Hands-On Explainable AI (XAI) with Python

Spis treści książki

• Preface
  • Who this book is for
  • What this book covers
  • To get the most out of this book
  • Get in touch
• Explaining Artificial Intelligence with Python
  • Defining explainable AI
    • Going from black box models to XAI white box models
    • Explaining and interpreting
  • Designing and extracting
    • The XAI executive function
  • The XAI medical diagnosis timeline
    • The standard AI program used by a general practitioner
      • Definition of a KNN algorithm
      • A KNN in Python
    • West Nile virus a case of life or death
      • How can a lethal mosquito bite go unnoticed?
      • What is the West Nile virus?
      • How did the West Nile virus get to Chicago?
    • XAI can save lives using Google Location History
    • Downloading Google Location History
      • Googles Location History extraction tool
    • Reading and displaying Google Location History data
      • Installation of the basemap packages
      • The import instructions
      • Importing the data
      • Processing the data for XAI and basemap
      • Setting up the plotting options to display the map
    • Enhancing the AI diagnosis with XAI
      • Enhanced KNN
    • XAI applied to the medical diagnosis experimental program
      • Displaying the KNN plot
      • Natural language explanations
      • Displaying the Location History map
      • Showing mosquito detection data and natural language explanations
      • A critical diagnosis is reached with XAI
  • Summary
  • Questions
  • References
  • Further reading
• White Box XAI for AI Bias and Ethics
  • Moral AI bias in self-driving cars
    • Life and death autopilot decision making
    • The trolley problem
    • The MIT Moral Machine experiment
    • Real life and death situations
    • Explaining the moral limits of ethical AI
  • Standard explanation of autopilot decision trees
    • The SDC autopilot dilemma
    • Importing the modules
    • Retrieving the dataset
    • Reading and splitting the data
    • Theoretical description of decision tree classifiers
    • Creating the default decision tree classifier
    • Training, measuring, and saving the model
    • Displaying a decision tree
  • XAI applied to an autopilot decision tree
    • Structure of a decision tree
      • The default output of the default structure of a decision tree
      • The customized output of a customized structure of a decision tree
      • The output of a customized structure of a decision tree
  • Using XAI and ethics to control a decision tree
    • Loading the model
    • Accuracy measurements
    • Simulating real-time cases
    • Introducing ML bias due to noise
    • Introducing ML ethics and laws
      • Case 1 not overriding traffic regulations to save four pedestrians
      • Case 2 overriding traffic regulations
      • Case 3 introducing emotional intelligence in the autopilot
  • Summary
  • Questions
  • References
  • Further reading
• Explaining Machine Learning with Facets
  • Getting started with Facets
    • Installing Facets on Google Colaboratory
    • Retrieving the datasets
    • Reading the data files
  • Facets Overview
    • Creating feature statistics for the datasets
      • Implementing the feature statistics code
      • Implementing the HTML code to display feature statistics
  • Sorting the Facets statistics overview
    • Sorting data by feature order
      • XAI motivation for sorting features
    • Sorting by non-uniformity
    • Sorting by alphabetical order
    • Sorting by amount missing/zero
    • Sorting by distribution distance
  • Facets Dive
    • Building the Facets Dive display code
    • Defining the labels of the data points
    • Defining the color of the data points
    • Defining the binning of the x axis and y axis
    • Defining the scatter plot of the x axis and the y axis
  • Summary
  • Questions
  • References
  • Further reading
• Microsoft Azure Machine Learning Model Interpretability with SHAP
  • Introduction to SHAP
    • Key SHAP principles
      • Symmetry
      • Null player
      • Additivity
    • A mathematical expression of the Shapley value
    • Sentiment analysis example
      • Shapley value for the first feature, "good"
      • Shapley value for the second feature, "excellent"
      • Verifying the Shapley values
  • Getting started with SHAP
    • Installing SHAP
      • Importing the modules
    • Importing the data
      • Intercepting the dataset
    • Vectorizing the datasets
  • Linear models and logistic regression
    • Creating, training, and visualizing the output of a linear model
    • Defining a linear model
    • Agnostic model explaining with SHAP
    • Creating the linear model explainer
    • Creating the plot function
    • Explaining the output of the model's prediction
      • Explaining intercepted dataset reviews with SHAP
      • Explaining the original IMDb reviews with SHAP
  • Summary
  • Questions
  • References
  • Further reading
    • Additional publications
• Building an Explainable AI Solution from Scratch
  • Moral, ethical, and legal perspectives
  • The U.S. census data problem
    • Using pandas to display the data
    • Moral and ethical perspectives
      • The moral perspective
      • The ethical perspective
      • The legal perspective
  • The machine learning perspective
    • Displaying the training data with Facets Dive
    • Analyzing the training data with Facets
    • Verifying the anticipated outputs
      • Using KMC to verify the anticipated results
      • Analyzing the output of the KMC algorithm
      • Conclusion of the analysis
    • Transforming the input data
  • WIT applied to a transformed dataset
  • Summary
  • Questions
  • References
  • Further reading
• AI Fairness with Google's What-If Tool (WIT)
  • Interpretability and explainability from an ethical AI perspective
    • The ethical perspective
