Tempering technology with care
نویسندگان
چکیده
منابع مشابه
Exact Sampling with Simulated Tempering
Multimodal structures in the sampling density (e.g. two competing phases) can be a serious problem for traditional Markov Chain Monte Carlo (MCMC), because correct sampling of the different structures can only be guaranteed for infinite sampling time. Samples may not decouple from the initial configuration for a long time and autocorrelation times may be hard to determine. We analyze a suitable...
متن کاملVariational Tempering
Variational inference (VI) combined with data subsampling enables approximate posterior inference with large data sets for otherwise intractable models, but suffers from poor local optima. We first formulate a deterministic annealing approach for the generic class of conditionally conjugate exponential family models. This algorithm uses a temperature parameter that deterministically deforms the...
متن کاملImproving Neonatal Care with Technology
The care for preterm infants has improved considerably in the last decades and although the outcome improved, prematurity is still a large global health issue and is ranked in the top 10 of the WHO list of leading causes of burden of disease (1). Prematurity still leads to organ injury, especially of the lungs and brain, and is responsible for 50% of perinatal mortality (2). Historically, much ...
متن کاملImportance tempering
Simulated tempering (ST) is an established Markov chain Monte Carlo (MCMC) method for sampling from a multimodal density π(θ). Typically, ST involves introducing an auxiliary variable k taking values in a finite subset of [0, 1] and indexing a set of tempered distributions, say πk(θ) ∝ π(θ) k. In this case, small values of k encourage better mixing, but samples from π are only obtained when the...
متن کاملDeep Tempering
Restricted Boltzmann Machines (RBMs) are one of the fundamental building blocks of deep learning. Approximate maximum likelihood training of RBMs typically necessitates sampling from these models. In many training scenarios, computationally efficient Gibbs sampling procedures are crippled by poor mixing. In this work we propose a novel method of sampling from Boltzmann machines that demonstrate...
متن کاملذخیره 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: The Journal of Spinal Cord Medicine
سال: 2012
ISSN: 1079-0268,2045-7723
DOI: 10.1179/1079026812z.00000000033