Algebraic Geometry and Statistical Learning Theory

Nonfiction, Computers, Advanced Computing, Engineering, Computer Vision, Science & Nature, Mathematics, General Computing
Cover of the book Algebraic Geometry and Statistical Learning Theory by Sumio Watanabe, Cambridge University Press
View on Amazon View on AbeBooks View on Kobo View on B.Depository View on eBay View on Walmart
Author: Sumio Watanabe ISBN: 9781107713963
Publisher: Cambridge University Press Publication: August 13, 2009
Imprint: Cambridge University Press Language: English
Author: Sumio Watanabe
ISBN: 9781107713963
Publisher: Cambridge University Press
Publication: August 13, 2009
Imprint: Cambridge University Press
Language: English

Sure to be influential, this book lays the foundations for the use of algebraic geometry in statistical learning theory. Many widely used statistical models and learning machines applied to information science have a parameter space that is singular: mixture models, neural networks, HMMs, Bayesian networks, and stochastic context-free grammars are major examples. Algebraic geometry and singularity theory provide the necessary tools for studying such non-smooth models. Four main formulas are established: 1. the log likelihood function can be given a common standard form using resolution of singularities, even applied to more complex models; 2. the asymptotic behaviour of the marginal likelihood or 'the evidence' is derived based on zeta function theory; 3. new methods are derived to estimate the generalization errors in Bayes and Gibbs estimations from training errors; 4. the generalization errors of maximum likelihood and a posteriori methods are clarified by empirical process theory on algebraic varieties.

View on Amazon View on AbeBooks View on Kobo View on B.Depository View on eBay View on Walmart

Sure to be influential, this book lays the foundations for the use of algebraic geometry in statistical learning theory. Many widely used statistical models and learning machines applied to information science have a parameter space that is singular: mixture models, neural networks, HMMs, Bayesian networks, and stochastic context-free grammars are major examples. Algebraic geometry and singularity theory provide the necessary tools for studying such non-smooth models. Four main formulas are established: 1. the log likelihood function can be given a common standard form using resolution of singularities, even applied to more complex models; 2. the asymptotic behaviour of the marginal likelihood or 'the evidence' is derived based on zeta function theory; 3. new methods are derived to estimate the generalization errors in Bayes and Gibbs estimations from training errors; 4. the generalization errors of maximum likelihood and a posteriori methods are clarified by empirical process theory on algebraic varieties.

More books from Cambridge University Press

Cover of the book West Germany, Cold War Europe and the Algerian War by Sumio Watanabe
Cover of the book Practical Procedures in Orthopaedic Trauma Surgery by Sumio Watanabe
Cover of the book Borges' Classics by Sumio Watanabe
Cover of the book Physics MCQs for the Part 1 FRCR by Sumio Watanabe
Cover of the book The Biafran War and Postcolonial Humanitarianism by Sumio Watanabe
Cover of the book From Philology to English Studies by Sumio Watanabe
Cover of the book Optimal Regulation and the Law of International Trade by Sumio Watanabe
Cover of the book Narrative and the Making of US National Security by Sumio Watanabe
Cover of the book Memory Laws, Memory Wars by Sumio Watanabe
Cover of the book Arbitration Law in America by Sumio Watanabe
Cover of the book Radiology for Anaesthesia and Intensive Care by Sumio Watanabe
Cover of the book Heat Transfer Physics by Sumio Watanabe
Cover of the book The Internet, Warts and All by Sumio Watanabe
Cover of the book Violence against Prisoners of War in the First World War by Sumio Watanabe
Cover of the book Iambic Poetics in the Roman Empire by Sumio Watanabe
We use our own "cookies" and third party cookies to improve services and to see statistical information. By using this website, you agree to our Privacy Policy