8.3.programando-una-GAN.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"accelerator":"GPU","colab":{"name":"8.3.programando-una-GAN.ipynb","version":"0.3.2","provenance":[{"file_id":"1lf1GEQUD-D-ilkh_kgdWEyfwJCVumuEt","timestamp":1566240158762},{"file_id":"10Pi7fY3yWf5cz4dCVW6SwZfp3e81cNwA","timestamp":1566239915206},{"file_id":"1-ARcRheRU7-bZSxw_uE2wC_UP2MfPBCy","timestamp":1566239315010},{"file_id":"https://github.com/tensorflow/docs/blob/master/site/en/r2/tutorials/generative/dcgan.ipynb","timestamp":1566022481822}],"collapsed_sections":[]},"kernelspec":{"name":"python3","display_name":"Python 3"}},"cells":[{"cell_type":"markdown","metadata":{"colab_type":"text","id":"rF2x3qooyBTI"},"source":["## 8.3 Deep Convolutional Generative Adversarial Network\n","### Deep Learning, Introducción práctica con Keras - Segunda parte"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"g5RstiiB8V-z","colab":{}},"source":["from __future__ import absolute_import, division, print_function, unicode_literals\n","\n","try:\n","  # %tensorflow_version only exists in Colab.\n","  %tensorflow_version 2.x\n","except Exception:\n","  pass\n"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"WZKbyU2-AiY-","outputId":"7c845b95-0e57-4ae7-9dc2-098fbe28e176","executionInfo":{"status":"ok","timestamp":1566247437675,"user_tz":-120,"elapsed":2019,"user":{"displayName":"Jordi Torres","photoUrl":"https://lh4.googleusercontent.com/-SfQ_VFnu6kE/AAAAAAAAAAI/AAAAAAAABzE/lQO8NYx2THo/s64/photo.jpg","userId":"14559872433417615139"}},"colab":{"base_uri":"https://localhost:8080/","height":34}},"source":["import tensorflow as tf\n","tf.__version__"],"execution_count":0,"outputs":[{"output_type":"execute_result","data":{"text/plain":["'2.0.0-beta1'"]},"metadata":{"tags":[]},"execution_count":21}]},{"cell_type":"code","metadata":{"colab_type":"code","id":"YfIk2es3hJEd","colab":{}},"source":["import imageio\n","import matplotlib.pyplot as plt\n","import numpy as np\n","import os\n","from tensorflow.keras import layers\n","import time\n"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"a4fYMGxGhrna","colab":{}},"source":["\n","(train_images, _), (_, _) = tf.keras.datasets.mnist.load_data()\n","\n","train_images = train_images.reshape(train_images.shape[0], 28, 28, 1).astype('float32')\n","\n","train_images = (train_images - 127.5) / 127.5 \n","\n","BUFFER_SIZE = 60000\n","BATCH_SIZE = 256\n","train_dataset = tf.data.Dataset.from_tensor_slices(train_images).shuffle(BUFFER_SIZE).batch(BATCH_SIZE)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"yZU3Ilr2gn5y","colab_type":"text"},"source":["## Modelos"]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"-tEyxE-GMC48"},"source":["### Generador"]},{"cell_type":"code","metadata":{"id":"QsDYOFtzKIHi","colab_type":"code","colab":{}},"source":["def make_generator_model():\n","    model = tf.keras.Sequential()\n","    model.add(layers.Dense(7*7*256, use_bias=False, input_shape=(100,)))\n","\n","    model.add(layers.Reshape((7, 7, 256)))\n","\n","    model.add(layers.Conv2DTranspose(128, (5, 5), strides=(2, 2), padding='same'))\n","    model.add(layers.BatchNormalization())\n","    model.add(layers.LeakyReLU(alpha=0.01))\n","\n","    model.add(layers.Conv2DTranspose(64, (5, 5), strides=(1, 1), padding='same'))\n","    model.add(layers.BatchNormalization())\n","    model.add(layers.LeakyReLU(alpha=0.01))\n","\n","    model.add(layers.Conv2DTranspose(1, (5, 5), strides=(2, 2), padding='same', activation='tanh'))\n","\n","    return model"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"-QlXbX2EwMCX","colab_type":"code","colab":{}},"source":["generator = make_generator_model()"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"gl7jcC7TdPTG","outputId":"0414a6cc-2394-4aa0-bd48-479fb8c13a0a","executionInfo":{"status":"ok","timestamp":1566247439130,"user_tz":-120,"elapsed":3457,"user":{"displayName":"Jordi Torres","photoUrl":"https://lh4.googleusercontent.com/-SfQ_VFnu6kE/AAAAAAAAAAI/AAAAAAAABzE/lQO8NYx2THo/s64/photo.jpg","userId":"14559872433417615139"}},"colab":{"base_uri":"https://localhost:8080/","height":286}},"source":["noise_dim = 100\n","noise = tf.random.normal([1, noise_dim])\n","generated_image = generator(noise, training=False)\n","\n","plt.imshow(generated_image[0, :, :, 0], cmap='gray')"],"execution_count":0,"outputs":[{"output_type":"execute_result","data":{"text/plain":["<matplotlib.image.AxesImage