This paper presents a new approach to robust Gaussian process regression, creating non-parametric Bayesian regression estimate outliers. Most existing approaches replace an outlier-prone likelihood with non-Gaussian induced from heavy tail distribution, such as the Laplace distribution and Student-t distribution. However, use of would incur need for computationally expensive approximate computa...

Related DatabasesWeb of Science You must be logged in with an active subscription to view this.Article DataHistorySubmitted: 22 October 2019Accepted: 17 September 2020Published online: February 2021Keywordsvariational data assimilation, asymptotic expansion, model error, parameter estimation, bias correction, LorenzAMS Subject Headings34A55, 65K10, 34E05Publication DataISSN (online): 1536-0040P...

This paper is the first to link economic theory with empirical life-satisfaction analyses referring to internal migration. We derive an extension of the Roback (1982) model to account for benefits from regional amenities in the utility function, while controlling for income, housing costs, and migration costs. Using highly disaggregated spatial panel information on people’s migration decisions ...

In this article, we propose a new probabilistic model for the distribution of ranks of elliptic curves in families of fixed Selmer rank, and compare the predictions with previous results, and with the databases of curves over the rationals that we have at our disposal. In addition, we document a phenomenon we refer to as Selmer bias that seems to play an important role in the data and in our mo...

This paper extends the analysis of the choice between internal and external R&D to consider the costs of internal R&D. The underlying hypothesis is that the choice of R&D mode is determined by their relative costs. Rather than merely estimating a reduced form probit model for R&D mode, we employ the Heckman two-stage estimator to estimate in addition the determinants of internal R&D unit cost (...

It is common in modern applications to use data-dependent model selection tools to select a promising model before drawing inference over the parameters of the selected model. However, this simple series of steps conceals a significant fault that is often left unattended: the act of selection biases the distributions of test statistics and makes standard inference procedures unsound. This is re...