZJelDSxbn2S7K2bZLeO/1PpVGplIW+e/pmZsMm0tP/FHBJk/aPRMS67HYHgKSzSa+de062zl9JKmUXS/84cClwNnBVtuyMGe7pO/TNzIZN5Bq5X5e0eoKPdzlwa0QMANslbQMuzOZti4hHACTdmi1737OueIJKiYd3zMxGm8qY/jslbcmGf5ZmbSuAnblldmVt47U3JWmDpD5Jff39/ZMqbjj0qw59M7O6yYb+9cDzgXXAHuBDWbuaLBvHaW8qIm6IiPURsb63t3dSBZaz0Pdhm2ZmI044vNNMRDxRn5Z0I/Dl7O4uYFVu0ZXA49n0eO0zopSk72c1D++YmQ2bVE9f0vLc3dcD9SN7NgJXSuqStAZYC3wH+C6wVtIaSZ2kH/ZunHzZJ1bKnpl7+mZmI07Y05d0C/ByoEfSLuD9wMslrSMdotkBvA0gIu6V9DnSD2grwDsiopo9zjuBrwAl4KaIuHfan03OcE/foW9mNmwiR+9c1aT5k8dZ/jrguibtdwB3PKvqpsA9fTOzsYp7Rm7W0/dx+mZmI4ob+j45y8xsjOKG/vAhm75koplZXWFDv36cvjPfzGxEYUPfPX0zs7EKH/o+OcvMbEThQ7/i794xMxtW+ND3t2yamY0ofuj7kE0zs2GFD32fkWtmNqKwoT9yyKZD38ysrrChn8g9fTOz0Qob+uWSe/pmZqMVNvRL7umbmY1R3ND3yVlmZmMUPvR9cpaZ2YjCh76P0zczG1HY0C/XL6Li4R0zs2EnDH1JN0l6UtLWXNufS3pA0hZJX5C0JGtfLemopLuz2ydy67xA0j2Stkn6mJR90jpDEl8u0cxsjIn09D8FXDKqbRNwbkScB/wQuDY37+GIWJfdrsm1Xw9sANZmt9GPOa3KvjC6mdkYJwz9iPg6sG9U279GRCW7+21g5fEeQ9JyYHFEfCsiAvgMcMXkSp4YH7JpZjbWdIzpvwX459z9NZK+L+lrkl6ata0AduWW2ZW1NSVpg6Q+SX39/f2TKqrkk7PMzMaYUuhLeh9QAW7OmvYAp0fE+cBvAf8gaTHQbPx+3DSOiBsiYn1ErO/t7Z1Ube7pm5mNVZ7sipKuBl4LvCIbsiEiBoCBbHqzpIeBs0h79vkhoJXA45Pd9kSMHLLpyyWamdVNqqcv6RLg94DXRcSRXHuvpFI2fSbpB7aPRMQe4KCki7Kjdt4MfGnK1R9HeTj0Z3IrZmYnlxP29CXdArwc6JG0C3g/6dE6XcCm7MjLb2dH6vws8EeSKkAVuCYi6h8Cv530SKB5pJ8B5D8HmHaJe/pmZmOcMPQj4qomzZ8cZ9nbgdvHmdcHnPusqpuiciKfnGVmllPYM3Ih7e37g1wzsxGFDv1yIh+yaWaWU+jQL8k9fTOzvGKHfkn+lk0zs5xih74c+mZmecUO/cShb2aWV+jQLzv0zcwaFDr0E4e+mVmDQoe+T84yM2tU6ND3yVlmZo0KHfrlRFSrDn0zs7pCh34iD++YmeUVOvTLPjnLzKxBoUO/lCQOfTOznGKHvnDom5nlFDr0y+7pm5k1KHToJ4l7+mZmeRMKfUk3SXpS0tZc2zJJmyQ9lP1cmrVL0sckbZO0RdIFuXWuzpZ/KLuw+owqJ4mP3jEzy5loT/9TwCWj2t4L3BkRa4E7s/sAl5JeEH0tsAG4HtI3CdLr674IuBB4f/2NYqb45Cwzs0YTCv2I+Dqwb1Tz5cCns+lPA1fk2j8TqW8DSyQtBy4GNkXEvoh4GtjE2DeSaZV+4ZovjG5mVjeVMf3nRsQegOznc7L2FcDO3HK7srbx2mdM+tXKM7kFM7OTy0x8kKsmbXGc9rEPIG2Q1Cepr7+/f9KFpBdRceqbmdVNJfSfyIZtyH4+mbXvAlblllsJPH6c9jEi4oaIWB8R63t7eyddoC+XaGbWaCqhvxGoH4FzNfClXPubs6N4LgL2Z8M/XwFeLWlp9gHuq7O2GePLJZqZNSpPZCFJtwAvB3ok7SI9Cud/Ap+T9FbgMeCN2eJ3AJcB24AjwK8CRMQ+Sf8d+G623B9FxOgPh6eVv0/fzKzRhEI/Iq4aZ9YrmiwbwDvGeZybgJsmXN0UJf5qZTOzBoU+I7fs4/TNzBoUOvRLiah5eMfMbFjhQ989fTOzEYUPfR+9Y2Y2otih70M2zcwaFDv0fXKWmVmDYoe+e/pmZg0KHfo+ZNPMrFGhQ7+UpE+v5uA3MwMKH/rpT/f2zcxSBQ/9rKfvE7TMzIDCh3760z19M7NUwUM/fXo+gsfMLFXs0M+u1eXQNzNLFTv0s/Gdii+ZaGYGFDz0y0na1Xfmm5mlCh36JaWh756+mVmq2KHvnr6ZWYNJh76kH5N0d+52QNJ7JH1A0u5c+2W5da6VtE3Sg5Iunp6nML566Lunb2aWmtA1cpuJiAeBdQCSSsBu4AukF0L/SER8ML+8pLOBK4FzgNOAr0o6KyKqk63hRIZ7+j45y8wMmL7hnVcAD0fEo8dZ5nLg1ogYiIjtwDbgwmnaflMjPX2HvpkZTF/oXwnckrv/TklbJN0kaWnWtgLYmVtmV9Y2hqQNkvok9fX390+6qOHQrzr0zcxgGkJfUifwOuAfs6brgeeTDv3sAT5UX7TJ6k3TOCJuiIj1EbG+t7d30rXVj97x8I6ZWWo6evqXAt+LiCcAIuKJiKhGRA24kZEhnF3Aqtx6K4HHp2H74yqVPLxjZpY3HaF/FbmhHUnLc/NeD2zNpjcCV0rqkrQGWAt8Zxq2P66Rk7Mc+mZmMIWjdwAkzQdeBbwt1/xnktaRDt3sqM+LiHslfQ64D6gA75jJI3cgf3KWQ9/MDKYY+hFxBDh1VNubjrP8dcB1U9nms1FyT9/MrEFbnJHrnr6ZWaotQt9frWxmlnLom5m1kbYIfQ/vmJmlCh36ZV8Y3cysQaFD3xdGNzNrVPDQz3r6Dn0zM6Dooe+Ts8zMGhQ79Ev1o3d8ERUzMyh66Kse+i0uxMxsjih26Cfu6ZuZ5RU69Ms+OcvMrEGhQz/xyVlmZg0KHfplXxjdzKxBoUPfX8NgZtaoLUK/6gujm5kBRQ/9+iGbHt4xMwMKHvpJIiQfvWNmVjfl0Je0Q9I9ku6W1Je1LZO0SdJD2c+lWbskfUzSNklbJF0w1e2fSDmRQ9/MLDNdPf3/GBHrImJ9dv+9wJ0RsRa4M7sPcCmwNrttAK6fpu2PK5FD38ysbqaGdy4HPp1Nfxq4Itf+mUh9G1giafkM1QC4p29mljcdoR/Av0raLGlD1vbciNgDkP18Tta+AtiZW3dX1tZA0gZJfZL6+vv7p1RcksiHbJqZZcrT8BgviYjHJT0H2CTpgeMsqyZtYxI5Im4AbgBYv379lBK7nIiKv3vHzAyYhp5+RDye/XwS+AJwIfBEfdgm+/lktvguYFVu9ZXA41Ot4XgWdXdw8FhlJjdhZnbSmFLoS1ogaVF9Gng1sBXYCFydLXY18KVseiPw5uwonouA/fVhoJly6sJO9h4amMlNmJmdNKY6vPNc4AtKT4IqA/8QEf8i6bvA5yS9FXgMeGO2/B3AZcA24Ajwq1Pc/gn1LOxi574jM70ZM7OTwpRCPyIeAX6qSftTwCuatAfwjqls89nqWdjF9x97ejY3aWY2ZxX6jFyAnoWd7Ds86MM2zcxoi9Dvohbw9JHBVpdiZtZyhQ/9Uxd2AvjDXDMz2iD0exZ2AbD3oHv6ZmZtE/pPHXZP38ysDUI/Hd7pP+jQNzMrfOifMq+DjpJ46rCHd8zMCh/6kjh1QRd73dM3Myt+6IO/isHMrK4tQr9nYZeHd8zMaKPQ9/COmVnbhH4new8Nkn71j5lZ+2qT0O9isFrj4IC/V9/M2ltbhP7wVzF4iMfM2lxbhP7wVzEc8oe5Ztbe2ir0n/Jhm2bW5tok9P1Nm2ZmMIXQl7RK0r9Jul/SvZLenbV/QNJuSXdnt8ty61wraZukByVdPB1PYCKWLehEgn4P75hZm5vK5RIrwG9HxPeyi6NvlrQpm/eRiPhgfmFJZwNXAucApwFflXRWRFSnUMOElEsJS+f7rFwzs0n39CNiT0R8L5s+CNwPrDjOKpcDt0bEQERsJ704+oWT3f6ztfyUbnY/fXS2NmdmNidNy5i+pNXA+cBdWdM7JW2RdJOkpVnbCmBnbrVdHP9NYlqt6VnA9r2HZ2tzZmZz0pRDX9JC4HbgPRFxALgeeD6wDtgDfKi+aJPVm54iK2mDpD5Jff39/VMtEYAzexey6+kjDFRmfDTJzGzOmlLoS+ogDfybI+LzABHxRERUI6IG3MjIEM4uYFVu9ZXA480eNyJuiIj1EbG+t7d3KiUOO7NnAbWAx546Mi2PZ2Z2MprK0TsCPgncHxEfzrUvzy32emBrNr0RuFJSl6Q1wFrgO5Pd/rO1pmcBAI94iMfM2thUjt55CfAm4B5Jd2dtvw9cJWkd6dDNDuBtABFxr6TPAfeRHvnzjtk4cqduTW8a+h7XN7N2NunQj4j/R/Nx+juOs851wHWT3eZULO7uoGdhF9v7Hfpm1r7a4ozcujN7FvDI3kOtLsPMrGXaKvR92KaZtbv2Cv3eBew9NMj+o0OtLsXMrCXaK/SzI3h2uLdvZm2qrUL/+T6Cx8zaXFuF/qpl80nkY/XNrH21Veh3lUucvmw+/75tL7WaL5