Gibbs Measures In Machine Learning

Gibbs Measures In Machine Learning book cover

Gibbs Measures In Machine Learning

Author(s): Abdullaev (Author), Rozikov (Author)

  • Publisher finelybook 出版社: WSPC
  • Publication Date 出版日期: November 25, 2025
  • Language 语言: English
  • Print length 页数: 380 pages
  • ISBN-10: 9819814561
  • ISBN-13: 9789819814565

Book Description

From the Ising model to large language models, Gibbs Measures in Machine Learning offers a complete journey through one of the most powerful concepts connecting statistical physics and modern AI. Starting with the mathematical foundations — measure theory, Markov chains, and configuration spaces — the book builds toward advanced applications in Bayesian inference, structured prediction, unsupervised learning, and deep neural networks. Along the way, it bridges classical models such as Potts and Solid-on-Solid with state-of-the-art techniques like attention mechanisms, diffusion models, and probabilistic programming. Readers will find clear, rigorous explanations of Gibbs measures and their probabilistic underpinnings, practical guidance on Gibbs sampling, MCMC, and interacting particle systems, case studies ranging from deep linear networks to transformer architectures, and insights into emerging trends, including modern associative memories and thermodynamics of autoregressive language modeling. Whether you are a researcher, graduate student, or experienced practitioner, this book provides the theoretical depth and practical tools needed to harness Gibbs measures for robust, efficient, and interpretable machine learning models.

Editorial Reviews

About the Author

Laziz U Abdullaev earned his bachelor’s degree in mathematics from the National University of Uzbekistan (NUUz) before commencing his PhD studies specializing in mathematics and machine learning in the National University of Singapore (NUS) in 2023.

During his prolific times at NUUz, Abdullaev has been an awardee in myriad international mathematics Olympiads for university students. Those together with his expertise in coding acquired through working for international IT companies have helped him win numerous graduate-level scholarships including the Erasmus and NUS Research Scholarship. His current research interests at NUS consist of developing mathematical interpretations of Large Language Models (LLMs), transformer models and attention mechanisms in particular, through the lens of well-established image processing and control theories.

Academician U A Rozikov earned his PhD in 1995 and his Doctor of Sciences degree in 2001 from the Institute of Mathematics in Tashkent. Since 2019, Rozikov has served as the Deputy Director of the Institute of Mathematics.

Rozikov is a globally recognized expert in probability, mathematical physics, and analysis, with a specialization in dynamical systems and statistical mechanics on graphs. His significant contributions include in-depth analyses of Gibbs measures in key models (Ising, Potts, SOS, HC) of statistical mechanics on trees. He has introduced innovative tools for studying Gibbs measures on graphs, including group representation theory, information flows, node-weighted random walks, contour methods on trees, and non-linear analysis.

Rozikov is an exceptionally prolific author, having published over 170 papers since 1995 in prestigious journals. His work includes 12 papers in the renowned Journal of Statistical Physics, 27 papers in Theoretical and Mathematical Physics, Lett. Math. Phys., Comm. Math. Phys., J. Stat. Mech. Theory Exp., J. Math. Anal. Appl., among others. Notably, he has published extensive review papers in Rev. Math. Phys. in 2013 and 2021. Additionally, Rozikov is the author of five research monographs, two of which are related to Gibbs measures.

Amazon Page

下载地址

PDF | 12 MB | 2026-01-15
下载地址 Download解决验证以访问链接!
打赏
未经允许不得转载:finelybook » Gibbs Measures In Machine Learning

评论 抢沙发

觉得文章有用就打赏一下文章作者

您的打赏,我们将继续给力更多优质内容

支付宝扫一扫

微信扫一扫