Patterns, Predictions, and Actions: Foundations of Machine Learning

Moritz Hardt and Benjamin Recht

An authoritative, up-to-date graduate textbook on machine learning that highlights its historical context and societal impacts

Hardcover

Price:: $59.00/£50.00
ISBN:
Published (US):: Oct 18, 2022
Published (UK):: Dec 13, 2022
Copyright:: 2022
Pages:: 320
Size:: 7 x 10 in.
Illus:: 41 b/w illus. 10 tables.

ebook

Price:: $59.00/£50.00
ISBN:
Published (US):: Oct 18, 2022
Published (UK):: Dec 13, 2022
Copyright:: 2022
Pages:: 320
Size:: 7 x 10 in.
Illus:: 41 b/w illus. 10 tables.

Buy This

Patterns, Predictions, and Actions introduces graduate students to the essentials of machine learning while offering invaluable perspective on its history and social implications. Beginning with the foundations of decision making, Moritz Hardt and Benjamin Recht explain how representation, optimization, and generalization are the constituents of supervised learning. They go on to provide self-contained discussions of causality, the practice of causal inference, sequential decision making, and reinforcement learning, equipping readers with the concepts and tools they need to assess the consequences that may arise from acting on statistical decisions.

Provides a modern introduction to machine learning, showing how data patterns support predictions and consequential actions
Pays special attention to societal impacts and fairness in decision making
Traces the development of machine learning from its origins to today
Features a novel chapter on machine learning benchmarks and datasets
Invites readers from all backgrounds, requiring some experience with probability, calculus, and linear algebra
An essential textbook for students and a guide for researchers

List of Figures
List of Tables
Preface
Acknowledgments
1 Introduction
- Ambitions of the twentieth century
- Pattern classiﬁcation
- Prediction and action
- Chapter notes
2 Fundamentals of Prediction
- Modeling knowledge
- Prediction via optimization
- Types of errors and successes
- The Neyman-Pearson Lemma
- Decisions that discriminate
- Chapter notes
3 Supervised Learning
- Sample versus population
- Supervised learning
- A ﬁrst learning algorithm: The perceptron
- Connection to empirical risk minimization
- Formal guarantees for the perceptron
- Chapter notes
4 Representations and Features
- Measurement
- Quantization
- Template matching
- Summarization and histograms
- Nonlinear predictors
- Chapter notes
5 Optimization
- Optimization basics
- Gradient descent
- Applications to empirical risk minimization
- Insights from quadratic functions
- Stochastic gradient descent
- Analysis of the stochastic gradient method
- Implicit convexity
- Regularization
- Squared loss methods and other optimization tools
- Chapter notes
6 Generalization
- Generalization gap
- Overparameterization: Empirical phenomena
- Theories of generalization
- Algorithmic stability
- Model complexity and uniform convergence
- Generalization from algorithms
- Looking ahead
- Chapter notes
7 Deep Learning
- Deep models and feature representation
- Optimization of deep nets
- Vanishing gradients
- Generalization in deep learning
- Chapter notes
8 Datasets
- The scientiﬁc basis of machine learning benchmarks
- A tour of datasets in diﬀerent domains
- Longevity of benchmarks
- Harms associated with data
- Toward better data practices
- Limits of data and prediction
- Chapter notes
9 Causality
- The limitations of observation
- Causal models
- Causal graphs
- Interventions and causal eﬀects
- Confounding
- Experimentation, randomization, potential outcomes
- Counterfactuals
- Chapter notes
10 Causal Inference in Practice
- Design and inference
- The observational basics: Adjustment and controls
- Reductions to model ﬁtting
- Quasi-experiments
- Limitations of causal inference in practice
- Chapter notes
11 Sequential Decision Making and Dynamic Programming
- From predictions to actions
- Dynamical systems
- Optimal sequential decision making
- Dynamic programming
- Computation
- Partial observation and the separation heuristic
- Chapter notes
12 Reinforcement Learning
- Exploration-exploitation trade-oﬀs: Regret and PAC-error
- Unknown models and approximate dynamic programming
- Certainty equivalence is often optimal
- The limits of learning in feedback loops
- Chapter notes
13 Epilogue
- Beyond pattern classiﬁcation?
14 Mathematical Background
- Common notation
- Multivariable calculus and linear algebra
- Probability
- Estimation
Bibliography
Index

"A thorough, very clearly written overview on the subject of machine learning for those with the prerequisite mathematical tools of calculus, linear algebra and probability."—Jonathan Shock, Mathemafrica

"Valuable."—J. Brzezinski, Choice

“This modern treatment of machine learning is notable for its coverage of emerging, important topics, from datasets and deep learning to optimization, causal inference, and social context, along the way pointing out the attendant perils that come from flawed predictions.”—David C. Parkes, Harvard University

“Patterns, Predictions, and Actions is not your everyday machine learning book; it’s rigorous not only in its mathematics but in its insistence on thinking about what the machines are really doing and what the subject is really about.”—Jordan Ellenberg, author of Shape: The Hidden Geometry of Information, Biology, Strategy, Democracy, and Everything Else

“Like Hansel and Gretel, machine learning students find themselves in an intellectual forest, struggling to digest a morass of ephemeral ideas. Hardt and Recht use history, theory, and society to expose the topology of this landscape. Patterns, Predictions, and Actions is a new foundational text for the field.”—Neil Lawrence, University of Cambridge

Patterns, Predictions, and Actions

Before you purchase audiobooks and ebooks

Support your local independent bookstore.

Patterns, Predictions, and Actions: Foundations of Machine Learning

Hardcover

ebook