fault_detection_semigan_.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/ZahraFarahzadii/PV-fault-detection-semigan/blob/main/fault_detection_semigan_.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from numpy.random import randint\n",
        "import numpy as np\n",
        "import os\n",
        "\n",
        "import cv2\n",
        "import matplotlib.pyplot as plt\n",
        "import tensorflow as tf\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Conv2D, Dense, Flatten\n",
        "from keras.datasets.mnist import load_data\n",
        "from keras.layers import Dropout\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "from numpy.random.mtrand import shuffle\n",
        "\n",
        "from numpy import expand_dims, zeros, ones, asarray\n",
        "from numpy.random import randn, randint\n",
        "import tensorflow as tf\n",
        "import keras\n",
        "import pandas as pd\n",
        "from keras.preprocessing.sequence import TimeseriesGenerator\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.layers import Dense, Dropout, LSTM\n",
        "from sklearn.model_selection import  train_test_split\n",
        "from sklearn.preprocessing import MinMaxScaler,StandardScaler\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "from keras.datasets.mnist import load_data\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "from keras.models import Model, Sequential\n",
        "\n",
        "from keras.layers import Input, Dense, Reshape, Flatten, Conv2D, Conv2DTranspose\n",
        "from keras.layers import LeakyReLU, Dropout, Lambda, Activation,ReLU\n",
        "\n",
        "\n",
        "from matplotlib import pyplot as plt\n",
        "from keras import backend as K\n",
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "import keras\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Conv2D, MaxPooling2D\n",
        "from keras.layers import Activation, Dropout, Flatten, Dense, BatchNormalization\n",
        "from keras import backend as K"
      ],
      "metadata": {
        "id": "Q4GsmqHeiG-t"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "khlR3kKbiJ16",
        "outputId": "555b41c5-1e84-41ee-be7d-543be3865b55"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "create train data"
      ],
      "metadata": {
        "id": "Mma8BkzooHar"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "training_data=[]\n",
        "def create_training_data():\n",
        "  DATADIR='/content/drive/MyDrive/data4class/train'\n",
        "  CATEGORIES=['class0','class1','class2','class3']\n",
        "  for category in CATEGORIES:\n",
        "    path=os.path.join(DATADIR,category)\n",
        "    class_num=CATEGORIES.index(category)\n",
        "    for img in os.listdir(path):\n",
        "      try:\n",
        "        img_array=cv2.imread(os.path.join(path,img))\n",
        "        # image_rgb = cv2.cvtColor(img_array, cv2.COLOR_BGR2RGB)\n",
        "        # new_array=cv2.resize(image_rgb,(384,384))\n",
        "        new_array=cv2.resize(img_array,(384,384))\n",
        "        training_data.append([new_array,class_num])\n",
        "      except Exception as e:\n",
        "        pass\n",
        "create_training_data()\n"
      ],
      "metadata": {
        "id": "aJNrLjnAiL3-"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import random\n",
        "rnd_seed=42\n",
        "random.seed(rnd_seed)\n",
        "random.shuffle(training_data)\n"
      ],
      "metadata": {
        "id": "Xqz5ty9hiOf8"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "xtrain=[]\n",
        "ytrain=[]\n",
        "for features, label in training_data:\n",
        "  xtrain.append(features)\n",
        "  ytrain.append(label)\n",
        "xtrain=np.array(xtrain).reshape(-1,384,384,3)\n",
        "ytrain=np.array(ytrain)"
      ],
      "metadata": {
        "id": "3vLyPGcDjgUn"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "create test data"
      ],
      "metadata": {
        "id": "0oy-jBNVoRh3"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "test_data=[]\n",
        "def create_test_data():\n",
        "  DATADIR='/content/drive/MyDrive/data4class/valid'\n",
        "  CATEGORIES=['class0','class1','class2','class3']\n",
        "  for category in CATEGORIES:\n",
        "    path=os.path.join(DATADIR,category)\n",
        "    class_num=CATEGORIES.index(category)\n",
        "    for img in os.listdir(path):\n",
        "      try:\n",
        "        img_array=cv2.imread(os.path.join(path,img))\n",
        "        # image_rgb = cv2.cvtColor(img_array, cv2.COLOR_BGR2RGB)\n",
        "        # new_array=cv2.resize(image_rgb,(384,384))\n",
        "        new_array=cv2.resize(img_array,(384,384))\n",
        "        test_data.append([new_array,class_num])\n",
        "      except Exception as e:\n",
        "        pass\n",
        "create_test_data()"
      ],
      "metadata": {
        "id": "s-_mOKodiQjG"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import random\n",
        "rnd_seed=42\n",
        "random.seed(rnd_seed)\n",
        "random.shuffle(test_data)"
      ],
      "metadata": {
        "id": "sQiUsnhdiVVO"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "xtest=[]\n",
        "ytest=[]\n",
        "for features, label in test_data:\n",
        "  xtest.append(features)\n",
        "  ytest.append(label)\n",
        "xtest=np.array(xtest).reshape(-1,384,384,3)\n",
        "ytest=np.array(ytest)\n"
      ],
      "metadata": {
        "id": "mAo_bD3DiVxe"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "semi supervised gan"
      ],
      "metadata": {
        "id": "NnNsEjJHvqNe"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# define the standalone generator model\n",
        "def define_generator(latent_dim):\n",
        "\n",
        "\tin_lat = Input(shape=(latent_dim,))\n",
        "\t#Start with enough dense nodes to be reshaped and ConvTransposed to 384x384x3\n",
        "\tn_nodes = 256 * 96 * 96\n",
        "\tX = Dense(n_nodes)(in_lat)\n",
        "\tX = LeakyReLU(alpha=0.2)(X)\n",
        "\tX = Reshape((96, 96, 256))(X)\n",
        "\n",
        "\tX = Conv2DTranspose(128, (3,3), strides=(2,2), padding='same')(X) #192x192x128\n",
        "\tX = LeakyReLU(alpha=0.2)(X)\n",
        "\n",
        "\tX = Conv2DTranspose(64, (3,3), strides=(1,1), padding='same')(X) #192x192x64\n",
        "\tX = LeakyReLU(alpha=0.2)(X)\n",
        "\t# output\n",
        "\tout_layer = Conv2DTranspose(3, (3,3), strides=(2,2), activation='tanh',\n",
        "                             padding='same')(X) #384x384x3\n",
        "# define model\n",
        "\tmodel = Model(in_lat, out_layer)\n",
        "\treturn model"
      ],
      "metadata": {
        "id": "iifpXrtEvwwr"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "gen_model=define_generator(100)\n",
        "from tensorflow.keras.utils import plot_model\n",
        "plot_model(gen_model,to_file='model_plot.png',show_shapes=True,\n",
