Perturbations, Optimization, and Statistics

— —

Tamir Hazan (Technion -- Israel Institute of Technology) George Papandreou (Research Scientist, Google, Inc.)

Perturbations, Optimization, and Statistics

Tamir Hazan (Technion -- Israel Institute of Technology), George Papandreou (Research Scientist, Google, Inc.), Daniel Tarlow (Researcher, Microsoft Research)

$165

Paperback

Not in-store but you can order this
How long will it take?

Availability Information

We source books from suppliers in Australia and overseas. For books we don't currently have in stock, the time it takes to get them from our suppliers can vary widely - from a few days to a few months - so we check each book with each supplier to determine the expected time it will take to be supplied to us.

We then advise you accordingly. If the time taken to get any book is too long for you, you can let us know and we will cancel or adjust your order, and refund as required.

To find out the anticipated arrival time for specific items prior to ordering, please contact us by phone or email:

Phone +61 2 9264 3111, or 1800 4 BOOKS (1800 4 26657) if outside Sydney:
option 1 Abbey's Bookshop (Crime, History, Science, Kids & more) • info@abbeys.com.au
option 2 Language Book Centre (ESL & Foreign Languages) • language@abbeys.com.au
option 3 Galaxy Bookshop (Sci-fi, Fantasy, Romance, Graphic Novels) • sf@galaxybooks.com.au

QTY:

English

MIT Press
05 December 2023

Optimisation; Machine learning

Series: Neural Information Processing series

A description of perturbation-based methods developed in machine learning to augment novel optimization methods with strong statistical guarantees.

A description of perturbation-based methods developed in machine learning to augment novel optimization methods with strong statistical guarantees.

In nearly all machine learning, decisions must be made given current knowledge. Surprisingly, making what is believed to be the best decision is not always the best strategy, even when learning in a supervised learning setting. An emerging body of work on learning under different rules applies perturbations to decision and learning procedures. These methods provide simple and highly efficient learning rules with improved theoretical guarantees. This book describes perturbation-based methods developed in machine learning to augment novel optimization methods with strong statistical guarantees, offering readers a state-of-the-art overview.

Chapters address recent modeling ideas that have arisen within the perturbations framework, including Perturb & MAP, herding, and the use of neural networks to map generic noise to distribution over highly structured data. They describe new learning procedures for perturbation models, including an improved EM algorithm and a learning algorithm that aims to match moments of model samples to moments of data. They discuss understanding the relation of perturbation models to their traditional counterparts, with one chapter showing that the perturbations viewpoint can lead to new algorithms in the traditional setting. And they consider perturbation-based regularization in neural networks, offering a more complete understanding of dropout and studying perturbations in the context of deep neural networks.

Edited by: Tamir Hazan (Technion -- Israel Institute of Technology), George Papandreou (Research Scientist, Google, Inc.), Daniel Tarlow (Researcher, Microsoft Research)
Imprint: MIT Press
Country of Publication: United States
Dimensions: Height: 254mm, Width: 203mm,
Weight: 369g
ISBN: 9780262549943
ISBN 10: 0262549948
Series: Neural Information Processing series
Pages: 412
Publication Date: 05 December 2023
Recommended Age: From 18 years
Audience: Professional and scholarly , Undergraduate
Format: Paperback
Publisher's Status: Active

Tamir Hazan is Assistant Professor at Technion, Israel Institute of Technology. George Papandreou is a Research Scientist for Google, Inc. Daniel Tarlow is a Researcher at Microsoft Research Cambridge, UK.