 Ver en TensorFlow.org |  Ejecutar en Google Colab |  Ver fuente en GitHub |  Descargar cuaderno |
Descripción general
TFL preparado de antemano agregados Modelos de función son formas rápidas y fáciles de construir TFL tf.keras.model casos para el aprendizaje de las funciones de agregación complejas. Esta guía describe los pasos necesarios para construir un modelo de función agregada prefabricado de TFL y entrenarlo / probarlo.
Configuración
Instalación del paquete TF Lattice:
pip install -q tensorflow-lattice pydot
Importación de paquetes requeridos:
import tensorflow as tf
import collections
import logging
import numpy as np
import pandas as pd
import sys
import tensorflow_lattice as tfl
logging.disable(sys.maxsize)
Descarga del conjunto de datos de Puzzles:
train_dataframe = pd.read_csv(
'https://raw.githubusercontent.com/wbakst/puzzles_data/master/train.csv')
train_dataframe.head()
test_dataframe = pd.read_csv(
'https://raw.githubusercontent.com/wbakst/puzzles_data/master/test.csv')
test_dataframe.head()
Extraer y convertir características y etiquetas
# Features:
# - star_rating rating out of 5 stars (1-5)
# - word_count number of words in the review
# - is_amazon 1 = reviewed on amazon; 0 = reviewed on artifact website
# - includes_photo if the review includes a photo of the puzzle
# - num_helpful number of people that found this review helpful
# - num_reviews total number of reviews for this puzzle (we construct)
#
# This ordering of feature names will be the exact same order that we construct
# our model to expect.
feature_names = [
'star_rating', 'word_count', 'is_amazon', 'includes_photo', 'num_helpful',
'num_reviews'
]
def extract_features(dataframe, label_name):
# First we extract flattened features.
flattened_features = {
feature_name: dataframe[feature_name].values.astype(float)
for feature_name in feature_names[:-1]
}
# Construct mapping from puzzle name to feature.
star_rating = collections.defaultdict(list)
word_count = collections.defaultdict(list)
is_amazon = collections.defaultdict(list)
includes_photo = collections.defaultdict(list)
num_helpful = collections.defaultdict(list)
labels = {}
# Extract each review.
for i in range(len(dataframe)):
row = dataframe.iloc[i]
puzzle_name = row['puzzle_name']
star_rating[puzzle_name].append(float(row['star_rating']))
word_count[puzzle_name].append(float(row['word_count']))
is_amazon[puzzle_name].append(float(row['is_amazon']))
includes_photo[puzzle_name].append(float(row['includes_photo']))
num_helpful[puzzle_name].append(float(row['num_helpful']))
labels[puzzle_name] = float(row[label_name])
# Organize data into list of list of features.
names = list(star_rating.keys())
star_rating = [star_rating[name] for name in names]
word_count = [word_count[name] for name in names]
is_amazon = [is_amazon[name] for name in names]
includes_photo = [includes_photo[name] for name in names]
num_helpful = [num_helpful[name] for name in names]
num_reviews = [[len(ratings)] * len(ratings) for ratings in star_rating]
labels = [labels[name] for name in names]
# Flatten num_reviews
flattened_features['num_reviews'] = [len(reviews) for reviews in num_reviews]
# Convert data into ragged tensors.
star_rating = tf.ragged.constant(star_rating)
word_count = tf.ragged.constant(word_count)
is_amazon = tf.ragged.constant(is_amazon)
includes_photo = tf.ragged.constant(includes_photo)
num_helpful = tf.ragged.constant(num_helpful)
num_reviews = tf.ragged.constant(num_reviews)
labels = tf.constant(labels)
# Now we can return our extracted data.
return (star_rating, word_count, is_amazon, includes_photo, num_helpful,
num_reviews), labels, flattened_features
train_xs, train_ys, flattened_features = extract_features(train_dataframe, 'Sales12-18MonthsAgo')
test_xs, test_ys, _ = extract_features(test_dataframe, 'SalesLastSixMonths')
# Let's define our label minimum and maximum.
min_label, max_label = float(np.min(train_ys)), float(np.max(train_ys))
min_label, max_label = float(np.min(train_ys)), float(np.max(train_ys))
Configuración de los valores predeterminados utilizados para el entrenamiento en esta guía:
LEARNING_RATE = 0.1
BATCH_SIZE = 128
NUM_EPOCHS = 500
MIDDLE_DIM = 3
MIDDLE_LATTICE_SIZE = 2
MIDDLE_KEYPOINTS = 16
OUTPUT_KEYPOINTS = 8
Configuraciones de funciones
Calibración de características y configuraciones per-función se establecen con tfl.configs.FeatureConfig . Configuraciones de características incluyen restricciones de monotonicidad, regularización per-función (consulte tfl.configs.RegularizerConfig ) y tamaños de celosía para modelos de celosía.
Tenga en cuenta que debemos especificar completamente la configuración de características para cualquier característica que queremos que nuestro modelo reconozca. De lo contrario, el modelo no tendrá forma de saber que existe tal característica. Para los modelos de agregación, estas características se considerarán automáticamente y se manejarán adecuadamente como irregulares.