        "           show_dtype=True, show_layer_names=False,rankdir='TB',\n",
        "           expand_nested=False,dpi=96,layer_range=None,show_layer_activations=True)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 976
        },
        "id": "adW60YdHLw1g",
        "outputId": "40dcb673-ae1d-4048-d0f7-1972c20483bf"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<IPython.core.display.Image object>"
            ],
            "image/png": "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\n"
          },
          "metadata": {},
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "\n",
        "# define the base discriminator model for sup and unsup discriminators\n",
        "#This is the base discriminator that supervised and unsupervised are going to share weights from.\n",
        "def define_discriminator(in_shape=(384,384,3), n_classes=4):\n",
        "    in_image = Input(shape=in_shape)\n",
        "    X = Conv2D(32, (3,3), strides=(2,2), padding='same')(in_image)\n",
        "    X = LeakyReLU(alpha=0.2)(X)\n",
        "\n",
        "    X = Conv2D(64, (3,3), strides=(2,2), padding='same')(X)\n",
        "    X = LeakyReLU(alpha=0.2)(X)\n",
        "\n",
        "    X = Conv2D(128, (3,3), strides=(2,2), padding='same')(X)\n",
        "    X = LeakyReLU(alpha=0.2)(X)\n",
        "\n",
        "    X = Flatten()(X)\n",
        "    X = Dropout(0.4)(X) #Consider adding more dropout layers to minimize overfitting\n",
        "    X = Dense(n_classes)(X)\n",
        "    model = Model(inputs=in_image, outputs=X)\n",
        "    return model"
      ],
      "metadata": {
        "id": "4OnHzLQzwKNz"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#Define the supervised discriminator.\n",
        "#Multiclass classification, so we will use softmax activation.\n",
        "#To avoid converting our labels to categorical, we will work with sparse categorical crossentropy loss.\n",
        "def define_sup_discriminator(disc):\n",
        "    model=Sequential()\n",
        "    model.add(disc)\n",
        "    model.add(Activation('softmax'))\n",
        "    # sparse categorical loss so we dont have to convert our Y to categorical\n",
        "    model.compile(optimizer=Adam(lr=0.0002, beta_1=0.5),\n",
        "                  loss=\"sparse_categorical_crossentropy\",metrics=['accuracy'])\n",
        "    return model"
      ],
      "metadata": {
        "id": "l97ayECbwQL4"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#Define the unsupervised discriminator\n",
        "#Takes the output of the supervised, just before the softmax activation.\n",
        "#Then, adds a layer with calculation of sum of exponential outputs.\n",
        "\n",
        "#This custom activation layer gives a value close to 0 for smaller activations\n",
        "#in the prior discriminator layer. It gives values close to 1 for large activations.\n",
        "#This way it gives low activation for fake images. No need for sigmoid anymore.\n",
        "\n",
        "# custom activation function for the unsupervised discriminator\n",
        "#D(x) = Z(x) / (Z(x) + 1) where Z(x) = sum(exp(l(x))). l(x) is the output from sup discr. prior to softmax\n",
        "def custom_activation(x):\n",
        "    Z_x = K.sum(K.exp(x), axis=-1, keepdims=True)\n",
        "    D_x = Z_x /(Z_x+1)\n",
        "\n",
        "    return D_x"
      ],
      "metadata": {
        "id": "XGwR-MYGwVgz"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "\n",
        "def define_unsup_discriminator(disc):\n",
        "    model=Sequential()\n",
        "    model.add(disc)\n",
        "    model.add(Lambda(custom_activation))\n",
        "    model.compile(loss='binary_crossentropy', optimizer=Adam(lr=0.0002, beta_1=0.5))\n",
        "    return model"
      ],
      "metadata": {
        "id": "PPA0-O0ywXD0"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "disc=define_discriminator()\n",
        "disc_sup=define_sup_discriminator(disc)\n",
        "disc_unsup=define_unsup_discriminator(disc)\n"
      ],
      "metadata": {
        "id": "Wznq6Ln849yu",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "89c9dcba-41d5-466c-acb1-9bd5096527d2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/keras/optimizer_v2/adam.py:105: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead.\n",
            "  super(Adam, self).__init__(name, **kwargs)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from tensorflow.keras.utils import plot_model\n",
        "plot_model(disc,to_file='model_plot.png',show_shapes=True,\n",
        "           show_dtype=True, show_layer_names=False,rankdir='TB',\n",
        "           expand_nested=False,dpi=96,layer_range=None,show_layer_activations=True)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "XX8zFspCMLaV",
        "outputId": "dd6a7bad-4cf0-4d0b-c2e1-f1add68ec82c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<IPython.core.display.Image object>"
            ],
            "image/png": "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\n"
          },
          "metadata": {},
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# define the combined generator and discriminator model, for updating the generator\n",
        "def define_gan(gen_model, disc_unsup):\n",
        "\n",
        "\tdisc_unsup.trainable = False # make unsup. discriminator not trainable\n",
        "\tgan_output = disc_unsup(gen_model.output) #Gen. output is the input to disc.\n",
        "\tmodel = Model(gen_model.input, gan_output)\n",
        "\tmodel.compile(loss='binary_crossentropy', optimizer=Adam(lr=0.0002, beta_1=0.5))\n",
        "\treturn model"
      ],
      "metadata": {
        "id": "wc58J4LMwamU"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# gan_model = define_gan(gen_model, disc_unsup)\n",
        "# print(gan_model.summary())"
      ],
      "metadata": {
        "id": "VLBaml6Z5F2-"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# load the images\n",
        "def load_real_samples(n_classes=4):\n",
        "    trainX=xtrain\n",
        "    trainy=ytrain\n",
        "    X = trainX.astype('float32')\n",
        "    X = (X - 127.5) / 127.5  # scale from [0,255] to [-1,1] as we will be using tanh activation in generator .\n",
        "    print(X.shape, trainy.shape)\n",
        "    return [X, trainy]"
      ],
      "metadata": {
        "id": "5rdeelR8wgRN"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#data = load_real_samples()\n",
        "\n",
        "#select subset of the dataset for supervised training\n",
        "# pick only 100 samples to be used in supervised training.\n",
        "#Also, we need to ensure we pick 10 samples per class to ensure a good balance\n",
        "def select_supervised_samples(dataset, n_samples=100, n_classes=4):\n",
        " \tX, y = dataset\n",
        " \tX_list, y_list = list(), list()\n",
        " \tn_per_class = int(n_samples / n_classes) #Number of amples per class.\n",
        " \tfor i in range(n_classes):\n",
