In [0]:
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
Objectifs d'apprentissage :
L'objectif est d'associer à chaque image d'entrée le chiffre numérique correct. Vous allez créer un réseau de neurones avec quelques couches cachées et, au sommet, une couche Softmax afin de sélectionner la classe qui l'emporte.
In [0]:
from __future__ import print_function
import glob
import math
import os
from IPython import display
from matplotlib import cm
from matplotlib import gridspec
from matplotlib import pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sns
from sklearn import metrics
import tensorflow as tf
from tensorflow.python.data import Dataset
tf.logging.set_verbosity(tf.logging.ERROR)
pd.options.display.max_rows = 10
pd.options.display.float_format = '{:.1f}'.format
mnist_dataframe = pd.read_csv(
"https://download.mlcc.google.com/mledu-datasets/mnist_train_small.csv",
sep=",",
header=None)
# Use just the first 10,000 records for training/validation.
mnist_dataframe = mnist_dataframe.head(10000)
mnist_dataframe = mnist_dataframe.reindex(np.random.permutation(mnist_dataframe.index))
mnist_dataframe.head()
La première colonne contient l'étiquette de classe. Les autres contiennent les valeurs de caractéristique ; une par pixel pour les 28×28=784 valeurs de pixel. La plupart de ces 784 valeurs de pixel sont égales à zéro ; vous pouvez prendre une minute pour vérifier qu'elles ne sont pas toutes nulles.
Ces exemples sont, en fait, des images à fort contraste et à relativement faible résolution de nombres écrits à la main. Chacun des dix chiffres 0-9 est représenté avec une étiquette de classe unique. Il s'agit donc d'un problème de classification à classes multiples avec 10 classes.
Nous allons, à présent, analyser les étiquettes et les caractéristiques, et observer quelques exemples. Notez l'utilisation de loc qui permet d'extraire des colonnes en fonction de l'emplacement d'origine, étant donné que cet ensemble de données est dépourvu de ligne d'en-tête.
In [0]:
def parse_labels_and_features(dataset):
"""Extracts labels and features.
This is a good place to scale or transform the features if needed.
Args:
dataset: A Pandas `Dataframe`, containing the label on the first column and
monochrome pixel values on the remaining columns, in row major order.
Returns:
A `tuple` `(labels, features)`:
labels: A Pandas `Series`.
features: A Pandas `DataFrame`.
"""
labels = dataset[0]
# DataFrame.loc index ranges are inclusive at both ends.
features = dataset.loc[:,1:784]
# Scale the data to [0, 1] by dividing out the max value, 255.
features = features / 255
return labels, features
In [0]:
training_targets, training_examples = parse_labels_and_features(mnist_dataframe[:7500])
training_examples.describe()
In [0]:
validation_targets, validation_examples = parse_labels_and_features(mnist_dataframe[7500:10000])
validation_examples.describe()
Affichez un exemple aléatoire et l'étiquette correspondante.
In [0]:
rand_example = np.random.choice(training_examples.index)
_, ax = plt.subplots()
ax.matshow(training_examples.loc[rand_example].values.reshape(28, 28))
ax.set_title("Label: %i" % training_targets.loc[rand_example])
ax.grid(False)
Commencez par créer un modèle de référence qui servira de base de comparaison. Le modèle LinearClassifier fournit un ensemble de k classificateurs un contre tous ; un pour chacune des k classes.
Outre l'indication de la justesse et la représentation graphique de la perte logistique au fil du temps, vous remarquerez qu'une matrice de confusion est également affichée. Cette matrice identifie les classes qui ont été classées de manière erronée. Quels chiffres ont été confondus avec d'autres ?
Notez également que la fonction log_loss est utilisée pour le suivi de l'erreur du modèle. Veillez à ne pas la confondre avec la fonction de perte interne au modèle LinearClassifier, laquelle est utilisée à des fins d'apprentissage.
In [0]:
def construct_feature_columns():
"""Construct the TensorFlow Feature Columns.