JuZu2nrUIf4JqXPZ/Njz7N9V97uNWlmJnNurYL/V984Spee95yPrzph/Tt2NfqcszMZlXbhb4k/uQNP8nKpfN41y3f5+nDvpqWmbWPtgt9gEXdHfyvq86n/9AAv3vbFiI8vm9m7aEtQx/gvJVLeO+lP8FX73+C//bFrXzjoX6ODFZaXZaZ2YyayrdsnvTe8pLV3Pv4fm6+6zFuvusx1vQs4PNv/2mWLuhsdWlmZjOibXv6kI7vf/gX1rHlA6/m4790AbufPso1f7+ZwUqt1aWZmc2Itg79usXdHbzmvOX82c+fx13b93Ht5++h6uP4zayA2np4Z7Qrzl/Bo08d4SNf/SFHhyr84evOZdN9T1CN4E0XndHq8szMpsyhP8q7X7mWBV0l/vif7ueOe3403H7Gsvn87FnTc71eM7NW8fBOE7/20jP5+C9dwFtesobb3/5iVp86n/dvvJeByqxd3dHMbEbMeuhLukTSg5K2SXrvbG9/ol5z3nL+4OfO5gVnLOMPLz+X7XsP88GvPMjW3ft5pP8Qe/YfZf+RIQYrtYbj/COi4YPgai144sCxhu/68XkBZtYqszq8I6kEfBx4FbAL+K6kjRFx32zW8Wy97KxeXvOTy7nxG9u58Rvbx8wvJWJ+R4lySRw8VqFSC3oWdtGzsJMdTx3m2FCNU+Z1cM5pi3niwDEefeoIZ/Yu4AVnLOWUeZ2UE5EkoiRRSshNi0SiXEp/lrL2JMmWy9rKycj80euWktxNYqBSZfczR3nq0CDdHSXmd5bo7igxr7PEvI4S3R0JidJ1kwRKEvO7yizoLCHEUK3G04cHeerwILVakCSiZ2EXy0/pZrBS46nDg3SVE06Z30FXOUnrzWoez+GBCvfvOUClFpx92mIWd3eMWaZSrVGpBZ2lhCTRmPsAg5Ua9+zez0NPHOT0ZfNZ+9xFnLqg87jbTmseIJHoLpfo7kzoLCVII+tUqumbeLmUEBEcHaoyVA3Ko/bt8bYzUUPVGuVEDdufbrVaMJhtp1xq7PdFBBGc8LnUOy4zWedMiAiODFYZqKSvydKz/J0dHazyt9/czq3f2cmL1izjN191FstP6Wbf4UF+dOAYTx4coHdhF2ufu5CucmlO7ifNZq9T0ouBD0TExdn9awEi4k/GW2f9+vXR19c3SxWOb6ha465H9nFooMKxoSpHBqscHapydLDC0ez+ULXGou407H60/xj9BwdY3bOAVUvn8cCPDnLfngM8b3E3py+bz0NPHuLunc9wZLBCtRa0w8FC6RvRyJtV+uYCBwcq5P8MF3WXGRiqUanVKCcJQTBUHVkgEQ37q5SIzlJCNQuzvERwyrwOapH+DivVYKhWo6OU0FVKODgw9oS8RDC/s8y8zhLHhqocPJYu01lKqNRq4/6uJIbfgKX0vzwh5nWWKCfi8GCFwUqNrnLaQRgYqlGNYFF3ma5ywv6jQxwbqpEI5mVvxJ2lhMODVQ4PVKhvVtm2hIYblKuh3q6GdlGp1RiqRsORafVtDwzVGKjUGKzWkGBRV5nOcsKhgQoDldpwR6KcheSxSo1aBIu7O1jYVSYiqEZQrUFtVKaMzphomNd8uYY1ounkxB83N6dSDSrZ808Ei+d1kGSBXN+vDN8bvQ/h8ECVQwMVLjh9CVt3HyAIJI05zDt9QxXHhmp0lhOWze+kXFK2/ZHfw1C2v7s7SnSVE7o7Sgg4NlRjUXeZTb/1MiZD0uaIWN9s3mx/kLsC2Jm7vwt40eiFJG0ANgCcfvrps1PZCXSUEn5mbc+MPX5EGvzpG0D6h1mtBbVa+mKq/6zWRm617EU2Mp2uV5/Or1OLoFINOsoJq5bO49QFXQxUatkbV/oGdiybrmW1RARDteDoYIXDA+nnGaVELF3QSc+CTkqJqNSC/oMDPL7/KPM6Sixb0MlgpcYzR4YYrNYa6q8FuemRgFgyv4OfXHEKpURs3b2fvdl/IeVEVLNX7/yOEqVS+uKqVIPOckIpEUPVGkPVGoOVGkkizl+1hB9/3mJ2Pn2Eh588NHx5zETp77BcSihn6w1Uaiyd30nvoi6Ahn2QvpFX6CqXWDI/DYYjg1U6SmJBVzmtrb5/q2N/N0H6wq8FHB1M//tb0FWms5QwUEn/U+juKFFK4NCxtOOwZH4nC7vKDFVrHB2scmSoysBQjQVdJeZ3liklaZjF8N9M9pORxsj9PUWuLQLKpfTNsaOU0FFWw++pq5zQVU6DpxbBgaNDDFaDxdmbQrX+N1lNn9u8jhKJYP/RIQ4OVNI38tx/ocNvSJnRHV01zGveC8435x/v+I/V/HHrU6VEnDKvg85ywr7D6d9Guk8b9xfk3zhG5pUSccX5K3jh6mXsfuYon/nmDhA8b3E3y0/ppndRNz/af4z79xxgsFqju6PEwFCVfYcHqdaCcin976oj+y+rnIgAjmV/ewOVtFMxryOZsZNEZ7un/0bg4oj4tez+m4ALI+I3xltnrvT0zcxOFsfr6c/2B7m7gFW5+yuBx2e5BjOztjXbof9dYK2kNZI6gSuBjbNcg5lZ25rVMf2IqEh6J/AVoATcFBH3zmYNZmbtbNbPyI2IO4A7Znu7ZmbmM3LNzNqKQ9/MrI049M3M2ohD38ysjczqyVmTIakfeHSSq/cAe6exnJngGqdurtcHrnG6uMaJOSMimn4X/JwP/amQ1DfeWWlzhWucurleH7jG6eIap87DO2ZmbcShb2bWRooe+je0uoAJcI1TN9frA9c4XVzjFBV6TN/MzBoVvadvZmY5Dn0zszZSyNCfixdfl7RK0r9Jul/SvZLenbUvk7RJ0kPZz6VzoNaSpO9L+nJ2f42ku7IaP5t9LXYr61si6TZJD2T788VzbT9K+s3s97xV0i2Sulu9HyXdJOlJSVtzbU33m1Ify15DWyRd0MIa/zz7XW+R9AVJS3Lzrs1qfFDSxa2oLzfvdySFpJ7sfkv24YkULvRzF1+/FDgbuErS2a2tCoAK8NsR8RPARcA7srreC9wZEWuBO7P7rfZu4P7c/T8FPpLV+DTw1pZUNeIvgH+JiB8Hfoq01jmzHyWtAN4FrI+Ic0m/RvxKWr8fPwVcMqptvP12KbA2u20Arm9hjZuAcyPiPOCHwLUA2evnSuCcbJ2/yl7/s10fklYBrwIeyzW3ah8eX3otzeLcgBcDX8ndvxa4ttV1NanzS6R/JA8Cy7O25cCDLa5rJemL/z8BXya9vOheoNxs/7agvsXAdrKDEHLtc2Y/MnIt6GWkX1/+ZeDiubAfgdXA1hPtN+CvgauaLTfbNY6a93rg5my64bVNep2OF7eiPuA20g7IDqCn1fvweLfC9fRpfvH1FS2qpSlJq4HzgbuA50bEHoDs53NaVxkAHwX+K1DL7p8KPBMRlex+q/fnmUA/8LfZENTfSFrAHNqPEbEb+CBpr28PsB/YzNzaj3Xj7be5+jp6C/DP2fScqFHS64DdEfGDUbPmRH2jFTH01aRtzhyXKmkhcDvwnog40Op68iS9FngyIjbnm5ss2sr9WQYuAK6PiPOBw8yNIbFh2bj45cAa4DRgAem/+qPNmb/LJuba7x1J7yMdJr253tRksVmtUdJ84H3AHzSb3aSt5b/zIob+nL34uqQO0sC/OSI+nzU/IWl5Nn858GSr6gNeArxO0g7gVtIhno8CSyTVr7LW6v25C9gVEXdl928jfROYS/vxlcD2iOiPiCHg88BPM7f2Y914+21OvY4kXQ28FvjlyMZKmBs1Pp/0zf0H2etmJfA9Sc+bI/WNUcTQn5MXX5ck4JPA/RHx4dysjcDV2fTVpGP9LRER10bEyohYTbrf/k9E/DLwb8DPZ4u1usYfATsl/VjW9ArgPubQfiQd1rlI0vzs916vcc7sx5zx9ttG4M3ZESgXAfvrw0CzTdIlwO8Br4uII7lZG4ErJXVJWvZdwnkAAADjSURBVEP6gel3ZrO2iLgnIp4TEauz180u4ILs73TO7MMGrf5QYYY+aLmM9FP+h4H3tbqerKafIf3Xbgtwd3a7jHTM/E7goeznslbXmtX7cuDL2fSZpC+mbcA/Al0trm0d0Jftyy8CS+fafgT+EHgA2Ar8HdDV6v0I3EL6GcMQaTi9dbz9Rjo08fHsNXQP6ZFIrapxG+nYeP1184nc8u/LanwQuLQV9Y2av4ORD3Jbsg9PdPPXMJiZtZEiDu+Ymdk4HPpmZm3EoW9m1kYc+mZmbcShb2bWRhz6ZmZtxKFvZtZG/j856KHQZrpQ+gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5506297777777778\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5767747222222223\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5714633888888889\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5839056666666667\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAEICAYAAACzliQjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