        "         X_with_class = X[y == i] # get all images for this class\n",
        "         ix = randint(0, len(X_with_class), n_per_class) # choose random images for each class\n",
        "         [X_list.append(X_with_class[j]) for j in ix] # add to list\n",
        "         [y_list.append(i) for j in ix]\n",
        " \treturn asarray(X_list), asarray(y_list) #Returns a list of 2 numpy arrays corresponding to X and Y"
      ],
      "metadata": {
        "id": "abVlsul2wx09"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Pick real samples from the dataset.\n",
        "#Return both images and corresponding labels in addition to y=1 indicating that the images are real.\n",
        "# we will not use the labels for unsupervised, only used for supervised.\n",
        "def generate_real_samples(dataset, n_samples):\n",
        "\n",
        "\timages, labels = dataset\n",
        "\tix = randint(0, images.shape[0], n_samples)\n",
        "\tX, labels = images[ix], labels[ix] #Select random images and corresponding labels\n",
        "\ty = ones((n_samples, 1)) #Label all images as 1 as these are real images. (for the discriminator training)\n",
        "\treturn [X, labels], y"
      ],
      "metadata": {
        "id": "rhrVvsWmw4Hg"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# generate latent points, to be used as inputs to the generator.\n",
        "def generate_latent_points(latent_dim, n_samples):\n",
        "\tz_input = randn(latent_dim * n_samples)\n",
        "\tz_input = z_input.reshape(n_samples, latent_dim) # reshape for input to the network\n",
        "\treturn z_input"
      ],
      "metadata": {
        "id": "L9Exw3sVw7L5"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Generate fake images using the generator and above latent points as input to it.\n",
        "def generate_fake_samples(generator, latent_dim, n_samples):\n",
        "\n",
        "\tz_input = generate_latent_points(latent_dim, n_samples)\n",
        "\tfake_images = generator.predict(z_input)\n",
        "\t# create class labels\n",
        "\ty = zeros((n_samples, 1)) #Label all images as 0 as these are fake images. (for the discriminator training)\n",
        "\treturn fake_images, y"
      ],
      "metadata": {
        "id": "Vdd7JG8bw8c9"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# report accuracy and save plots & the model periodically.\n",
        "def summarize_performance(step, gen_model, disc_sup, latent_dim, dataset, n_samples=100):\n",
        "\t# Generate fake images\n",
        "\tX, _ = generate_fake_samples(gen_model, latent_dim, n_samples)\n",
        "\n",
        "\tX = (X + 1) / 2.0 # scale to [0,1] for plotting\n",
        "\t# plot images\n",
        "\tfor i in range(100):\n",
        "\t\tplt.subplot(10, 10, 1 + i)\n",
        "\t\tplt.axis('off')\n",
        "\t\tplt.imshow(X[i, :, :, 0], cmap='gray_r')\n",
        "\t# save plot to drive\n",
        "\tfilename1 = 'generated_plot_%04d.png' % (step+1)\n",
        "\tplt.savefig(filename1)\n",
        "\tplt.close()\n",
        "\n",
        "\t# evaluate the discriminator\n",
        "\n",
        "\tX, y = dataset\n",
        "\t_, acc = disc_sup.evaluate(X, y, verbose=0)\n",
        "\tprint('Discriminator Accuracy: %.3f%%' % (acc * 100))\n",
        "\t# save the generator model\n",
        "\tfilename2 = 'gen_model_%04d.h5' % (step+1)\n",
        "\tgen_model.save(filename2)\n",
        "\t# save the Discriminator (classifier) model\n",
        "\tfilename3 = 'disc_sup_%04d.h5' % (step+1)\n",
        "\tdisc_sup.save(filename3)\n",
        "\tprint('>Saved: %s, %s, and %s' % (filename1, filename2, filename3))"
      ],
      "metadata": {
        "id": "8kKPXOnBw_IR"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# train the generator and discriminator\n",
        "def train(gen_model, disc_unsup, disc_sup, gan_model, dataset, latent_dim, n_epochs=20, n_batch=60):\n",
        "\n",
        "    # select supervised dataset for training.\n",
        "    #Remember that we are not using all 60k images, just a subset (100 images, 10 per class. )\n",
        "\tX_sup, y_sup = select_supervised_samples(dataset)\n",
        "\t#print(X_sup.shape, y_sup.shape)\n",
        "\n",
        "\tbat_per_epo = int(dataset[0].shape[0] / n_batch)\n",
        "\t# iterations\n",
        "\tn_steps = bat_per_epo * n_epochs\n",
        "\n",
        "\thalf_batch = int(n_batch / 2)\n",
        "\tprint('n_epochs=%d, n_batch=%d, 1/2=%d, b/e=%d, steps=%d' % (n_epochs,\n",
        "                                                              n_batch, half_batch,\n",
        "                                                              bat_per_epo, n_steps))\n",
        "\t #  enumerate epochs\n",
        "\tfor i in range(n_steps):\n",
        "\t\t\t\t# update supervised discriminator (disc_sup) on real samples.\n",
        "        # we use real labels to train as this is supervised.\n",
        "\t\t\t[Xsup_real, ysup_real], _ = generate_real_samples([X_sup, y_sup], half_batch)\n",
        "\t\t\tsup_loss, sup_acc = disc_sup.train_on_batch(Xsup_real, ysup_real)\n",
        "\t\t\t\t\t# update unsupervised discriminator (disc_unsup) - just like in our regular GAN.\n",
        "\t\t\t\t\t# we will not train on labels as this is unsupervised, just binary as in our regular GAN.\n",
        "\t\t\t\t\t#The y_real below indicates 1s telling the discriminator that these images are real.\n",
        "\t\t\t[X_real, _], y_real = generate_real_samples(dataset, half_batch)\n",
        "\t\t\td_loss_real = disc_unsup.train_on_batch(X_real, y_real)\n",
        "\t\t\t\t\t#Now train on fake.\n",
        "\t\t\tX_fake, y_fake = generate_fake_samples(gen_model, latent_dim, half_batch)\n",
        "\t\t\td_loss_fake = disc_unsup.train_on_batch(X_fake, y_fake)\n",
        "\t\t\t\t\t# update generator (gen) - like we do in regular GAN.\n",
        "\t\t\tX_gan, y_gan = generate_latent_points(latent_dim, n_batch), ones((n_batch, 1))\n",
        "\t\t\tgan_loss = gan_model.train_on_batch(X_gan, y_gan)\n",
        "\n",
        "\t\t\t# summarize loss on this batch\n",
        "\t\t\tprint('>%d, c[%.3f,%.0f], d[%.3f,%.3f], g[%.3f]' % (i+1, sup_loss, sup_acc*100, d_loss_real, d_loss_fake, gan_loss))\n",
        "\t\t\t# evaluate the model performance periodically\n",
        "\t\t\tif (i+1) % (bat_per_epo * 1) == 0:\n",
        "\t\t\t\tsummarize_performance(i, gen_model, disc_sup, latent_dim, dataset)"
      ],
      "metadata": {
        "id": "IT9zOQJgxF9T"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# TRAIN\n",
        "latent_dim = 100\n",
        "\n",
        "# create the discriminator models\n",
        "disc=define_discriminator()\n",
        "disc_sup=define_sup_discriminator(disc) #Supervised discriminator model\n",
        "disc_unsup=define_unsup_discriminator(disc) #Unsupervised discriminator model.\n",
        "\n",
        "gen_model = define_generator(latent_dim) #Generator\n",
        "gan_model = define_gan(gen_model, disc_unsup) #GAN\n",
