Ejemplo de Colab de indicadores de equidad de FaceSSD

Ver en TensorFlow.org Ejecutar en Google Colab Ver en GitHub Descargar cuaderno

Descripción general

En esta actividad, usará los indicadores de equidad para explorar las predicciones de FaceSSD en el conjunto de datos de Caras etiquetadas en el medio silvestre . Los indicadores de equidad son un conjunto de herramientas creadas sobre el análisis de modelos de TensorFlow que permiten la evaluación regular de las métricas de equidad en las canalizaciones de productos.

Acerca del conjunto de datos

En este ejercicio, trabajará con el conjunto de datos de predicción de FaceSSD, aproximadamente 200.000 predicciones de imágenes diferentes y verdades básicas generadas por la API de FaceSSD.

Acerca de las herramientas

TensorFlow Model Analysis es una biblioteca para evaluar modelos de aprendizaje automático de TensorFlow y no TensorFlow. Permite a los usuarios evaluar sus modelos en grandes cantidades de datos de manera distribuida, computando en gráficos y otras métricas sobre diferentes segmentos de datos y visualizarlos en cuadernos.

La validación de datos de TensorFlow es una herramienta que puede utilizar para analizar sus datos. Puede usarlo para encontrar problemas potenciales en sus datos, como valores perdidos y desequilibrios de datos, que pueden generar disparidades de equidad.

Con los indicadores de equidad , los usuarios podrán:

  • Evaluar el rendimiento del modelo, dividido en grupos definidos de usuarios.
  • Siéntase seguro de los resultados con intervalos de confianza y evaluaciones en múltiples umbrales.

Importador

Ejecute el siguiente código para instalar la biblioteca fairness_indicators. Este paquete contiene las herramientas que usaremos en este ejercicio. Se puede solicitar reiniciar el tiempo de ejecución, pero no es necesario.

pip install -q -U pip==20.2
pip install fairness-indicators
import os
import tempfile
import apache_beam as beam
import numpy as np
import pandas as pd
from datetime import datetime

import tensorflow_hub as hub
import tensorflow as tf
import tensorflow_model_analysis as tfma
import tensorflow_data_validation as tfdv
from tensorflow_model_analysis.addons.fairness.post_export_metrics import fairness_indicators
from tensorflow_model_analysis.addons.fairness.view import widget_view
from tensorflow_model_analysis.model_agnostic_eval import model_agnostic_predict as agnostic_predict
from tensorflow_model_analysis.model_agnostic_eval import model_agnostic_evaluate_graph
from tensorflow_model_analysis.model_agnostic_eval import model_agnostic_extractor

from witwidget.notebook.visualization import WitConfigBuilder
from witwidget.notebook.visualization import WitWidget

Descargue y comprenda los datos

Caras etiquetadas en la naturaleza es un conjunto de datos de referencia público para la verificación facial, también conocido como emparejamiento de pares. LFW contiene más de 13.000 imágenes de rostros recopiladas de la web.

Ejecutamos predicciones de FaceSSD en este conjunto de datos para predecir si una cara está presente en una imagen determinada. En este Colab, dividiremos los datos según el género para observar si hay diferencias significativas entre el rendimiento del modelo para diferentes grupos de género.

Si hay más de una cara en una imagen, el género se etiqueta como "FALTA".

Hemos alojado el conjunto de datos en Google Cloud Platform para mayor comodidad. Ejecute el siguiente código para descargar los datos de GCP, los datos tardarán aproximadamente un minuto en descargarse y analizarse.