nO3de3RcZ33u8e+zZ0YjS7ItO5ZDYjvYIeaShEtSERJoDyxCIaGshLawmpye4kLOSi+05dKekpS1CrSnLSy6uOS0TZsmKaFQAk1p46Zp05xACxyaFBtC7ond4MSOb3J81X0uv/PH3iOPZMmyJUujzDyftWZp9rvf0fxmS/PMO+/es0cRgZmZtYak0QWYmdn8ceibmbUQh76ZWQtx6JuZtRCHvplZC3Hom5m1EIe+vaBI2ibpLSfYNySdM8P7mfFtZ3BfH5f0peOsf1TSm+ajFmt++UYXYGbHFxHnNboGax4e6ZuZtRCHvr1gSbpI0n9IOihpl6Q/kdQ2odvbJT0taZ+kT0tK6m7/PkmPSzog6R5JL57jej8i6TlJRyQ9KenSutVtkr6YrXtUUm/d7camtLKpoDskfTXr+31Jr57Luq25OPTthawCfAhYAVwCXAr86oQ+Pw30AhcCVwLvA5D0TuB3gJ8BeoBvA185kTuV9GfZC81kl4emuM3LgF8DXhsRi4G3AdvqulwB3A50AxuBPzlOCVcCfwssB/4G+AdJhROp3cyhby9YEbE5Iu6PiHJEbAP+AnjjhG6fioj9EfEs8Dng6qz9l4A/iojHI6IM/CHwmhMZ7UfEr0ZE9xSXV01xswpQBM6VVIiIbRHxX3XrvxMRd0dEBfhr4Hij980RcUdElIDPAO3AxdPVbQYOfXsBk/RSSXdJ2i3pMGlwr5jQbXvd9WeAM7PrLwY+XxuhA/sBAavmotaI2Ap8EPg4sFfS7ZLOrOuyu+76INAuaaoDLcYeU0RUgR0cfVxmx+XQtxeyG4EngPURsYR0ukYT+qypu34WsDO7vh34pQmj9EUR8d3p7lTSn0vqn+Ly6FS3i4i/iYgfJ33BCeBTJ/5QJ39M2T6K1XWPy+y4HPr2QrYYOAz0S3o58CuT9PlfkpZJWgN8APhq1v7nwPWSzgOQtFTSu0/kTiPilyOia4rLpIdXSnqZpDdLKgLDwBDplM9M/Jikn8neCXwQGAHun+Hvshbj0LcXst8C/jtwBPhLjgZ6vTuBzcCDwD8BtwBExN+TjrRvz6aGHgEun8Nai8AngX2kUzkrSd+ZzMSdwM8BB4BfAH4mm983m5b8JSpmLxySPg6cExH/o9G12AuTR/pmZi3EoW9m1kI8vWNm1kKmHelLulXSXkmPTLLut7KzEa7IliXpBklbJT0k6cK6vhskbckuG07twzAzsxNxImfZ/ALpR8K/WN+YHQL3k8Czdc2XA+uzy+tIj6N+naTlwMdIPw4fwGZJGyPiwPHueMWKFbF27doTeiBmZpbavHnzvojomWzdtKEfEd+StHaSVZ8Ffpv08LGaK4EvRjpndL+kbklnAG8C7o2I/QCS7gUuY5pznaxdu5ZNmzZNV6KZmdWR9MxU62a0I1fSFcBzEfHDCatWMf5j7zuytqnaJ/vd10raJGlTX1/fTMozM7MpnHToS+oAPgr87mSrJ2mL47Qf2xhxU0T0RkRvT8+k707MzGyGZjLSfwmwDvihpG2k5/34vqQXkY7g6891UjsnyFTtZmY2j0469CPi4YhYGRFrI2ItaaBfGBG7Sc8D/p7sKJ6LgUMRsQu4B3hrdg6UZcBbszYzM5tHJ3LI5leA/wBeJmmHpGuO0/1u4GlgK+m5UH4VINuB+/vA97LL79V26pqZ2fxZ0B/O6u3tDR+9Y2Z2ciRtjojeydb5NAxmZi2kKUN/YKTMZ/71SR7cfrDRpZiZLShNGfoj5So3fGMrP3Tom5mN05Shn8+lHwsoVxfu/gozs0ZoztBPstCvVBtciZnZwtKkoZ8+LI/0zczGa8rQL2TTOyWP9M3MxmnK0JdELhHlikf6Zmb1mjL0gTT0Pb1jZjZO04Z+IZF35JqZTdC0oZ/PJR7pm5lN0LShX8jJO3LNzCZo2tDPJaLikb6Z2ThNG/r5JKHko3fMzMZp2tAv5ES56ukdM7N6TRv6+Vzi4/TNzCZo3tBPPNI3M5uoeUM/50/kmplN1LyhnySUfPSOmdk4TRv6hZw/kWtmNlHThr7PvWNmdqxpQ1/SrZL2Snqkru3Tkp6Q9JCkv5fUXbfueklbJT0p6W117ZdlbVslXXfqH8p4hVzikb6Z2QQnMtL/AnDZhLZ7gfMj4lXAU8D1AJLOBa4Czstu82eScpJywJ8ClwPnAldnfedM3iN9M7NjTBv6EfEtYP+Etn+NiHK2eD+wOrt+JXB7RIxExI+ArcBF2WVrRDwdEaPA7VnfOZPP+RO5ZmYTnYo5/fcB/5xdXwVsr1u3I2ubqv0Ykq6VtEnSpr6+vhkXlU9Excfpm5mNM6vQl/RRoAx8udY0Sbc4TvuxjRE3RURvRPT29PTMuDZ/ItfM7Fj5md5Q0gbgHcClEVFL1x3Amrpuq4Gd2fWp2udEIRElj/TNzMaZ0Uhf0mXAR4ArImKwbtVG4CpJRUnrgPXAfwLfA9ZLWiepjXRn78bZlX58/kSumdmxph3pS/oK8CZghaQdwMdIj9YpAvdKArg/In45Ih6V9DXgMdJpn/dHRCX7Pb8G3APkgFsj4tE5eDxjcom/OcvMbKJpQz8irp6k+Zbj9P8D4A8mab8buPukqpsFfyLXzOxYTfuJ3HziHblmZhM1begXct6Ra2Y2UdOGfi7xjlwzs4maNvTzuXRH7tGjSc3MrGlDv5Cknwer+AgeM7MxTRv6+Vz60HzYppnZUc0b+tlIv+TDNs3MxjRv6Oc8vWNmNlETh3760Hx6ZTOzo5o29Gs7css+Vt/MbEzThn6uFvoe6ZuZjWna0C/46B0zs2M0bejXduT6pGtmZkc1b+gn3pFrZjZRE4e+d+SamU3UvKFfm97xnL6Z2ZimDf2xHbme3jEzG9O0oT82veMduWZmY5o39LPpnZKnd8zMxjRv6GdH71S8I9fMbMy0oS/pVkl7JT1S17Zc0r2StmQ/l2XtknSDpK2SHpJ0Yd1tNmT9t0jaMDcP56ixkb7n9M3MxpzISP8LwGUT2q4D7ouI9cB92TLA5cD67HItcCOkLxLAx4DXARcBH6u9UMwV78g1MzvWtKEfEd8C9k9ovhK4Lbt+G/DOuvYvRup+oFvSGcDbgHsjYn9EHADu5dgXklMq5+P0zcyOMdM5/dMjYhdA9nNl1r4K2F7Xb0fWNlX7MSRdK2mTpE19fX0zLA8KiUf6ZmYTneoduZqkLY7TfmxjxE0R0RsRvT09PTMu5OiHszzSNzOrmWno78mmbch+7s3adwBr6vqtBnYep33OeEeumdmxZhr6G4HaETgbgDvr2t+THcVzMXAom/65B3irpGXZDty3Zm1zJj82veORvplZTX66DpK+ArwJWCFpB+lROJ8EvibpGuBZ4N1Z97uBtwNbgUHgvQARsV/S7wPfy/r9XkRM3Dl8SvncO2Zmx5o29CPi6ilWXTpJ3wDeP8XvuRW49aSqm4WxHbkOfTOzMc37iVx/iYqZ2TGaN/QT78g1M5uoaUNfErlEPmTTzKxO04Y+pKN9z+mbmR3V1KFfyCX+RK6ZWZ2mDv1cIu/INTOr09ShX8jJX6JiZlanqUM/nyRUPL1jZjamuUM/J0o+esfMbExzh34i78g1M6vT3KGfS3ycvplZneYOfY/0zczGaerQL+QSfzjLzKxOU4d+LhElH6dvZjamqUO/kPP0jplZvaYO/XySUPH0jpnZmOYOfR+nb2Y2TnOHvo/eMTMbp7lDP5d4R66ZWZ2mDv1CTp7TNzOr09Shn098nL6ZWb1Zhb6kD0l6VNIjkr4iqV3SOkkPSNoi6auS2rK+xWx5a7Z+7al4AMeT93H6ZmbjzDj0Ja0CfgPojYjzgRxwFfAp4LMRsR44AFyT3eQa4EBEnAN8Nus3p/I+Tt/MbJzZTu/kgUWS8kAHsAt4M3BHtv424J3Z9SuzZbL1l0rSLO//+MX5hGtmZuPMOPQj4jngj4FnScP+ELAZOBgR5azbDmBVdn0VsD27bTnrf9rE3yvpWkmbJG3q6+ubaXkAFPzF6GZm48xmemcZ6eh9HXAm0AlcPknXWupONqo/JpEj4qaI6I2I3p6enpmWB0Au8Rejm5nVm830zluAH0VEX0SUgK8Drwe6s+kegNXAzuz6DmANQLZ+KbB/Fvc/rULOO3LNzOrNJvSfBS6W1JHNzV8KPAZ8E3hX1mcDcGd2fWO2TLb+GxExp8PwfM7TO2Zm9WYzp/8A6Q7Z7wMPZ7/rJuAjwIclbSWds78lu8ktwGlZ+4eB62ZR9wmpnXBtjl9bzMxeMPLTd5laRHwM+NiE5qeBiybpOwy8ezb3d7LySboboVwNCrk5PVDIzOwFobk/kZtLH5535pqZpZo69Guje59e2cws1dShX5veqXikb2YGNHno57LpHY/0zcxSTR36hdqOXI/0zcyAJg9978g1MxuvqUO/tiPXJ10zM0s1dejn6o7TNzOzJg/9fJLtyPX5d8zMgCYP/bHpHc/pm5kBTR76YztyPb1jZgY0e+iPHbLp6R0zM2iV0PdI38wMaPbQz3lHrplZvaYO/dqO3IpH+mZmQJOHfu04/ZKP3jEzA5o89AtjR+94esfMDJo89PM+4ZqZ2ThNHfoFH6dvZjZOU4d+zsfpm5mNM6vQl9Qt6Q5JT0h6XNIlkpZLulfSluznsqyvJN0