Returns:
A set of feature columns
"""
# There are 784 pixels in each image.
return set([tf.feature_column.numeric_column('pixels', shape=784)])
Vous allez maintenant créer des fonctions d'entrée distinctes pour l'apprentissage et la prédiction. Vous allez les imbriquer dans create_training_input_fn() et create_predict_input_fn(), respectivement, afin de pouvoir les invoquer pour renvoyer les fonctions _input_fn correspondantes qui doivent être transmises aux appels .train() et .predict().
In [0]:
def create_training_input_fn(features, labels, batch_size, num_epochs=None, shuffle=True):
"""A custom input_fn for sending MNIST data to the estimator for training.
Args:
features: The training features.
labels: The training labels.
batch_size: Batch size to use during training.
Returns:
A function that returns batches of training features and labels during
training.
"""
def _input_fn(num_epochs=None, shuffle=True):
# Input pipelines are reset with each call to .train(). To ensure model
# gets a good sampling of data, even when number of steps is small, we
# shuffle all the data before creating the Dataset object
idx = np.random.permutation(features.index)
raw_features = {"pixels":features.reindex(idx)}
raw_targets = np.array(labels[idx])
ds = Dataset.from_tensor_slices((raw_features,raw_targets)) # warning: 2GB limit
ds = ds.batch(batch_size).repeat(num_epochs)
if shuffle:
ds = ds.shuffle(10000)
# Return the next batch of data.
feature_batch, label_batch = ds.make_one_shot_iterator().get_next()
return feature_batch, label_batch
return _input_fn
In [0]:
def create_predict_input_fn(features, labels, batch_size):
"""A custom input_fn for sending mnist data to the estimator for predictions.
Args:
features: The features to base predictions on.
labels: The labels of the prediction examples.
Returns:
A function that returns features and labels for predictions.
"""
def _input_fn():
raw_features = {"pixels": features.values}
raw_targets = np.array(labels)
ds = Dataset.from_tensor_slices((raw_features, raw_targets)) # warning: 2GB limit
ds = ds.batch(batch_size)
# Return the next batch of data.
feature_batch, label_batch = ds.make_one_shot_iterator().get_next()
return feature_batch, label_batch
return _input_fn
In [0]:
def train_linear_classification_model(
learning_rate,
steps,
batch_size,
training_examples,
training_targets,
validation_examples,
validation_targets):
"""Trains a linear classification model for the MNIST digits dataset.
In addition to training, this function also prints training progress information,
a plot of the training and validation loss over time, and a confusion
matrix.
Args:
learning_rate: A `float`, the learning rate to use.
steps: A non-zero `int`, the total number of training steps. A training step
consists of a forward and backward pass using a single batch.
batch_size: A non-zero `int`, the batch size.
training_examples: A `DataFrame` containing the training features.
training_targets: A `DataFrame` containing the training labels.
validation_examples: A `DataFrame` containing the validation features.
validation_targets: A `DataFrame` containing the validation labels.
Returns:
The trained `LinearClassifier` object.
"""
periods = 10
steps_per_period = steps / periods
# Create the input functions.
predict_training_input_fn = create_predict_input_fn(
training_examples, training_targets, batch_size)
predict_validation_input_fn = create_predict_input_fn(
validation_examples, validation_targets, batch_size)
training_input_fn = create_training_input_fn(
training_examples, training_targets, batch_size)
# Create a LinearClassifier object.
my_optimizer = tf.train.AdagradOptimizer(learning_rate=learning_rate)
my_optimizer = tf.contrib.estimator.clip_gradients_by_norm(my_optimizer, 5.0)
classifier = tf.estimator.LinearClassifier(
feature_columns=construct_feature_columns(),
n_classes=10,
optimizer=my_optimizer,
config=tf.estimator.RunConfig(keep_checkpoint_max=1)
)
# Train the model, but do so inside a loop so that we can periodically assess
# loss metrics.
print("Training model...")
