A unified performance analysis of likelihood-informed subspace methods

نویسندگان

چکیده

The likelihood-informed subspace (LIS) method offers a viable route to reducing the dimensionality of high-dimensional probability distributions arising in Bayesian inference. LIS identifies an intrinsic low-dimensional linear where target distribution differs most from some tractable reference distribution. Such can be identified using leading eigenvectors Gram matrix gradient log-likelihood function. Then, original is approximated through various forms marginalization likelihood function, which only has support on subspace. This approximation enables design inference algorithms that scale sub-linearly with apparent problem. Intuitively, accuracy approximation, and hence performance algorithms, are influenced by three factors—the dimension truncation error identifying subspace, Monte Carlo estimating matrices, constructing marginalizations. work establishes unified framework analyze each these factors their interplay. Under mild technical assumptions, we establish bounds for range existing reduction techniques based principle LIS. Our also provide useful insights into methods. In addition, integration sampling methods such as Markov Chain (MCMC) sequential (SMC). We demonstrate applicability our analysis inverse problem Gaussian prior, shows all estimates dimension-independent if prior covariance trace-class operator. Finally, aspects theoretical claims two nonlinear problems.

برای دانلود باید عضویت طلایی داشته باشید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Unified Subspace Analysis for Face Recognition

We propose a face difference model that decomposes face difference into three components, intrinsic difference, transformation difference, and noise. Using the face difference model and a detailed subspace analysis on the three components we develop a unified framework for subspace analysis. Using this framework we discover the inherent relationship among different subspace methods and their un...

متن کامل

Unified Linear Subspace Approach to Semantic Analysis

The Basic Vector Space Model (BVSM) is well known in information retrieval. Unfortunately, it can not retrieve all relevant documents since it is based on literal term matching. The Generalized Vector Space Model (GVSM) and Latent Semantic Indexing (LSI) are two famous semantic retrieval methods, for which we assume there is some underlying latent semantic structure in a dataset. However, their...

متن کامل

a swot analysis of the english program of a bilingual school in iran

با توجه به جایگاه زبان انگلیسی به عنوان زبانی بین المللی و با در نظر گرفتن این واقعیت که دولت ها و مسئولان آموزش و پرورش در سراسر جهان در حال حاضر احساس نیاز به ایجاد موقعیتی برای کودکان جهت یاد گیری زبان انگلیسی درسنین پایین در مدارس دو زبانه می کنند، تحقیق حاضر با استفاده از مدل swot (قوت ها، ضعف ها، فرصتها و تهدیدها) سعی در ارزیابی مدرسه ای دو زبانه در ایران را دارد. جهت انجام این تحقیق در م...

15 صفحه اول

A performance analysis of subspace-based methods in the presence of model error. II. Multidimensional algorithms

This is the second of a two-part paper dealing with the performance of subspace-based algorithms for narrow-hand direction-of-arrival (DOA) estimation when the array manifold and noise covariance are not correctly modeled. In Part I, the performance of the MUSIC algorithm was investigated. In Part 11, we extend this analysis to multidimensional (MD) subspacebased algorithms including determinis...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

ژورنال

عنوان ژورنال: Bernoulli

سال: 2022

ISSN: ['1573-9759', '1350-7265']

DOI: https://doi.org/10.3150/21-bej1437