Automatically Tuned General-Purpose MCMC via New Adaptive Diagnostics
نویسندگان
چکیده
Adaptive Markov Chain Monte Carlo (MCMC) algorithms attempt to ‘learn’ from the results of past iterations so the Markov chain can converge quicker. Unfortunately, adaptive MCMC algorithms are no longer Markovian, so their convergence is difficult to guarantee. In this paper, we develop new diagnostics to determine whether the adaption is still improving the convergence. We present an algorithm which automatically stops adapting once it determines further adaption will not increase the convergence speed. Our algorithm allows the computer to tune a ‘good’ Markov chain through multiple phases of adaption, and then run conventional non-adaptive MCMC. In this way, the efficiency gains of adaptive MCMC can be obtained while still ensuring convergence to the target distribution.
منابع مشابه
Identification and Adaptive Position and Speed Control of Permanent Magnet DC Motor with Dead Zone Characteristics Based on Support Vector Machines
In this paper a new type of neural networks known as Least Squares Support Vector Machines which gained a huge fame during the recent years for identification of nonlinear systems has been used to identify DC motor with nonlinear dead zone characteristics. The identified system after linearization in each time span, in an online manner provide the model data for Model Predictive Controller of p...
متن کاملAdaptive Tunable Vibration Absorber using Shape Memory Alloy
This study presents a new approach to control the nonlinear dynamics of an adaptive absorber using shape memory alloy (SMA) element. Shape memory alloys are classified as smart materials that can remember their original shape after deformation. Stress and temperature-induced phase transformations are two typical behaviors of shape memory alloys. Changing the stiffness associated with phase tran...
متن کاملOn the e ciency of adaptive MCMC algorithms
Abstract We study a class of adaptive Markov Chain Monte Carlo (MCMC) processes which aim at behaving as an optimal target process via a learning procedure. We show, under appropriate conditions, that the adaptive process and the optimal (nonadaptive) MCMC process share identical asymptotic properties. The special case of adaptive MCMC algorithms governed by stochastic approximation is consider...
متن کاملAn Adaptive Sequential Monte Carlo Sampler
Sequential Monte Carlo (SMC) methods are not only a popular tool in the analysis of state–space models, but offer an alternative to Markov chain Monte Carlo (MCMC) in situations where Bayesian inference must proceed via simulation. This paper introduces a new SMC method that uses adaptive MCMC kernels for particle dynamics. The proposed algorithm features an online stochastic optimization proce...
متن کاملExamples of Adaptive MCMC by Gareth
We investigate the use of adaptive MCMC algorithms to automatically tune the Markov chain parameters during a run. Examples include the Adaptive Metropolis (AM) multivariate algorithm of Haario et al. (2001), Metropolis-within-Gibbs algorithms for non-conjugate hierarchical models, regionally adjusted Metropolis algorithms, and logarithmic scalings. Computer simulations indicate that the algori...
متن کامل