data_location = tf.keras.utils.get_file('lfw_dataset.tf', 'https://storage.googleapis.com/facessd_dataset/lfw_dataset.tfrecord')

stats = tfdv.generate_statistics_from_tfrecord(data_location=data_location)
tfdv.visualize_statistics(stats)
Downloading data from https://storage.googleapis.com/facessd_dataset/lfw_dataset.tfrecord
200835072/200828483 [==============================] - 1s 0us/step
WARNING:apache_beam.runners.interactive.interactive_environment:Dependencies required for Interactive Beam PCollection visualization are not available, please use: `pip install apache-beam[interactive]` to install necessary dependencies to enable all data visualization features.
WARNING:apache_beam.io.tfrecordio:Couldn't find python-snappy so the implementation of _TFRecordUtil._masked_crc32c is not as fast as it could be.
WARNING:tensorflow:From /tmpfs/src/tf_docs_env/lib/python3.6/site-packages/tensorflow_data_validation/utils/stats_util.py:247: tf_record_iterator (from tensorflow.python.lib.io.tf_record) is deprecated and will be removed in a future version.
Instructions for updating:
Use eager execution and: 
`tf.data.TFRecordDataset(path)`
WARNING:tensorflow:From /tmpfs/src/tf_docs_env/lib/python3.6/site-packages/tensorflow_data_validation/utils/stats_util.py:247: tf_record_iterator (from tensorflow.python.lib.io.tf_record) is deprecated and will be removed in a future version.
Instructions for updating:
Use eager execution and: 
`tf.data.TFRecordDataset(path)`

Definición de constantes

BASE_DIR = tempfile.gettempdir()

tfma_eval_result_path = os.path.join(BASE_DIR, 'tfma_eval_result')

compute_confidence_intervals = True

slice_key = 'object/groundtruth/Gender'
label_key = 'object/groundtruth/face'
prediction_key = 'object/prediction/face'

feature_map = {
    slice_key:
        tf.io.FixedLenFeature([], tf.string, default_value=['none']),
    label_key:
        tf.io.FixedLenFeature([], tf.float32, default_value=[0.0]),
    prediction_key:
        tf.io.FixedLenFeature([], tf.float32, default_value=[0.0]),
}

Configuración agnóstica del modelo para TFMA

model_agnostic_config = agnostic_predict.ModelAgnosticConfig(
    label_keys=[label_key],
    prediction_keys=[prediction_key],
    feature_spec=feature_map)

model_agnostic_extractors = [
    model_agnostic_extractor.ModelAgnosticExtractor(
        model_agnostic_config=model_agnostic_config, desired_batch_size=3),
    tfma.extractors.slice_key_extractor.SliceKeyExtractor(
          [tfma.slicer.SingleSliceSpec(),
           tfma.slicer.SingleSliceSpec(columns=[slice_key])])
]

Devolución de llamadas de equidad y métricas de equidad informática

# Helper class for counting examples in beam PCollection
class CountExamples(beam.CombineFn):
    def __init__(self, message):
      self.message = message

    def create_accumulator(self):
      return 0

    def add_input(self, current_sum, element):
      return current_sum + 1

    def merge_accumulators(self, accumulators): 
      return sum(accumulators)

    def extract_output(self, final_sum):
      if final_sum:
        print("%s: %d"%(self.message, final_sum))
metrics_callbacks = [
  tfma.post_export_metrics.fairness_indicators(
      thresholds=[0.1, 0.3, 0.5, 0.7, 0.9],
      labels_key=label_key,
      target_prediction_keys=[prediction_key]),
  tfma.post_export_metrics.auc(
      curve='PR',
      labels_key=label_key,
      target_prediction_keys=[prediction_key]),
]

eval_shared_model = tfma.types.EvalSharedModel(
    add_metrics_callbacks=metrics_callbacks,
    construct_fn=model_agnostic_evaluate_graph.make_construct_fn(
        add_metrics_callbacks=metrics_callbacks,
        config=model_agnostic_config))

with beam.Pipeline() as pipeline:
  # Read data.
  data = (
      pipeline
      | 'ReadData' >> beam.io.ReadFromTFRecord(data_location))

  # Count all examples.
  data_count = (
      data | 'Count number of examples' >> beam.CombineGlobally(
          CountExamples('Before filtering "Gender:MISSING"')))

