deep_learning/space_image/TF2Keras_MNIST_Classification.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"TF2Keras_MNIST_Classification.ipynb","provenance":[],"collapsed_sections":[]},"kernelspec":{"name":"python3","display_name":"Python 3"},"accelerator":"GPU"},"cells":[{"cell_type":"markdown","metadata":{"id":"mEdHLo4qbGKf","colab_type":"text"},"source":["# Installing and Loading Packages"]},{"cell_type":"code","metadata":{"id":"BP-YI31ZZ8Ug","colab_type":"code","colab":{}},"source":["!pip install -q tf-nightly"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"Ip2uXQszaKRr","colab_type":"code","colab":{}},"source":["import tensorflow as tf\n","from tensorflow.python.client import device_lib\n","\n","print(\"Tensorflow Version: {}\".format(tf.__version__))\n","print(\"Device List: {}\".format(device_lib.list_local_devices()))"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"OKgEM0SZeN8h","colab_type":"text"},"source":["# Quickstarts for Beginners"]},{"cell_type":"markdown","metadata":{"id":"U-DSP5IvbMT-","colab_type":"text"},"source":["## Data Preparation"]},{"cell_type":"code","metadata":{"id":"VJCEZXPIa4ZT","colab_type":"code","colab":{}},"source":["mnist = tf.keras.datasets.mnist"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"W82yS879bhQB","colab_type":"code","colab":{}},"source":["(x_train, t_train), (x_test, t_test) = mnist.load_data()"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"WMLxJM31cHBW","colab_type":"code","colab":{}},"source":["print(x_train.shape, t_test.shape)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"k0FWZOmmcQSO","colab_type":"text"},"source":["### Data Preprocessing\n","\n","Normalize the image into range from 0 to 1."]},{"cell_type":"code","metadata":{"id":"rlIahk3hcKtO","colab_type":"code","colab":{}},"source":["x_train, x_test = x_train / 255.0, x_test / 255.0"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"ZHgA0x79ci79","colab_type":"text"},"source":["## Building Models"]},{"cell_type":"code","metadata":{"id":"Y6lV-aUBchch","colab_type":"code","colab":{}},"source":["model_seq = tf.keras.Sequential([\n","  tf.keras.layers.Flatten(input_shape=(28,28)),\n","  tf.keras.layers.Dense(128, activation='relu'),\n","  tf.keras.layers.Dropout(0.2),\n","  tf.keras.layers.Dense(10, activation='softmax')\n","])"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"Q2Es5l52dCmm","colab_type":"code","colab":{}},"source":["model_seq.compile(optimizer='adam', \n","                  loss='sparse_categorical_crossentropy', \n","                  metrics=['accuracy'])"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"iOK6SrjcdSuR","colab_type":"text"},"source":["## Train and Evaluate Models"]},{"cell_type":"code","metadata":{"id":"niTLFn1adQhh","colab_type":"code","colab":{}},"source":["model_seq.fit(x_train, t_train, epochs=10)"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"-A6-tugfdgUS","colab_type":"code","colab":{}},"source":["model_seq.evaluate(x_test, t_test, verbose=2)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"ueu2p_9Feapz","colab_type":"text"},"source":["# Quickstarts for experts"]},{"cell_type":"markdown","metadata":{"id":"lUUwljtue05M","colab_type":"text"},"source":["## Data Preparation"]},{"cell_type":"code","metadata":{"id":"jLs6yfQWd5Li","colab_type":"code","colab":{}},"source":["mnist = tf.keras.datasets.mnist"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"l74pmtIpe7T9","colab_type":"code","colab":{}},"source":["(x_train, y_train), (x_test, y_test) = mnist.load_data()"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"JgvYfWFlfONf","colab_type":"text"},"source":["### Data Preprocessing"]},{"cell_type":"code","metadata":{"id":"g3o-A6bcfM7u","colab_type":"code","colab":{}},"source":["x_train, x_test = x_train / 255.0, x_test / 255.0"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"Wgu-6JvofV3U","colab_type":"code","colab":{}},"source":["print(x_train.shape, y_train.shape)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"lSB12F7jfaR7","colab_type":"text"},"source":["Expand a new axis to the x_train and x_test. Here we are going to build a convolution neural network."]},{"cell_type":"code","metadata":{"id":"zubFoY2lfY5q","colab_type":"code","colab":{}},"source":["x_train = x_train[..., tf.newaxis]\n","x_test = x_test[..., tf.newaxis]\n","print(x_train.shape, x_test.shape)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"TsQ2ShOHgBut","colab_type":"text"},"source":["Use `tf.data` to batch and shuffle the dataset."]},{"cell_type":"code","metadata":{"id":"bMtK9XWYf5KM","colab_type":"code","colab":{}},"source":["train_ds = tf.data.Dataset.from_tensor_slices((x_train, y_train)).shuffle(10000).batch(32)\n","test_ds = tf.data.Dataset.from_tensor_slices((x_test, y_test)).batch(32)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"Pw6bYKD3glvz","colab_type":"text"},"source":["## Building Models\n","\n","Build the `tf.keras` model via Keras model APIs."]