aggregate_function_models.ipynb

Kernel: Python 3

Copyright 2020 The TensorFlow Authors.

#@title 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.

Modelos de função agregada do TF Lattice

Ver em TensorFlow.org

Executar no Google Colab

Ver fonte no GitHub

Baixar notebook

Visão geral

Os modelos de função agregada pré-fabricados do TFL são maneiras rápidas e fáceis de criar instâncias tf.keras.model do TFL para o aprendizado de funções de agregação complexas. Este guia descreve as etapas necessárias para construir e treinar/testar um modelo de função agregada pré-fabricado do TFL.

Configuração

Instale o pacote do TF Lattice:

#@test {"skip": true}
!pip install tensorflow-lattice pydot

Importe os pacotes necessários:

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)

Baixe o dataset 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()

Extraia e converta características e rótulos:

# 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))

Configure os valores padrão usados para treinamento neste guia:

LEARNING_RATE = 0.1
BATCH_SIZE = 128
NUM_EPOCHS = 500
MIDDLE_DIM = 3
MIDDLE_LATTICE_SIZE = 2
MIDDLE_KEYPOINTS = 16
OUTPUT_KEYPOINTS = 8

Configurações de características

A calibração de características e as configurações por característica são definidas usando tfl.configs.FeatureConfig. As configurações de características incluem restrições de monotonicidade, regularização por recurso (ver tfl.configs.RegularizerConfig) e tamanhos de lattice para modelos lattice.

É necessário especificar totalmente a configuração para qualquer característica que queremos que o modelo reconheça. Caso contrário, o modelo não terá como saber se essa característica existe. Para modelos de agregação, essas características serão automaticamente consideradas e corretamente tradas como irregulares.

Compute os quantis

Embora a configuração padrão para pwl_calibration_input_keypoints em tfl.configs.FeatureConfig seja "quantis", é preciso definir manualmente os keypoints de entrada para modelos pré-fabricados. Para isso, primeiro definimos nossa própria função helper para computar quantis.

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)

Defina as configurações das características

Agora que os quantis podem ser computados, defina uma configuração para cada característica que o modelo deve aceitar 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 função agregada

Para criar um modelo pré-fabricado do TFL, primeiro construa uma configuração de modelo a partir de tfl.configs. Um modelo de função agregada é construído usando a tfl.configs.AggregateFunctionConfig. Ela aplica a calibração linear por partes e categórica e, em seguida, um modelo lattice em cada dimensão da entrada irregular. Depois, ela aplica uma camada de agregação em cima da saída de cada dimensão, além de uma calibração linear por partes de saída 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')

A saída de cada camada de agregação é a saída média de um lattice calibrado em cima das entradas irregulares. Aqui está o modelo usado dentro da primeira camada de agregação:

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')

Agora, como com qualquer outro tf.keras.Model, é preciso compilar e ajustar o modelo aos dados.

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)

Após o treinamento do modelo, é possível avaliá-lo no dataset de teste.

print('Test Set Evaluation...')
print(aggregate_function_model.evaluate(test_xs, test_ys))