gaaukhyRdeGoewtTyY1+X6JG+mRnMfqT/eeBfIuLlwKuBx4HrgPsiYj1wX7YMcDmwPrtcC9w4y/ueViGpnU/fI30zM5hF6EtaAvw34BaAiBiNiIPAlcBtWbfbgHdm168Evhip+4FuSWfMuPITkPcJ18zMxpnNSP9soA/4K0k/kHSzpE7g9IjYBZD9XJn1XwVsr7v9jqxtHEnXStokaVNfX98syoO2fPrwRj3SNzMDZhf6eeBC4MaIuAAY4OhUzmQ0SdsxQ/CIuCkieiOit6enZxblQVsuIZ+I/pHyrH6PmVmzmE3o7wB2RMQD2fIdpC8Ce2rTNtnPvXX919TdfjWwcxb3Py1JdLXnGXDom5kBswj9iNgNbJf0sqzpUuAxYCOwIWvbANyZXd8IvCc7iudi4FBtGmgudbblPdI3M8vkZ3n7Xwe+LKkNeBp4L+kLydckXQM8C7w763s38HZgKzCY9Z1zXUWP9M3MamYV+hHxINA7yapLJ+kbwPtnc38z0VnMeaRvZpZp6k/kAnQW8/SPVBpdhpnZgtD0oe/pHTOzoxz6ZmYtpOlDP53eceibmUELhH5tpJ/uRzYza21NH/qdxTzVgKGSd+aamTV96HcVcwCe4jEzoxVCvz39KMKAD9s0M2v+0O9sq4W+R/pmZk0f+l3FNPQ9vWNm1gKh31n0SN/MrKZlQt8jfTOzFgh9T++YmR3V/KHf7ukdM7Oapg/9jkLtOH0fsmlm1vShnySisy3nkb6ZGS0Q+pDuzHXom5m1SOh3FfMcceibmbVG6Hukb2aWaonQ9xepmJmlZh36knKSfiDprmx5naQHJG2R9FVJbVl7MVvemq1fO9v7PlH+nlwzs9SpGOl/AHi8bvlTwGcjYj1wALgma78GOBAR5wCfzfrNi66ij94xM4NZhr6k1cBPATdnywLeDNyRdbkNeGd2/cpsmWz9pVn/Oec5fTOz1GxH+p8DfhuoZsunAQcjopawO4BV2fVVwHaAbP2hrP+c89E7ZmapGYe+pHcAeyNic33zJF3jBNbV/95rJW2StKmvr2+m5Y3TWcwzWq5SqlSn72xm1sRmM9J/A3CFpG3A7aTTOp8DuiXlsz6rgZ3Z9R3AGoBs/VJg/8RfGhE3RURvRPT29PTMoryjunx6ZTMzYBahHxHXR8TqiFgLXAV8IyJ+Hvgm8K6s2wbgzuz6xmyZbP03IuKYkf5c8Jk2zcxSc3Gc/keAD0vaSjpnf0vWfgtwWtb+YeC6ObjvSR39IhUftmlmrS0/fZfpRcS/Af+WXX8auGiSPsPAu0/F/Z2szmLtTJse6ZtZa2uZT+SCQ9/MrCVC39+Ta2aWaonQ90jfzCzVUqHvkb6ZtbqWCP3a9E7/sEPfzFpbS4R+Wz6hsy3HgcFSo0sxM2uolgh9gJ7FRfr6RxpdhplZQ7VW6B8ZbnQZZmYN1WKh75G+mbW21gn9riL7+kcbXYaZWUO1TugvLnJoqMRI2effMbPW1VKhD3i0b2YtrWVCf0VXGvqe1zezVtYyoV8b6Tv0zayVOfTNzFpIy4T+aZ0OfTOzlgn9tnzCso4Cff3+gJaZta6WCX1Ip3j2HfHRO2bWulou9H3+HTNrZa0V+l0+FYOZtbbWCv3s/DsR0ehSzMwaYsahL2mNpG9KelzSo5I+kLUvl3SvpC3Zz2VZuyTdIGmrpIckXXiqHsSJWtFVZKhUYWDUp2Iws9Y0m5F+GfjNiHgFcDHwfknnAtcB90XEeuC+bBngcmB9drkWuHEW9z0jPlbfzFrdjEM/InZFxPez60eAx4FVwJXAbVm324B3ZtevBL4YqfuBbklnzLjyGXDom1mrOyVz+pLWAhcADwCnR8QuSF8YgJVZt1XA9rqb7cjaJv6uayVtkrSpr6/vVJQ3xqFvZq1u1qEvqQv4O+CDEXH4eF0naTtmj2pE3BQRvRHR29PTM9vyxunpqp1p06FvZq1pVqEvqUAa+F+OiK9nzXtq0zbZz71Z+w5gTd3NVwM7Z3P/J2tZRxu5ROw57E/lmllrms3ROwJuAR6PiM/UrdoIbMiubwDurGt/T3YUz8XAodo00HxJErH2tA6e2tM/n3drZrZg5Gdx2zcAvwA8LOnBrO13gE8CX5N0DfAs8O5s3d3A24GtwCDw3lnc94ydv2opDzy9vxF3bWbWcDMO/Yj4DpPP0wNcOkn/AN4/0/s7VV65ail3PriTviMjYzt2zcxaRUt9IhfgvDOXAvDIzkMNrsTMbP61XuivWgLAo8859M2s9bRc6C9pL7D2tA4eduibWQtqudCHdGfuI88d7yMFZmbNqWVD/7mDQxwY8BeqmFlracnQf+Uq78w1s9bUkqF/3pnpztwfPHuwwZWYmc2vlgz97o42el+8jD/5xlbueXR3o8sxM5s3LRn6ADdv6OW8VUv4lS9t5q/+34+oVv1tWmbW/Fo29Ls72vjSNa/jjS/t4RP/+BhX/+X9bNs30OiyzMzmVMuGPkBnMc+tv/haPvWzr+SxnYe57PPf4uZvP03Fo34za1ItHfoAkvi5157FvR9+I294yQr+9z89zs/e+F227DnS6NLMzE65lg/9mhctbefmDb18/qrXsO35AX7qhu/wpfufaXRZZmanlEO/jiSufM0q7v3QG7n4JafxsY2P8uB2H9ZpZs3DoT+JnsVF/s/VF7BycZEPf+1BhkuVRpdkZnZKOPSnsHRRgU+/69U83TfAp+95stHlmJmdEg794/jx9Su4+qKzuO2723w4p5k1BYf+ND70lvXkc+Jz//epRpdiZjZrDv1prFzSzi++fh13/nAnT+z26ZjNXii++B/buOG+LY0uY8GZzRejt4xffuPZfPmBZ7jmC5s4u6eT9kKORYUcSxcVWLeikzXLO1i6qMCSRXkWtxdY0p6nsy1Pkhz7FcKlSpXdh4ZZvWwR0lRfMWxms7FlzxE+8Y+PUakGb375Ss7PzqxrDQh9SZcBnwdywM0R8cn5ruFkdXe08el3vYovfHcbR4bL9B0ZYaRcZV//CEeGy5PeJhF0tuXJ5UQhl7Cso0Axn+PJPUcYLVc5c2k7P7G+h77+EX60b4CeriJrV3SwbkUXa0/roLujjY62HM8PjPDcwWEODY7SP1Jh3YoOXv+SFVO+aFSqQd+RkfRFqaMw1r6vf4Tfv+sxdh8a5pM/+yrWreics+1l1kgRwe/d9RidbTkk8cf/+iRfeO9FjS5rwVDE/J1yQFIOeAr4SWAH8D3g6oh4bLL+vb29sWnTpnmr72RFBM8PjPLcgSGODJc5PFzi8FBp7Hr/SJlKNRgtV9k/MMpQqcLLX7SY1cs6+PaWfTzwo+dZ1b2Is3s62XdklKf3DbCvf2TK+8snopydIqItl7C0o0AE9I+UxtoGRytjfc5Z2cU5PV10FvN844k9DIxUKBYSqtVgw+vXMlyq0tc/Qt+RYcqVYP3pizlnZRfLOwssKuQYGKkwMFrmyHCZ/pEy/cNlRstVzuhu58yliyhXg8HRMkOjFQZLFYZGK4yUK/Qsbues5R3kEhgpVRkpVxktV1mzvIMLzuqmu6NAtQqVCCrVoJhPKOYTdh0a5nvb9jM4WuGMpe2s6CrSXsjx7P4B/u3JPg4OlnjlqqWsWraI/uEylQhO62xjuFzlO1v62HN4hMvPfxFvOfd0qtVIax45WvtIuUpXMc+SRXmWtBfoLOYJoFIJStUq1WrQ1Z6ne1Eb5WqVodEKxUKOrmKe/uEyff3DREB7IZddEgq5hFwiEolETPpCHBGMlKsMlyoMlSr0ZwOFM7oX0dmWY2C0wp7Dwzy1+wjPHRwin4iu9gLnr1rC+pWLySUiIthzeIQ9h4d50dJ2Vi4uMlqpcmS4TLUa1J7F1QgGRsr0j1RYuqjAysVF9h4Z4em+fha3F1i7ooOuYp4Ixm5TzKePA6BaDQazGtvy6WBFEkOjFUYrVZa058c9xtr/92ilyvP9Izz83CH2Hh7hgrO6edXqbnKJGBgts3VvP7sPDfOKM5aw9rQOJFGtxqTvhodLFf79qT7+/ak+1izr4E0v62H9yi7yuaOz0RHBaKVKPkm3f025UuUfH9rJh776Q373HecyWqnyyX9+gq/90iVctG752G0r1SCR0BR/s5MREQyMVijkRFsuOeHfV60G5WpQjfRnpRJUIth5cIgndh8hl8BPX7B6RjVJ2hwRvZOum+fQvwT4eES8LVu+HiAi/miy/gs99OfCkeESzzw/yOGhEoOjFZZ1trF62SK6Owq05RK27u3n/qef57mDwxwcHEUSnW05kkSMlCp0FvOc2b2IQ0MlNj9zgO37BxkYKbN2RSefuOI8Oot5Pvy1B7n/6f10FfP0LC6yoqsNSTy5+wiHhkqT1pVPxOL2PPlcwr7+ESb+2xRyYlEhR1s+x/6BEU729EX1L2iTWVTI0d1RYNeh4UnXL2nPs6yzjWeeHzy5Oz7FEpG+ACQiJ1GJNBSP1/942yqfiLZ8QqWavnDU5BKd0nNEdbblABgYHf+ZlEWFHIWcOJy9UOUTsagtR6mSvpDPpITFxTzlajBUqtDRlmNJe4FKBCOlCiPl6tjjrL0gAkiwrKMNAUOlCsOlyth9txcSuop5Oot59h4eYahUYf3KLu7+wE9QrgRv/PQ3OTRUIp+IUiV9sZio9nervQiMLXO0PUnql9M+ARwcHKVUSYvJJaKjkGNRtj0r1Rgb3NRCvtY2XfSed+YS/uk3fuLkNzDHD/35nt5ZBWyvW94BvG6ea1jQFrcXjjv/uP70xaw/ffGs7uP2ay9huFShvZAb1x4RHBgscWioxOBoeeyJ1FXMU8wfHcEMlyr0HRmhLZ/QXsjR0ZYbGykCjJar7Dw4BEBbNorPJwlb+/p5aMdBBkcrJBK5JH0C1UasPV1FLlq3fCzcDw6WGCpVWNZR4LVrl9NeyNF3ZIR9/SMszkacz/ePIMS5Zy4hETy4/SCbth1gUVs6Qq89hsXt6WPoHylzeLjM4aH0nVgiyCUJhVz6RO4fLnNwaJS2XEKxkGOkVOHIcJnF7ekLZC5JR721UXupEtnIkeyJPP5JLUF7Pn1nUMwnY3VVI9h1aJgjwyWWtBdY0VXkpacv5qzlHVQj2D84ysM7DrFlbzodKIk1yxZx+pJ29hweZteh4TQwFxXGRroiDaf0b5bj4GCJ3YeHWdFV5CU9nfSPVNi2b4ChUgWR1pb+PatjL/a123YW84yUqjx3cIhSpcrpS9op5hP2D4wyOFoZe3dQyCW05dPtt6S9wHmrlrBycTubnznA47sOk0i0FxLO7uni9CVFHt15mMd3HaaYT+hoy6d/j6ES+Vz6f1IsJLTnc7x27XJed/Zy9g+M8u0t+9i+f3DsXXBtn1p7IcnebVboHykzMFJmWUcbF5zVzZteujKrDz77c6/hXx7ZndWZ1ptPRET6ziiy//1qBNUgfRc0Ybma/W2rAUGtPU3tpYvaWNZRGHvnO5j9fwDZ/3n6TjCfiFwuHQzkk3RgkE9ELknIJen/YU5wWleRV5yxZM6mYOd7pP9u4G0R8T+z5V8ALoqIX6/rcy1wLcBZZ531Y8884/PfmJmdjOON9Of7kM0dwJq65dXAzvoOEXFTRPRGRG9PT8+8Fmdm1uzmO/S/B6yXtE5SG3AVsHGeazAza1nzOqcfEWVJvwbcQ3rI5q0R8eh81mBm1srm/Tj9iLgbuHu+79fMzHwaBjOzluLQNzNrIQ59M7MW4tA3M2sh8/rhrJMlqQ+YzaezVgD7TlE5c2Wh17jQ6wPXeKq4xlNjIdT44oiY9INOCzr0Z0vSpqk+lbZQLPQaF3p94BpPFdd4aiz0Gj29Y2bWQhz6ZmYtpNlD/6ZGF3ACFnqNC70+cI2nims8NRZ0jU09p29mZuM1+0jfzMzqOPTNzFpIU4a+pMskPSlpq6TrGl0PgKQ1kr4p6XFJj0r6QNa+XNK9krZkP5ctgFpzkn4g6a5seZ2kB7Iav5qdFruR9XVLukPSE9n2vGQhbUdJH8r+xo9I+oqk9oWwDSXdKmmvpEfq2ibdbkrdkD2HHpJ0YYPq+3T2d35I0t9L6q5bd31W35OS3jbX9U1VY92635IUklZky/O+DU9E04V+9uXrfwpcDpwLXC3p3MZWBUAZ+M2IeAVwMfD+rK7rgPsiYj1wX7bcaB8AHq9b/hTw2azGA8A1DanqqM8D/xIRLwdeTVrrgtiOklYBvwH0RsT5pKcQv4qFsQ2/AFw2oW2q7XY5sD67XAvc2KD67gXOj4hXAU8B1wNkz52rgPOy2/xZ9txvRI1IWgP8JLLJuckAAAMuSURBVPBsXXMjtuH0Ivvux2a5AJcA99QtXw9c3+i6JqnzTtJ/kieBM7K2M4AnG1zXatIn/5uBuwCRfrowP9n2bUB9S4AfkR2EUNe+ILYjR78HejnpqcvvAt62ULYhsBZ4ZLrtBvwFcPVk/eazvgnrfhr4cnZ93POa9Ds6LmnENsza7iAdgGwDVjRyG053abqRPpN/+fqqBtUyKUlrgQuAB4DTI2IXQPZzZeMqA+BzwG8D1Wz5NOBgRJSz5UZvz7OBPuCvsimomyV1skC2Y0Q8B/wx6YhvF3AI2MzC2ob1ptpuC/F59D7gn7PrC6Y+SVcAz0XEDyesWjA11mvG0NckbQvmuFRJXcDfAR+MiMONrqeepHcAeyNic33zJF0buT3zwIXAjRFxATDAwpgSAyCbE78SWAecCXSSvs2faMH8T05hQf3dJX2UdIr0y7WmSbrNe32SOoCPAr872epJ2hr+d2/G0J/2y9cbRVKBNPC/HBFfz5r3SDojW38GsLdR9QFvAK6QtA24nXSK53NAt6Tat6w1envuAHZExAPZ8h2kLwILZTu+BfhRRPRFRAn4OvB6FtY2rDfVdlswzyNJG4B3AD8f2TwJC6e+l5C+wP8we96sBr4v6UUsnBrHacbQX5Bfvi5JwC3A4xHxmbpVG4EN2fUNpHP9DRER10fE6ohYS7rdvhERPw98E3hX1q3RNe4Gtkt6WdZ0KfAYC2c7PgtcLKkj+5vX6lsw23CCqbbbRuA92REoFwOHatNA80nSZcBHgCsiYrBu1UbgKklFSetId5b+53zXFxEPR8TKiFibPW92ABdm/6cLYhseo9E7FeZoR8vbSff0/xfw0UbXk9X046Rv7R4CHswubyedM78P2JL9XN7oWrN63wTclV0/m/QJtRX4W6DY4NpeA2zKtuU/AMsW0nYEPgE8ATwC/DVQXAjbEPgK6X6GEmk4XTPVdiOdmvjT7Dn0MOnRSI2obyvpvHjtOfPndf0/mtX3JHB5o7bhhPXbOLojd9634YlcfBoGM7MW0ozTO2ZmNgWHvplZC3Hom5m1EIe+mVkLceibmbUQh76ZWQtx6JuZtZD/D3FoDfJFvt+bAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6780762222222223\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6674378333333334\n"
     ]
    }
   ],
   "source": [
    "objective = 'soft-boundary'\n",
    "i = 0\n",
    "nu = 0.2\n",
    "outlier_nu = 0.2\n",
    "label_list = []\n",
    "auc_list = []\n",
    "time_list = []\n",
    "confusionmatrix_list = []\n",
    "\n",
    "for label_num in np.arange(10):\n",
    "    ## 資料設定\n",
    "    train_dataset, x_train, x_test, y_train, y_test = load_data(label_num, batch_size = 200,\n",
    "                                                                outlier_nu = outlier_nu)\n",
    "    ## 訓練 autoencoder\n",
    "    pre_time = time.time()\n",
    "    ae = Autoencoder(label_num)\n",
    "    ae.train_ae(350, train_dataset)\n",
    "    encoder = ae.encoder\n",
    "    ## 訓練 SVDD\n",
    "    SVDD = train_SVDD(train_dataset = train_dataset, label_num = label_num , objective = objective, encoder = encoder)\n",
    "    SVDD.train(x_train, n_epochs = 150)\n",
    "    radius_end =SVDD.train_result(x_train, y_train)\n",
    "    test_auc, confusionmatrix  = SVDD.test(x_test, y_test, radius_end)\n",
    "    train_time = time.time() - pre_time\n",
    "    print('auc = ', test_auc)\n",
    "    label_list.append(label_num)\n",
    "    auc_list.append(test_auc)\n",
    "    time_list.append(train_time)\n",
    "    confusionmatrix_list.append(confusionmatrix)\n",
    "    i += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "label =  airplane\n",
      "auc =  0.6174309444444445\n",
      "time =  288.3424835205078\n",
      "[[ 190 8810]\n",
      " [  35  965]]\n",
      "label =  automobile\n",
      "auc =  0.5945320000000001\n",
      "time =  285.9426381587982\n",
      "[[2379 6621]\n",
      " [ 142  858]]\n",
      "label =  bird\n",
      "auc =  0.4962631111111111\n",
      "time =  286.2885935306549\n",
