We briefly review the two classical problems of distribution estimation and identity testing (in the context of property testing), then propose to extend them to a Markovian setting. We will see that the sample complexity depends not only on the number of states, but also on the stationary and mixing ...

Regularization by Noise for Stochastic Differential and Stochastic Partial Differential Equations The regularizing effects of noisy perturbations of differential equations is a central subject of stochastic analysis. Recent breakthroughs initiated...

Theory and Methods for Inference in Multi-armed Bandit Problems Multi-armed bandit (MAB) algorithms have been argued for decades as useful to conduct adaptively-randomized experiments. By skewing the allocation of the arms towards the more efficie...

Selection of Priors in Bayesian Structural Equation Modelling Structural equation modeling (SEM) is an important framework within the social sciences that encompasses a wide variety of statistical models. Traditionally, estimation of SEMs has reli...

Recent Advances in Approximate Bayesian Inference In approximate Bayesian computation, likelihood function is intractable and needs to be itself estimated using forward simulations of the statistical model (Beaumont et al., 2002; Marin et al., 201...

Recent Advancements in Applied Instrumental Variable Methods Instrumental variables (IV) is one of most important and widespread research designs in economics and statistics, as it can identify causal effects in the presence of unobserved confound...

Overview Bayesian nonparametric (BNP) models are a prominent tool for performing flexible inference with a natural quantification of uncertainty. Traditionallly, flexible inference within a homogeneous sample is performed with exchangeable models ...

This is an updated version of a blog post on RIKEN AIP Approximate Bayesian Inference team webpage: https://team-approx-bayes.github.io/blog/mmd/ INTRODUCTION A very old and yet very exciting problem in statistics is the definition of a universal estimator \(\hat{\theta}\). An estimation procedure that would work all ...

Overview Suppose \(\{X_t: t\in \mathbb{Z}\}\) is a second order stationary time series where \(c(r) = \text{cov}(X_{t+r},X_t)\) and \(f(\omega) = \sum_{r\in\mathbb{Z}}c(r)e^{ir\omega}\) are the corresponding autocovariance and spectral density fun...

Inclusion Process and Sticky Brownian Motions The ninth “One World webinar” organized by YoungStatS will take place on February 9th, 2022. Inclusion process (IP) is a stochastic lattice gas where particles perform random walks subjected to mutual ...

Problem Setting In our forthcoming paper on Annals of Applied Statistics, we propose a new method – which we call Bayesian Causal Forest with Instrumental Variable (BCF-IV) – to interpretably discover the subgroups with the largest or smallest cau...

In frozen percolation on a graph, there is a barrier located on each edge. Initially, the barriers are closed and they are assigned i.i.d. uniformly distributed activation times. At its activation time, a barrier opens, provided it is not frozen. At a fixed set \(\Xi\) of freezing times, ...

Novel Algebraic Approaches to Maximum Likelihood Estimation The seventh “One World webinar” organized by YoungStatS will take place on November 17th, 2021. Maximum likelihood estimation (MLE) is a tool in data analysis to estimate a probability di...

Advancements in Symbolic Data Analysis The sixth “One World webinar” organized by YoungStatS will take place on October 27th, 2021. With the development of digital systems, very large datasets have become routine. However, standard statistical app...

Advances in Difference-in-Differences in Econometrics The eighth “One World webinar” organized by YoungStatS will take place on December 13th, 2021. The difference-in-differences design is a quasi-experimental identification strategy for estimatin...

Disclosure risk for microdata Protection against disclosure is a legal and ethical obligation for agencies releasing microdata files for public use. Consider a microdata sample \({X}_n=(X_{1},\ldots,X_{n})\) of size \(n\) from a finite population ...

Figure 1: Depth quantile functions for the wine data (d=13), class 2 vs class 3. Blue curves correspond to between class comparisons, red/pink correspond to within class comparisons. A common technique in modern statistics is the so-called kernel ...

The fifth “One World webinar” organized by YoungStatS will take place on September 15th, 2021. Selected young European researchers active in the areas of probability and machine learning will present their recent contributions. The webinar is joint...

In recent years, a surprising number of scientific results have failed to hold up to continued scrutiny. Part of this ‘replicability crisis’ may be caused by practices that ignore the assumptions of traditional (frequentist) statistical methods (John, Loewenstein, and Prelec 2012). One of these assumptions is that the experimental protocol should ...

In recent years, a surprising number of scientific results have failed to hold up to continued scrutiny. Part of this ‘replicability crisis’ may be caused by practices that ignore the assumptions of traditional (frequentist) statistical methods (John, Loewenstein, and Prelec 2012). One of these assumptions is that the experimental protocol should ...