at 0x7f1037002be0>"]},"metadata":{"tags":[]},"execution_count":27},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n5hdJlJpfJFFqfpFE/R8JdbGWJ2lgOQAAAABJRU5ErkJggg==\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"D0IKnaCtg6WE"},"source":["### Discriminador"]},{"cell_type":"code","metadata":{"id":"mypJk88So23v","colab_type":"code","colab":{}},"source":["def make_discriminator_model():\n","    model = tf.keras.Sequential()\n","    model.add(layers.Conv2D(32, (5, 5), strides=(2, 2), padding='same',\n","                                     input_shape=[28, 28, 1]))\n","    model.add(layers.LeakyReLU(alpha=0.01))\n","    \n","    model.add(layers.Conv2D(64, (5, 5), strides=(2, 2), padding='same'))\n","    model.add(layers.BatchNormalization())\n","    model.add(layers.LeakyReLU(alpha=0.01))\n","    \n","    model.add(layers.Conv2D(128, (5, 5), strides=(2, 2), padding='same'))\n","    model.add(layers.BatchNormalization())\n","    model.add(layers.LeakyReLU(alpha=0.01))\n","\n","    model.add(layers.Flatten())\n","    model.add(layers.Dense(1, activation='sigmoid'))\n","\n","    return model\n","\n"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"colab_type":"code","id":"gDkA05NE6QMs","outputId":"76b91a81-4181-402d-e7a5-f18b5a3d1eca","executionInfo":{"status":"ok","timestamp":1566247439132,"user_tz":-120,"elapsed":3447,"user":{"displayName":"Jordi Torres","photoUrl":"https://lh4.googleusercontent.com/-SfQ_VFnu6kE/AAAAAAAAAAI/AAAAAAAABzE/lQO8NYx2THo/s64/photo.jpg","userId":"14559872433417615139"}},"colab":{"base_uri":"https://localhost:8080/","height":510}},"source":["discriminator = make_discriminator_model()\n","decision = discriminator(generated_image)\n","print (decision)\n","\n","discriminator.summary ()"],"execution_count":0,"outputs":[{"output_type":"stream","text":["tf.Tensor([[0.4995159]], shape=(1, 1), dtype=float32)\n","Model: \"sequential_3\"\n","_________________________________________________________________\n","Layer (type)                 Output Shape              Param #   \n","=================================================================\n","conv2d_3 (Conv2D)            (None, 14, 14, 32)        832       \n","_________________________________________________________________\n","leaky_re_lu_7 (LeakyReLU)    (None, 14, 14, 32)        0         \n","_________________________________________________________________\n","conv2d_4 (Conv2D)            (None, 7, 7, 64)          51264     \n","_________________________________________________________________\n","batch_normalization_6 (Batch (None, 7, 7, 64)          256       \n","_________________________________________________________________\n","leaky_re_lu_8 (LeakyReLU)    (None, 7, 7, 64)          0         \n","_________________________________________________________________\n","conv2d_5 (Conv2D)            (None, 4, 4, 128)         204928    \n","_________________________________________________________________\n","batch_normalization_7 (Batch (None, 4, 4, 128)         512       \n","_________________________________________________________________\n","leaky_re_lu_9 (LeakyReLU)    (None, 4, 4, 128)         0         \n","_________________________________________________________________\n","flatten_1 (Flatten)          (None, 2048)              0         \n","_________________________________________________________________\n","dense_3 (Dense)              (None, 1)                 2049      \n","=================================================================\n","Total params: 259,841\n","Trainable params: 259,457\n","Non-trainable params: 384\n","_________________________________________________________________\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"0FMYgY_mPfTi"},"source":["## Funciones de *Loss* y optimizadores\n"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"psQfmXxYKU3X","colab":{}},"source":["cross_entropy = tf.keras.losses.BinaryCrossentropy(from_logits=True)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"PKY_iPSPNWoj"},"source":["### Discriminator loss"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"wkMNfBWlT-PV","colab":{}},"source":["def discriminator_loss(real_output, fake_output):\n","    real_loss = cross_entropy(tf.ones_like(real_output), real_output)\n","    fake_loss = cross_entropy(tf.zeros_like(fake_output), fake_output)\n","    