      "[[ 183 8817]\n",
      " [  36  964]]\n",
      "label =  cat\n",
      "auc =  0.510377388888889\n",
      "time =  288.0035471916199\n",
      "[[ 757 8243]\n",
      " [  64  936]]\n",
      "label =  deer\n",
      "auc =  0.5506297777777778\n",
      "time =  289.2272641658783\n",
      "[[ 714 8286]\n",
      " [  65  935]]\n",
      "label =  dog\n",
      "auc =  0.5767747222222223\n",
      "time =  280.7212426662445\n",
      "[[ 374 8626]\n",
      " [  21  979]]\n",
      "label =  frog\n",
      "auc =  0.5714633888888889\n",
      "time =  289.76577615737915\n",
      "[[1160 7840]\n",
      " [  61  939]]\n",
      "label =  horse\n",
      "auc =  0.5839056666666667\n",
      "time =  289.4905722141266\n",
      "[[2169 6831]\n",
      " [ 144  856]]\n",
      "label =  ship\n",
      "auc =  0.6780762222222223\n",
      "time =  282.35827827453613\n",
      "[[  25 8975]\n",
      " [   0 1000]]\n",
      "label =  truck\n",
      "auc =  0.6674378333333334\n",
      "time =  278.9040505886078\n",
      "[[3013 5987]\n",
      " [ 145  855]]\n"
     ]
    }
   ],
   "source": [
    "for i in range(len(names)):\n",
    "    print('label = ', names[i])\n",
    "    print('auc = ', auc_list[i])\n",
    "    print('time = ', time_list[i])\n",
    "    print(confusionmatrix_list[i])"
   ]
  },
  {
   "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": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6069626666666667\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5580329444444445\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAEICAYAAACzliQjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAdxklEQVR4nO3de5Bcd3nm8e9zTvf03GTNyJKMrYtljLiYVDCO1ohkk5B1ANtJYcgutXYuOAlbzm6cClDJbuyQhE2yZEllNwEqYOIEB5IiXEIgaImDcRm2kmzAsQyOkTHGwtiyLNmSrbs0l768+8c5Peq5acbSaLp1zvOp6uruX5/pfvvM9NO/ec/p04oIzMysHJJuF2BmZsvHoW9mViIOfTOzEnHom5mViEPfzKxEHPpmZiXi0LdzjqTHJf3oIpcNSS86zcc57Z89xX3OW7ukH5T0yPO8v49I+h9LU52VQaXbBZhZJiL+EXhJt+uwYvNM3+wcIMkTNFsSDn07p0m6UtJXJB2StFfSH0vqm7HYtZIek/SspD+QlHT8/M9LeljSQUl3Sbp4Gcr+N5K+mT/mn0vqz2t5jaTdHbU9LunXJD0IHJdUkfRKSV+TdFTSJ4H+ZajXCsShb+e6JvAOYDXwauAq4BdnLPMmYAtwBXAd8PMAkt4I/DrwE8Aa4B+Bjy/mQSV9MH+jmev04AI//lPA64FLgRcDv3GKZW8AfgwYIXu9/i3wl8Aq4K+Bf7+Yes3aHPp2TouI+yPiqxHRiIjHgT8BfnjGYr8fEQciYhfwXrIgBfgF4H9GxMMR0QB+D7h8MbP9iPjFiBiZ5/S9C/z4H0fEkxFxAHh3Rz1zeX++7BiwFagC742IekR8GrhvoVrNOjn07Zwm6cWSPi/paUlHyIJ79YzFnuy4/ARwUX75YuB97Rk6cAAQsO4slz1fPQstexHwVEw/SuITS1mYFZ9D3851twHfAjZHxHlk7RrNWGZDx+WNwJ788pPAL8yYpQ9ExD8v9KCSPiTp2Dynhxb48fnqmUtnwO8F1knqfH4bF6rVrJND3851K4AjwDFJLwX+yxzL/FdJo5I2AG8DPpmPfwi4VdLLASStlPTmxTxoRPzniBie5/TyBX78ZknrJa0ie5P65ALLt30FaAC/nG/U/QngykX+rBng0Ldz368CPwkcBf6UuQP0c8D9wAPA3wEfBoiIzwK/D3wibw3tAK5Zhpr/Cvgi8Fh+WtSHqyJikmyj888CB4H/CHzm7JRoRSV/iYqZWXl4pm9mViIOfTOzEnHom5mViEPfzKxEevogTqtXr45NmzZ1uwwzs3PK/fff/2xErJnrtp4O/U2bNrF9+/Zul2Fmdk6RNO8ntd3eMTMrEYe+mVmJOPTNzErEoW9mViIOfTOzEnHom5mViEPfzKxEChn6xyca/OEXH+Hruw52uxQzs55SyNCfaLR4/5d28uDuw90uxcyspxQy9NMk+za5RsvfFWBm1qmQoV/JQ7/ZanW5EjOz3lLI0G/P9OtNz/TNzDoVMvRPzvQd+mZmnQoZ+u7pm5nNrZChL4lKIvf0zcxmKGToQzbb90zfzGy6woZ+JRENb8g1M5umuKGfJt6Qa2Y2Q3FDPxEN9/TNzKYpbOiniTzTNzObobChX0nkD2eZmc2wYOhL2iDpy5IelvSQpLfl46sk3S3p0fx8NB+XpPdL2inpQUlXdNzXjfnyj0q68ew9LUhTz/TNzGZazEy/AfxKRLwM2ArcLOky4BbgnojYDNyTXwe4Bticn24CboPsTQJ4F/Aq4ErgXe03irOhmiTeZdPMbIYFQz8i9kbE1/LLR4GHgXXAdcBH88U+Crwxv3wd8BeR+SowIulC4PXA3RFxICIOAncDVy/ps+mQ+sNZZmazPK+evqRNwCuBe4ELImIvZG8MwNp8sXXAkx0/tjsfm2985mPcJGm7pO379+9/PuVNk3o/fTOzWRYd+pKGgb8B3h4RR0616BxjcYrx6QMRt0fElojYsmbNmsWWN0sl9SdyzcxmWlToS6qSBf7HIuIz+fAzeduG/HxfPr4b2NDx4+uBPacYPytS9/TNzGZZzN47Aj4MPBwRf9hx0zagvQfOjcDnOsbfku/FsxU4nLd/7gJeJ2k034D7unzsrKi6p29mNktlEcv8APAzwDckPZCP/TrwHuBTkt4K7ALenN92J3AtsBM4AfwcQEQckPS7wH35cr8TEQeW5FnMwT19M7PZFgz9iPgn5u7HA1w1x/IB3DzPfd0B3PF8CjxdlVRM1D3TNzPrVNhP5KZJQt09fTOzaQob+u7pm5nNVtjQd0/fzGy2woZ+xcfeMTObpbCh7/30zcxmK2zo+0tUzMxmK3ToN93TNzObprih72PvmJnNUtjQ99clmpnNVtjQryQJ9aZ7+mZmnQoc+p7pm5nNVNjQT93TNzObpbCh75m+mdlshQ399oezsoN+mpkZFDj0K0l2NGjP9s3MTipu6KdZ6Luvb2Z2UnFD3zN9M7NZChv6aZI9Nc/0zcxOKmzot2f6DX9Ay8xsSmFDP3V7x8xslsKGftUbcs3MZils6Ld7+p7pm5mdVNjQb/f0fdA1M7OTChv67umbmc1W2NB3T9/MbLbChr57+mZmsxU29Kf203fom5lNKWzop/5wlpnZLIUNfc/0zcxmK27op+7pm5nNVNjQTz3TNzObpbChf/LQyu7pm5m1FTb006lP5Hqmb2bWVtjQr7qnb2Y2S2FD3z19M7PZChv67umbmc22YOhLukPSPkk7Osb+u6SnJD2Qn67tuO1WSTslPSLp9R3jV+djOyXdsvRPZTr39M3MZlvMTP8jwNVzjP9RRFyen+4EkHQZcD3w8vxnPigplZQCHwCuAS4DbsiXPWsqqY+yaWY2U2WhBSLiHyRtWuT9XQd8IiImgO9K2glcmd+2MyIeA5D0iXzZbz7vihep4i9GNzOb5Ux6+r8k6cG8/TOaj60DnuxYZnc+Nt/4WTPV0/exd8zMppxu6N8GXApcDuwF/nc+rjmWjVOMzyLpJknbJW3fv3//aZYHqY+nb2Y2y2mFfkQ8ExHNiGgBf8rJFs5uYEPHouuBPacYn+u+b4+ILRGxZc2aNadTHuADrpmZzeW0Ql/ShR1X3wS09+zZBlwvqSbpEmAz8C/AfcBmSZdI6iPb