    • The legal perspective
    • Explaining and interpreting
    • Preparing an ethical dataset
  • Getting started with WIT
    • Importing the dataset
    • Preprocessing the data
    • Creating data structures to train and test the model
  • Creating a DNN model
    • Training the model
  • Creating a SHAP explainer
    • The plot of Shapley values
  • Model outputs and SHAP values
  • The WIT datapoint explorer and editor
    • Creating WIT
    • The datapoint editor
    • Features
    • Performance and fairness
      • Ground truth
      • Cost ratio
      • Slicing
      • Fairness
      • The ROC curve and AUC
      • The PR curve
      • The confusion matrix
  • Summary
  • Questions
  • References
  • Further reading
• A Python Client for Explainable AI Chatbots
  • The Python client for Dialogflow
    • Installing the Python client for Google Dialogflow
    • Creating a Google Dialogflow agent
    • Enabling APIs and services
    • The Google Dialogflow Python client
  • Enhancing the Google Dialogflow Python client
    • Creating a dialog function
    • The constraints of an XAI implementation on Dialogflow
    • Creating an intent in Dialogflow
      • The training phrases of the intent
      • The response of an intent
      • Defining a follow-up intent for an intent
    • The XAI Python client
      • Inserting interactions in the MDP
      • Interacting with Dialogflow with the Python client
  • A CUI XAI dialog using Google Dialogflow
    • Dialogflow integration for a website
    • A Jupyter Notebook XAI agent manager
    • Google Assistant
  • Summary
  • Questions
  • Further reading
• Local Interpretable Model-Agnostic Explanations (LIME)
  • Introducing LIME
    • A mathematical representation of LIME
  • Getting started with LIME
    • Installing LIME on Google Colaboratory
    • Retrieving the datasets and vectorizing the dataset
  • An experimental AutoML module
    • Creating an agnostic AutoML template
    • Bagging classifiers
    • Gradient boosting classifiers
    • Decision tree classifiers
    • Extra trees classifiers
  • Interpreting the scores
  • Training the model and making predictions
    • The interactive choice of classifier
    • Finalizing the prediction process
      • Interception functions
  • The LIME explainer
    • Creating the LIME explainer
    • Interpreting LIME explanations
      • Explaining the predictions as a list
      • Explaining with a plot
      • Conclusions of the LIME explanation process
  • Summary
  • Questions
  • References
  • Further reading
• The Counterfactual Explanations Method
  • The counterfactual explanations method
    • Dataset and motivations
    • Visualizing counterfactual distances in WIT
    • Exploring data point distances with the default view
    • The logic of counterfactual explanations
      • Belief
      • Truth
      • Justification
      • Sensitivity
  • The choice of distance functions
    • The L1 norm
    • The L2 norm
    • Custom distance functions
  • The architecture of the deep learning model
    • Invoking WIT
    • The custom prediction function for WIT
    • Loading a Keras model
    • Retrieving the dataset and model
  • Summary
  • Questions
  • References
  • Further reading
• Contrastive XAI
  • The contrastive explanations method
  • Getting started with the CEM applied to MNIST
    • Installing Alibi and importing the modules
    • Importing the modules and the dataset
      • Importing the modules
      • Importing the dataset
      • Preparing the data
  • Defining and training the CNN model
    • Creating the CNN model
    • Training the CNN model
    • Loading and testing the accuracy of the model
  • Defining and training the autoencoder
    • Creating the autoencoder
    • Training and saving the autoencoder
    • Comparing the original images with the decoded images
  • Pertinent negatives
    • CEM parameters
    • Initializing the CEM explainer
    • Pertinent negative explanations
  • Summary
  • Questions
  • References
  • Further reading
• Anchors XAI
  • Anchors AI explanations
    • Predicting income
    • Classifying newsgroup discussions
  • Anchor explanations for ImageNet
    • Installing Alibi and importing the modules
    • Loading an InceptionV3 model
    • Downloading an image
    • Processing the image and making predictions
    • Building the anchor image explainer
    • Explaining other categories
    • Other images and difficulties
  • Summary
  • Questions
  • References
  • Further reading
• Cognitive XAI
  • Cognitive rule-based explanations
    • From XAI tools to XAI concepts
    • Defining cognitive XAI explanations
    • A cognitive XAI method
      • Importing the modules and the data
      • The dictionaries
      • The global parameters
      • The cognitive explanation function
    • The marginal contribution of a feature
      • A mathematical perspective
      • The Python marginal cognitive contribution function
  • A cognitive approach to vectorizers
    • Explaining the vectorizer for LIME
    • Explaining the IMDb vectorizer for SHAP
  • Human cognitive input for the CEM
    • Rule-based perspectives
  • Summary
  • Questions
  • Further reading
• Answers to the Questions
  • Chapter 1, Explaining Artificial Intelligence with Python
  • Chapter 2, White Box XAI for AI Bias and Ethics
  • Chapter 3, Explaining Machine Learning with Facets
  • Chapter 4, Microsoft Azure Machine Learning Model Interpretability with SHAP
  • Chapter 5, Building an Explainable AI Solution from Scratch
  • Chapter 6, AI Fairness with Google's What-If Tool (WIT)
  • Chapter 7, A Python Client for Explainable AI Chatbots
  • Chapter 8, Local Interpretable Model-Agnostic Explanations (LIME)
  • Chapter 9, The Counterfactual Explanations Method
  • Chapter 10, Contrastive XAI
  • Chapter 11, Anchors XAI
  • Chapter 12, Cognitive XAI
• Other Books You May Enjoy
• Index

pokaż cały spis treści

ukryj recenzje