        "dataset = load_real_samples() #Define the dataset by loading real samples. (This will be a list of 2 numpy arrays, X and y)\n",
        "\n",
        "# train the model\n",
        "\n",
        "train(gen_model, disc_unsup, disc_sup, gan_model, dataset, latent_dim, n_epochs=20, n_batch=60)\n"
      ],
      "metadata": {
        "id": "JxjgPWuDxNqP",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "61a39829-c661-414c-8180-ffa0276b8835"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/keras/optimizer_v2/adam.py:105: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead.\n",
            "  super(Adam, self).__init__(name, **kwargs)\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(1384, 384, 384, 3) (1384,)\n",
            "n_epochs=20, n_batch=60, 1/2=30, b/e=23, steps=460\n",
            ">1, c[1.393,17], d[0.152,1.620], g[0.223]\n",
            ">2, c[2.180,27], d[0.007,1.602], g[0.229]\n",
            ">3, c[2.500,33], d[0.004,1.563], g[0.246]\n",
            ">4, c[1.876,27], d[0.002,1.479], g[0.284]\n",
            ">5, c[1.515,30], d[0.001,1.320], g[0.367]\n",
            ">6, c[1.616,27], d[0.001,1.074], g[0.538]\n",
            ">7, c[1.346,50], d[0.001,0.774], g[0.843]\n",
            ">8, c[1.346,30], d[0.001,0.468], g[1.343]\n",
            ">9, c[1.339,33], d[0.001,0.238], g[1.971]\n",
            ">10, c[1.273,47], d[0.002,0.107], g[2.700]\n",
            ">11, c[1.306,30], d[0.000,0.069], g[3.069]\n",
            ">12, c[1.272,27], d[0.000,0.056], g[3.297]\n",
            ">13, c[1.345,33], d[0.000,0.032], g[3.802]\n",
            ">14, c[1.533,17], d[0.001,0.030], g[4.195]\n",
            ">15, c[1.193,57], d[0.000,0.020], g[4.510]\n",
            ">16, c[1.226,40], d[0.002,0.011], g[5.053]\n",
            ">17, c[1.380,57], d[0.000,0.006], g[5.390]\n",
            ">18, c[1.452,47], d[0.000,0.006], g[5.514]\n",
            ">19, c[1.244,47], d[0.000,0.007], g[5.438]\n",
            ">20, c[1.301,43], d[0.000,0.006], g[5.651]\n",
            ">21, c[1.290,40], d[0.001,0.004], g[5.978]\n",
            ">22, c[1.437,40], d[0.001,0.003], g[6.066]\n",
            ">23, c[1.400,23], d[0.000,0.003], g[6.052]\n",
            "Discriminator Accuracy: 46.026%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0023.png, gen_model_0023.h5, and disc_sup_0023.h5\n",
            ">24, c[1.358,37], d[0.000,0.005], g[5.842]\n",
            ">25, c[1.332,37], d[0.000,0.007], g[5.818]\n",
            ">26, c[1.538,50], d[0.005,0.005], g[5.995]\n",
            ">27, c[2.409,30], d[0.000,0.003], g[6.277]\n",
            ">28, c[1.533,27], d[0.000,0.005], g[6.438]\n",
            ">29, c[1.675,37], d[0.002,0.004], g[6.733]\n",
            ">30, c[1.208,47], d[0.001,0.005], g[7.200]\n",
            ">31, c[1.333,43], d[0.000,0.003], g[7.498]\n",
            ">32, c[1.171,57], d[0.000,0.003], g[7.759]\n",
            ">33, c[1.273,30], d[0.000,0.002], g[7.899]\n",
            ">34, c[1.067,57], d[0.000,0.006], g[9.182]\n",
            ">35, c[1.262,37], d[0.000,0.000], g[9.220]\n",
            ">36, c[1.166,50], d[0.000,0.002], g[8.337]\n",
            ">37, c[1.116,50], d[0.000,0.001], g[8.370]\n",
            ">38, c[1.088,53], d[0.000,0.001], g[8.618]\n",
            ">39, c[1.078,60], d[0.000,0.003], g[10.383]\n",
            ">40, c[1.090,57], d[0.000,0.000], g[10.196]\n",
            ">41, c[1.044,57], d[0.000,0.000], g[8.830]\n",
            ">42, c[0.902,70], d[0.000,0.002], g[10.191]\n",
            ">43, c[1.078,40], d[0.000,0.000], g[10.816]\n",
            ">44, c[0.924,63], d[0.000,0.000], g[9.124]\n",
            ">45, c[1.007,53], d[0.000,0.003], g[11.614]\n",
            ">46, c[1.005,57], d[0.000,0.000], g[12.740]\n",
            "Discriminator Accuracy: 31.286%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0046.png, gen_model_0046.h5, and disc_sup_0046.h5\n",
            ">47, c[1.136,47], d[0.000,0.000], g[10.517]\n",
            ">48, c[1.076,53], d[0.000,0.001], g[9.181]\n",
            ">49, c[1.205,43], d[0.000,0.001], g[10.455]\n",
            ">50, c[1.099,43], d[0.001,0.000], g[10.035]\n",
            ">51, c[0.984,67], d[0.000,0.000], g[9.736]\n",
            ">52, c[1.253,47], d[0.000,0.000], g[9.964]\n",
            ">53, c[1.094,40], d[0.000,0.000], g[9.850]\n",
            ">54, c[1.086,43], d[0.000,0.000], g[9.941]\n",
            ">55, c[0.897,63], d[0.000,0.000], g[10.591]\n",
            ">56, c[0.888,57], d[0.000,0.000], g[10.147]\n",
            ">57, c[0.860,70], d[0.000,0.001], g[12.976]\n",
            ">58, c[1.095,43], d[0.000,0.000], g[11.747]\n",
            ">59, c[0.881,60], d[0.000,0.001], g[12.829]\n",
            ">60, c[0.950,50], d[0.002,0.002], g[11.021]\n",
            ">61, c[1.361,40], d[0.000,0.000], g[9.995]\n",
            ">62, c[1.067,47], d[0.000,0.001], g[11.004]\n",
            ">63, c[1.294,43], d[0.001,0.001], g[10.014]\n",
            ">64, c[0.845,63], d[0.003,0.257], g[15.425]\n",
            ">65, c[2.196,27], d[0.003,0.000], g[15.425]\n",
            ">66, c[1.872,47], d[0.002,0.000], g[15.425]\n",
            ">67, c[1.390,27], d[0.000,0.000], g[15.425]\n",
            ">68, c[0.930,60], d[0.000,0.000], g[15.425]\n",
            ">69, c[0.975,60], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 38.873%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0069.png, gen_model_0069.h5, and disc_sup_0069.h5\n",
            ">70, c[1.105,53], d[0.000,0.000], g[15.425]\n",
            ">71, c[1.069,50], d[0.000,0.000], g[15.425]\n",
            ">72, c[0.953,57], d[0.000,0.000], g[15.425]\n",
            ">73, c[1.061,53], d[0.000,0.000], g[15.425]\n",
            ">74, c[1.144,43], d[0.000,0.000], g[15.425]\n",
            ">75, c[0.820,63], d[0.000,0.000], g[15.425]\n",
            ">76, c[1.291,43], d[0.000,0.000], g[15.425]\n",
            ">77, c[0.993,57], d[0.000,0.000], g[15.425]\n",
            ">78, c[0.957,50], d[0.000,0.000], g[15.425]\n",
            ">79, c[0.736,73], d[0.000,0.000], g[15.425]\n",
            ">80, c[1.103,50], d[0.000,0.000], g[15.425]\n",
            ">81, c[0.986,47], d[0.000,0.000], g[15.425]\n",
            ">82, c[0.907,63], d[0.000,0.000], g[15.425]\n",
            ">83, c[1.038,47], d[0.000,0.000], g[15.425]\n",
            ">84, c[0.978,43], d[0.000,0.000], g[15.425]\n",
            ">85, c[1.006,53], d[0.000,0.000], g[15.425]\n",
            ">86, c[0.888,60], d[0.000,0.000], g[15.425]\n",
            ">87, c[0.797,63], d[0.000,0.000], g[15.425]\n",
            ">88, c[0.747,70], d[0.000,0.000], g[15.425]\n",
            ">89, c[1.235,33], d[0.000,0.000], g[15.425]\n",
            ">90, c[0.925,57], d[0.000,0.000], g[15.425]\n",
            ">91, c[0.917,67], d[0.000,0.000], g[15.425]\n",
            ">92, c[0.623,73], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 33.309%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0092.png, gen_model_0092.h5, and disc_sup_0092.h5\n",
            ">93, c[0.672,70], d[0.000,0.000], g[15.425]\n",
            ">94, c[1.057,57], d[0.000,0.000], g[15.425]\n",