print("LogLoss error (on validation data):")
training_errors = []
validation_errors = []
for period in range (0, periods):
# Train the model, starting from the prior state.
classifier.train(
input_fn=training_input_fn,
steps=steps_per_period
)
# Take a break and compute probabilities.
training_predictions = list(classifier.predict(input_fn=predict_training_input_fn))
training_probabilities = np.array([item['probabilities'] for item in training_predictions])
training_pred_class_id = np.array([item['class_ids'][0] for item in training_predictions])
training_pred_one_hot = tf.keras.utils.to_categorical(training_pred_class_id,10)
validation_predictions = list(classifier.predict(input_fn=predict_validation_input_fn))
validation_probabilities = np.array([item['probabilities'] for item in validation_predictions])
validation_pred_class_id = np.array([item['class_ids'][0] for item in validation_predictions])
validation_pred_one_hot = tf.keras.utils.to_categorical(validation_pred_class_id,10)
# Compute training and validation errors.
training_log_loss = metrics.log_loss(training_targets, training_pred_one_hot)
validation_log_loss = metrics.log_loss(validation_targets, validation_pred_one_hot)
# Occasionally print the current loss.
print(" period %02d : %0.2f" % (period, validation_log_loss))
# Add the loss metrics from this period to our list.
training_errors.append(training_log_loss)
validation_errors.append(validation_log_loss)
print("Model training finished.")
# Remove event files to save disk space.
_ = map(os.remove, glob.glob(os.path.join(classifier.model_dir, 'events.out.tfevents*')))
# Calculate final predictions (not probabilities, as above).
final_predictions = classifier.predict(input_fn=predict_validation_input_fn)
final_predictions = np.array([item['class_ids'][0] for item in final_predictions])
accuracy = metrics.accuracy_score(validation_targets, final_predictions)
print("Final accuracy (on validation data): %0.2f" % accuracy)
# Output a graph of loss metrics over periods.
plt.ylabel("LogLoss")
plt.xlabel("Periods")
plt.title("LogLoss vs. Periods")
plt.plot(training_errors, label="training")
plt.plot(validation_errors, label="validation")
plt.legend()
plt.show()
# Output a plot of the confusion matrix.
cm = metrics.confusion_matrix(validation_targets, final_predictions)
# Normalize the confusion matrix by row (i.e by the number of samples
# in each class).
cm_normalized = cm.astype("float") / cm.sum(axis=1)[:, np.newaxis]
ax = sns.heatmap(cm_normalized, cmap="bone_r")
ax.set_aspect(1)
plt.title("Confusion matrix")
plt.ylabel("True label")
plt.xlabel("Predicted label")
plt.show()
return classifier
Exercez-vous, pendant cinq minutes, à obtenir une valeur de justesse élevée avec un modèle linéaire de ce type. Pour cet exercice, limitez-vous à expérimenter avec les hyperparamètres relatifs à la taille du lot, au taux d'apprentissage et aux nombres d'étapes.
Arrêtez-vous dès que vous obtenez une valeur de justesse supérieure à environ 0,9.
In [0]:
classifier = train_linear_classification_model(
learning_rate=0.02,
steps=100,
batch_size=10,
training_examples=training_examples,
training_targets=training_targets,
validation_examples=validation_examples,
validation_targets=validation_targets)
Voici un ensemble de paramètres avec lequel la justesse devrait avoisiner 0,9.
In [0]:
_ = train_linear_classification_model(
learning_rate=0.03,
steps=1000,
batch_size=30,
training_examples=training_examples,
training_targets=training_targets,
validation_examples=validation_examples,
validation_targets=validation_targets)
Remplacez le modèle LinearClassifier ci-dessus par un modèle DNNClassifier et identifiez une combinaison de paramètres offrant une justesse de 0,95 ou plus.
Vous pouvez essayer d'autres méthodes de régularisation, telles qu'une régularisation par abandon. Ces méthodes sont documentées dans les commentaires de la classe DNNClassifier.
In [0]:
#
# YOUR CODE HERE: Replace the linear classifier with a neural network.
#
Dès que vous disposez d'un modèle satisfaisant, vérifiez que l'ensemble de validation n'a pas été surappris en évaluant les données de test qui seront chargées ci-après.
In [0]:
mnist_test_dataframe = pd.read_csv(
"https://download.mlcc.google.com/mledu-datasets/mnist_test.csv",
sep=",",
header=None)
test_targets, test_examples = parse_labels_and_features(mnist_test_dataframe)
test_examples.describe()
In [0]:
#
# YOUR CODE HERE: Calculate accuracy on the test set.
#
Le code ci-dessous est pratiquement identique au code d'apprentissage LinearClassifer d'origine, à l'exception de la configuration spécifique au réseau de neurones, comme l'hyperparamètre des unités cachées.