total_loss = real_loss + fake_loss\n","    return total_loss"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"Jd-3GCUEiKtv"},"source":["### Generator loss"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"90BIcCKcDMxz","colab":{}},"source":["def generator_loss(fake_output):\n","    return cross_entropy(tf.ones_like(fake_output), fake_output)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"MgIc7i0th_Iu"},"source":["### Optimizadores"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"iWCn_PVdEJZ7","colab":{}},"source":["generator_optimizer = tf.keras.optimizers.Adam(1e-4)\n","\n","discriminator_optimizer = tf.keras.optimizers.Adam(1e-4)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"UuXySZBahFjt","colab_type":"text"},"source":["## Entrenamiento"]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"Rw1fkAczTQYh"},"source":["###Training step\n","\n"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"3t5ibNo05jCB","colab":{}},"source":["@tf.function\n","def train_step(images):\n","    noise = tf.random.normal([BATCH_SIZE, noise_dim])\n","\n","    with tf.GradientTape() as gen_tape, tf.GradientTape() as disc_tape:\n","      generated_images = generator(noise, training=True)\n","\n","      real_output = discriminator(images, training=True)\n","      fake_output = discriminator(generated_images, training=True)\n","\n","      gen_loss = generator_loss(fake_output)\n","      disc_loss = discriminator_loss(real_output, fake_output)\n","\n","    gradients_of_generator = gen_tape.gradient(gen_loss, generator.trainable_variables)\n","    gradients_of_discriminator = disc_tape.gradient(disc_loss, discriminator.trainable_variables)\n","\n","    generator_optimizer.apply_gradients(zip(gradients_of_generator, generator.trainable_variables))\n","    discriminator_optimizer.apply_gradients(zip(gradients_of_discriminator, discriminator.trainable_variables))"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"Nts43kLUdPG-","colab_type":"text"},"source":["### Training loop"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"2M7LmLtGEMQJ","colab":{}},"source":["grid_size_x= 10\n","grid_size_y= 10\n","seed = tf.random.normal([grid_size_x*grid_size_y , noise_dim])\n","\n","def train(dataset, epochs):\n","  for epoch in range(epochs):\n","    start = time.time()\n","\n","    for image_batch in dataset:\n","      train_step(image_batch)\n","\n","    generate_images(generator,seed)\n","    print ('Time for epoch {} is {} sec'.format(epoch + 1, time.time()-start))\n","\n","  generate_images(generator, seed)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"2aFF7Hk3XdeW"},"source":["### Visualización de las imagenes (en realidad solo las que genera el Generador para ver que va aprendiendo)\n","\n"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"RmdVsmvhPxyy","colab":{}},"source":["\n","def generate_images(model, test_input):\n","  \n","  predictions = model(test_input, training=False)\n","\n","  fig = plt.figure(figsize=(grid_size_x,grid_size_y))\n","  for i in range(predictions.shape[0]):\n","      plt.subplot(grid_size_x, grid_size_y, i+1)\n","      plt.imshow(predictions[i, :, :, 0] * 127.5 + 127.5, cmap='gray')\n","      plt.axis('off')            \n","  plt.show()"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"colab_type":"text","id":"dZrd4CdjR-Fp"},"source":["### Entrenar el model"]},{"cell_type":"code","metadata":{"colab_type":"code","id":"Ly3UN0SLLY2l","outputId":"0ca58f5a-1c43-49ae-e967-63662a6c5132","executionInfo":{"status":"ok","timestamp":1566250960274,"user_tz":-120,"elapsed":3524569,"user":{"displayName":"Jordi Torres","photoUrl":"https://lh4.googleusercontent.com/-SfQ_VFnu6kE/AAAAAAAAAAI/AAAAAAAABzE/lQO8NYx2THo/s64/photo.jpg","userId":"14559872433417615139"}},"colab":{"base_uri":"https://localhost:8080/","height":1000,"output_embedded_package_id":"1O8bS1NTiLOCyoE8M8orUou7msus69BR-"}},"source":["%%time\n","EPOCHS = 100\n","train(train_dataset, EPOCHS)"],"execution_count":0,"outputs":[{"output_type":"display_data","data":{"text/plain":"Output hidden; open in https://colab.research.google.com to view."},"metadata":{}}]},{"cell_type":"code","metadata":{"id":"C6LjGXEk1d4R","colab_type":"code","colab":{}},"source":["import os, signal\n","\n","os.kill( os.getpid() , \n","         signal.SIGKILL\n","       )"],"execution_count":0,"outputs":[]}]}