Several recent works have explored stochastic gradient methods for variational inference that exploit the geometry of the variational-parameter space. However, the theoretical properties of these methods are not well-understood and these methods typically only apply to conditionallyconjugate models. We present a new stochastic method for variational inference which exploits the geometry of the ...

This paper introduces a new empirical strategy for the characterization of business cycles. It combines non-parametric decoding methods that classify a series into expansions and recessions but does not require specication of the underlying stochastic process generating the data. It then uses network analysis to combine the signals obtained from di erent economic indicators to generate a uniqu...

—Gaussian processes such as Brownian motion and the Ornstein-Uhlenbeck process have been popular 1 models for the evolution of quantitative traits and are widely used in phylogenetic comparative methods. How2 ever, they have drawbacks which limit their utility. Here I describe new, non-Gaussian stochastic differential 3 equation (diffusion) models of quantitative trait evolution. I present gene...

We focus on the use of two stable and accurate explicit finite difference schemes in order to approximate the solution of stochastic partial differential equations of It¨o type, in particular, parabolic equations. The main properties of these deterministic difference methods, i.e., convergence, consistency, and stability, are separately developed for the stochastic cases.

Probabilistic or stochastic programming is a framework for modeling optimization problems that involve uncertainty.In this paper, we focus on multi-objective linear programmingproblems in which the coefficients of constraints and the righthand side vector are fuzzy random variables. There are several methodsin the literature that convert this problem to a stochastic or<b...