  # If there are more than one face in image, the gender feature is 'MISSING'
  # and we are filtering that image out.
  def filter_missing_gender(element):
    example = tf.train.Example.FromString(element)
    if example.features.feature[slice_key].bytes_list.value[0] != b'MISSING':
      yield element

  filtered_data = (
      data
      | 'Filter Missing Gender' >> beam.ParDo(filter_missing_gender))

  # Count after filtering "Gender:MISSING".
  filtered_data_count = (
      filtered_data | 'Count number of examples after filtering'
      >> beam.CombineGlobally(
          CountExamples('After filtering "Gender:MISSING"')))

  # Because LFW data set has always faces by default, we are adding
  # labels as 1.0 for all images.
  def add_face_groundtruth(element):
    example = tf.train.Example.FromString(element)
    example.features.feature[label_key].float_list.value[:] = [1.0]
    yield example.SerializeToString()

  final_data = (
      filtered_data
      | 'Add Face Groundtruth' >> beam.ParDo(add_face_groundtruth))

  # Run TFMA.
  _ = (
      final_data
      | 'ExtractEvaluateAndWriteResults' >>
       tfma.ExtractEvaluateAndWriteResults(
                 eval_shared_model=eval_shared_model,
                 compute_confidence_intervals=compute_confidence_intervals,
                 output_path=tfma_eval_result_path,
                 extractors=model_agnostic_extractors))

eval_result = tfma.load_eval_result(output_path=tfma_eval_result_path)
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring send_type hint: <class 'NoneType'>
WARNING:apache_beam.typehints.typehints:Ignoring return_type hint: <class 'NoneType'>
WARNING:tensorflow:From /tmpfs/src/tf_docs_env/lib/python3.6/site-packages/tensorflow_model_analysis/post_export_metrics/post_export_metrics.py:178: auc (from tensorflow.python.ops.metrics_impl) is deprecated and will be removed in a future version.
Instructions for updating:
The value of AUC returned by this may race with the update so this is deprecated. Please use tf.keras.metrics.AUC instead.
WARNING:tensorflow:From /tmpfs/src/tf_docs_env/lib/python3.6/site-packages/tensorflow_model_analysis/post_export_metrics/post_export_metrics.py:178: auc (from tensorflow.python.ops.metrics_impl) is deprecated and will be removed in a future version.
Instructions for updating:
The value of AUC returned by this may race with the update so this is deprecated. Please use tf.keras.metrics.AUC instead.
Before filtering "Gender:MISSING": 13836
After filtering "Gender:MISSING": 11544
WARNING:apache_beam.io.filebasedsink:Deleting 1 existing files in target path matching: 
WARNING:apache_beam.io.filebasedsink:Deleting 1 existing files in target path matching: 
WARNING:apache_beam.io.filebasedsink:Deleting 1 existing files in target path matching:

Representar indicadores de equidad

Renderice el widget Indicadores de equidad con los resultados de la evaluación exportados.

A continuación, verá gráficos de barras que muestran el rendimiento de cada segmento de los datos en las métricas seleccionadas. Puede ajustar el segmento de comparación de la línea de base, así como los umbrales mostrados mediante los menús desplegables en la parte superior de la visualización.

Una métrica relevante para este caso de uso es la tasa positiva verdadera, también conocida como recuperación. Utilice el selector en el lado izquierdo para elegir el gráfico de true_positive_rate. Estos valores métricos coinciden con los valores que se muestran en la tarjeta del modelo .

Para algunas fotos, el género se etiqueta como joven en lugar de masculino o femenino, si la persona en la foto es demasiado joven para anotarla con precisión.

widget_view.render_fairness_indicator(eval_result=eval_result,
                                      slicing_column=slice_key)
FairnessIndicatorViewer(slicingMetrics=[{'sliceValue': 'Overall', 'slice': 'Overall', 'metrics': {'post_export…