},{"cell_type":"code","metadata":{"id":"p-_e7xKega3U","colab_type":"code","colab":{}},"source":["class MyModel(tf.keras.Model):\n","  def __init__(self):\n","    super(MyModel, self).__init__()\n","    self.conv1 = tf.keras.layers.Conv2D(32, 3, activation='relu')\n","    self.flatten = tf.keras.layers.Flatten()\n","    self.d1 = tf.keras.layers.Dense(128, activation='relu')\n","    self.d2 = tf.keras.layers.Dense(10, activation='softmax')\n","\n","  def call(self, x):\n","    x = self.conv1(x)\n","    x = self.flatten(x)\n","    x = self.d1(x)\n","    return self.d2(x)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"jh7tgk3Ch1FA","colab_type":"text"},"source":["Instantiate a model object."]},{"cell_type":"code","metadata":{"id":"9WNTY9aGhumd","colab_type":"code","colab":{}},"source":["model = MyModel()"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"LSQjunVYQKAt","colab_type":"text"},"source":["You can build a model via a more forward way."]},{"cell_type":"code","metadata":{"id":"olr5V65wBkQ7","colab_type":"code","colab":{}},"source":["def CNNModel(inputs):\n","  x = tf.keras.layers.Conv2D(filters=32, kernel_size=(3,3), strides=(2,2), padding='same')(inputs)\n","  x = tf.keras.layers.Flatten()(x)\n","  x = tf.keras.layers.Dense(128, activation='relu')(x)\n","  x = tf.keras.layers.Dense(10, activation='softmax', name=\"output\")(x)\n","  return x"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"DA3ibT06CkCn","colab_type":"code","colab":{}},"source":["inputs = tf.keras.Input(shape=(28, 28, 1), dtype=tf.float32)\n","outputs = CNNModel(inputs)\n","model = tf.keras.Model(inputs, outputs)\n","model.summary()"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"0gIRiNkLmBg5","colab_type":"text"},"source":["## Train and evaluate the model."]},{"cell_type":"markdown","metadata":{"id":"oK2loIOMiQWs","colab_type":"text"},"source":["Define a loss function and an optimizer."]},{"cell_type":"code","metadata":{"id":"fWrGByfch0Sp","colab_type":"code","colab":{}},"source":["loss_object = tf.keras.losses.SparseCategoricalCrossentropy()"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"_-72scmtiZAg","colab_type":"code","colab":{}},"source":["optimizer = tf.keras.optimizers.Adam()"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"fdBiDWQfjz3k","colab_type":"text"},"source":["Define metrics to measure the loss and accuracy of a model."]},{"cell_type":"code","metadata":{"id":"42arEJEoiexX","colab_type":"code","colab":{}},"source":["train_loss = tf.keras.metrics.Mean(name='train_loss')\n","train_accuracy = tf.keras.metrics.SparseCategoricalAccuracy(name='train_accuracy')"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"UzB0ftZGkHQO","colab_type":"code","colab":{}},"source":["test_loss = tf.keras.metrics.Mean(name='test_loss')\n","test_accuracy = tf.keras.metrics.SparseCategoricalAccuracy(name='test_accuracy')"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"y6QrRIgfka9G","colab_type":"text"},"source":["Use `tf.GradientTape` to train the model."]},{"cell_type":"code","metadata":{"id":"SiSYZbWAkX85","colab_type":"code","colab":{}},"source":["@tf.function\n","def train_step(images, labels):\n","  with tf.GradientTape() as tape:\n","    predictions = model(images)\n","    loss = loss_object(labels, predictions)\n","  gradients = tape.gradient(loss, model.trainable_variables)\n","  optimizer.apply_gradients(zip(gradients, model.trainable_variables))\n","\n","  train_loss(loss)\n","  train_accuracy(labels, predictions)"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"g4tAhwTYlZZV","colab_type":"text"},"source":["Test the model."]},{"cell_type":"code","metadata":{"id":"PN9h1GkAlYCe","colab_type":"code","colab":{}},"source":["@tf.function\n","def test_step(images, labels):\n","  predictions = model(images)\n","  t_loss = loss_object(labels, predictions)\n","\n","  test_loss(t_loss)\n","  test_accuracy(labels, predictions)"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"ZwFOabGbl85p","colab_type":"code","colab":{}},"source":["EPOCHS = 10"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"5NKuiNdimFa0","colab_type":"code","colab":{}},"source":["for epoch in range(EPOCHS):\n","  # Reset the metrics at the start of the next epoch\n","  train_loss.reset_states()\n","  train_accuracy.reset_states()\n","  test_loss.reset_states()\n","  test_accuracy.reset_states()\n","\n","  for images, labels in train_ds:\n","    train_step(images, labels)\n","\n","  for test_images, test_labels in test_ds:\n","    test_step(test_images, test_labels)\n","\n","  template = 'Epoch {}, Loss: {}, Accuracy: {}, Test Loss: {}, Test Accuracy: {}'\n","  print(template.format(epoch+1,\n","                        train_loss.result(),\n","                        train_accuracy.result()*100,\n","                        test_loss.result(),\n","                        test_accuracy.result()*100))"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"rcXtyaU0L2-W","colab_type":"code","colab":{}},"source":[""],"execution_count":0,"outputs":[]}]}