Calcular cuantiles
Aunque la configuración predeterminada para pwl_calibration_input_keypoints en tfl.configs.FeatureConfig es cuantiles '', para los modelos prefabricados tenemos que definir manualmente los puntos clave de entrada. Para hacerlo, primero definimos nuestra propia función auxiliar para calcular cuantiles.
def compute_quantiles(features,
num_keypoints=10,
clip_min=None,
clip_max=None,
missing_value=None):
# Clip min and max if desired.
if clip_min is not None:
features = np.maximum(features, clip_min)
features = np.append(features, clip_min)
if clip_max is not None:
features = np.minimum(features, clip_max)
features = np.append(features, clip_max)
# Make features unique.
unique_features = np.unique(features)
# Remove missing values if specified.
if missing_value is not None:
unique_features = np.delete(unique_features,
np.where(unique_features == missing_value))
# Compute and return quantiles over unique non-missing feature values.
return np.quantile(
unique_features,
np.linspace(0., 1., num=num_keypoints),
interpolation='nearest').astype(float)
Definición de nuestras configuraciones de funciones
Ahora que podemos calcular nuestros cuantiles, definimos una configuración de características para cada característica que queremos que nuestro modelo tome como entrada.
# Feature configs are used to specify how each feature is calibrated and used.
feature_configs = [
tfl.configs.FeatureConfig(
name='star_rating',
lattice_size=2,
monotonicity='increasing',
pwl_calibration_num_keypoints=5,
pwl_calibration_input_keypoints=compute_quantiles(
flattened_features['star_rating'], num_keypoints=5),
),
tfl.configs.FeatureConfig(
name='word_count',
lattice_size=2,
monotonicity='increasing',
pwl_calibration_num_keypoints=5,
pwl_calibration_input_keypoints=compute_quantiles(
flattened_features['word_count'], num_keypoints=5),
),
tfl.configs.FeatureConfig(
name='is_amazon',
lattice_size=2,
num_buckets=2,
),
tfl.configs.FeatureConfig(
name='includes_photo',
lattice_size=2,
num_buckets=2,
),
tfl.configs.FeatureConfig(
name='num_helpful',
lattice_size=2,
monotonicity='increasing',
pwl_calibration_num_keypoints=5,
pwl_calibration_input_keypoints=compute_quantiles(
flattened_features['num_helpful'], num_keypoints=5),
# Larger num_helpful indicating more trust in star_rating.
reflects_trust_in=[
tfl.configs.TrustConfig(
feature_name="star_rating", trust_type="trapezoid"),
],
),
tfl.configs.FeatureConfig(
name='num_reviews',
lattice_size=2,
monotonicity='increasing',
pwl_calibration_num_keypoints=5,
pwl_calibration_input_keypoints=compute_quantiles(
flattened_features['num_reviews'], num_keypoints=5),
)
]
Modelo de función agregada
Para construir un modelo premade TFL, construir primero una configuración de modelo de tfl.configs . Un modelo de función de agregado se construye utilizando el tfl.configs.AggregateFunctionConfig . Aplica una calibración categórica y lineal por partes, seguida de un modelo de celosía en cada dimensión de la entrada irregular. Luego aplica una capa de agregación sobre la salida para cada dimensión. A continuación, le sigue una calibración lineal por partes de salida opcional.
# Model config defines the model structure for the aggregate function model.
aggregate_function_model_config = tfl.configs.AggregateFunctionConfig(
feature_configs=feature_configs,
middle_dimension=MIDDLE_DIM,
middle_lattice_size=MIDDLE_LATTICE_SIZE,
middle_calibration=True,
middle_calibration_num_keypoints=MIDDLE_KEYPOINTS,
middle_monotonicity='increasing',
output_min=min_label,
output_max=max_label,
output_calibration=True,
output_calibration_num_keypoints=OUTPUT_KEYPOINTS,
output_initialization=np.linspace(
min_label, max_label, num=OUTPUT_KEYPOINTS))
# An AggregateFunction premade model constructed from the given model config.
aggregate_function_model = tfl.premade.AggregateFunction(
aggregate_function_model_config)
# Let's plot our model.
tf.keras.utils.plot_model(
aggregate_function_model, show_layer_names=False, rankdir='LR')
La salida de cada capa de agregación es la salida promedio de una celosía calibrada sobre las entradas irregulares. Aquí está el modelo utilizado dentro de la primera capa de agregación:
aggregation_layers = [
layer for layer in aggregate_function_model.layers
if isinstance(layer, tfl.layers.Aggregation)
]
tf.keras.utils.plot_model(
aggregation_layers[0].model, show_layer_names=False, rankdir='LR')
Ahora, al igual que con cualquier otra tf.keras.Model , compilamos y ajustar el modelo a los datos.
aggregate_function_model.compile(
loss='mae',
optimizer=tf.keras.optimizers.Adam(LEARNING_RATE))
aggregate_function_model.fit(
train_xs, train_ys, epochs=NUM_EPOCHS, batch_size=BATCH_SIZE, verbose=False)
<tensorflow.python.keras.callbacks.History at 0x7fee7d3033c8>
Después de entrenar nuestro modelo, podemos evaluarlo en nuestro conjunto de prueba.
print('Test Set Evaluation...')
print(aggregate_function_model.evaluate(test_xs, test_ys))
Test Set Evaluation... 7/7 [==============================] - 2s 3ms/step - loss: 53.4633 53.4632682800293