            ">95, c[1.006,53], d[0.000,0.000], g[15.425]\n",
            ">96, c[0.783,57], d[0.001,0.000], g[15.425]\n",
            ">97, c[1.022,47], d[0.000,0.000], g[15.425]\n",
            ">98, c[0.893,50], d[0.000,0.000], g[15.425]\n",
            ">99, c[1.027,50], d[0.000,0.000], g[15.425]\n",
            ">100, c[0.947,53], d[0.000,0.000], g[15.425]\n",
            ">101, c[0.923,47], d[0.000,0.000], g[15.425]\n",
            ">102, c[0.925,53], d[0.000,0.000], g[15.425]\n",
            ">103, c[0.918,53], d[0.000,0.000], g[15.425]\n",
            ">104, c[0.905,57], d[0.000,0.000], g[15.425]\n",
            ">105, c[0.844,53], d[0.000,0.000], g[15.425]\n",
            ">106, c[0.943,50], d[0.000,0.000], g[15.425]\n",
            ">107, c[0.732,60], d[0.000,0.000], g[15.425]\n",
            ">108, c[0.897,53], d[0.000,0.000], g[15.425]\n",
            ">109, c[0.695,67], d[0.000,0.000], g[15.425]\n",
            ">110, c[0.844,57], d[0.000,0.000], g[15.425]\n",
            ">111, c[0.906,53], d[0.000,0.000], g[15.425]\n",
            ">112, c[0.877,60], d[0.000,0.000], g[15.425]\n",
            ">113, c[0.842,63], d[0.000,0.000], g[15.425]\n",
            ">114, c[0.759,67], d[0.000,0.000], g[15.425]\n",
            ">115, c[0.900,60], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 44.364%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0115.png, gen_model_0115.h5, and disc_sup_0115.h5\n",
            ">116, c[0.722,63], d[0.000,0.000], g[15.425]\n",
            ">117, c[0.634,70], d[0.000,0.000], g[15.425]\n",
            ">118, c[0.817,67], d[0.000,0.000], g[15.425]\n",
            ">119, c[0.780,67], d[0.000,0.000], g[15.425]\n",
            ">120, c[0.754,70], d[0.000,0.000], g[15.425]\n",
            ">121, c[1.001,60], d[0.000,0.000], g[15.425]\n",
            ">122, c[0.825,57], d[0.000,0.000], g[15.425]\n",
            ">123, c[0.815,57], d[0.000,0.000], g[15.425]\n",
            ">124, c[0.756,70], d[0.000,0.000], g[15.425]\n",
            ">125, c[0.997,50], d[0.000,0.000], g[15.425]\n",
            ">126, c[0.514,83], d[0.000,0.000], g[15.425]\n",
            ">127, c[1.078,50], d[0.000,0.000], g[15.425]\n",
            ">128, c[0.662,73], d[0.000,0.000], g[15.425]\n",
            ">129, c[0.827,67], d[0.000,0.000], g[15.425]\n",
            ">130, c[0.812,63], d[0.000,0.000], g[15.425]\n",
            ">131, c[0.823,63], d[0.000,0.000], g[15.425]\n",
            ">132, c[1.116,37], d[0.000,0.000], g[15.425]\n",
            ">133, c[0.768,80], d[0.000,0.000], g[15.425]\n",
            ">134, c[0.751,67], d[0.000,0.000], g[15.425]\n",
            ">135, c[0.971,37], d[0.000,0.000], g[15.425]\n",
            ">136, c[1.124,53], d[0.000,0.000], g[15.425]\n",
            ">137, c[1.055,40], d[0.000,0.000], g[15.425]\n",
            ">138, c[0.914,47], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 50.361%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0138.png, gen_model_0138.h5, and disc_sup_0138.h5\n",
            ">139, c[0.808,77], d[0.000,0.000], g[15.425]\n",
            ">140, c[0.824,70], d[0.000,0.000], g[15.425]\n",
            ">141, c[0.694,63], d[0.000,0.000], g[15.425]\n",
            ">142, c[0.945,50], d[0.000,0.000], g[15.425]\n",
            ">143, c[0.866,60], d[0.000,0.000], g[15.425]\n",
            ">144, c[0.768,63], d[0.000,0.000], g[15.425]\n",
            ">145, c[0.806,60], d[0.000,0.000], g[15.425]\n",
            ">146, c[1.297,50], d[0.000,0.000], g[15.425]\n",
            ">147, c[1.144,47], d[0.000,0.000], g[15.425]\n",
            ">148, c[0.891,50], d[0.000,0.000], g[15.425]\n",
            ">149, c[0.784,63], d[0.000,0.000], g[15.425]\n",
            ">150, c[0.914,57], d[0.000,0.000], g[15.425]\n",
            ">151, c[0.847,57], d[0.000,0.000], g[15.425]\n",
            ">152, c[0.777,77], d[0.000,0.000], g[15.425]\n",
            ">153, c[0.734,80], d[0.000,0.000], g[15.425]\n",
            ">154, c[0.700,63], d[0.000,0.000], g[15.425]\n",
            ">155, c[0.908,53], d[0.000,0.000], g[15.425]\n",
            ">156, c[0.798,60], d[0.000,0.000], g[15.425]\n",
            ">157, c[0.895,50], d[0.000,0.000], g[15.425]\n",
            ">158, c[0.782,67], d[0.000,0.000], g[15.425]\n",
            ">159, c[0.875,50], d[0.000,0.000], g[15.425]\n",
            ">160, c[0.870,50], d[0.000,0.000], g[15.425]\n",
            ">161, c[0.970,63], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 40.462%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0161.png, gen_model_0161.h5, and disc_sup_0161.h5\n",
            ">162, c[0.669,63], d[0.000,0.000], g[15.425]\n",
            ">163, c[0.970,53], d[0.000,0.000], g[15.425]\n",
            ">164, c[0.982,50], d[0.000,0.000], g[15.425]\n",
            ">165, c[0.789,70], d[0.000,0.000], g[15.425]\n",
            ">166, c[0.771,63], d[0.000,0.000], g[15.425]\n",
            ">167, c[0.929,63], d[0.000,0.000], g[15.425]\n",
            ">168, c[0.620,77], d[0.000,0.000], g[15.425]\n",
            ">169, c[0.777,63], d[0.000,0.000], g[15.425]\n",
            ">170, c[1.018,53], d[0.000,0.000], g[15.425]\n",
            ">171, c[0.782,63], d[0.000,0.000], g[15.425]\n",
            ">172, c[1.369,40], d[0.000,0.000], g[15.425]\n",
            ">173, c[0.424,80], d[0.000,0.000], g[15.425]\n",
            ">174, c[1.304,40], d[0.002,0.000], g[15.425]\n",
            ">175, c[0.911,57], d[0.000,0.000], g[15.425]\n",
            ">176, c[1.168,40], d[0.000,0.000], g[15.425]\n",
            ">177, c[1.089,50], d[0.000,0.000], g[15.425]\n",
            ">178, c[0.807,60], d[0.000,0.000], g[15.425]\n",
            ">179, c[0.895,63], d[0.000,0.000], g[15.425]\n",
            ">180, c[0.877,67], d[0.000,0.000], g[15.425]\n",
            ">181, c[0.752,57], d[0.000,0.000], g[15.425]\n",
            ">182, c[0.680,73], d[0.000,0.000], g[15.425]\n",
            ">183, c[0.565,87], d[0.000,0.000], g[15.425]\n",
            ">184, c[0.766,67], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 34.682%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0184.png, gen_model_0184.h5, and disc_sup_0184.h5\n",
            ">185, c[0.612,80], d[0.000,0.000], g[15.425]\n",
            ">186, c[0.669,63], d[0.000,0.000], g[15.425]\n",
            ">187, c[0.773,70], d[0.000,0.000], g[15.425]\n",
            ">188, c[0.838,60], d[0.000,0.000], g[15.425]\n",
            ">189, c[0.818,53], d[0.000,0.000], g[15.425]\n",
            ">190, c[0.783,67], d[0.000,0.000], g[15.425]\n",
            ">191, c[0.946,50], d[0.000,0.000], g[15.425]\n",
            ">192, c[0.663,83], d[0.000,0.000], g[15.425]\n",
            ">193, c[0.757,57], d[0.000,0.000], g[15.425]\n",
            ">194, c[0.842,50], d[0.000,0.000], g[15.425]\n",
            ">195, c[0.799,67], d[0.000,0.000], g[15.425]\n",
            ">196, c[0.707,63], d[0.000,0.000], g[15.425]\n",
            ">197, c[0.776,60], d[0.000,0.000], g[15.425]\n",
            ">198, c[0.801,67], d[0.000,0.000], g[15.425]\n",