In [0]:
def train_nn_classification_model(
learning_rate,
steps,
batch_size,
hidden_units,
training_examples,
training_targets,
validation_examples,
validation_targets):
"""Trains a neural network classification model for the MNIST digits dataset.
In addition to training, this function also prints training progress information,
a plot of the training and validation loss over time, as well as a confusion
matrix.
Args:
learning_rate: A `float`, the learning rate to use.
steps: A non-zero `int`, the total number of training steps. A training step
consists of a forward and backward pass using a single batch.
batch_size: A non-zero `int`, the batch size.
hidden_units: A `list` of int values, specifying the number of neurons in each layer.
training_examples: A `DataFrame` containing the training features.
training_targets: A `DataFrame` containing the training labels.
validation_examples: A `DataFrame` containing the validation features.
validation_targets: A `DataFrame` containing the validation labels.
Returns:
The trained `DNNClassifier` object.
"""
periods = 10
# Caution: input pipelines are reset with each call to train.
# If the number of steps is small, your model may never see most of the data.
# So with multiple `.train` calls like this you may want to control the length
# of training with num_epochs passed to the input_fn. Or, you can do a really-big shuffle,
# or since it's in-memory data, shuffle all the data in the `input_fn`.
steps_per_period = steps / periods
# Create the input functions.
predict_training_input_fn = create_predict_input_fn(
training_examples, training_targets, batch_size)
predict_validation_input_fn = create_predict_input_fn(
validation_examples, validation_targets, batch_size)
training_input_fn = create_training_input_fn(
training_examples, training_targets, batch_size)
# Create the input functions.
predict_training_input_fn = create_predict_input_fn(
training_examples, training_targets, batch_size)
predict_validation_input_fn = create_predict_input_fn(
validation_examples, validation_targets, batch_size)
training_input_fn = create_training_input_fn(
training_examples, training_targets, batch_size)
# Create feature columns.
feature_columns = [tf.feature_column.numeric_column('pixels', shape=784)]
# Create a DNNClassifier object.
my_optimizer = tf.train.AdagradOptimizer(learning_rate=learning_rate)
my_optimizer = tf.contrib.estimator.clip_gradients_by_norm(my_optimizer, 5.0)
classifier = tf.estimator.DNNClassifier(
feature_columns=feature_columns,
n_classes=10,
hidden_units=hidden_units,
optimizer=my_optimizer,
config=tf.contrib.learn.RunConfig(keep_checkpoint_max=1)
)
# Train the model, but do so inside a loop so that we can periodically assess
# loss metrics.
print("Training model...")
print("LogLoss error (on validation data):")
training_errors = []
validation_errors = []
for period in range (0, periods):
# Train the model, starting from the prior state.
classifier.train(
input_fn=training_input_fn,
steps=steps_per_period
)
# Take a break and compute probabilities.
training_predictions = list(classifier.predict(input_fn=predict_training_input_fn))
training_probabilities = np.array([item['probabilities'] for item in training_predictions])
training_pred_class_id = np.array([item['class_ids'][0] for item in training_predictions])
training_pred_one_hot = tf.keras.utils.to_categorical(training_pred_class_id,10)
validation_predictions = list(classifier.predict(input_fn=predict_validation_input_fn))
validation_probabilities = np.array([item['probabilities'] for item in validation_predictions])
validation_pred_class_id = np.array([item['class_ids'][0] for item in validation_predictions])
validation_pred_one_hot = tf.keras.utils.to_categorical(validation_pred_class_id,10)
# Compute training and validation errors.
training_log_loss = metrics.log_loss(training_targets, training_pred_one_hot)
validation_log_loss = metrics.log_loss(validation_targets, validation_pred_one_hot)
# Occasionally print the current loss.
print(" period %02d : %0.2f" % (period, validation_log_loss))
# Add the loss metrics from this period to our list.
training_errors.append(training_log_loss)
validation_errors.append(validation_log_loss)
print("Model training finished.")
