book

Interpretable AI

Name: Interpretable AI
Author: Ajay Thampi
ISBN: 9781617297649

by Ajay Thampi

July 2022

Intermediate to advanced

328 pages

10h 17m

English

Manning Publications

Audiobook available

Read now

Unlock full access

inside front cover
Interpretable AI
Copyright
dedication
brief content
contents
front matter
prefaceacknowledgmentsabout this bookWho should read this bookHow this book is organized: a roadmapAbout the codeliveBook discussion forumabout the authorabout the cover illustration
Part 1. Interpretability basics
1 Introduction
1.1 Diagnostics+ AI—an example AI system1.2 Types of machine learning systems1.2.1 Representation of data1.2.2 Supervised learning1.2.3 Unsupervised learning1.2.4 Reinforcement learning1.2.5 Machine learning system for Diagnostics+ AI1.3 Building Diagnostics+ AI1.4 Gaps in Diagnostics+ AI1.4.1 Data leakage1.4.2 Bias1.4.3 Regulatory noncompliance1.4.4 Concept drift1.5 Building a robust Diagnostics+ AI system1.6 Interpretability vs. explainability1.6.1 Types of interpretability techniques1.7 What will I learn in this book?1.7.1 What tools will I be using in this book?1.7.2 What do I need to know before reading this book?Summary
2 White-box models
2.1 White-box models2.2 Diagnostics+—diabetes progression2.3 Linear regression2.3.1 Interpreting linear regression2.3.2 Limitations of linear regression2.4 Decision trees2.4.1 Interpreting decision trees2.4.2 Limitations of decision trees2.5 Generalized additive models (GAMs)2.5.1 Regression splines2.5.2 GAM for Diagnostics+ diabetes2.5.3 Interpreting GAMs2.5.4 Limitations of GAMs2.6 Looking ahead to black-box modelsSummary

Part 2. Interpreting model processing
3 Model-agnostic methods: Global interpretability
3.1 High school student performance predictor3.1.1 Exploratory data analysis3.2 Tree ensembles3.2.1 Training a random forest3.3 Interpreting a random forest3.4 Model-agnostic methods: Global interpretability3.4.1 Partial dependence plots3.4.2 Feature interactionsSummary
4 Model-agnostic methods: Local interpretability
4.1 Diagnostics+ AI: Breast cancer diagnosis4.2 Exploratory data analysis4.3 Deep neural networks4.3.1 Data preparation4.3.2 Training and evaluating DNNs4.4 Interpreting DNNs4.5 LIME4.6 SHAP4.7 AnchorsSummary
5 Saliency mapping
5.1 Diagnostics+ AI: Invasive ductal carcinoma detection5.2 Exploratory data analysis5.3 Convolutional neural networks5.3.1 Data preparation5.3.2 Training and evaluating CNNs5.4 Interpreting CNNs5.4.1 Probability landscape5.4.2 LIME5.4.3 Visual attribution methods5.5 Vanilla backpropagation5.6 Guided backpropagation5.7 Other gradient-based methods5.8 Grad-CAM and guided Grad-CAM5.9 Which attribution method should I use?Summary
Part 3. Interpreting model representations
6 Understanding layers and units
6.1 Visual understanding6.2 Convolutional neural networks: A recap6.3 Network dissection framework6.3.1 Concept definition6.3.2 Network probing6.3.3 Quantifying alignment6.4 Interpreting layers and units6.4.1 Running network dissection6.4.2 Concept detectors6.4.3 Concept detectors by training task6.4.4 Visualizing concept detectors6.4.5 Limitations of network dissectionSummary
7 Understanding semantic similarity
7.1 Sentiment analysis7.2 Exploratory data analysis7.3 Neural word embeddings7.3.1 One-hot encoding7.3.2 Word2Vec7.3.3 GloVe embeddings7.3.4 Model for sentiment analysis7.4 Interpreting semantic similarity7.4.1 Measuring similarity7.4.2 Principal component analysis (PCA)7.4.3 t-distributed stochastic neighbor embedding (t-SNE)7.4.4 Validating semantic similarity visualizationsSummary
Part 4. Fairness and bias
8 Fairness and mitigating bias
8.1 Adult income prediction8.1.1 Exploratory data analysis8.1.2 Prediction model8.2 Fairness notions8.2.1 Demographic parity8.2.2 Equality of opportunity and odds8.2.3 Other notions of fairness8.3 Interpretability and fairness8.3.1 Discrimination via input features8.3.2 Discrimination via representation8.4 Mitigating bias8.4.1 Fairness through unawareness8.4.2 Correcting label bias through reweighting8.5 Datasheets for datasetsSummary
9 Path to explainable AI
9.1 Explainable AI9.2 Counterfactual explanationsSummary
Appendix A. Getting set up
A.1 PythonA.2 Git code repositoryA.3 Conda environmentA.4 Jupyter notebooksA.5 Docker
Appendix B. PyTorch
B.1 What is PyTorch?B.2 Installing PyTorchB.3 TensorsB.3.1 Data typesB.3.2 CPU and GPU tensorsB.3.3 OperationsB.4 Dataset and DataLoaderB.5 ModelingB.5.1 Automatic differentiationB.5.2 Model definitionB.5.3 Training
index
inside back cover

Content preview from Interpretable AI

Part 2. Interpreting model processing

This part of the book focuses on black-box models and understanding how the model processes the inputs and arrives at the final prediction.

In chapter 3, you’ll learn about a class of black-box models called tree ensembles. You will learn about their characteristics and what makes them black-box. You’ll also learn how to interpret them using post hoc model-agnostic methods that are global in scope, such as partial dependence plots (PDPs) and feature interaction plots.

In chapter 4, you’ll learn about deep neural networks, specifically the vanilla fully connected neural networks. You will learn about characteristics that make these models black-box and how to interpret them using post hoc model-agnostic methods ...

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Publisher Resources

ISBN: 9781617297649Publisher Support Publisher Website Errata Page

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Interpretable AI

by Ajay Thampi

Part 2. Interpreting model processing

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.