            ">199, c[0.623,90], d[0.000,0.000], g[15.425]\n",
            ">200, c[0.801,70], d[0.000,0.000], g[15.425]\n",
            ">201, c[0.918,67], d[0.000,0.000], g[15.425]\n",
            ">202, c[0.773,63], d[0.000,0.000], g[15.425]\n",
            ">203, c[0.768,60], d[0.000,0.000], g[15.425]\n",
            ">204, c[0.839,53], d[0.000,0.000], g[15.425]\n",
            ">205, c[0.865,53], d[0.000,0.000], g[15.425]\n",
            ">206, c[0.601,77], d[0.000,0.000], g[15.425]\n",
            ">207, c[0.642,70], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 54.769%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0207.png, gen_model_0207.h5, and disc_sup_0207.h5\n",
            ">208, c[0.631,70], d[0.000,0.000], g[15.425]\n",
            ">209, c[0.660,77], d[0.000,0.000], g[15.425]\n",
            ">210, c[0.640,67], d[0.000,0.000], g[15.425]\n",
            ">211, c[0.692,70], d[0.000,0.000], g[15.425]\n",
            ">212, c[0.616,67], d[0.000,0.000], g[15.425]\n",
            ">213, c[1.017,47], d[0.000,0.000], g[15.425]\n",
            ">214, c[0.548,80], d[0.000,0.000], g[15.425]\n",
            ">215, c[0.664,70], d[0.000,0.000], g[15.425]\n",
            ">216, c[0.888,60], d[0.000,0.000], g[15.425]\n",
            ">217, c[0.815,53], d[0.000,0.000], g[15.425]\n",
            ">218, c[0.749,70], d[0.000,0.000], g[15.425]\n",
            ">219, c[0.614,73], d[0.000,0.000], g[15.425]\n",
            ">220, c[0.757,53], d[0.000,0.000], g[15.425]\n",
            ">221, c[0.964,50], d[0.000,0.000], g[15.425]\n",
            ">222, c[0.906,53], d[0.000,0.000], g[15.425]\n",
            ">223, c[0.931,50], d[0.000,0.000], g[15.425]\n",
            ">224, c[0.824,47], d[0.000,0.000], g[15.425]\n",
            ">225, c[1.281,50], d[0.000,0.000], g[15.425]\n",
            ">226, c[0.973,43], d[0.000,0.000], g[15.425]\n",
            ">227, c[0.874,47], d[0.000,0.000], g[15.425]\n",
            ">228, c[0.770,67], d[0.000,0.000], g[15.425]\n",
            ">229, c[0.737,67], d[0.000,0.000], g[15.425]\n",
            ">230, c[0.885,50], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 56.142%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0230.png, gen_model_0230.h5, and disc_sup_0230.h5\n",
            ">231, c[0.724,70], d[0.000,0.000], g[15.425]\n",
            ">232, c[0.710,63], d[0.000,0.000], g[15.425]\n",
            ">233, c[0.906,50], d[0.000,0.000], g[15.425]\n",
            ">234, c[0.671,67], d[0.000,0.000], g[15.425]\n",
            ">235, c[0.807,60], d[0.000,0.000], g[15.425]\n",
            ">236, c[0.687,80], d[0.000,0.000], g[15.425]\n",
            ">237, c[0.920,57], d[0.000,0.000], g[15.425]\n",
            ">238, c[0.626,73], d[0.000,0.000], g[15.425]\n",
            ">239, c[0.540,83], d[0.000,0.000], g[15.425]\n",
            ">240, c[0.683,67], d[0.000,0.000], g[15.425]\n",
            ">241, c[0.733,67], d[0.000,0.000], g[15.425]\n",
            ">242, c[0.731,70], d[0.000,0.000], g[15.425]\n",
            ">243, c[0.665,77], d[0.000,0.000], g[15.425]\n",
            ">244, c[0.657,63], d[0.000,0.000], g[15.425]\n",
            ">245, c[0.686,63], d[0.000,0.000], g[15.425]\n",
            ">246, c[0.834,57], d[0.000,0.000], g[15.425]\n",
            ">247, c[0.872,53], d[0.000,0.000], g[15.425]\n",
            ">248, c[0.621,73], d[0.000,0.000], g[15.425]\n",
            ">249, c[0.690,63], d[0.000,0.000], g[15.425]\n",
            ">250, c[0.874,70], d[0.000,0.000], g[15.425]\n",
            ">251, c[0.632,70], d[0.000,0.000], g[15.425]\n",
            ">252, c[0.694,80], d[0.000,0.000], g[15.425]\n",
            ">253, c[0.532,80], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 42.847%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0253.png, gen_model_0253.h5, and disc_sup_0253.h5\n",
            ">254, c[0.891,53], d[0.000,0.000], g[15.425]\n",
            ">255, c[0.556,77], d[0.000,0.000], g[15.425]\n",
            ">256, c[0.739,57], d[0.000,0.000], g[15.425]\n",
            ">257, c[0.459,77], d[0.000,0.000], g[15.425]\n",
            ">258, c[0.827,50], d[0.000,0.000], g[15.425]\n",
            ">259, c[0.517,77], d[0.000,0.000], g[15.425]\n",
            ">260, c[0.654,67], d[0.000,0.000], g[15.425]\n",
            ">261, c[0.639,63], d[0.000,0.000], g[15.425]\n",
            ">262, c[0.665,60], d[0.000,0.000], g[15.425]\n",
            ">263, c[0.649,70], d[0.000,0.000], g[15.425]\n",
            ">264, c[0.547,70], d[0.000,0.000], g[15.425]\n",
            ">265, c[0.481,80], d[0.000,0.000], g[15.425]\n",
            ">266, c[0.389,87], d[0.000,0.000], g[15.425]\n",
            ">267, c[0.562,77], d[0.000,0.000], g[15.425]\n",
            ">268, c[0.710,70], d[0.000,0.000], g[15.425]\n",
            ">269, c[0.761,60], d[0.000,0.000], g[15.425]\n",
            ">270, c[0.750,63], d[0.000,0.000], g[15.425]\n",
            ">271, c[0.590,77], d[0.000,0.000], g[15.425]\n",
            ">272, c[0.562,73], d[0.000,0.000], g[15.425]\n",
            ">273, c[0.581,77], d[0.000,0.000], g[15.425]\n",
            ">274, c[0.610,70], d[0.000,0.000], g[15.425]\n",
            ">275, c[0.777,70], d[0.000,0.000], g[15.425]\n",
            ">276, c[0.650,77], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 48.410%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0276.png, gen_model_0276.h5, and disc_sup_0276.h5\n",
            ">277, c[0.600,70], d[0.000,0.000], g[15.425]\n",
            ">278, c[1.042,60], d[0.000,0.000], g[15.425]\n",
            ">279, c[0.725,70], d[0.000,0.000], g[15.425]\n",
            ">280, c[0.699,63], d[0.000,0.000], g[15.425]\n",
            ">281, c[0.495,80], d[0.000,0.000], g[15.425]\n",
            ">282, c[0.544,83], d[0.000,0.000], g[15.425]\n",
            ">283, c[0.547,90], d[0.000,0.000], g[15.425]\n",
            ">284, c[0.602,70], d[0.000,0.000], g[15.425]\n",
            ">285, c[0.712,53], d[0.000,0.000], g[15.425]\n",
            ">286, c[0.738,60], d[0.000,0.000], g[15.425]\n",
            ">287, c[0.653,63], d[0.000,0.000], g[15.425]\n",
            ">288, c[0.572,87], d[0.000,0.000], g[15.425]\n",
            ">289, c[0.756,60], d[0.000,0.000], g[15.425]\n",
            ">290, c[0.607,70], d[0.000,0.000], g[15.425]\n",
            ">291, c[0.682,60], d[0.000,0.000], g[15.425]\n",
            ">292, c[0.740,63], d[0.000,0.000], g[15.425]\n",
            ">293, c[0.798,63], d[0.000,0.000], g[15.425]\n",
            ">294, c[0.711,63], d[0.000,0.000], g[15.425]\n",
            ">295, c[0.549,73], d[0.000,0.000], g[15.425]\n",
            ">296, c[0.506,80], d[0.000,0.000], g[15.425]\n",
            ">297, c[0.718,60], d[0.000,0.000], g[15.425]\n",
            ">298, c[0.641,63], d[0.000,0.000], g[15.425]\n",
            ">299, c[0.719,67], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 49.061%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0299.png, gen_model_0299.h5, and disc_sup_0299.h5\n",
            ">300, c[0.528,77], d[0.000,0.000], g[15.425]\n",
            ">301, c[0.538,73], d[0.000,0.000], g[15.425]\n",