# Remove event files to save disk space.
_ = map(os.remove, glob.glob(os.path.join(classifier.model_dir, 'events.out.tfevents*')))
# Calculate final predictions (not probabilities, as above).
final_predictions = classifier.predict(input_fn=predict_validation_input_fn)
final_predictions = np.array([item['class_ids'][0] for item in final_predictions])
accuracy = metrics.accuracy_score(validation_targets, final_predictions)
print("Final accuracy (on validation data): %0.2f" % accuracy)
# Output a graph of loss metrics over periods.
plt.ylabel("LogLoss")
plt.xlabel("Periods")
plt.title("LogLoss vs. Periods")
plt.plot(training_errors, label="training")
plt.plot(validation_errors, label="validation")
plt.legend()
plt.show()
# Output a plot of the confusion matrix.
cm = metrics.confusion_matrix(validation_targets, final_predictions)
# Normalize the confusion matrix by row (i.e by the number of samples
# in each class).
cm_normalized = cm.astype("float") / cm.sum(axis=1)[:, np.newaxis]
ax = sns.heatmap(cm_normalized, cmap="bone_r")
ax.set_aspect(1)
plt.title("Confusion matrix")
plt.ylabel("True label")
plt.xlabel("Predicted label")
plt.show()
return classifier
In [0]:
classifier = train_nn_classification_model(
learning_rate=0.05,
steps=1000,
batch_size=30,
hidden_units=[100, 100],
training_examples=training_examples,
training_targets=training_targets,
validation_examples=validation_examples,
validation_targets=validation_targets)
Vous allez ensuite vérifier la justesse de l'ensemble d'évaluation.
In [0]:
mnist_test_dataframe = pd.read_csv(
"https://download.mlcc.google.com/mledu-datasets/mnist_test.csv",
sep=",",
header=None)
test_targets, test_examples = parse_labels_and_features(mnist_test_dataframe)
test_examples.describe()
In [0]:
predict_test_input_fn = create_predict_input_fn(
test_examples, test_targets, batch_size=100)
test_predictions = classifier.predict(input_fn=predict_test_input_fn)
test_predictions = np.array([item['class_ids'][0] for item in test_predictions])
accuracy = metrics.accuracy_score(test_targets, test_predictions)
print("Accuracy on test data: %0.2f" % accuracy)
Passez quelques minutes à explorer le réseau de neurones pour savoir ce qu'il a appris. Pour cela, vous allez accéder à l'attribut weights_ du modèle.
La couche d'entrée du modèle compte 784 pondérations correspondant aux images d'entrée 28×28. La première couche cachée aura 784×N pondérations, où N représente le nombre de nœuds de cette couche. Vous pouvez reconvertir ces pondérations en images 28×28 en remodelant chacun des N tableaux de pondérations 1×784 en N tableaux d'une taille de 28×28.
Exécutez la cellule suivante pour représenter graphiquement les pondérations. Pour cette cellule, notez qu'un modèle DNNClassifier appelé "classifier" doit déjà avoir été entraîné.
In [0]:
print(classifier.get_variable_names())
weights0 = classifier.get_variable_value("dnn/hiddenlayer_0/kernel")
print("weights0 shape:", weights0.shape)
num_nodes = weights0.shape[1]
num_rows = int(math.ceil(num_nodes / 10.0))
fig, axes = plt.subplots(num_rows, 10, figsize=(20, 2 * num_rows))
for coef, ax in zip(weights0.T, axes.ravel()):
# Weights in coef is reshaped from 1x784 to 28x28.
ax.matshow(coef.reshape(28, 28), cmap=plt.cm.pink)
ax.set_xticks(())
ax.set_yticks(())
plt.show()
La première couche cachée du réseau de neurones doit normalement modéliser des caractéristiques de bas niveau. Lorsque vous visualiserez les pondérations, vous ne verrez donc probablement que quelques blobs flous ou peut-être quelques parties de chiffres. Vous constaterez également que certains neurones sont essentiellement des données bruitées ; soit ils n'ont pas convergé, soit ils sont ignorés par les couches supérieures.
Il peut être intéressant d'interrompre l'apprentissage après un certain nombre d'itérations afin d'examiner le résultat.
Effectuez l'apprentissage du classificateur pour 10, 100 et 1 000 pas, puis exécutez à nouveau cette visualisation.
Quelles différences remarquez-vous pour les différents niveaux de convergence ?