Conformal prediction : a gentle introduction /
Black-box machine learning models are now routinely used in high-risk settings, like medical diagnostics, which demand uncertainty quantification to avoid consequential model failures. Conformal prediction is a user-friendly paradigm for creating statistically rigorous uncertainty sets/intervals for...
Saved in:
Main Authors: | |
---|---|
Group Author: | |
Published: |
Now Publishers,
|
Publisher Address: | Norwell, MA : |
Publication Dates: | [2023] |
Literature type: | Book |
Language: | English |
Series: |
Foundations and Trends® in Machine Learning,
16:4 |
Subjects: | |
Summary: |
Black-box machine learning models are now routinely used in high-risk settings, like medical diagnostics, which demand uncertainty quantification to avoid consequential model failures. Conformal prediction is a user-friendly paradigm for creating statistically rigorous uncertainty sets/intervals for the predictions of such models. One can use conformal prediction with any pre-trained model, such as a neural network, to produce sets that are guaranteed to contain the ground truth with a user-specified probability, such as 90%. It is easy-to-understand, easy-to-use, and in general, applies naturally to problems arising in the fields of computer vision, natural language processing, deep reinforcement learning, amongst others.In this hands-on introduction the authors provide the reader with a working understanding of conformal prediction and related distribution-free uncertainty quantification techniques. They lead the reader through practical theory and examples of conformal prediction and describe its extensions to complex machine learning tasks involving structured outputs, distribution shift, time-series, outliers, models that abstain, and more. Throughout, there are many explanatory illustrations, examples, and code samples in Python. With each code sample comes a Jupyter notebook implementing the method on a real-data example.This hands-on tutorial, full of practical and accessible examples, is essential reading for all students, practitioners and researchers working on all types of systems deploying machine learning techniques. |
Item Description: |
"Now Publishers" 1. Conformal Prediction2. Examples of Conformal Procedures3. Evaluating Conformal Prediction4. Extensions of Conformal Prediction5. Worked Examples6. Full Conformal Prediction7. Historical Notes on Conformal Prediction8. AcknowledgementsAppendicesReferences |
Carrier Form: | 106 pages : color illustrations ; 24 cm. |
Bibliography: | Includes bibliographical references (pages 95-106). |
ISBN: | 9781638281580 |
Index Number: | QA279 |
CLC: | O211.67 |
Call Number: | O211.67/A584 |
Contents: | Intro -- Conformal Prediction -- Examples of Conformal Procedures -- Evaluating Conformal Prediction -- Extensions of Conformal Prediction -- Worked Examples -- Full Conformal Prediction -- Historical Notes on Conformal Prediction -- Acknowledgements -- Appendices -- Distribution-Free Control of General Risks -- Examples of Distribution-Free Risk Control -- Concentration Properties of the Empirical Coverage -- Theorem and Proof: Coverage Property of Conformal Prediction -- References |