            ">302, c[0.732,63], d[0.000,0.000], g[15.425]\n",
            ">303, c[0.483,83], d[0.000,0.000], g[15.425]\n",
            ">304, c[0.629,63], d[0.000,0.000], g[15.425]\n",
            ">305, c[0.653,60], d[0.000,0.000], g[15.425]\n",
            ">306, c[0.565,77], d[0.000,0.000], g[15.425]\n",
            ">307, c[0.888,60], d[0.000,0.000], g[15.425]\n",
            ">308, c[0.668,77], d[0.000,0.000], g[15.425]\n",
            ">309, c[0.747,67], d[0.000,0.000], g[15.425]\n",
            ">310, c[0.722,63], d[0.000,0.000], g[15.425]\n",
            ">311, c[0.711,67], d[0.000,0.000], g[15.425]\n",
            ">312, c[0.705,67], d[0.000,0.000], g[15.425]\n",
            ">313, c[0.601,70], d[0.000,0.000], g[15.425]\n",
            ">314, c[0.743,67], d[0.000,0.000], g[15.425]\n",
            ">315, c[1.238,43], d[0.000,0.000], g[15.425]\n",
            ">316, c[0.599,63], d[0.000,0.000], g[15.425]\n",
            ">317, c[0.916,53], d[0.000,0.000], g[15.425]\n",
            ">318, c[0.650,77], d[0.000,0.000], g[15.425]\n",
            ">319, c[0.601,70], d[0.000,0.000], g[15.425]\n",
            ">320, c[0.695,67], d[0.000,0.000], g[15.425]\n",
            ">321, c[0.683,63], d[0.000,0.000], g[15.425]\n",
            ">322, c[0.685,73], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 52.818%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0322.png, gen_model_0322.h5, and disc_sup_0322.h5\n",
            ">323, c[0.648,73], d[0.000,0.000], g[15.425]\n",
            ">324, c[0.587,70], d[0.000,0.000], g[15.425]\n",
            ">325, c[0.859,53], d[0.000,0.000], g[15.425]\n",
            ">326, c[0.766,80], d[0.000,0.000], g[15.425]\n",
            ">327, c[0.737,60], d[0.000,0.000], g[15.425]\n",
            ">328, c[0.619,63], d[0.000,0.000], g[15.425]\n",
            ">329, c[0.521,80], d[0.000,0.000], g[15.425]\n",
            ">330, c[0.742,67], d[0.000,0.000], g[15.425]\n",
            ">331, c[0.617,83], d[0.000,0.000], g[15.425]\n",
            ">332, c[0.474,87], d[0.000,0.000], g[15.425]\n",
            ">333, c[0.729,67], d[0.000,0.000], g[15.425]\n",
            ">334, c[0.587,70], d[0.000,0.000], g[15.425]\n",
            ">335, c[0.396,80], d[0.000,0.000], g[15.425]\n",
            ">336, c[0.667,70], d[0.000,0.000], g[15.425]\n",
            ">337, c[0.495,77], d[0.000,0.000], g[15.425]\n",
            ">338, c[0.594,73], d[0.000,0.000], g[15.425]\n",
            ">339, c[0.601,67], d[0.000,0.000], g[15.425]\n",
            ">340, c[0.498,90], d[0.000,0.000], g[15.425]\n",
            ">341, c[0.877,60], d[0.000,0.000], g[15.425]\n",
            ">342, c[0.720,67], d[0.000,0.000], g[15.425]\n",
            ">343, c[0.769,63], d[0.000,0.000], g[15.425]\n",
            ">344, c[0.555,80], d[0.000,0.000], g[15.425]\n",
            ">345, c[0.508,80], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 48.699%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0345.png, gen_model_0345.h5, and disc_sup_0345.h5\n",
            ">346, c[0.884,57], d[0.000,0.000], g[15.425]\n",
            ">347, c[0.721,63], d[0.000,0.000], g[15.425]\n",
            ">348, c[0.745,60], d[0.000,0.000], g[15.425]\n",
            ">349, c[0.637,63], d[0.000,0.000], g[15.425]\n",
            ">350, c[0.584,77], d[0.000,0.000], g[15.425]\n",
            ">351, c[0.441,83], d[0.000,0.000], g[15.425]\n",
            ">352, c[0.594,77], d[0.000,0.000], g[15.425]\n",
            ">353, c[0.460,80], d[0.000,0.000], g[15.425]\n",
            ">354, c[0.670,73], d[0.000,0.000], g[15.425]\n",
            ">355, c[0.571,77], d[0.000,0.000], g[15.425]\n",
            ">356, c[0.548,83], d[0.000,0.000], g[15.425]\n",
            ">357, c[0.654,63], d[0.000,0.000], g[15.425]\n",
            ">358, c[0.618,60], d[0.000,0.000], g[15.425]\n",
            ">359, c[0.449,73], d[0.000,0.000], g[15.425]\n",
            ">360, c[0.426,87], d[0.000,0.000], g[15.425]\n",
            ">361, c[0.462,80], d[0.000,0.000], g[15.425]\n",
            ">362, c[0.491,77], d[0.000,0.000], g[15.425]\n",
            ">363, c[0.659,70], d[0.000,0.000], g[15.425]\n",
            ">364, c[0.653,60], d[0.000,0.000], g[15.425]\n",
            ">365, c[0.625,67], d[0.000,0.000], g[15.425]\n",
            ">366, c[0.761,57], d[0.000,0.000], g[15.425]\n",
            ">367, c[0.606,77], d[0.000,0.000], g[15.425]\n",
            ">368, c[0.304,90], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 46.893%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0368.png, gen_model_0368.h5, and disc_sup_0368.h5\n",
            ">369, c[0.693,73], d[0.000,0.000], g[15.425]\n",
            ">370, c[0.490,87], d[0.000,0.000], g[15.425]\n",
            ">371, c[0.466,80], d[0.000,0.000], g[15.425]\n",
            ">372, c[0.644,73], d[0.000,0.000], g[15.425]\n",
            ">373, c[0.548,83], d[0.000,0.000], g[15.425]\n",
            ">374, c[0.546,73], d[0.000,0.000], g[15.425]\n",
            ">375, c[0.680,70], d[0.000,0.000], g[15.425]\n",
            ">376, c[0.717,63], d[0.000,0.000], g[15.425]\n",
            ">377, c[0.467,77], d[0.000,0.000], g[15.425]\n",
            ">378, c[0.516,90], d[0.000,0.000], g[15.425]\n",
            ">379, c[0.420,80], d[0.000,0.000], g[15.425]\n",
            ">380, c[0.637,63], d[0.000,0.000], g[15.425]\n",
            ">381, c[0.779,77], d[0.000,0.000], g[15.425]\n",
            ">382, c[0.771,63], d[0.000,0.000], g[15.425]\n",
            ">383, c[0.933,57], d[0.000,0.000], g[15.425]\n",
            ">384, c[0.606,80], d[0.000,0.000], g[15.425]\n",
            ">385, c[0.898,50], d[0.000,0.000], g[15.425]\n",
            ">386, c[0.414,87], d[0.000,0.000], g[15.425]\n",
            ">387, c[0.713,70], d[0.000,0.000], g[15.425]\n",
            ">388, c[0.651,70], d[0.000,0.000], g[15.425]\n",
            ">389, c[0.559,67], d[0.000,0.000], g[15.425]\n",
            ">390, c[0.511,80], d[0.000,0.000], g[15.425]\n",
            ">391, c[0.570,77], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 64.017%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0391.png, gen_model_0391.h5, and disc_sup_0391.h5\n",
            ">392, c[0.627,63], d[0.000,0.000], g[15.425]\n",
            ">393, c[0.608,70], d[0.000,0.000], g[15.425]\n",
            ">394, c[0.533,80], d[0.000,0.000], g[15.425]\n",
            ">395, c[0.780,63], d[0.000,0.000], g[15.425]\n",
            ">396, c[0.855,70], d[0.000,0.000], g[15.425]\n",
            ">397, c[0.596,73], d[0.000,0.000], g[15.425]\n",
            ">398, c[0.714,60], d[0.000,0.000], g[15.425]\n",
            ">399, c[0.507,77], d[0.000,0.000], g[15.425]\n",
            ">400, c[0.621,73], d[0.000,0.000], g[15.425]\n",
            ">401, c[0.547,80], d[0.000,0.000], g[15.425]\n",
            ">402, c[0.653,70], d[0.000,0.000], g[15.425]\n",
            ">403, c[0.670,80], d[0.000,0.000], g[15.425]\n",
            ">404, c[0.479,90], d[0.000,0.000], g[15.425]\n",
            ">405, c[0.563,77], d[0.000,0.000], g[15.425]\n",
            ">406, c[0.479,80], d[0.000,0.000], g[15.425]\n",
