Practical Guide to Applied Conformal Prediction in Python : Learn and Apply the Best Uncertainty Frameworks to Your Industry Applications / Valery Manokhin and Agus Sudjianto.
"Take your machine learning skills to the next level by mastering the best framework for uncertainty quantification - Conformal Prediction Key Features Master Conformal Prediction, a fast-growing ML framework, with Python applications. Explore cutting-edge methods to measure and manage uncertainty in industry applications. The book will explain how Conformal Prediction differs from traditional machine learning. Book Description In the rapidly evolving landscape of machine learning, the ability to accurately quantify uncertainty is pivotal. ""Practical Guide to Applied Conformal Prediction in Python"" addresses this need by offering an in-depth exploration of Conformal Prediction, a cutting-edge framework set to revolutionize uncertainty management in various ML applications. Embark on a comprehensive journey through Conformal Prediction, exploring its fundamentals and practical applications in binary classification, regression, time series forecasting, imbalanced data, computer vision, and NLP. Each chapter delves into specific aspects, offering hands-on insights and best practices for enhancing prediction reliability. The book concludes with a focus on multi-class classification nuances, providing expert-level proficiency to seamlessly integrate Conformal Prediction into diverse industries. Practical examples in Python using real-world datasets reinforce intuitive explanations, ensuring you acquire a robust understanding of this modern framework for uncertainty quantification. This guide is a beacon for mastering Conformal Prediction in Python, providing a blend of theory and practical application. It serves as a comprehensive toolkit to enhance machine learning skills, catering to professionals from data scientists to ML engineers. What you will learn The fundamental concepts and principles of conformal prediction Learn how conformal prediction differs from traditional ML methods Apply real-world examples to your own industry applications Explore advanced topics - imbalanced data and multi-class CP Dive into the details of the conformal prediction framework Boost your career as a data scientist, ML engineer, or researcher Learn to apply conformal prediction to forecasting and NLP Who this book is for Ideal for readers with a basic understanding of machine learning concepts and Python programming, this book caters to data scientists, ML engineers, academics, and anyone keen on advancing their skills in uncertainty quantification in ML.".
Notes
Includes index.
Source of description
Description based on print version record.
Contents
Intro
Title Page
Copyright and Credits
Foreword
Contributors
Table of Contents
Preface
Part 1: Introduction
Chapter 1: Introducing Conformal Prediction
Technical requirements
Introduction to conformal prediction
Understanding conformity measures
The origins of conformal prediction
The future of conformal prediction
How conformal prediction differs from traditional machine learning
The p-value and its role in conformal prediction
Summary
Chapter 2: Overview of Conformal Prediction
Understanding uncertainty quantification
Aleatoric uncertainty
Epistemic uncertainty
Different ways to quantify uncertainty
Quantifying uncertainty using conformal prediction
Part 2: Conformal Prediction Framework
Chapter 3: Fundamentals of Conformal Prediction
Fundamentals of conformal prediction
Definition and principles
Basic components of a conformal predictor
Types of nonconformity measures
Chapter 4: Validity and Efficiency of Conformal Prediction
The validity of probabilistic predictors
Classifier calibration
The efficiency of probabilistic predictors
Chapter 5: Types of Conformal Predictors
Understanding classical predictors
Applying TCP for classification problems
Applying TCP for regression problems
Advantages
Understanding inductive conformal predictors
Choosing the right conformal predictor
Transductive conformal predictors
Inductive conformal predictors
Part 3: Applications of Conformal Prediction
Chapter 6: Conformal Prediction for Classification
Understanding the concepts of classifier calibration
Evaluating calibration performance
Various approaches to classifier calibration
Histogram binning
Platt scaling
Isotonic regression.
Conformal prediction for classifier calibration
Venn-ABERS conformal prediction
Comparing calibration methods
Open source tools for conformal prediction in classification problems
Nonconformist
Chapter 7: Conformal Prediction for Regression
Uncertainty quantification for regression problems
Understanding the types and sources of uncertainty in regression modeling
The concept of prediction intervals
Why do we need prediction intervals?
How is it different from a confidence interval?
Conformal prediction for regression problems
Building prediction intervals and predictive distributions using conformal prediction
Mechanics of CQR
Quantile regression
CQR
Jackknife+
Jackknife regression
Jackknife+ regression
Conformal predictive distributions
Chapter 8: Conformal Prediction for Time Series and Forecasting
UQ for time series and forecasting problems
The importance of UQ
The history of UQ
Early statistical methods - the roots of UQ in time series
Modern machine learning approaches
The concept of PIs in forecasting applications
Definition and construction
The importance of forecasting applications
Challenges and considerations
Various approaches to producing PIs
Parametric approaches
Non-parametric approaches
Bayesian approaches
Machine learning approaches
Conformal prediction
Conformal prediction for time series and forecasting
Ensemble batch PIs (EnbPIs)
NeuralProphet
Chapter 9: Conformal Prediction for Computer Vision
Uncertainty quantification for computer vision
Why does uncertainty matter?
Types of uncertainty in computer vision
Quantifying uncertainty
Why does deep learning produce miscalibrated predictions?
Post-2012 - the deep learning surge.
The "calibration crisis" in deep learning - a turning point in 2017
Overconfidence in modern deep learning computer vision models
Various approaches to quantify uncertainty in computer vision problems
The superiority of conformal prediction in uncertainty quantification
Conformal prediction for computer vision
Uncertainty sets for image classifiers using conformal prediction
Building computer vision classifiers using conformal prediction
Naïve Conformal prediction
Adaptive Prediction Sets (APS)
Regularized Adaptive Prediction Sets (RAPS)
Chapter 10: Conformal Prediction for Natural Language Processing
Uncertainty quantification for NLP
What is uncertainty in NLP?
Benefits of quantifying uncertainty in NLP
The challenges of uncertainty in NLP
Understanding why deep learning produces miscalibrated predictions
Introduction to deep learning in NLP
Challenges with deep learning predictions in NLP
The implications of miscalibration
Various approaches to quantify uncertainty in NLP problems
Bootstrap methods and ensemble techniques
Out-of-distribution (OOD) detection
Conformal prediction for NLP
How conformal prediction works in NLP
Practical applications of conformal prediction in NLP
Advantages of using conformal prediction in NLP
Part 4: Advanced Topics
Chapter 11: Handling Imbalanced Data
Introducing imbalanced data
Why imbalanced data problems are complex to solve
Methods for solving imbalanced data
The methods for solving imbalanced data
Solving imbalanced data problems by applying conformal prediction
Addressing imbalanced data with Venn-Abers predictors
Key insights from the Credit Card Fraud Detection notebook
Chapter 12: Multi-Class Conformal Prediction
Multi-class classification problems.
Algorithms for multi-class classification
One-vs-all and one-vs-one strategies
Metrics for multi-class classification problems
Confusion matrix
Precision
Recall
F1 score
Macro- and micro-averaged metrics
Area Under Curve (AUC-ROC)
Log loss and its application in measuring calibration of multi-class models
Brier score and its application in measuring the calibration of multi-class models
How conformal prediction can be applied to multi-class classification problems
Multi-class probabilistic classification using inductive and cross-Venn-ABERS predictors
Index
Other Books You May Enjoy
Blank Page.
ISBN
9781805120919
1805120913
OCLC
1414457176
Statement on responsible collection description
Princeton University Library aims to describe library materials in a manner that is respectful to the individuals and communities who create, use, and are represented in the collections we manage. Read more...