2Lvt9MtemL8u0cxstgU35Er6OPAaYLWk3cC7gNdIupysRfM48AsAEfGQpE+RbaBtADdHRDO/n18C7gJS4I6IeGjJn02HantDrnfZNDObspi9d26YY/jDp1j+3cC75xi/E7jzeVV3BpJESP5wlplZp8J+Iheyvn7d7R0zsymFDv00kXv6ZmYdCh361SRxT9/MrEOhQz9N5Z6+mVmHQod+JZH30zcz61Do0E8Tub1jZtah0KFfSRLP9M3MOhQ79N3TNzObptChn7qnb2Y2TaFDv+L99M3Mpil06KdJ4q9LNDPrUOjQr7qnb2Y2TaFD3z19M7PpCh367umbmU1X6ND3h7PMzKYrdOhnH85yT9/MrK3YoZ+6vWNm1qnYoe8NuWZm0xQ69P0lKmZm0xU69CtJQr3pnr6ZWVuhQ98zfTOz6Qod+pXUPX0zs07FDn3P9M3Mpil06PuAa2Zm0xU69LOZvjfkmpm1FTv03dM3M5um2KHvnr6Z2TSFDv3UX4xuZjZNoUO/koiGP5xlZjal0KGfJqIV0PJs38wMKHjoV1MB0AyHvpkZFDz00yR7et6Ya2aWKXToV5Jspu+DrpmZZQod+mke+p7pm5llCh367Z6+d9s0M8sUOvTd0zczm67Qod/u6Xumb2aWWTD0Jd0haZ+kHR1jqyTdLenR/Hw0H5ek90vaKelBSVd0/MyN+fKPSrrx7Dyd6do9fX9Ay8wss5iZ/keAq2eM3QLcExGbgXvy6wDXAJvz003AbZC9SQDvAl4FXAm8q/1GcTZV3NM3M5tmwdCPiH8ADswYvg74aH75o8AbO8b/IjJfBUYkXQi8Hrg7Ig5ExEHgbma/kSy5inv6ZmbTnG5P/4KI2AuQn6/Nx9cBT3Ystzsfm298Fkk3Sdouafv+/ftPs7zMyfaOQ9/MDJZ+Q67mGItTjM8ejLg9IrZExJY1a9acUTEV76dvZjbN6Yb+M3nbhvx8Xz6+G9jQsdx6YM8pxs+qNO/p1/3tWWZmwOmH/jagvQfOjcDnOsbfku/FsxU4nLd/7gJeJ2k034D7unzsrPJM38xsuspCC0j6OPAaYLWk3WR74bwH+JSktwK7gDfni98JXAvsBE4APwcQEQck/S5wX77c70TEzI3DS669Idc9fTOzzIKhHxE3zHPTVXMsG8DN89zPHcAdz6u6M9TeZdMzfTOzTKE/kdvee8c9fTOzTKFDf6qn7/aOmRlQ+NDPe/pu75iZAUUPfff0zcymKXToT30i1z19MzOg4KFf8WEYzMymKXTo++sSzcymK3ToV1NvyDUz61To0D8503dP38wMCh767Z5+3T19MzOg4KHvnr6Z2XSFDv2+Svb0JhrNLldiZtYbCh36tUpKNRXHJx36ZmZQ8NAHGKpVODbe6HYZZmY9ofih31fh+IRD38wMShD6K/orHHPom5kBJQj9oVqF45MOfTMzKEnoH5vwhlwzMyhB6A/XUvf0zcxyhQ/9oT7vvWNm1lb40B/u9947ZmZtxQ/9fENuhA/FYGZW+NAfqlVoBYzVvTHXzKwUoQ94X30zM0oQ+sO1FMAbc83MKEXoVwE47n31zcyKH/pD7Zm+2ztmZsUP/eG8p+/dNs3MShD67Q25Pv6OmVkJQr890z/qDblmZuUJfbd3zMxKEPqDfSmSQ9/MDEoQ+pKyg655l00zs+KHPmS7bXqmb2ZWktAfrlU45r13zMxKFPree8fMrByhP1TzMfXNzOAMQ1/S45K+IekBSdvzsVWS7pb0aH4+mo9L0vsl7ZT0oKQrluIJLEb2PbkOfTOzpZjp/0hEXB4RW/LrtwD3RMRm4J78OsA1wOb8dBNw2xI89qK0v0jFzKzszkZ75zrgo/nljwJv7Bj/i8h8FRiRdOFZePxZhmsVH2XTzIwzD/0Avijpfkk35WMXRMRegPx8bT6+Dniy42d352PTSLpJ0nZJ2/fv33+G5WWGvCHXzAyAyhn+/A9ExB5Ja4G7JX3rFMtqjrFZX1wbEbcDtwNs2bJlSb7YdriWMtlsMdlo0VcpxbZrM7M5nVECRsSe/Hwf8FngSuCZdtsmP9+XL74b2NDx4+uBPWfy+Is15OPvmJkBZxD6koYkrWhfBl4H7AC2ATfmi90IfC6/vA14S74Xz1bgcLsNdLb5e3LNzDJn0t65APispPb9/FVEfEHSfcCnJL0V2AW8OV/+TuBaYCdwAvi5M3js52WFj6lvZgacQehHxGPAK+YYfw64ao7xAG4+3cc7E1MzfW/MNbOSK8VWTbd3zMwypQj9k1+k4n31zazcShH6Q7UU8N47ZmalCP0VtSrg9o6ZWSlCvz3Td+ibWdmVIvQracKqoT6ePjLe7VLMzLqqFKEPsGF0gCcPnOh2GWZmXVWa0F8/OshTB8e6XYaZWVeVJ/RXDbD74Bit1pIcw83M7JxUmtDfMDrIZLPFvqMT3S7FzKxrShP660cHAHjyoPv6ZlZepQn9DasGAdjt0DezEitN6K8byWf6B7wx18zKqzSh319NWbui5t02zazUShP6kLV4dnu3TTMrsXKF/uiAN+SaWamVKvTXjw6y9/A4jWar26WYmXVFqUJ/w6oBmq1g72Efg8fMyqlUob9+NNtt0y0eMyurUoX+hjz0d3u3TTMrqVKF/oUj/aSJ+NbTR7tdiplZV5Qq9KtpwtXf8wI+du8T7HrOLR4zK59ShT7Ab/7YZVQS8a5tO4jwETfNrFxKF/ovWNnPO177Yr78yH7ueuiZbpdjZrasShf6AD/7/Zt40dph3nfPo57tm1mplDL0K2nCTT/4Qh7ee4SvfOe5bpdjZrZsShn6AG+4/CJWD/fxZ//03W6XYma2bEob+v3VlJ/ZuokvfWsfO/d5F04zK4fShj7AT2/dSF8l4bf/zzcZrze7XY6Z2VlX6tA/f7jGb7/h5fzTzmd5yx3/wpHxerdLMjM7q0od+gA3XLmR913/Sr72xEHe9IH/x0N7Dne7JDOzs6bS7QJ6wRtekW3UffsnHuBNH/hnfuKKdWxYNcirLz2fKzaOdrs8M7Ml49DPff+lq/nC23+I3/zcDu566GkOnshaPTdcuZFbrnkpKweqXa7QzOzMOfQ7rBrq4wM/eQUAR8fr/PGXdvKn//gYf/fgHm64ciPfd/EoD+89ylAt5ae3Xkx/NaXZCo5PNjiv328KZtb71MufSN2yZUts3769qzXseOowt/3f7/D3O/bSCpAgAi5ZPcRrL7uAbQ/s4ekj41x24Xn86MvW8p9+6IV+AzCzrpJ0f0RsmfM2h/7i7Dk0xt7D47z0BSv4+q5D/MbffoMnDpzgh1+8hlesH+He7z7Hvd89wPlDffzM1k3sOTTGkwdPMDJYZc1wjRddsIKX5KeVg1VareDIeJ2hWoVqWvrt6Wa2hHoq9CVdDbwPSIE/i4j3zLdsL4X+TPVmixMTTVYOnpzVf2P3YX5r2w6+vusQq4b6uPj8QY6M1XnmyATHJhpTy60cqHJsokGzla370cEql64Z5kVrh3n22CS7DhwHYKCaUqumDFRTvnf9Sq562QWsHx2Yuh911LOiv0pfxW8eZtZDoS8pBb4NvBbYDdwH3BAR35xr+V4O/flEBIdO1BkZrCJpamzv4XEeeeYo3376KLsOZP8BjA72cWyiwTNHJti57yiP7T/O6uEaG88fJBGM11uM15scm2jw8N4jtE7xq5JgzXCNi0YGWDcywESjxXf2HyMi2HzBCi44r0ZE9p0C5w1UGaim1Jst6s0Wk80WEdBfSahVU/qrKX2pqDeDyWaLeqNFK2DlQIWVg1UioNkKBvpShvoqDPalDNUqDNWyyxHQaLU4eLzOgROTrByoctFIPxFw6ESdw2PZqa+SsHq4j+FahTQR1TTJzpOENBVjk02ePTbBeL1JXyVhRa3K2vx5fPuZoxweq7NudIC1K2okOvkW2GgFR8bqTDRarB7uY+VAlWYrGG+0mKg3mWy2EGKs3uQr33mOr+06yMZVg7xiwwirh/tYUasyVEun/gtLxNTvsv37qKYJK2oVWhEcHW8QQH81ob+SkiQiIjgynr2xD1RTapWEJDlZY0S2bicb2akVMNCXUknE8YkGY/UmaSLSRFSSJD/XtHNJM/8MptSbLVJp1mM2WzH1d1RNT30fZ8Nko8WhsUlGB/vm/Q93stGikkyvfS6NZrbeFlo2W79BfzU95f01W8GJyQbDtcqyr5d2Di/V454q9Jd7Q+6VwM6IeAxA0ieA64A5Q/9cJInRob5ZYxeNDHDRyAA/8pK1p3W/B49P8g+P7ufIWLZXUWf+R8DBE5NTLaiH9x6hkoqXXbgCIR555ij3P3GQRNkL4OhEg/Z7vQR9aYLyN5lzQXu7ylItPzpY5dBYfcH7nHk/891vXyWh2Yqp/+TaapWESqLsjbR55pOtdMabQCURicTxycbU77L9399c9aSJGK5ViAjG8k+kt4NxvN5EiOH+CrX8Pjqfa+R/gZ1jrYBWxFQYN1uR1ZZmtbUCDhyfBCBRtuPEZKPFeKPFcK3Civ7K1KQgUfbfayURwfRQTKSpN9+2RFBJEiqppk0gJupNjow38ueWMNhXyd74WkGjFVPrIBHsOzpBI3+TPn+4j/QUbyTz3bJQaLefR0xdhxOTDQ6dqNPI11clzSY+l28c4S/f+qpT