            ">407, c[0.648,67], d[0.000,0.000], g[15.425]\n",
            ">408, c[0.666,80], d[0.000,0.000], g[15.425]\n",
            ">409, c[0.730,60], d[0.000,0.000], g[15.425]\n",
            ">410, c[0.414,77], d[0.000,0.000], g[15.425]\n",
            ">411, c[0.551,80], d[0.000,0.000], g[15.425]\n",
            ">412, c[0.532,77], d[0.000,0.000], g[15.425]\n",
            ">413, c[0.638,67], d[0.000,0.000], g[15.425]\n",
            ">414, c[0.615,77], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 55.708%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0414.png, gen_model_0414.h5, and disc_sup_0414.h5\n",
            ">415, c[0.532,77], d[0.000,0.000], g[15.425]\n",
            ">416, c[0.478,77], d[0.000,0.000], g[15.425]\n",
            ">417, c[0.531,73], d[0.000,0.000], g[15.425]\n",
            ">418, c[0.574,67], d[0.000,0.000], g[15.425]\n",
            ">419, c[0.501,87], d[0.000,0.000], g[15.425]\n",
            ">420, c[0.605,70], d[0.000,0.000], g[15.425]\n",
            ">421, c[0.487,77], d[0.000,0.000], g[15.425]\n",
            ">422, c[0.566,70], d[0.000,0.000], g[15.425]\n",
            ">423, c[0.654,73], d[0.000,0.000], g[15.425]\n",
            ">424, c[0.486,77], d[0.000,0.000], g[15.425]\n",
            ">425, c[0.689,60], d[0.000,0.000], g[15.425]\n",
            ">426, c[0.534,70], d[0.000,0.000], g[15.425]\n",
            ">427, c[0.458,80], d[0.000,0.000], g[15.425]\n",
            ">428, c[0.731,63], d[0.000,0.000], g[15.425]\n",
            ">429, c[0.436,87], d[0.000,0.000], g[15.425]\n",
            ">430, c[0.776,63], d[0.000,0.000], g[15.425]\n",
            ">431, c[0.691,70], d[0.000,0.000], g[15.425]\n",
            ">432, c[0.516,77], d[0.000,0.000], g[15.425]\n",
            ">433, c[0.586,73], d[0.000,0.000], g[15.425]\n",
            ">434, c[0.532,77], d[0.000,0.000], g[15.425]\n",
            ">435, c[0.516,70], d[0.000,0.000], g[15.425]\n",
            ">436, c[0.437,77], d[0.000,0.000], g[15.425]\n",
            ">437, c[0.468,87], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 56.792%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0437.png, gen_model_0437.h5, and disc_sup_0437.h5\n",
            ">438, c[0.641,57], d[0.000,0.000], g[15.425]\n",
            ">439, c[0.518,73], d[0.000,0.000], g[15.425]\n",
            ">440, c[0.521,77], d[0.000,0.000], g[15.425]\n",
            ">441, c[0.433,83], d[0.000,0.000], g[15.425]\n",
            ">442, c[0.557,77], d[0.000,0.000], g[15.425]\n",
            ">443, c[0.565,73], d[0.000,0.000], g[15.425]\n",
            ">444, c[0.480,77], d[0.000,0.000], g[15.425]\n",
            ">445, c[0.535,77], d[0.000,0.000], g[15.425]\n",
            ">446, c[0.544,73], d[0.000,0.000], g[15.425]\n",
            ">447, c[0.633,77], d[0.000,0.000], g[15.425]\n",
            ">448, c[0.547,70], d[0.000,0.000], g[15.425]\n",
            ">449, c[0.725,63], d[0.000,0.000], g[15.425]\n",
            ">450, c[0.612,70], d[0.000,0.000], g[15.425]\n",
            ">451, c[0.460,83], d[0.000,0.000], g[15.425]\n",
            ">452, c[0.494,80], d[0.000,0.000], g[15.425]\n",
            ">453, c[0.590,70], d[0.000,0.000], g[15.425]\n",
            ">454, c[0.500,70], d[0.000,0.000], g[15.425]\n",
            ">455, c[0.430,90], d[0.000,0.000], g[15.425]\n",
            ">456, c[0.449,80], d[0.000,0.000], g[15.425]\n",
            ">457, c[0.588,67], d[0.000,0.000], g[15.425]\n",
            ">458, c[0.468,87], d[0.000,0.000], g[15.425]\n",
            ">459, c[0.568,80], d[0.000,0.000], g[15.425]\n",
            ">460, c[0.490,83], d[0.000,0.000], g[15.425]\n",
            "Discriminator Accuracy: 52.673%\n",
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            ">Saved: generated_plot_0460.png, gen_model_0460.h5, and disc_sup_0460.h5\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#EVALUATE THE SUPERVISED DISCRIMINATOR ON TEST DATA\n",
        "# This is the model we want as a classifier.\n",
        "\n",
        "from keras.models import load_model\n",
        "# load the model\n",
        "disc_sup_trained_model = load_model('disc_sup_0391.h5')\n",
        "\n",
        "# load the dataset\n",
        "testX=xtest\n",
        "testy = ytest\n",
        "\n",
        "\n",
        "testX = expand_dims(testX, axis=-1)\n",
        "\n",
        "# convert from ints to floats\n",
        "testX = testX.astype('float32')\n",
        "\n",
        "# scale from [0,255] to [-1,1]\n",
        "testX = (testX - 127.5) / 127.5"
      ],
      "metadata": {
        "id": "s49SJs-_xPtZ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "2bd3cfc2-8fc5-492b-8e05-95256c003c35"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "WARNING:tensorflow:Error in loading the saved optimizer state. As a result, your model is starting with a freshly initialized optimizer.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# evaluate the model\n",
        "_, test_acc = disc_sup_trained_model.evaluate(testX, testy, verbose=0)\n",
        "print('Test Accuracy: %.3f%%' % (test_acc * 100))\n",
        "\n",
        "# Predicting the Test set results\n",
        "y_pred_test = disc_sup_trained_model.predict(testX)\n",
        "prediction_test = np.argmax(y_pred_test, axis=1)\n",
        "\n",
        "# Confusion Matrix\n",
        "from sklearn.metrics import confusion_matrix\n",
        "import seaborn as sns\n",
        "cm = confusion_matrix(testy, prediction_test)\n",
        "sns.heatmap(cm, annot=True)\n"
      ],
      "metadata": {
        "id": "ofzYtah6xSHM",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 300
        },
        "outputId": "dbde6538-2d5e-4f08-fa73-6e1004c7b39e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Test Accuracy: 59.884%\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc7a36837d0>"
            ]
          },
          "metadata": {},
          "execution_count": 29
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#PREDICT / GENERATE IMAGES using the generator, just for fun.\n",
        "##################################################################\n",
        "\n",
        "# Plot generated images\n",
        "def show_plot(examples, n):\n",
        "\tfor i in range(n * n):\n",
        "\t\tplt.subplot(n, n, 1 + i)\n",
        "\t\tplt.axis('off')\n",
        "\t\tplt.imshow(examples[i, :, :, :], cmap='gray')\n",
        "\tplt.show()\n",
        "gen_trained_model = load_model('gen_model_0092.h5') #Model trained for 100 epochs\n",
        "# generate images\n",
        "latent_points = generate_latent_points(100, 25)  #Latent dim and n_samples\n",
        "# generate images\n",
        "X = gen_trained_model.predict(latent_points)\n",
        "# scale from [-1,1] to [0,1]\n",
        "X = (X + 1) / 2.0\n",
        "\n",
        "X = (X*255).astype(np.uint8)\n",
        "\n",
        "# plot the result\n",
        "show_plot(X, 2)\n"
      ],
      "metadata": {
        "id": "Y3eHkdmaxV5Q"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}