3t/pWO7QXwc82XF9N7D0z6qARof6uO7ydUtyX81WMNlo0VdJpv1hRwQTjRYT9Wz235cmVCuiL5+RHR6rc2S8kc18EeONJscnGpyY7DifbCCyF/nKwSqrhvo4dKLOnkNjpIkYGawyMtDHeQMVJhotnj06wYnJJs38RdhotWg0s4CqVRNWD9cYqKZMNlt5q2ycZgte8oJhRgb7eOrgGM8dn5j2/BKJ8/J217PHJjh0Ivuvor+a0F9NqaYJEZAm8H0Xj3LpmmGOTjT45p4jHDoxybGJ7Pm0W3CtyGbHkc8Wh2sV6nk9aZKwoj8LjbH8P7PxepNKKkYH+6gkYrzRYmyyyXijSaMZ9FUS+tJk2rkEY5NNGq1guFZhoJpmAZqHdXbeys6b84znp6FaheFaJf/vpjn1+0g7ThHBeL3FkfE6icRAXxb2Y5NZ+A/0ZY9/fKLBRL1FO8vUEXdTY1ND0x8nETRbTNUHsGZFjdHBPp47NsG+oxPUKgn9fSnHJxocGWtMbQNr/76b7bBHSNmbSrOVtT7PG8j+E6s3WzRbQb15cl008nXUl4pVQzUqqTg8VufEZINUIk0S0iSr79hEnUYzuHCkn/P6q+w7OsGB45PzHnZ9vrfrU00agpNvFCfXJfm6rjAyWKWaJjSaWf31ZouLVg7McU9nbrlDf663wWmrStJNwE0AGzduXI6aSidNTr7IO0miP2/vzOX84RrnD9fOdnldcV5/la0vPL/bZZiddcu95W83sKHj+npgT+cCEXF7RGyJiC1r1qxZ1uLMzIpuuUP/PmCzpEsk9QHXA9uWuQYzs9Ja1vZORDQk/RJwF9kum3dExEPLWYOZWZkt+2EYIuJO4M7lflwzM/Ohlc3MSsWhb2ZWIg59M7MSceibmZVITx9lU9J+4IkzuIvVwLNLVM7Z0us19np94BqXimtcGr1Q48URMecHnXo69M+UpO3zHXSoV/R6jb1eH7jGpeIal0av1+j2jplZiTj0zcxKpOihf3u3C1iEXq+x1+sD17hUXOPS6OkaC93TNzOz6Yo+0zczsw4OfTOzEilk6Eu6WtIjknZKuqXb9QBI2iDpy5IelvSQpLfl46sk3S3p0fx8tAdqTSV9XdLn8+uXSLo3r/GT+WGxu1nfiKRPS/pWvj5f3UvrUdI78t/xDkkfl9TfC+tQ0h2S9kna0TE253pT5v35a+hBSVd0qb4/yH/PD0r6rKSRjttuzet7RNLrz3Z989XYcduvSgpJq/Pry74OF6NwoZ9/+foHgGuAy4AbJF3W3aoAaAC/EhEvA7YCN+d13QLcExGbgXvy6932NuDhjuu/D/xRXuNB4K1dqeqk9wFfiIiXAq8gq7Un1qOkdcAvA1si4nvIDiF+Pb2xDj8CXD1jbL71dg2wOT/dBNzWpfruBr4nIr4X+DZwK0D+2rkeeHn+Mx/MX/vdqBFJG4DXArs6hruxDhcWEYU6Aa8G7uq4fitwa7frmqPOz5H9kTwCXJiPXQg80uW61pO9+P8d8Hmyr7h8FqjMtX67UN95wHfJd0LoGO+J9cjJ74FeRXbo8s8Dr++VdQhsAnYstN6APwFumGu55axvxm1vAj6WX572uib7jo5Xd2Md5mOfJpuAPA6s7uY6XOhUuJk+c3/5+tJ8o/gSkbQJeCVwL3BBROwFyM/Xdq8yAN4L/DeglV8/HzgUEY38erfX5wuB/cCf5y2oP5M0RI+sx4h4CvhfZDO+vcBh4H56ax12mm+99eLr6OeBv88v90x9kt4APBUR/zrjpp6psVMRQ3/BL1/vJknDwN8Ab4+II92up5OkHwf2RcT9ncNzLNrN9VkBrgBui4hXAsfpjZYYAHlP/DrgEuAiYIjs3/yZeuZvch499XuX9E6yFunH2kNzLLbs9UkaBN4J/NZcN88x1vXfexFDf8EvX+8WSVWywP9YRHwmH35G0oX57RcC+7pVH/ADwBskPQ58gqzF815gRFL7W9a6vT53A7sj4t78+qfJ3gR6ZT3+KPDdiNgfEXXgM8D301vrsNN8661nXkeSbgR+HPipyPsk9E59l5K9wf9r/rpZD3xN0gvonRqnKWLo9+SXr0sS8GHg4Yj4w46btgE35pdvJOv1d0VE3BoR6yNiE9l6+1JE/BTwZeA/5It1u8angSclvSQfugr4Jr2zHncBWyUN5r/zdn09sw5nmG+9bQPeku+BshU43G4DLSdJVwO/BrwhIk503LQNuF5STdIlZBtL/2W564uIb0TE2ojYlL9udgNX5H+nPbEOZ+n2RoWztKHlWrIt/d8B3tntevKa/i3Zv3YPAg/kp2vJeub3AI/m56u6XWte72uAz+eXX0j2gtoJ/DVQ63JtlwPb83X5t8BoL61H4LeBbwE7gL8Ear2wDoGPk21nqJOF01vnW29krYkP5K+hb5DtjdSN+naS9cXbr5kPdSz/zry+R4BrurUOZ9z+OCc35C77OlzMyYdhMDMrkSK2d8zMbB4OfTOzEnHom5mViEPfzKxEHPpmZiXi0DczKxGHvplZifx/+aYuSLHywWIAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.47320249999999997\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5252887222222222\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5492582777777778\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5599119444444445\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.5463498888888889\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6176732222222223\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6470263333333333\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "auc =  0.6511520000000001\n"
     ]
    }
   ],
   "source": [
    "objective = 'soft-boundary'\n",
    "i = 0\n",
    "nu = 0.3\n",
    "outlier_nu = 0.3\n",
    "label_list = []\n",
    "auc_list = []\n",
    "time_list = []\n",
    "confusionmatrix_list = []\n",
    "\n",
    "for label_num in np.arange(10):\n",
    "    ## 資料設定\n",
    "    train_dataset, x_train, x_test, y_train, y_test = load_data(label_num, batch_size = 200,\n",
    "                                                                outlier_nu = outlier_nu)\n",
    "    ## 訓練 autoencoder\n",
    "    pre_time = time.time()\n",
    "    ae = Autoencoder(label_num)\n",
    "    ae.train_ae(350, train_dataset)\n",
    "    encoder = ae.encoder\n",
    "    ## 訓練 SVDD\n",
    "    SVDD = train_SVDD(train_dataset = train_dataset, label_num = label_num , objective = objective, encoder = encoder)\n",
    "    SVDD.train(x_train, n_epochs = 150)\n",
    "    radius_end =SVDD.train_result(x_train, y_train)\n",
    "    test_auc, confusionmatrix  = SVDD.test(x_test, y_test, radius_end)\n",
    "    train_time = time.time() - pre_time\n",
    "    print('auc = ', test_auc)\n",
    "    label_list.append(label_num)\n",
    "    auc_list.append(test_auc)\n",
    "    time_list.append(train_time)\n",
    "    confusionmatrix_list.append(confusionmatrix)\n",
    "    i += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "SVDD_ver2_test.ipynb",
   "provenance": []
  },
  "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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}